UvA · ASSESSMENT OF PHARMACOVIGILANCE APPROACHES FOR MONITORING THE SAFETY OF ANTIMALARIAL DRUGS...

UvA-DARE is a service provided by the library of the University of Amsterdam (http://dare.uva.nl)

UvA-DARE (Digital Academic Repository)

Assessment of pharmacovigilance approaches for monitoring the safety of antimalarial drugsin pregnancy

Dellicour, S.O.M.C.

Link to publication

Citation for published version (APA):Dellicour, S. O. M. C. (2014). Assessment of pharmacovigilance approaches for monitoring the safety ofantimalarial drugs in pregnancy. Alblasserdam: Dutch University Press.

General rightsIt is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s),other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).

Disclaimer/Complaints regulationsIf you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, statingyour reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Askthe Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam,The Netherlands. You will be contacted as soon as possible.

Download date: 24 Aug 2020

https://dare.uva.nl/personal/pure/en/publications/assessment-of-pharmacovigilance-approaches-for-monitoring-the-safety-of-antimalarial-drugs-in-pregnancy(7caae066-a726-467b-a66b-3ecee07cc7d2).html

Assessment of Pharmacovigilance Approaches

for Monitoring the Safety of Antimalarial

Drugs in Pregnancy

Stéphanie Dellicour

Assessm

ent of Pharmacovigilance A

pproaches for Monitoring the Safety of A

ntimalarial D

rugs in PregnancyStéphanie D

ellicour

Assessmentofpharmacovigilanceapproachesformonitoringthesafetyofantimalarial

drugsinpregnancy

2

© Stephanie Dellicour, 2014

All rights reserved

Dutch University Press

Alblasserdam

ISBN 978 90 361 0408 1

NUR 870

ASSESSMENTOFPHARMACOVIGILANCEAPPROACHESFOR

MONITORINGTHESAFETYOFANTIMALARIALDRUGSINPREGNANCY

ACADEMISCH PROEFSCHRIFT

ter verkrijging van de graad van doctor

aan de Universiteit van Amsterdam

op gezag van de Rector Magnificus

prof. dr. D.C. van den Boom

ten overstaan van een door het college voor promoties ingestelde

commissie, in het openbaar te verdedigen in de Aula der Universiteit

op woensdag 22 oktober 2014, te 13:00 uur

door Stéphanie Ondine Marie Cornélie Dellicour

geboren te Uccle, België

4

Promotiecommissie Promotores: Prof. dr F.O. ter Kuile

Prof. dr M. Boele van Hensbroek

Copromotor: Dr M. Desai Overige leden: Prof. dr R.C. Pool Prof. dr F.G.J. Cobelens Dr P.F. Mens Prof. dr M.W. Borgdorff Dr J. Webster Dr M.J. Rijken Faculteit der Geneeskunde

To Per and Lea

6

Contents Page Number Chapter 1: General Introduction 7 Chapter 2: Dellicour S, Hall S, Chandramohan D, Greenwood B. The safety of

artemisinins during pregnancy: a pressing question. Malaria Journal 2007, 6:15.

23

Chapter 3: Dellicour S, Tatem AJ, Guerra CA, Snow RW, ter Kuile FO. Quantifying

the Number of Pregnancies at Risk of Malaria in 2007: A Demographic Study. PLoS Medicine 2010, 7(1):e1000221.

37

Chapter 4: Dellicour S, ter Kuile FO, Stergachis A. Pregnancy Exposure Registries

for Assessing Antimalarial Drug Safety in Pregnancy in Malaria‐Endemic Countries. PLoS Medicine 2008, 5(9):e187.

65

Chapter 5: Dellicour S, Brasseur P, Thorn P, Gaye O, Olliaro P, Badiane M,

Stergachis A, ter Kuile FO. Probabilistic Record Linkage for Monitoring the Safety of Artemisinin‐Based Combination Therapy in the First Trimester of Pregnancy in Senegal. Drug Safety 2013, 36(7):505‐513.

79

Chapter 6: Dellicour S, Desai M, Mason L, Odidi B, Aol G, Phillips‐Howard PA,

Laserson KF, Ter Kuile FO. Exploring Risk Perception and Attitudes to Miscarriage and Congenital Anomaly in Rural Western Kenya. PLoS One 2013, 8(11):e80551.

97

Chapter 7: Riley C, Dellicour S, Ouma P, Kioko U, ter Kuile FO, Ahmeddin O,

Kariuki S, Buff AM, Desai M, Gutman J. Assessment of Knowledge and Adherence to the National Guidelines for Malaria Case Management in Pregnancy among Healthcare Providers and Drug Outlet Dispensers in rural, Western Kenya. To be submitted

113

Chapter 8: Dellicour S, Desai M, Aol G, Oneko M, Ouma P, Bigogo G, Burton D,

Breiman R, Hamel M, Slutsker L, Feikin D, Kariuki S, Odhiambo F, Pandit J, Laserson KF, Calip G, Stergachis A, ter Kuile FO. Risk of inadvertent exposure to artemisinin derivatives in the first trimester of pregnancy and its association with miscarriage: a prospective study in Western Kenya. To be submitted

141

Chapter 9: Discussion and Summary 179 Samenvatting 191 Acknowledgements 197 About the author 199

7

Chapter1

GeneralIntroduction

Chapter 1

8

PharmacovigilanceAll drugs have the potential to cause harm and their risk‐benefit profiles need to be determined to

maximize therapeutic outcomes.[1] Attention to the importance of drug safety monitoring was

brought to light by the thalidomide tragedy in the 1960s.[2,3] Thalidomide was used as an

antiemetic and sedative which was considered safe including during pregnancy. More than 10,000

malformed children with phocomelia (with limbs severely underdeveloped or absent) were born

before a safety signal was identified and thalidomide was withdrawn.[4] This had a profound impact

on drug regulatory processes from registration to post‐marketing surveillance.[5]

Pharmacovigilance, sometimes referred to as drug monitoring or post‐marketing surveillance, is

defined by the World Health Organization (WHO) as “the science and activities relating to the

detection, assessment, understanding, and prevention of adverse reactions or any other possible

drug‐related problems”.[1] Fundamental aims are early detection of unknown safety issues,

quantifying risk and risk factors associated with adverse drug reactions (ADRs) and primarily

preventing unnecessary harm to patients.

Before a drug is marketed, drugs are tested pre‐clinically in‐vitro and in animal studies for toxicity. If

a compound passes these tests it goes through a series of clinical trials (see figure 1). Drug

regulatory authorities make their decision based on evidence regarding efficacy, safety and quality

deemed adequate for a drug to be approved. By the time a drug is marketed, information on safety

and efficacy has been derived from limited number of subjects (between 500‐5000) who are

typically carefully selected and followed under controlled conditions for a relatively short amount of

time.[6] This limits the possibility of detecting rare adverse events (the rule of three suggests that by

the time of registration, safety information is typically only available for adverse events occurring at

frequencies of 0.1 to 1%1[7]), or events with slow onset such as cancer or events occurring in

vulnerable groups such as pregnant women, children and the elderly or people with co‐morbid

conditions and concomitant medications which are usually excluded from pre‐registration clinical

trials.

Passive surveillance through spontaneous reporting is the backbone of post‐marketing surveillance.

This entails identification of a suspected ADR following drug exposure by healthcare professionals,

assessments of severity and causality and reporting according to standard procedures. Passive

surveillance is useful for hypothesis generation and identification of new safety signals but has its

limitations such as under‐reporting, variable quality and completeness of the reported

data, tendency for reporting of known reactions and false causality attribution.[8] Active surveillance

approaches are essential to assess and evaluate the nature and rates of adverse events as they can

provide denominator data. The most common pharmacovigilance active surveillance involves

identifying individuals exposed to a drug of interest and following them systematically to assess for

adverse events (e.g. Prescription Event Monitoring system used in the UK [9]). Other approaches

using pharmacoepidemiological methods, include cohort or case control designs, are used for safety

signal confirmation or characterization.[10,11] Deciding on an approach should be based on the

safety specification of a product and whether the purpose of the investigation is hypothesis or signal

1 The rule of three is commonly used in pharmacovigilance to estimate the sample size needed to detect adverse events. It

is based on the assumption that there is 95% chance of observing one occurence of an event in a population 3 times the size of the event’s rate (e.g. to pick up a rare event occuring at a rate of 1 /10 000, 30 000 individuals will need to be monitored to be 95% certain of detecting this event).

Introduction

9

generating; hypothesis strengthening or a confirmatory study, (i.e. evaluation of a known safety

signal).

Figure 1 Clinical development of medicines adapted from “WHO Policy Perspectives on Medicines — Pharmacovigilance: ensuring the safe use of medicines”.[6]

PharmacovigilanceinresourceconstrainedsettingsThe public health burden of ADRs and medication errors is difficult to quantify but it is likely to be

worse in resource constrained settings because of weak health systems, overstretched and often

inadequately trained healthcare staff, unreliable supply and quality of drugs, polypharmacy (use of

multiple drugs concomitantly), concomitant use of herbal remedies as well as availability of most

prescription drugs from the informal market.[12] Few low and middle‐income countries have

functioning national pharmacovigilance systems. Currently, less than 2% of ADR reports collated at

the Uppsala Monitoring Centre of the WHO Programme for International Drug Monitoring (UMC,

which collates reports from national pharmacovigilance centres globally [13]) come from low or

middle‐income countries.[14] Indeed, most of these countries make national treatment or

prevention policy decisions with reliance on drug safety data from industrialised countries where

pharmacovigilance systems and regulatory authorities are better established. However, this often

comes at the expense of no or limited safety data for drugs targeting tropical diseases which are

seldom encountered in these industrialised settings.

While there is increasing access to medicines to combat HIV, malaria and tuberculosis in the tropics,

systems are needed to assess the risk‐benefit balance of these much‐needed interventions.[15] Drug

safety monitoring cannot rely on passive spontaneous reporting systems for ADRs due to the

limitations mentioned above. Targeted pharmacovigilance approaches are required in situations

where no systematic recording of drug exposure and suspected ADRs exist particularly where health

resources are limited. Passive surveillance has been proposed at selected sentinel sites where

healthcare professionals are trained and incentivised to report suspected severe ADRs to specific

drugs of interest.[16] Building on existing infrastructure and surveillance systems or studies would

provide more cost‐effective ways to collect reliable and timely data. Demographic surveillance sites

Preclinical Animal Experiments

Phase I > Phase II > Phase IIPhase IV

Post Approval

Development

Registration

Post registration

Phase I20‐50 healthy volunteers to

gather preliminary data

Animal experiments for acute toxicity, organ damage, dose dependence,

metabolism, kinetics, carcinogenicity, mutagenicity & teratogenicity

Phase III250‐4000 varied patient groups to determine short term safety and

efficacy

Phase II150‐350 patient with disease to determine safety and dosage

Phase IVPost‐approval studies to

determine specific safety issues

Chapter 1

10

are increasingly being considered for use as platforms for post‐marketing surveillance. Such sites

monitor vital events (births, deaths and migration) through regular censuses of a pre‐defined

population (typically 2 to 4 times per year), most also have links to clinical and treatment data from

local health facilities. They have well‐defined and characterized populations, and are valuable to

monitor public health interventions where health information systems are weak as they provide

denominator data, a framework to identify specific groups (such as pregnant women or children)

outside the healthcare setting as well as human resources and infrastructure for research.[17] Five

African sites that are part of the International Network for the Continuous Demographic Evaluation

of Populations and Their Health (INDEPTH) are conducting Phase IV studies to monitor antimalarial

effectiveness and safety in “real life” settings.[18] Antimalarial safety is being monitored through a

Spontaneous Adverse Events Reporting System (passive surveillance) and Cohort Event Monitoring

(CEM; active surveillance). The spontaneous reporting system is being strengthened in the selected

sites ensuring health workers are sensitised and familiar with the ADR reporting procedures. CEM

protocols involve active follow up of patients prescribed antimalarials on days 3 and 7 following

treatment and systematic recording of all clinical events during that time. This enables the

characterisation of known ADRs in terms of incidence as a denominator is available (which is not the

case for systems based on spontaneous reporting) and identification of risk factors as well as

detection of unknown safety signals. Another approach is to capitalize on planned studies, such as

those conducted by the Malaria in Pregnancy Consortium (MiPc) and the ACT consortium (ACTc)

which have set up a centralized safety database collating ADRs collected across studies using

standardized reporting procedures and tools.[19,20] Such systems could then be extended to other

drugs and expanded to additional sites through new collaborations.

DrugsafetyinpregnancySpecial considerations apply to the assessment of drug safety in pregnancy, particularly for drugs

used to treat diseases that are only endemic in resource constrained settings. Most drugs are

marketed with limited information on their safety when used during pregnancy. Pre‐approval

studies have inherent limitations in determining the safety of drugs used during pregnancy. Animal

reproductive toxicology studies have ambiguous predictive value for human teratogenesis due to

variations in species‐specific effects [21] and pregnant women are routinely excluded from pre‐

licensure clinical trials for fear of harming the mother or the developing fetus.[22,23] Physiological

changes associated with pregnancy limit the inference of pharmacokinetic and pharmacodynamic

data from non‐pregnant subjects.[24] Consequently, most drugs are not recommended for use

during pregnancy due to the lack of information on their risk‐benefit profile. Yet drugs are widely

used by pregnant women. Medication often cannot be avoided for chronic diseases, such as asthma,

epilepsy, malignant diseases, HIV, or other illnesses which may harm the mother and the unborn

baby if left untreated. Furthermore, many women are likely to be inadvertently exposed to drugs at

the most vulnerable time of embryo development early in gestation before pregnancy status is

recognised. Healthcare providers, pregnant women and policy‐makers need valid information in

order to make informed decisions about the use of drugs during pregnancy.

Introduction

11

PharmacovigilanceapproachesfordrugsusedinpregnancyThere are several methodological challenges to the study of safety of drugs in pregnancy, including

those common to overall pharmacovigilance methods such as the need for large sample sizes to

minimise the possibility that the observed association between a drug and a rare outcome occurred

due to chance; the possibility of confounding by indication which implies studies should take into

account the contributing risk of the underlying maternal illness; the general lack of background rates

needed to put signals into context and the potential self‐referral bias introduced by voluntary

reporting.[25,26] Special considerations are also required for assessment of outcome and drug

exposure. Monitoring drug safety in pregnancy necessitates systematic recording of pregnancy

outcomes, examining the newborns and possibly following up the infants to detect anomalies not

detectable at birth. Where deliveries occur outside health facilities, as is the case in many rural

settings in low and middle income countries and particularly for early pregnancy loss, systems need

to be put in place to capture all pregnancy outcomes. Ascertainment of drug exposure requires

reliable information on the drug and gestational age at the time of exposure. This is important as the

impact on the fetus from drugs used in pregnancy depends on the stage of pregnancy at the time of

exposure as different tissues and organs have specific developmental timelines.[27,28] These

methodological considerations are described in detail in chapter 4 based on the case of artemisinin

combination therapies used for the treatment of malaria.

Passive surveillance (i.e. spontaneous reports of suspected ADRs) for detecting drug with potential

embryo‐fetal adverse effects relies on judicious clinicians to make the association between a drug

exposure in pregnancy and an adverse event which will often occur months and sometimes years

later (for congenital anomalies not detectable at birth). In resource constrained settings, such an

approach has limited use due to under‐ascertainment of outcomes such as miscarriages and birth

defects in communities where birth anomalies are considered taboos, as well as the general lack of

awareness around pharmacovigilance and the need to report suspected ADRs.

Pregnancy exposure registries, a type of cohort design, are the most common approach to monitor

drug safety in pregnancy in industrialised countries (e.g. the Antiretroviral Pregnancy Registry[29] or

the Antiepileptic Drug (AED) Pregnancy Registry[30]). The U.S. Food and Drug Administration (FDA)

and the European Medicines Agency (EMA) recommend pregnancy exposure registries for products

that are likely to be used during pregnancy or by women of reproductive age, particularly if there

have been case reports of adverse pregnancy outcomes following exposure, drugs in the same

pharmacological class are known to pose risk during pregnancy or pre‐clinical animal data suggests

potential teratogenic risk.[31,32] A pregnancy exposure registry involves active identification and

recruitment of pregnant women exposed to a drug or class of drug of interest and following these

women throughout pregnancy. This approach has many advantages such as active enrolment and

pregnancy outcome ascertainment to minimise loss to follow up, centralised and prospective

ascertainment of exposed pregnancies which reduces biases. Drawbacks of most pregnancy

registries are the limited sample size to detect moderate effects, the lack of outcome validation and

standardisation of birth defect assessments.[27] To overcome the sample size requirement

pharmacoepidemiologic studies using existing databases are often used in high income countries.

Cohort studies use large databases such as medical insurance claim databases (e.g. the Medicaid

database in the US[33]), electronic medical records (e.g. GPRD in the UK[34]), or data from

teratology information services (e.g. the Motherisk Program in Canada[35]). Birth defect registries

and case‐control surveillance systems are also used to assess associations between medications and

Chapter 1

12

congenital malformations (e.g. the Birth Defects Study by Slone Epidemiology Centre[36]).

Applicability of these methods in low and middle income countries is unknown.

Malariainpregnancy

EpidemiologyandburdenMalaria is a major public health problem affecting an estimated 3.4 billion people worldwide (half of

the world’s population) which live in areas at risk of malaria. In 2012, an estimated 207 million cases

of malaria occurred globally, resulting in over half a million deaths.[37] This life threatening disease

is preventable and treatable through proper use of antimalarial drugs, insecticide treated bednets

and indoor residual spraying of insecticides. Since 2000, a significant decrease in malaria mortality

rates have been observed due to the vast increase in the financing and coverage of malaria control

programmes.[37]

Pregnant women and children under five are most at risk of malaria. Malaria in pregnancy (MiP) can

have devastating consequences for the mother and fetus, including severe maternal anaemia and

mortality, miscarriage, intrauterine growth retardation and low birthweight (LBW), preterm birth

and stillbirth.[38,39] It is estimated to cause as many as 900,000 low birth weight births, and

100,000 infant deaths every year. Furthermore, up to a quarter of maternal mortality in endemic

countries could be attributable to malaria.[39,40,41,42,43]

PreventionguidelinesWHO recommends two approaches for the prevention of MiP in areas with stable malaria

transmission (where the population is exposed to a fairly constant, high rate of malaria infected

mosquito bites [>10 per person per year]): intermittent preventative treatment in pregnancy (IPTp)

with sulfadoxine‐pyrimethamine (SP) at each antenatal clinic visit following quickening and the use

of insecticide‐treated bed nets (ITNs) to protect pregnant women from the bites of infected

mosquitoes.[44,45] Coverage with these prevention strategies remains very low. In 2007 it was

estimated that only about 39% and 25% of pregnant women in sub‐Saharan Africa used ITNs and

received at least two doses of IPTp, respectively.[46]

TreatmentguidelinesPregnant women, when infected, require rapid diagnosis and case management with safe and

effective antimalarial drugs to prevent progression to severe disease or death, or to prevent

asymptomatic infections from becoming chronic leading to fetal growth restriction and malaria‐

related maternal anaemia.[38,39] Established antimalarials like chloroquine and SP are no longer

recommended for the case management of malaria illness due to drug resistance of the parasite.

Artemisinin‐based combination therapies (ACTs) are now the recommended treatment for

uncomplicated malaria in sub‐Saharan Africa.[47] Artemisinin derivatives have been used for

centuries in China but they have only been deployed for use as ACTs since 2001.[47,48] Based on the

level of drug resistance to the partner medicine, the following fixed‐dose combination ACTs are

recommended: artemether plus lumefantrine; artesunate plus amodiaquine; artesunate plus

mefloquine; artesunate plus SP and dihydroartemisinin plus piperaquine.

In the first trimester of pregnancy an ACT is indicated only if an alternative antimalarial is not

immediately available, or if treatment with 7‐day quinine (without, or combined with, clindamycin)

fails or there is uncertainty of compliance with the 7‐day quinine regiment. The recommendation for

treatment of pregnant women with severe malaria includes parenteral artesunate because it is

Introduction

13

faster acting than parenteral quinine and is associated with improved survival in patients with severe

malaria.[49,50,51]

Antimalarialsafetyinpregnancy

AntimalarialsconsideredsafethroughoutpregnancyAntimalarials recommended for pregnant patients must be safe for the mother and her unborn

baby. There is insufficient information on the safety of most antimalarials in pregnancy, particularly

for exposure in the first trimester. A retrospective study in Thailand, reported the deleterious effect

of even a single episode of malaria in the first trimester and its association with an increase in the

risk of miscarriage, emphasising the need for safe and efficacious drugs in this critical period.[52]

This is further supported by a recent modelling study reporting that up to two thirds of placental

infections occur by the end of the first trimester.[43] Antimalarial medicines considered safe in the

first trimester of pregnancy are quinine (with or without clindamycin), chloroquine, proguanil and

more recently mefloquine although this is based on limited evidence. As mentioned above,

chloroquine is no longer recommended for treatment of falciparum malaria. Proguanil and

chlorproguanil although deemed very safe in pregnancy, have limited value in resource constrained

settings as the current formulations available are either too expensive (atovaquone‐proguanil) or

carries the risk of acute haemolysis in patients with glucose‐6‐phosphate dehydrogenase (G6PD)

deficiency (which has a prevalence of up to 30% in some African countries[53]) when in combination

with dapsone. This combination is no longer available since its marketing authorization holder

withdrew it from the market in 2008.[54] Animal studies (in rodents, dogs and primates) did not find

quinine to have embryo‐fetal toxicity except one study which reported congenital malformations in

5% of pups born to rats receiving quinine.[55,56,57] There was no evidence of embryo‐fetal toxicity

in published human data. Quinine is often thought to have abortifacient properties at high dosage.

The author of a commonly cited publication from 1908 reports the beneficial effect of quinine as “It

may be laid down as an almost invariable rule that if in a pregnant woman suffering from malarial

fever uterine action has begun, the quinine will probably hasten the abortion; but that if uterine

action has not begun, it will not start it, but will, on the contrary, probably be the means of saving

the pregnancy”.[58] A recent observational study in Tanzania reported an increased risk of

pregnancy loss in women exposed to quinine in the first trimester [59] however it is not clear

whether the observed effect could be explained by the underlying malaria infection, especially

because of the potential for poor compliance with the 7‐day quinine regimen given the poor

tolerability of quinine due to cinchonism (a syndrome which includes symptoms of ringing of the

ears, blurred vision, headache, nausea and dizziness) and hypoglycaemia.[60,61] Mefloquine has

been recommended as a prophylactic for pregnant travellers by WHO and CDC. Recently after

review of existing evidence the FDA and CDC changed their recommendation for the use of

mefloquine in pregnancy both as a prophylactic and for treatment.[62] Mefloquine has an

acceptable reproductive toxicity profile in animal studies at standard doses and data on 1000 first

trimester exposures suggests no increase in risk to adverse pregnancy outcomes. Although 1

retrospective study found an increase in the risk of stillbirths associated with mefloquine treatment

doses in pregnancy this finding has not been confirmed by subsequent studies.[60,61,63] The result

of a multi‐centre randomised controlled trial in over 4000 second and third trimester pregnant

women did not find an increased risk of stillbirth with two or three 15 mg/kg treatment doses of

mefloquine compared to SP when used as IPTp. However tolerability was low, even when 15 mg/kg

Chapter 1

14

was given as a split dose over two days, with 30% of pregnant women reporting vomiting and

dizziness which could limit the use of mefloquine for IPTp.[64]

AntimalarialscontraindicatedinthefirsttrimesterofpregnancySP, which has limited use due to increasing drug resistance, is still used for the prevention of malaria

in pregnancy through IPTp in most areas with high malaria transmission. As an anti‐folate it is not

recommended in the first trimester of pregnancy due to risk of neural tube defects.[65] Animal

studies found embryotoxicity at high doses including cleft palate in rat models.[66] Documented use

in over two thousand pregnancies in the second and third trimester of pregnancy found no evidence

of embryotoxicity.[60,61] Amodiaquine is considered safe in pregnancy although there are neither

documented data on exposure in the first trimester of pregnancy nor animal reprotoxicology

studies.[67,68] Lumefantrine is only available as an ACT in combination with artemether which is

contraindicated in the first trimester of pregnancy (see below) and data from animal studies did not

show any embryotoxic effect.[60,61] Piperaquine, which is also only available as an ACT, has a safety

profile expected to be similar to that of chloroquine. Animal reprotoxicology studies of piperaquine

found no safety concerns except that gestation was prolonged in exposed rats.[69,70] The few

studies on the use of the combination dihydroartemisinin‐piperaquine in second and third trimesters

of pregnancy reported favourable outcomes for pregnant women.[71,72,73] Further trials are

underway with women in their second and third trimesters.[74,75,76,77,78]

ArtemisininderivativesinearlypregnancyAlthough ACTs are highly effective in treating malaria the safety of artemisinin derivatives during

early pregnancy remains to be determined. Animal reprotoxicology studies showed that artemisinin

derivatives have embryotoxic effects in all species studied (i.e. rat, rabbit and monkey) at low dose

ranges.[79,80] Pre‐clinical studies showed that the mechanism of embryotoxicity was through insult

to immature red blood cells (primitive erythroblasts) causing severe anaemia in the embryo and

leading to either embryolethality or malformations, skeletal (shortened or bent long bones and

scapulae, misshapen ribs, cleft sternebrae and incompletely ossified pelvic bones) and

cardiovascular (ventricular septal and vessel defects).[79,81] These studies also predicted that the

main window for insult to the fetus will occur early in pregnancy (between 4 and 10 weeks post

conception).[79] This is the period when the nucleated, metabolically active primitive erythroblasts

predominate in the blood. However, the exact moment when humans are most sensitive is unknown

as the primitive erythroblasts are gradually replaced by enucleated mature erythrocytes (which are

less sensitive to the effects of artemisinins) over several weeks. Information regarding risks

associated with the use of antimalarials in pregnancy in humans is sparse. Although the limited data

on human exposures is reassuring for first trimester exposures 2 , further information is

required.[61,82,83] With the widespread deployment of ACTs, the potential for inadvertent

exposure early in pregnancy is high when neither the physician nor the patient is aware of the

pregnancy.

It is essential to study the risks of artemisinins and ACT drugs in early pregnancy, and that vigilant

attempts are made to establish systems for the systematic collection of pregnancy‐drug exposure

data. It is unknown how best this can achieved in resource constrained malaria endemic countries.

The specialized nature of the reliable assessment of drug exposures, pregnancy outcome and

2 At the time of the first review in chapter 2 there were 123 documented exposures this number is now close to 760 as discussed in the last chapter of this thesis.

Introduction

15

malformations is not easily achievable from routine surveillance systems such as national

pharmacovigilance programmes, where they exist, and will require sentinel sites with enhanced

active surveillance or dedicated studies, good record keeping and follow‐up systems, and training of

staff to examine newborns.

ThesisAimandOutlineThe aim of this thesis is to develop and evaluate such targeted pharmacovigilance systems to assess

the safety of the ACTs in early pregnancy.

This thesis is divided into two components: desk‐based reviews (chapters 2‐4) and field‐based

studies (chapters 5‐8) with the overall aim to assess pharmacovigilance approaches to provide better

estimates of the risk‐benefit profiles of ACTs used in early pregnancy.

Objectivesandoutline1) Review existing evidence on human exposure to ACTs in pregnancy with the emphasis on

first trimester exposures (chapter 2)

2) Derive global estimates of the number of pregnancies at risk of malaria based on a

contemporary map of malaria transmission and demographic data for pregnancy and

fertility rates to provide an estimate of the scale of the problem posed by MiP (chapter 3)

3) Describe the methodological considerations for setting up antimalarial pregnancy exposure

registries in resource constrained settings (chapter 4)

4) Assess the feasibility of record linkage using routinely collected healthcare data as a

pragmatic means of monitoring antimalarial safety in early pregnancy. A study involving

extraction of data from health records from a dispensary in south‐western Senegal is

reported (chapter 5)

5) Explore community perceptions of miscarriage and congenital anomalies. Findings from 10

focus group discussions carried out in western Kenya are described which provide insight

for studies of pregnancy outcomes in similar rural African settings (chapter 6)

6) Assess prescribing behaviour and knowledge of malaria treatment guidelines for pregnant

women among healthcare providers and drug outlet dispensers in an area of high malaria

transmission. The results from a cross‐sectional survey in rural western Kenya are

presented providing insight on the scale of ACT prescribing in early pregnancy (chapter 7)

7) Assess the risk of miscarriage associated with exposure to ACTs in the first trimester of

pregnancy. The findings from a prospective cohort study of women of childbearing age in

Western Kenya are presented (chapter 8)

8) Draw conclusions on the potential of the proposed pharmacovigilance approaches for drugs

used by woman of childbearing age and pregnant women in resource constrained settings

based on all the evidence presented in this thesis (chapter 9)

Descriptionofstudysites

SenegalThe study described in chapter 5 took place in a private mission dispensary based in Mlomp, a village

of approximately 8,000 inhabitants in the District of Oussouye, Casamance, south‐western Senegal.

The dispensary offers outpatient, antenatal clinic (ANC running once a week), and maternity (since

1968) services. The dispensary is well attended, offers high quality services and is equipped to

perform simple laboratory tests including microscopy evaluations of malaria slides. Dispensary

Chapter 1

16

registers for antenatal care, delivery, child welfare clinics and a general register for outpatient visits

have been meticulously kept since 1993. Nearly all pregnant women attend ANC and health facility

deliveries have increased from 50% in 1961 to 99% in 1999. [84,85] The district hospital is situated in

Oussouye (the closest town) about 10km away with limited public transport option and the regional

hospital is in Ziguinchor (about 50km away).

Malaria occurs year‐round and peaks during the rainy season (July to December) in this area. A

recent study showed that malaria transmission intensity in southern Senegal has been decreasing

significantly in the past 15 years.[86] The area is rural, there is no running water or electricity and

rice cultivation is the main economic activity. Mlomp has been under yearly demographic

surveillance by the French National Institute of Demographic Studies (INED) since 1985.[87] Several

research studies on malaria and malaria chemotherapy have been conducted in the study

area.[88,89,90,91]

This setting, with a relatively enclosed population and comprehensive records on malaria episodes,

treatment and pregnancy outcomes, provided a good opportunity to pilot utilisation of routine

healthcare data for monitoring antimalarial safety in pregnancy.

Figure 2. Study site in south‐western Senegal (study presented in chapter 5).

KenyaStudies described in chapters 6 to 8 were conducted within the Health and Demographic

Surveillance System (HDSS) in western Kenya under a long‐standing collaboration between the

Kenya Medical Research Institute (KEMRI) and the US‐based Centers for Disease Control and

Prevention (CDC).[92,93] The HDSS operates a quarterly survey covering an area of about 700 km2

and 225,000 people living in 385 villages. Data on pregnancies, births, deaths, cause of deaths via

verbal autopsies and migrations are collected.[94] This is an integrated field site designed to manage

the longitudinal follow up of residential units, households and individuals. The field operations also

involve surveillance of paediatric out‐patient visits in peripheral health facilities, and monitoring of

paediatric in‐patient visits at two District Hospitals. In addition, household socioeconomic and

Introduction

17

educational status surveys are conducted annually to complement the morbidity and demographic

data. Extensive laboratory facilities have been established to support diagnostic work in parasitic

and bacterial diseases as well as HIV; basic immunology and molecular biology research in these

areas is also conducted. The centre has conducted a number of large studies and trials (e.g., the

large community‐based, group‐randomised, controlled trial of permethrin‐treated bed nets (ITNs)

carried out between 1996 ‐1999, the Phase 3 trial of the RTS,S malaria vaccine and a Phase 2B trial of

a tuberculosis vaccine, among others).

Malaria transmission is perennial and holo‐endemic, although transmission has been greatly reduced

following provision of free ITNs.[95,96] The prevalence of malaria among individuals over 15 years of

age ranged between 10–20% in the period 2006 to 2008. [KEMRI/CDC, unpublished observations]

There is a high rate of HIV infection (in 2008: 15.4% overall: 20.5% among females and 10.2% among

males while the National HIV prevalence is around 7%).[97] The prevalence of TB in individuals over

15 years of age was 600/100,000 and geohelminth prevalence in pregnant women was recorded to

be as high as 76.2%.[98,99] Consequently, the area has mortality figures that reflect this burden of

infectious diseases ‐ infant mortality rate of 111 per 1,000 live births and a life expectancy at birth of

45 years in 2008.[93] The maternal mortality ratio is 669 per 100,000 live births which is higher than

the national estimate of 488 per 100,000 live births reported by the Kenya Demographic and Health

Survey in 2008/2009 for approximately the same time period.[100]

Nyanza Province

Figure 3. Study sites in western Kenya part of the KEMRI/CDC Health and Demographic Surveillance sites (studies presented in Chapters 6‐8).

References1. WHO (2002) The importance of pharmacovigilance. Safety monitoring of medicinal products. Geneva. 2. Lenz W (1988) A short history of thalidomide embryopathy. Teratology 38: 203‐215. 3. McBride W (1961) Thalidomide and congenital abnormalities. Lancet 278: 1358. 4. Zwingenberger K, Wnendt S (1995) Immunomodulation by thalidomide: systematic review of the literature

and of unpublished observations. J Inflamm 46: 177‐211. 5. Eriksson T (2014) Thalidomide in Mann’s Pharmacovigilance. In: Moore EBAaN, editor. Mann’s

Pharmacovigilance: John Wiley & Sons.

Chapter 1

18

6. WHO (2004) Pharmacovigilance: ensuring the safe use of medicines. Geneva. 7. Eypasch E, Lefering R, Kum CK, Troidl H (1995) Probability of adverse events that have not yet occurred: a

statistical reminder. BMJ 311: 619‐620. 8. Edwards IR (1999) Spontaneous reporting‐‐of what? Clinical concerns about drugs. Br J Clin Pharmacol 48:

138‐141. 9. Mann RD, Wilton LV, Pearce GL, Mackay FJ, Dunn NR (1997) Prescription‐event monitoring (PEM) in 1996‐a

method of non‐interventional observational cohort pharmacovigilance. Pharmacoepidemiol Drug Saf 6 Suppl 3: S5‐11.

10. Mann R, Andrews E (2007) Pharmacovigilance. Chichester: John Wiley & Sons Ltd. 11. Strom BL (2000) Pharmacoepidemiology. Sussex: John Wiley & Sons, Ltd. 12. Isah AO, Pal SN, Olsson S, Dodoo A, Bencheikh RS (2012) Specific features of medicines safety and

pharmacovigilance in Africa. Ther Adv Drug Saf 3: 25‐34. 13. UMC the Uppsala Monitoring Centre. 14. UMC (2014) Nine Million. Uppsala. 15. Pirmohamed M, Atuah KN, Dodoo AN, Winstanley P (2007) Pharmacovigilance in developing countries.

BMJ 335: 462. 16. Talisuna AO, Staedke SG, D'Alessandro U (2006) Pharmacovigilance of antimalarial treatment in Africa: is it

possible? Malar J 5: 50. 17. Chandramohan D, Shibuya K, Setel P, Cairncross S, Lopez AD, et al. (2008) Should data from demographic

surveillance systems be made more widely available to researchers? PLoS Med 5: e57. 18. INDEPTH The INDEPTH Effectiveness and Safety Studies of Anti‐malarial Drugs in Africa (INESS). 19. ACT consortium Collating drug safety data from ACT Consortium studies. 20. MiPc Malaria in Pregnancy Consortium Pharmacogivilance Activities. 21. Warkary J (1974) Problems in applying teratologic observations in animals to man. Paediatrics 53: 820. 22. Mitchell A (1994) Special considerations in studies of drug‐induced birth defects. In: Strom B, editor.

Pharmacoepidemiology. Chichester: Wiley. 23. White NJ, McGready RM, Nosten FH (2008) New medicines for tropical diseases in pregnancy: catch‐22.

PLoS Med 5: e133. 24. Little BB (1999) Pharmacokinetics during pregnancy: evidence‐based maternal dose formulation. Obstet

Gynecol 93: 858‐868. 25. Kallen BA (2005) Methodological issues in the epidemiological study of the teratogenicity of drugs.

Congenit Anom (Kyoto) 45: 44‐51. 26. Khoury MJ, Holtzman NA (1987) On the ability of birth defects monitoring to detect new teratogens. Am J

Epidemiol 126: 136‐143. 27. Mitchell AA (2003) Systematic identification of drugs that cause birth defects‐‐a new opportunity. N Engl J

Med 349: 2556‐2559. 28. Moore, Persaud (1998) The Developing Human: Clinically Oriented Embryology. sixth ed. Philadelphia: W.B.

Saunders Company. pp. 548. 29. Tilson HH, Doi PA, Covington DL, Parker A, Shields K, et al. (2007) The antiretrovirals in pregnancy registry:

a fifteenth anniversary celebration. Obstet Gynecol Surv 62: 137‐148. 30. Holmes LB, Wyszynski DF, Lieberman E (2004) The AED (antiepileptic drug) pregnancy registry: a 6‐year

experience. Arch Neurol 61: 673‐678. 31. CDER, CBER (2002) Guidance for Industry: Establishing Pregnancy Exposure Registries. Rockville: FDA. 32. European Medicines Agency (2005) Guideline on the exposure to medicinal products during pregnancy:

need for post‐authorisation data. 33. Rosa F (1999) Databases in the assessment of the effects of drugs during pregnancy. J Allergy Clin Immunol

103: S360‐361. 34. Charlton RA, Cunnington MC, de Vries CS, Weil JG (2008) Data resources for investigating drug exposure

during pregnancy and associated outcomes: the General Practice Research Database (GPRD) as an alternative to pregnancy registries. Drug Saf 31: 39‐51.

35. Pastuszak A, Pinelli M, Koren G (1993) Pregnancy outcome following fetal exposure to tiaprofenic acid in the first trimester. Am J Perinatol 10: 354‐357.

36. Louik C, Kerr S, Mitchell AA (2014) First‐trimester exposure to bupropion and risk of cardiac malformations. Pharmacoepidemiol Drug Saf.

37. WHO (2013) World Malaria Report: 2013. Geneva: World Health Organisation. 38. Steketee RW, Nahlen BL, Parise ME, Menendez C (2001) The burden of malaria in pregnancy in malaria‐

endemic areas. Am J Trop Med Hyg 64: 28‐35.

Introduction

19

39. Desai M, ter Kuile FO, Nosten F, McGready R, Asamoa K, et al. (2007) Epidemiology and burden of malaria in pregnancy. Lancet Infect Dis 7: 93‐104.

40. Guyatt HL, Snow RW (2001) The epidemiology and burden of Plasmodium falciparum‐related anemia among pregnant women in sub‐Saharan Africa. Am J Trop Med Hyg 64: 36‐44.

41. Brabin B, Verhoeff F (2002) The contribution of malaria. In: Maclean AB, Nielson J, editors. Maternal morbidity and mortality. London: Royal College of Obstetricians and Gynaecologists. pp. 65‐78.

42. Guyatt HL, Snow RW (2001) Malaria in pregnancy as an indirect cause of infant mortality in sub‐Saharan Africa. Trans R Soc Trop Med Hyg 95: 569‐576.

43. Walker PG, Ter Kuile FO, Garske T, Menendez C, Ghani AC (2014) Estimated risk of placental infection and low birthweight attributable to Plasmodium falciparum malaria in Africa in 2010: a modelling study. Lancet Glob Health 2: e460‐e467.

44. WHO (2004) A strategic framework for malaria prevention and control during pregnancy in the Africa region. . Brazzaville.

45. WHO (2012) Updated WHO Policy Recommendation (October 2012): Intermittent Preventive Treatment of malaria in pregnancy using Sulfadoxine‐Pyrimethamine (IPTp‐SP). Geneva: World Health Organization.

46. van Eijk AM, Hill J, Alegana VA, Kirui V, Gething PW, et al. (2011) Coverage of malaria protection in pregnant women in sub‐Saharan Africa: a synthesis and analysis of national survey data. Lancet Infect Dis 11: 190‐207.

47. WHO (2001) Antimalarial drug combination therapy: Report of a Technical Consultation. Geneva: World Health Organization.

48. Mutabingwa TK (2005) Artemisinin‐based combination therapies (ACTs): best hope for malaria treatment but inaccessible to the needy! Acta Trop 95: 305‐315.

49. WHO (2010) Guidelines for the treatment of malaria. Second Edition. Geneva. 50. Dondorp A, Nosten F, Stepniewska K, Day N, White N (2005) Artesunate versus quinine for treatment of

severe falciparum malaria: a randomised trial. Lancet 366: 717‐725. 51. Sinclair D, Donegan S, Isba R, Lalloo DG (2012) Artesunate versus quinine for treating severe malaria.

Cochrane Database Syst Rev 6: CD005967. 52. McGready R, Lee SJ, Wiladphaingern J, Ashley EA, Rijken MJ, et al. (2012) Adverse effects of falciparum and

vivax malaria and the safety of antimalarial treatment in early pregnancy: a population‐based study. Lancet Infect Dis 12: 388‐396.

53. Howes RE, Piel FB, Patil AP, Nyangiri OA, Gething PW, et al. (2012) G6PD deficiency prevalence and estimates of affected populations in malaria endemic countries: a geostatistical model‐based map. PLoS Med 9: e1001339.

54. Luzzatto L (2010) The rise and fall of the antimalarial Lapdap: a lesson in pharmacogenetics. Lancet 376: 739‐741.

55. Savini E, Moulin M, Herrou M (1971) Experimental study of the effects of quinine on the rat, rabbit and dog fetus. Therapie 26: 563‐574.

56. Colley J, Edwards J, Heywood R, Purser D (1989) Toxicity studies with quinine hydrochloride. Toxicology 54: 219‐226.

57. Lapointe G, Nosal G (1979) Saccharin‐ or quinine‐induced changes in the rat pups following prolonged ingestion by the dam. Biol Neonate 36: 273‐276.

58. Maxwell J (1908) The use of quinine during pregnancy, labour, and the puerperium. Transaction of the Royal Society of Tropical Medicine and Hygiene 1: 229‐235.

59. Mosha D, Mazuguni F, Mrema S, Sevene E, Abdulla S, et al. (2014) Safety of artemether‐lumefantrine exposure in first trimester of pregnancy: an observational cohort. Malar J 13: 197.

60. Nosten F, McGready R, d'Alessandro U, Bonell A, Verhoeff F, et al. (2006) Antimalarial drugs in pregnancy: a review. Curr Drug Saf 1: 1‐15.

61. Ward SA, Sevene EJ, Hastings IM, Nosten F, McGready R (2007) Antimalarial drugs and pregnancy: safety, pharmacokinetics, and pharmacovigilance. Lancet Infect Dis 7: 136‐144.

62. CDC (2011) Update: New Recommendations for Mefloquine Use in Pregnancy. Atlanta: Centres for Disease Control and Prevention.

63. Gonzalez R, Hellgren U, Greenwood B, Menendez C (2014) Mefloquine safety and tolerability in pregnancy: a systematic literature review. Malar J 13: 75.

64. WHO (2013) WHO Evidence Review Group on Intermittent Preventive Treatment (IPT) of malaria in pregnancy Geneva

65. Hernandez‐Diaz S, Werler MM, Walker AM, Mitchell AA (2000) Folic acid antagonists during pregnancy and the risk of birth defects. N Engl J Med 343: 1608‐1614.

Chapter 1

20

66. Kato T, Kitagwa S (1973) Production of congenital abnormalities in fetuses of rats and mice with various sulfonamides. Congenital Anomalies 13: 7‐13.

67. Tagbor H, Bruce J, Browne E, Randal A, Greenwood B, et al. (2006) Efficacy, safety, and tolerability of amodiaquine plus sulphadoxine‐pyrimethamine used alone or in combination for malaria treatment in pregnancy: a randomised trial. Lancet 368: 1349‐1356.

68. Tagbor HK, Chandramohan D, Greenwood B (2007) The safety of amodiaquine use in pregnant women. Expert Opin Drug Saf 6: 631‐635.

69. Longo M, Pace S, Messina M, Ferraris L, Brughera M, et al. (2012) Piperaquine phosphate: reproduction studies. Reprod Toxicol 34: 584‐597.

70. Batty KT, Moore BR, Stirling V, Ilett KF, Page‐Sharp M, et al. (2010) Investigation of reproductive toxicity of piperaquine in mice. Reprod Toxicol 29: 206‐213.

71. Poespoprodjo JR, Fobia W, Kenangalem E, Lampah DA, Sugiarto P, et al. (2014) Dihydroartemisinin‐piperaquine treatment of multidrug resistant falciparum and vivax malaria in pregnancy. PLoS One 9: e84976.

72. Rijken MJ, McGready R, Boel ME, Barends M, Proux S, et al. (2008) Dihydroartemisinin‐piperaquine rescue treatment of multidrug‐resistant Plasmodium falciparum malaria in pregnancy: a preliminary report. Am J Trop Med Hyg 78: 543‐545.

73. Tarning J, Rijken MJ, McGready R, Phyo AP, Hanpithakpong W, et al. (2012) Population pharmacokinetics of dihydroartemisinin and piperaquine in pregnant and nonpregnant women with uncomplicated malaria. Antimicrob Agents Chemother 56: 1997‐2007.

74. D'Alessandro U, Van Geertruyden J, T. M, Halidou T (2009) Safe and efficacious artemisinin‐based combination treatments for African pregnant women with malaria (PREGACT). ClinicalTrialsgov: Institute of Tropical Medicine, Belgium.

75. Desai M, Katana A, Gutman J (2012) Efficacy of intermittent screening and treatment or intermittent preventive treatment (IPT) with dihydroartemisinin‐piperaquine versus IPT with sulfadoxine‐pyrimethamine for the control of malaria in pregnancy in Kenya (STOP MiP Kenya). ClinicalTrialsgov: Kenya Medical Research Institute.

76. Nosten F, Cheah PY, McGready R (2010) Randomised Trial of 3 Artemisinin Combination Therapy for Malaria in Pregnancy (DMA). Registered study: University of Oxford.

77. Osarfo J (2010) Efficacy, safety and tolerability of dihydroartemisinin‐piperaquine for treatment of uncomplicated malaria in pregnancy in Ghana. ClinicalTrialsgov: Kwame Nkrumah University of Science and Technology.

78. Ter Kuile F (2013) STOPMiP: Intermittent screening and treatment or intermittent preventive therapy for the control of malaria in pregnancy in Indonesia. MiP Library: Liverpool School of Tropical Medicine.

79. Clark RL (2009) Embryotoxicity of the artemisinin antimalarials and potential consequences for use in women in the first trimester. Reprod Toxicol 28: 285‐296.

80. WHO, TDR (2006) Assessment of the safety of artemisinin compounds in pregnancy: report of two joint informal consultations convened in 2006. Geneva: WHO. WHO/GMP/TDR/Artemisinin/07.1. WHO/GMP/TDR/Artemisinin/07.1.

81. Finaurini S, Ronzoni L, Colancecco A, Cattaneo A, Cappellini MD, et al. (2010) Selective toxicity of dihydroartemisinin on human CD34+ erythroid cell differentiation. Toxicology 276: 128‐134.

82. Dellicour S, Hall S, Chandramohan D, Greenwood B (2007) The safety of artemisinins during pregnancy: a pressing question. Malar J 6: 15.

83. Nosten F, McGready R, Alessandro U, Bonell A, Verhoeff F, et al. (2006) Antimalarial Drugs in Pregnancy: A Review. Current Drug Safety 1: 1‐15.

84. Enel C, Pison G, Lefebvre M (1993) De l'accouchement traditionnel à l'accouchement moderne au Sénégal Cahiers Santé 3: 441‐446.

85. Pison G, Kodio B, Guyavarch E, Etard J‐F (2000) La mortalité maternelle en milieu rural au Sénégal. Population 55: 1003‐1018.

86. Brasseur P, Badiane M, Cisse M, Agnamey P, Vaillant MT, et al. (2011) Changing patterns of malaria during 1996‐2010 in an area of moderate transmission in southern Senegal. Malar J 10: 203.

87. INDEPTH (2011) Mlomp DSS. 88. Agnamey P, Brasseur P, Cisse M, Gaye O, Dumoulin J, et al. (2005) Economic evaluation of a policy change

from single‐agent treatment for suspected malaria to artesunate‐amodiaquine for microscopically confirmed uncomplicated falciparum malaria in the Oussouye District of south‐western Senegal. Trop Med Int Health 10: 926‐933.

Introduction

21

89. Agnamey P, Brasseur P, de Pecoulas PE, Vaillant M, Olliaro P (2006) Plasmodium falciparum in vitro susceptibility to antimalarial drugs in Casamance (southwestern Senegal) during the first 5 years of routine use of artesunate‐amodiaquine. Antimicrob Agents Chemother 50: 1531‐1534.

90. Brasseur P, Agnamey P, Gaye O, Vaillant M, Taylor WR, et al. (2007) Efficacy and safety of artesunate plus amodiaquine in routine use for the treatment of uncomplicated malaria in Casamance, southern Senegal. Malar J 6: 150.

91. Duthé G (2008) Malaria resurgence in Senegal: measuring malaria mortality in Mlomp. Population 63: 198. 92. Sankoh O, de Savigny D, Binka F (2004) Generating Empirical Population and Health Data in Resource

constrained Countries in the Developing World. INDEPTH working paper series 1. 93. Odhiambo FO, Laserson KF, Sewe M, Hamel MJ, Feikin DR, et al. (2012) Profile: the KEMRI/CDC Health and

Demographic Surveillance System‐‐Western Kenya. Int J Epidemiol 41: 977‐987. 94. Adazu K, Lindblade KA, Rosen DH, Odhiambo F, Ofware P, et al. (2005) Health and demographic

surveillance in rural western Kenya: a platform for evaluating interventions to reduce morbidity and mortality from infectious diseases. Am J Trop Med Hyg 73: 1151‐1158.

95. Ouma P, van Eijk AM, Hamel MJ, Parise M, Ayisi JG, et al. (2007) Malaria and anaemia among pregnant women at first antenatal clinic visit in Kisumu, western Kenya. Trop Med Int Health 12: 1515‐1523.

96. ter Kuile FO, Terlouw DJ, Phillips‐Howard PA, Hawley WA, Friedman JF, et al. (2003) Reduction of malaria during pregnancy by permethrin‐treated bed nets in an area of intense perennial malaria transmission in western Kenya. Am J Trop Med Hyg 68: 50‐60.

97. Amornkul PN, Vandenhoudt H, Nasokho P, Odhiambo F, Mwaengo D, et al. (2009) HIV prevalence and associated risk factors among individuals aged 13‐34 years in Rural Western Kenya. PLoS One 4: e6470.

98. van Eijk AM, Lindblade KA, Odhiambo F, Peterson E, Rosen DH, et al. (2009) Geohelminth Infections among pregnant women in rural western Kenya; a cross‐sectional study. PLoS Negl Trop Dis 3: e370.

99. van't Hoog AH, Laserson KF, Githui WA, Meme HK, Agaya JA, et al. (2011) High prevalence of pulmonary tuberculosis and inadequate case finding in rural western Kenya. Am J Respir Crit Care Med 183: 1245‐1253.

100. Desai M, Phillips‐Howard PA, Odhiambo FO, Katana A, Ouma P, et al. (2013) An analysis of pregnancy‐related mortality in the KEMRI/CDC health and demographic surveillance system in western Kenya. PLoS One 8: e68733.

23

Chapter2

Thesafetyofartemisininsduringpregnancy:apressingquestion

StephanieDellicour1,SusanHall2,DanielChandramohan1and

BrianGreenwood1

1 Department of Infectious and Tropical Diseases, London School of Hygiene and Tropical

Medicine, London, United Kingdom, 2 Worldwide Epidemiology, GlaxoSmithKline, Research

Triangle Park, United States of America

Malaria Journal 2007, 6:15

Copyright© BioMed Central Ltd

Chapter 2

24

AbstractBackground: An increasing number of countries in sub‐Saharan Africa are changing to

artemisinins combination therapy (ACT) as first or second line treatment for malaria. There is an

urgent need to assess the safety of these drugs in pregnant women who may be inadvertently

exposed to or actively treated with ACTs.

Objectives: To examine existing published evidence on the relationship between artemisinin

compounds and adverse pregnancy outcomes and consider the published evidence with regard to

the safety of these compounds when administered during pregnancy.

Methods: Studies on ACT use in pregnancy were identified via searches of MEDLINE, EMBASE,

Cochrane and Current Contents databases. Data on study characteristics, maternal adverse

events, pregnancy outcomes and infant follow up were extracted.

Results: Fourteen relevant studies (nine descriptive/case reports and five controlled trials) were

identified. Numbers of participants in these studies ranged from six to 461. Overall there were

reports on 945 women exposed to an artemisinin during pregnancy, 123 in the 1st trimester and

822 in 2nd or 3rd trimesters. The primary end points for these studies were drug efficacy and

parasite clearance. Secondary endpoints were birth outcomes including low birth weight, pre‐

term birth, pregnancy loss, congenital anomalies and developmental milestones. While none of

the studies found evidence for an association between the use of artemisinin compounds and

increased risk of adverse pregnancy outcomes, none were of sufficient size to detect small

differences in event rates that could be of public health importance. Heterogeneity between

studies in the artemisinin and comparator drugs used, and in definitions of adverse pregnancy

outcomes, limited any pooled analysis.

Conclusion: The limited data available suggest that artemisinins are effective and unlikely to be

cause of foetal loss or abnormalities, when used in late pregnancy. However, none of these studies

had adequate power to rule out rare serious adverse events, even in 2nd and 3rd trimesters and

there is not enough evidence to effectively assess the risk‐benefit profile of artemisinin compounds

for pregnant women particularly for 1st trimester exposure. Methodologically rigorous, larger

studies and post‐marketing pharmacovigilance are urgently required.

The safety of artemisinins during pregnancy

25

BackgroundEvery year over 50 million pregnancies occur in areas where malaria is endemic, mostly in sub‐

Saharan Africa. Malaria can have serious consequences for the mother and her baby. In regions

where malaria transmission is unstable, pregnant women are at high risk of developing severe

malaria, spontaneous abortion, stillbirth or premature delivery [1]. In high transmission regions,

infected pregnant women are often asymptomatic but parasitaemia can cause maternal anaemia and

low birth weight (LBW), a leading cause of infant morbidity and mortality. About 8–14% of LBWs and

3–8% of infant mortality in sub‐Saharan Africa are attributable to pregnancy‐associated malaria

(PAM)[2].

The WHO recommends a three‐pronged approach to control malaria during pregnancy which includes

effective case management, intermittent preventive treatment (IPT) and insecticide‐treated nets

(ITN). Until recently, WHO recommendations for case management during any trimester of pregnancy

were chloroquine (CQ), or sulphadoxine‐pyrimethamine (SP) in CQ‐resistant areas and, alternatively,

quinine in areas where neither were effective. There is limited information on the safety profile of

most antimalarials used for treatment or prevention of malaria in pregnancy. In a period of increasing

parasite resistance to inexpensive, conventional antimalarials such as chloroquine, SP and quinine,

there is a pressing need to identify drugs which have the most favourable harm‐benefit balance for

these vulnerable patients [3,4]. Artemisinins are a potentially valuable alternative as they are highly

effective, act rapidly and are well‐tolerated. In addition, they have the potential to reduce the

transmission of malaria and to slow development of resistance [5]. In 2002, after a detailed review of

published and unpublished data, a WHO expert committee concluded that artemisinins could be used

during the second or third trimesters if no suitable alternative was available [6] However, treatment

in the first trimester was not recommended unless the life of the woman was at risk because of

concerns raised by animal experiments which suggested that artemisinins might be teratogenic and

cause foetal resorption. Further studies have confirmed the embryotoxic effects of artemisinin and its

derivatives in animals, including primates, with risk being confined to a defined period of gestation

[7,8]. However it remains unknown how these findings translate to man. Because of these safety

'signals' from animal models, there is an urgent need to establish the safety profile of this class of

drugs in pregnancy. By June 2006, 37 countries in Africa had adopted ACTs as the first or second line

treatment policy [9] and consequently pregnant women are likely to be increasingly exposed to ACTs;

many women will be exposed in the first trimester before they are aware of their pregnancy. This

review examines all published and ongoing studies reporting exposure to artemisinins during

pregnancy and discusses their safety during pregnancy.

MethodsElectronic searches were made using MEDLINE/PubMed, EMBASE (1980 – July 2006) databases and

through the Cochrane Central Register of Controlled Trials (CENTRAL). Reference lists from identified

articles and conference abstracts, including those of the Multilateral Initiative on Malaria (MIM)

conference, 2005, and the American Society of Tropical Medicine and Hygiene (ASTMH) annual

meeting, 2005, were reviewed. The National Institute of Health's list of clinical trials was searched.

Studies on exposure of pregnant women with Plasmodium falciparum malaria to any artemisinin

derivative in, irrespective of whether the participants were symptomatic, have been included in this

review. The following data were extracted from reports: setting, year, study design, sample size,

characteristics of study participants, drug regimen, and outcomes of interest. Outcomes of interests

Chapter 2

26

were: (1) maternal serious adverse events (fatal, life‐ threatening or requiring hospitalization); (2)

maternal non‐serious adverse events (3) adverse pregnancy outcomes (miscarriage, stillbirth, preterm

delivery, low birth weight, neonatal death, congenital abnormality, developmental delay).

The internal and external validity of each study was assessed based on reporting comprehensiveness,

the representativness of the study population, how bias and confounding factors were accounted for,

and the power of the study.

DescriptionofstudiesSixty‐nine studies of artemisinin‐related drugs were identified. However, 55 reports were excluded

because they were review papers, animal studies, or because study participants were not pregnant

women. Among the fourteen studies that had reported maternal and foetal outcomes of treatment

with an artemisinin derivative in pregnancy, five were randomized trials (RCT) and nine were

descriptive non‐randomized studies. Overall these represent 1,121 women who had been exposed to

an artemisinin compound during pregnancy (242 in RCTs and 879 in observational studies). Taking

into account inclusion of women in more than one of the studies reported from the Thai‐Burmese

border, reports were found of 945 pregnancies exposed to an artemisinin compound (123 in the 1st

trimester and 822 in 2nd or 3rd trimesters). The design, characteristics of the study population, type

of intervention and outcomes of each study included in the review are shown in Table 1.

RandomizedcontrolledtrialsFive RCTs of artemisinin in pregnant patients have been reported [10‐14]. A randomized trial in

Nigeria studied 45 pregnant women treated with artemether alone or in combination with

mefloquine, but provided only limited information regarding the safety of artemether because of

inconsistent infant follow up [14]. In Thailand, an effectiveness and safety trial compared artesunate

plus mefloquine with quinine in 57 pregnant women. This study excluded women who had taken

antimalarials within the previous 28 days and those who had over 4% parasitized red cells. The study

population comprised predominantly milder cases. Specific neurological evaluation of mothers was

performed weekly using established tests and infants were assessed for physical and neurological

development at 12 or 24 months [10]. Three of the comparative trials were conducted in antenatal

clinics (ANCs) on the Thai Burmese Border. They compared different drug regimens of artemisinin or

its derivatives to quinine in second or third trimester pregnancies. In 2000, a report of 108 pregnant

women randomized to receive quinine or artesunate plus mefloquine was published [12]. This was

followed in 2001 by a report of a randomized trial of artesunate versus quinine plus clindamycin in

129 pregnant women [13]. More recently, the results of a trial of 81 pregnant women randomized to

receive quinine or artesunate‐atovaquone‐proguanil has been published [11]. Women who presented

with a first malaria episode were recruited excluding those with complicated or severe malaria.

Detailed information on outcome was available. A high level of efficacy of the artemisinin derivatives

was reported in each study and the drugs were well tolerated.

DescriptivestudiesThe first descriptive report of the use of artemisinin derivatives during pregnancy came from China.

Wang et al reported the outcomes of six pregnancies exposed to either artemisinin or artemether.

Children of these women were examined 5 to 10 years later for congenital malformation, physical and

neurodevelopmental evaluation. No adverse events were reported for the mother or the infants [15].


27

McGready et al reported the outcome of case‐series pregnancies exposed to artemisinins on the Thai‐

Burmese border in three succeeding publications [16‐18]. The latest publication of 2001 encompasses

all cases seen between 1992–2000 (461 women with 539 artemisinin‐based treatments) including

women described in previous publications [12,13]. Overall, there was 11% loss to follow up. A

subgroup of 44 women was exposed inadvertently during their first trimester. In addition, two

publications report on the treatment of 27 pregnant patients with atovaquone‐proguanil plus

artesunate [19,20].

The second publication [20], focuses on the pharmacokinetics properties of the drugs in 24 of these

women. This group of studies provides the largest and most detailed record on the use of

artemisinins in pregnancy.

During a mass drug administration campaign in the Gambia, 287 pregnant women were accidentally

exposed to treatment with artesunate plus SP, including 77 who were exposed during the first

trimester of pregnancy [21]. Both active and passive surveillance were used to maximize

ascertainment of pregnancy outcomes. Reliable case ascertainment and outcome measurement was

available for births notified within seven days of delivery, as these newborns were thoroughly

examined by a paediatrician.

In Sudan, Adam et al have reported on two studies [22,23]. The first consisted of the follow‐up of 28

symptomatic pregnant women who received intramuscular artemether after previous treatment

failure with CQ or quinine. Women were assessed for neurological defects. There was one first

trimester exposure which resulted in a normal newborn. The second study described 32

symptomatic pregnant women treated with artesunate plus SP. In both studies, women and their

infants were followed up systematically (at birth and one year of age).

Ashley et al reported accidental exposure to artemisinin compounds during pregnancy in three

clinical trials for the treatment of uncomplicated P. falciparum malaria [24]: they documented the

inadvertent exposures to dihydroartemisinin‐piperaquine of one woman at 11 weeks gestation and

another at 18 weeks gestation [25,26]. In Uganda, four pregnant women were accidentally exposed

in their 1st trimester to artemether‐lumefantrine during a trial [27]. All these women delivered

normal babies.

ResultsMaternal adverse events and pregnancy outcomes in the artemisinin group and comparison groups

(where applicable) are summarized in Tables 2 and 3 respectively.

MaternaladverseeventsOverall, artemisinin derivatives were well tolerated; none of the studies reported serious adverse

event (SAE) attributed to the use of artemisinin. There were seven maternal deaths reported in total:

three women exposed to an artemisinin (two on the Thai‐Burmese border and one during the mass

drug administration campaign in the Gambia), three with unknown exposure (two reports from RCTs

on the Thai‐Burmese border and one in the Gambia) and one not exposed from the Gambia study.

Cause of death was known for only five women [severe malaria and anaemia (1), liver abscess (1),

post‐partum haemorrhage (3)]. Other maternal adverse events were described clearly only in the

studies conducted at the Thai‐Burmese border.

Chapter 2

28

Table 1: Description of the studies included in the review on the use of artemisinin in pregnancy.

Study site, year Reference Study Population Antimalarial treatment Primary Outcomes

Randomized trials*

Nigeria, 1994–1997 [14] Pregnant women infected with P. falciparum malaria after treatment failure§

1) Artemether IM + mefloquine n = 22 2) Artemether IM n = 23 Artemether overall dosage: 3.2–9.6 mg/kg

Parasitaemia at day 28; AEs Foetal outcomes; infant neurodevelopment

Thailand, 1995–1998 [10] Pregnant women infected with P. falciparum malaria§

1)Quinine n = 29 2)Artesunate+ mefloquine n = 28 Artesunate overall dose: 12 mg/kg

Time to parasite & fever clearance; anaemia; AEs; neurological examination Perinatal assessment; birth outcomes; infant physical and neurological assessment

TBB, 1995–1997 [12] Pregnant women with uncomplicated P. falciparum malaria

1)Quinine n = 42 2)Artesunate+mefloquine n = 66 Artesunate overall dose: 12 mg/kg

Parasite clearance; anaemia; AEs; neurological deficit Birth outcomes; infant developmental milestones

TBB, 1997–2000 [13] Pregnant women with uncomplicated P. falciparum malaria

1)Quinine+clindamycin n = 65 2) Artesunate n = 64 Artesunate overall dose: 12 mg/kg

Treatment failure at day 42 or before delivery; anaemia; AEs Birth outcomes; infant developmental milestones

TBB, 2001–2003 [11] Pregnant women with uncomplicated P. falciparum or mixed malaria

1)Quinine n = 42 2) Artesunate atovaquone-proguanil n = 39 Artesunate overall dose: 12 mg/kg

Fever, parasite clearance; treatment failure at day 63; anaemia; AEs Birth outcomes; infant developmental milestones

Case series

China, 1976–1980 [15] Pregnant women infected with P. falciparum or P vivax malaria§

Trimesters (n): 3rd (1), 2nd (5)

Artemisinin IM or Artemether Overall artemisinins dose: 500–900 mg

Therapeutic effects and adverse reactions for mother and child

TBB, 1992–1996 [17] Pregnant women with uncomplicated multidrug resistant P. falciparum or mixed malarial infection Trimesters (n): 3rd & 2nd (74), 1st (16)

Artesunate alone or with mefloquine; or Artemether+mefloquine Overall artemisinins dose: 12–16 mg/kg

Treatment failure at day 42; anaemia; ADR; neurological deficit Birth outcome; infant neurological development

TBB, 1991–1996 [18] Pregnant women with uncomplicated multidrug resistant P. falciparum malarial infection after treatment failure Trimesters (n): 3rd & 2nd (74), 1st (16)

Artesunate alone or with mefloquine; or Artemether+mefloquine Overall artemisinins dose: 12–16 mg/kg

Treatment failure at day 42; anaemia; ADR; neurological deficit Birth outcome; infant neurological development

TBB, 1992–2000 [16] Pregnant women with uncomplicated multidrug resistant P. falciparum or mixed malarial infection Trimesters (n): 3rd (211), 2nd (201), 1st (42)

Artesunate alone or with mefloquine; clindamycin atovaquone-proguanil;or coartemether, artemether IM Overall artemisinins dose: 12–16 mg/kg

Treatment failure at day 42; anaemia; ADR; neurological deficit Birth outcome; infant developmental milestones

TBB 1999–2001 [19] Pregnant women with multi-drug resistant P. falciparum or mixed malarial infection Trimesters (n): 3rd & 2nd (24), 1st (3)

Artesunate+atovaquone-proguanil Artesunate overall dose: 12–14 mg/kg

Treatment failure at day 42; parasite clearance; AEs Birth outcomes

TBB 2000–2001 [20] Pharmacokinetic study: pregnant women with multi-drug resistant P. falciparum or mixed malarial infection Trimesters (n): 3rd (13) & 2nd (11)

Artesunate+atovaquone-proguanil Artesunate overall dose: 12 mg/kg

Pharmacokinetic parameters; AEs (including ECG); parasite clearance Birth outcomes

Gambia, 1999 [41] Pregnant women participating in mass prevention campaign Trimesters (n): 1st (77)

Artesunate + SP Artesunate overall dose: 200 mg

MDA 1° aim: malaria transmission Pregnancy outcomes

Sudan, 1997–2001 [22] Pregnant women infected with P. falciparum malaria after treatment failure§

Trimesters (n): 3rd (15), 2nd (12), 1st (1)

Artemether IM Artemether overall dose: 480 mg

Treatment failure at day 28; anaemia; neurological deficit Birth outcomes

Sudan, 2004–2005 [23] Pregnant women infected with uncomplicated P. falciparum malaria§

Trimesters (n): 3rd (22), 2nd (10)

Artesunate + SP Artesunate overall dose: 200 mg

Treatment failure at day 28; anaemia Birth outcomes

* 2nd and 3rd trimesters only §100% symptomatic/febrile Abbreviation: ADR: adverse drug reaction; AE: adverse event; ECG: electrocardiograph; IM: intramuscular injection; NR: not reported; Q: quinine; A: artemisinin derivative; A+M: artesunate+mefloquine; TBB: Thai-Burmese border

In these studies, adverse events that could potentially be associated with the study drug were defined

as the occurrence of symptoms after treatment not present at baseline and which occurred before a

recrudescence of parasitaemia or a second infection. Data from Thailand and the Thai Burmese

border showed that tinnitus and dizziness were significantly less common in the artemisinin

treatment group (range 9–64%) compared to the quinine arm (45–79%). Although nausea and

vomiting were also significantly less frequent in the artesunate‐mefloquine treatment arm in the

Thailand study, no significant difference was reported in the Thai‐Burmese border trials. Other

adverse events were reported with a similar frequency in each treatment group. The most frequent

complaints according to these studies were dizziness (range 42–45% for the artemisinin group versus

49– 87% for the quinine arm); nausea (range 22–57% and 8– 93% respectively); headache (range 21–


29

30% and 30–50% respectively); anorexia (range 7–33% and 0–47% respectively) and muscle and joint

pain (range 12–31% and 12– 27% respectively).

All of these symptoms were also reported on presentation (dizziness 57–61%; headache 72–75%;

anorexia 46–62%, nausea 30%–40% and muscle/joint pain 59–68%) and so the apparent adverse

events may have been due to the underlying malarial episode. Specific neurological evaluation of

mothers was performed in four studies [10,12,17,22]. This included Romberg's test, assessment of

heel‐toe ataxia, fine finger dexterity, auditory acuity and an assessment for nystagmus. There was no

report of a woman with an adverse neurological event associated with drug administration. The

studies were too heterogeneous in terms of severity of disease, drug treatment and outcome

measurements, and numbers to few to pool data for a meta‐analysis.

Table 2: Safety outcomes: maternal adverse events and pregnancy outcomes in women exposed to artemisinin compounds. Randomized trials

Study site, year Ref

Antimalarial treatment N (% follow up) Outcomes of interests

Maternal Safety¥ Pregnancy outcomes

Nigeria, 1994–1997 [14] 1) Artemether IM + mefloquine n = 22 2) Artemether IM n = 23

45 (100%) Minimal, only A+M: abdominal discomfort (9%) and dizziness (9%)

Neonatal jaundice (n = 2 A & n = 1 A+M) 6 (13%) followed up to 1 year

Thailand, 1995–1998 [10] 1) Quinine n = 29 2) Artesunate+ mefloquine n = 28

60 (95%) Neurological exam: all normal. Nausea (16%), vomiting (12%), vertigo (12%), tinnitus (18%), and hypoglycaemia (3%) more frequent in Q (p < 0.05). Other no difference: Palpitation (6%), blurring vision (11%).

Neonatal jaundice (n = 5 Q & n = 1 A+M) 46 (81%) followed up to 1 year

TBB, 1995–1997 [12] 1) Quinine n = 42 2) Artesunate+ mefloquine n = 66

108 (85%) Neurological exam 1 maternal death* Tinnitus (15%) and dizziness (42%) more frequent in Q (p < 0.05). Headache (21%), nausea (45%), abdominal pain (28%), vertigo (12%), muscle/joint pain (32%), and anorexia (35%) no difference with Q (p > 0.05).

Abortions (n = 2 A+M) 46 (49%) followed up to 1 year Neonatal deaths (n = 2 A+M & n = 1 Q)

TBB, 1997–2000 [13] 1) Quinine)+ clindamycin n = 65 2) Artesunate n = 64

129 (91%) Tinnitus (9%) more frequent in Q+C (p < 0.05). Headache (30%), dizziness (41%) nausea (25%), vomiting (8%), abdominal pain (18%), rash (9%), contractions (35%), muscle/joint pain (12%), and anorexia (42%) no difference with Q+C (p > 0.05).

Stillbirths (n = 1 A & n = 1 Q+C) Congenital abnormality: midline epidermoid cyst (n = 1 Q+C) Neonatal deaths (n = 1 A & n = 2 Q+C) 72 (62%) followed up to 1 year

TBB, 2001–2003 [11] 1) Quinine n = 42 2) Artesunate atovaquone-proguanil (AAP) n = 39

81 (91%) 1 maternal death** Tinnitus (24%) more frequent in Q (p < 0.05).

Stillbirth (n = 1 not specified maternal death) Congenital abnormalities: polythelia (n = 1 AAP); cleft lip & palate (n = 1 AAP); aural atresia (n = 1 Q) Neonatal deaths (n = 1 A & n = 2 Q+C) 59 (78%) followed up to 1 year Developmentally delayed (n = 1 AAP)

TOTAL No 1st trimester, P. falciparum or mixed malaria 17% to 100% symptomatic 242 artemisinins exposures

423 (94%) 2 maternal deaths Neurological exam in 2 studies

Stillbirths (n = 1 A; n = 1 Q+C & n = 1 unknown) Congenital abnormality: (n = 2 A & n = 2 Q) Neonatal deaths (n = 4 A+M & n = 5 Q) 229 (59%) infant followed up to 1 year

¥ Prevalence of possible ADRs are only reported for the artemisinin treatment groups * cause unrelated to malaria (treatment group NR) ** caused by a ruptured liver abscess (treatment group NR) Abbreviation: ADR: adverse drug reaction; AE: adverse event; IM: intramuscular injection; NR: not reported; Q: quinine; A: artemisinin derivative; A+M: artesunate+mefloquine; TBB: Thai-Burmese border

PregnancyoutcomesNinety‐six percent of the 945 women exposed to an artemisinin in pregnancy were followed up to

delivery. Twenty (2.1%) had miscarriages, 19 (2%) stillbirths and 11 (1.2%) neonatal deaths. Six (0.6%)

congenital abnormalities were reported (one left aural atresia, one polythelia, one epidermoid cyst

Chapter 2

30

and three not described); whereas the expected rate of birth defects in developing countries is about

6% [28]. Of the 214 infants examined up to at least one year of age only one was reported to be

developmentally delayed. The mother of this infant was treated with artesunate‐atovaquone‐

proguanil and he was assessed for motor and neuro‐developmental milestones (no details provided).

Table 3: Safety outcomes: maternal adverse events and pregnancy outcomes in women exposed to artemisinin compounds: Descriptive studies.

Study site, year Ref Antimalarial treatment N (% Follow-up) Outcomes of interests

Maternal Safety Pregnancy outcomes

Overall 1st trimester exposures

China, 1976–1980 [15] Artemisinin IM or Artemether 6 (100%) No adverse effect 6 (100%) followed up at 5–9 year

None

TBB, 1992–2000 [16] Artesunate alone or with mefloquine; clindamycin atovaquone-proguanil; coartemether, artemether IM

461 (89%) No ADRs (pruritus) Maternal deaths (n = 2)*

Abortions 4.8% (n = 20) Stillbirth 1.8% (n = 7) Congenital abnormalities 0.8%: anencephaly (n = 1); midline epidermoid cyst (n = 1)

Abortions 23% (n = 10)

TBB 1999–2001 [19] Artesunate+atovaquone- proguanil

27 (100%) Symptoms possible ADRs: Headache (42%); muscle/joint pain (30%); abdominal pain (42%); anorexia (40%); nausea (25%); vomiting (10%); rash (15%); dizziness (70%), tinnitus (24%); contraction (15%) and sleep disturbance (8%)

Neonatal deaths (n = 1) Normal deliveries and healthy newborns

Gambia, 1999 [41] Artesunate + SP 325 (88%) Maternal deaths (n = 1)** Stillbirth (n = 11) Congenital abnormalities: umbilical hernia (n = 1); undescended testis; (n = 1) Neonatal deaths (n = 8)

Not reported

Sudan, 1997–2001 [22] Artemether IM 28 (100%) NR Neonatal deaths (n = 1) Normal delivery and healthy newborn

Sudan, 2004–2005 [23] Artesunate + SP 32 (100%) Giddiness and nausea (13%) Neurological exam.

Neonatal deaths (n = 1) None

TOTAL 123 1st trimester, P. falciparum or mixed malaria 7% to 100% symptomatic 945 artemisinin exposures

879 (90%) 3 maternal deaths Neurological exam in 2 studies

Abortions n = 20 Stillbirths (n = 18) Congenital abnormality: (n = 4) Neonatal deaths (n = 11) 65 (15%) infant followed up to 1 year

Abortions n = 10

* One maternal death was due to severe malaria and anaemia, the other died of causes unrelated to malaria. ** For 1 maternal death the exposure status could not be confirmed; verbal autopsies indicate that the deaths were due to postpartum haemorrhage. Abbreviation: ADR: adverse drug reaction; AE: adverse event; IM: intramuscular injection; NR: not reported; Q: quinine; A: artemisinin derivative; A+M: artesunate+mefloquine; TBB: Thai-Burmese border

DiscussionNone of the studies included in this review revealed an increased risk of serious maternal adverse

effects, adverse birth outcomes, or neuro‐development deficits associated with the use of an

artemisinin drug during pregnancy. However, these studies were not designed to assess safety

endpoints and, although they were sufficiently powered to answer the original study objective, the

studies were under‐powered to detect rare safety outcomes, or small differences in adverse event

rates, between the comparison groups.

The very low prevalence of congenital anomalies reported can partly be explained by the complex

nature of birth defect ascertainment; ideally a dysmorphologist should have assessed all newborns

for abnormalities that would not be detected by an untrained physician. It was also not possible to

assess the expected rate of adverse birth outcomes in any of the study settings due to the lack of


31

background population data on rates for abortions, stillbirths and congenital abnormalities. Four

studies which looked for neurological damage following artemisinin exposure revealed no indication

of neurotoxicity. However, in the light of recent debates about the potential ototoxicity of the

artemisinins, further studies conducting thorough auditory evaluation before and after treatment

may be needed [29‐31].

The studies reviewed were highly heterogeneous in terms of treatment used, outcomes assessment,

population and follow up rate. Different artemisinin drug regimens were used with varying control

treatments as a comparator and six studies did not have a control group but compared the rates of

adverse birth outcome to community rates derived from separate studies. Most of the studies used

an artemisinin derivative combined with another drug, which adds to the difficulty of teasing out

individual drug effects and restricts comparison between studies. Four studies used artesunate in

combination with mefloquine; use of the latter in pregnancy has caused concern because of an

increased risk of stillbirth in women who received this drug during pregnancy in one study [32]

although this was not found in others [33‐36]. The methods used to monitor adverse event and birth

outcomes also differed widely between the studies. Under‐estimates of adverse outcomes cannot

be ruled out, particularly for early pregnancy loss in the mass drug administration campaign in the

Gambia. There are no statistics on expected rate of early pregnancy loss, and determining the

prevalence of spontaneous abortion is difficult. In western countries, miscarriages occur in 10 to15%

of pregnancies, mostly during the first few weeks. The prevalence of miscarriages is likely to be even

higher in resource poor setting. When assessing maternal AEs, there is the methodological challenge

of differentiating between the effect of malaria and its treatment which is difficult to do.

Furthermore, the severity of malaria episode varied between the different study populations since

three included women who had already experienced a treatment failure whilst nine others included

some asymptomatic women. Furthermore, the study in the Gambia was preventive; as the majority

of the participants in this study are unlikely to have had malaria, lower rates of adverse birth

outcomes would be expected compared to those of studies of malaria case management. The

allocation of interventions could have been influenced by many factors particularly for the non‐

randomized studies, such as prognostic factors (severity or malaria attack rates, parasite resistance,

mother's age, gravidity, other drugs etc.), which could themselves influence birth outcome and

treatment response. There is, therefore, a potential for selection and detection bias. Maternal

malaria has been associated with stillbirth, abortion and LBW and these adverse end‐points are also

considered as possible indicators of an adverse event related to drug administration [1].

Furthermore, none of these studies controlled for other drug use, which could potentially influence

pregnancy outcomes.

Most of the information on artemisinin exposure during pregnancy comes from studies conducted in

Southeast Asia, at the antenatal clinics of the Shoklo Malaria Research Unit among women of the

Karen ethnic minority. The RCTs had robust design and randomly allocated treatment groups,

although the investigators were not blinded to the treatment group. Follow up of infants varied

between these three studies (49 to 80%) and only 46 infants were followed up to one year of age

and examined for developmental abnormalities in the non‐comparative studies. The women

enrolled in these studies are likely to be representative of this population group (Karen ethnic

minority) as over 90% of the women in the studied Thai‐ Burmese border region attend ANCs. These

findings cannot be extrapolated directly to sub‐Saharan Africa considering geographic differences in

parasite species, resistance pattern, transmission intensity and host immunity.

Chapter 2

32

A number of studies of artemisinin treatment in pregnancy are in progress (Table 4), which will

contribute further to knowledge on the safety of these drugs. However, except for the phase IV

study being conducted by the Centres for Disease Control in Tanzania, these studies focus on 2nd

and 3rd trimester pregnancies. The results of animal studies suggest that if there is a safety issue

related to the administration of artemisinins in pregnancy this is likely to occur very early in

pregnancy. Only 123 documented first trimester exposures have been reported and this does not

provide enough evidence to determine safety. Moreover, an "all or nothing rule" seems to apply for

exposure during the first weeks of pregnancy. Animal experiments suggest that congenital

abnormalities occur only after exposure over a narrow dose range and over a limited period of time

and that exposure usually lead to a normal pregnancy outcome or death of the foetus.

Developmental toxicity studies in the monkey confirmed findings from rodent studies; with embryo

death induced at therapeutic dose ranges [37]. The teratogenic effect is thought to involve red blood

cells production, erythropoiesis, which implies the human sensitive period would be within the first

trimester of pregnancy [38]. An effect of ACTs on early pregnancy loss will be difficult to detect,

especially in communities where artemisinins are likely to be used most frequently. More extensive

studies are needed that will be able to detect rarer outcomes and any congenital abnormalities that

might result from exposure in the first trimester of pregnancy. Ethical constraints will prohibit

randomized trials of artemisinins in the first trimester of pregnancy in most communities where

alternatives exist until more information on safety has been obtained. Thus, data will largely have to

come from observational studies.

Unfortunately, most countries in sub‐Saharan Africa do not have the infrastructure and resources for

routine pharmacovigilance and very few have a formal system for routine collection of data on

possible drug related adverse effects [39]. In industrialized countries, a variety of post‐marketing

surveillance techniques is used. Case reports and case series are the first source of information to

detect adverse events in pregnancy. Although these are useful in the generation of hypotheses of a

possible safety signal, specific pharmaco‐epidemiological studies using a cohort or case‐control

approach are required to evaluate teratogenic risk. The main factors impeding the implementation

of pharmacovigilance in poor countries include limited access to healthcare facilities, availability of

most prescription drugs from the informal market, poor labelling of medicines, counterfeit and sub‐

standard pharmaceutical products, a high level of illiteracy, poor record keeping and a shortage of

qualified healthcare professionals. Special pharmaco‐epidemiological studies will be needed to

assess the safety profile of a product's use outside the controlled environment of clinical trials.

Active surveillance systems could use the existing sentinel demographic surveillance sites (DSS),

which already undertake regular household visits to obtain more information on drug usage and

adverse effects during pregnancy. However, these studies are expensive and not sustainable in the

long term and should be restricted to newer products or following identification of new safety

concerns for an older drug. Antenatal clinics (ANC) could make a useful platform for routine

surveillance, as a high proportion of pregnant women attend an ANC at least once during pregnancy

in sub‐Saharan Africa [40]. It is necessary to develop a more pragmatic pharmacovigilance system

that can be linked to a routine health information system and ideally a mix of different approaches

should be used to assess the safety profile of individual drugs for pregnant patients.


33

ConclusionMalaria in pregnancy is a major public health issue and infected pregnant women need prompt

treatment with effective drugs. Although artemisinin derivatives and combinations have an excellent

efficacy profile, there is very limited data on the safety of artemisinin use during pregnancy

particularly in sub‐Saharan Africa. Although a few studies of the safety and efficacy of artemisinins

during pregnancy are currently underway, these will not produce data on the safety of artemisinins

during the first trimester of pregnancy. Post‐marketing pharmacovigilance of artemisinin use during

pregnancy is needed urgently as ACT is implemented in almost all countries in Africa. Innovative

pharmacovigilance tools, methods and systems are needed to monitor the safety of artemisinins and

other antimalarials.

Authors' contributionsSD carried out the literature review and wrote the first draft of the paper. SH, DC and BG contributed

to the structure and content and were involved in re‐drafting the paper.

AcknowledgementsWe thank Jenny Hill and Feiko ter Kuile, from the MiP working group, for sharing with us the

information collated for the MiP database. We are also grateful to John McArthur for sharing with us

the outline for the phase IV study in Tanzania, Rose McGready for clarifying the overlap between

publications for the studies on the Thai‐Burmese border and Steve Bowling for reviewing the

manuscript. Stephanie Dellicour is supported by GlaxoSmithKline.

Chapter 2

34

Table 4: Ongoing Studies (Information extracted from Malaria in Pregnancy (MiP)* Database [42] on 05/02/2007).

Study Setting Design & Drug Regimen Outcome Status (start date-completion date)

Safety/Efficacy prevention trial in pregnancy

Ifakara, Tanzania CDC/IHRDC-IMPACT Randomised open label, n = 1200 (400 per arm) IPTp Control: SP 2 doses Intervention: 1. SP monthly 2. SP+artesunate monthly

1°: Placental parasitaemia and AEs 2°: Maternal illness and parasitaemia at delivery, birth outcome (BW, Gestational age, foetal and infant health), childhood developmental milestones

Recruitment concluded; ongoing follow up (01/03-ongoing)

Safety/Efficacy Treatment trial in Pregnancy

ANC at Muheza Hospital, Tanzania GMP Randomised open label, target (Phase III) n = 350 2nd or 3rd trimesters Control: SP Intervention: SP+amodiaquine, Amodiaquine+artesunate, Chloroporguanil-dapsone

1°: Treatment failure at day 28; Treatment outcome (parasite/fever clearance, parasite recrudescence) 2°: Foetal viability and birth outcomes (preterm delivery, foetal death, perinatal/ neonatal mortality, neonatal abnormality); maternal AE (hypoglycaemia)

Recruitment completed (01/04–07/06)

Shoklo Malaria Research Unit (SMRU) ANC, Thailand UNICEF-UNDP-World Bank-WHO-TDR

Randomized intervention trial n = 250 2nd or 3rd trimesters Group 1: Artesunate Group 2: Co-artemether (artemether/ lumefantrine)

1°: Treatment outcome at day 42 or at delivery (parasite/fever clearance, parasite recrudescence) 2°:Gametocyte carriage; pharmacokinetic parameters; histo- pathology examination of the placenta

Currently recruiting (06/02/2004–31/12/ 2008)

Bangladesh WHO

Randomised controlled trial n = 684 Control: placebo rectal capsule Intervention: Artesunate rectal capsule

Pregnancy outcomes Currently recruiting (10/11/2003- ongoing)

Malawi; Prof Meshnick, UNC Randomised open label, n = 141 2nd or 3rd trimesters Control: SP Intervention: SP+artesunate or SP+azithromycin

1°: Parasitological failure rates; parasite clearance time; fever clearance times and incidence rate of adverse events 2°: Prevalence rate of abortions; still births; peripheral parasitaemia at delivery; placental malaria and of maternal anaemia

Recruitment completed (09/2003–10/ 2005)

Efficacy/Pharmacokinetic trial in Pregnancy

Mozambique UCT, South Africa

Non-Randomized openLabel, target n = 30 2nd or 3rd trimester pregnant HistoricalControl Intervention: SP+artesunate

1°: Pharmacokinetic parameters 2°: gametocyte carriage, maternal AE & birth outcomes

Currently recruiting (03/2006–09/2008)

Kinshasa, DRC, NIH-NICHD Dose-equivalence trial: part of investigational new drug application n = 60 2nd or 3rd trimester Control: SP Intervention: Artesunate-mefloquine combinations

Pharmacokinetic parameters Recruitment completed (07/2005–12/ 2005)

Pharmacovigilance: Post-marketing surveillance

Tanzania, CDC Pharmacovigilance surveillance system: part of a large ongoing study to look at district wide use of ACTs 1st trimester Control: SP Intervention: SP+Artesunate

Pregnancy outcome and status of child Ongoing (2005–2007)

A partnership between Novartis WHO- TDR and the Government of Zambia. [43]

Pregnancy Registry Prospective active surveillance cohort. Expected n = 1600 Control: SP Intervention: artemether-lumefantrine

Maternal and neonatal outcomes examined

Ongoing (2005)

* MiP is a consortium of experts in the field of malaria funded by the Bill and Melinda Gates Foundation to review current research and develop future research strategy for malaria in pregnancy. One of the key objectives of MiP is to create a database containing all published and unpublished research and a trial registry on malaria in pregnancy. Abbreviations: ACT: artemisinin combination therapy; AE: adverse events; ANC: antenatal care; BW: birthweight; CDC: Centers for Disease Control & Prevention; GMP: Gates Malaria Partenership; DRC: Democratic Republic of the Congo; IHRDC: Ifakara Health Research and Development Centre; IPTp: intermittent presumptive treatment for pregnancy; NIH-NICHD: National Institutes of Health-The National Institute of Child Health and Human Development; RCT: randomized controlled trials; SP: sulphadoxine-pyrmethamine; TBB: Thai-Burmese Border; UNC: University of North Carolina; UCT: University of Cape Town; WHO-TDR: World Health Organization – The Special Programme for Research and Training in Tropical Diseases


35

References1. Luxemburger C, McGready R, Kham A, Morison L, Cho T, Chongsu‐ phajaisiddhi T, White NJ, Nosten F:

Effects of malaria during pregnancy on infant mortality in an area of low malaria transmission. Am J Epidemiol 2001, 154:459‐465.

2. Steketee RW, Nahlen BL, Parise ME, Menendez C: The burden of malaria in pregnancy in malaria‐endemic areas. Am J Trop Med Hyg 2001, 64:28‐35.

3. Newman RD, Parise ME, Slutsker L, Nahlen B, Steketee RW: Safety, efficacy and determinants of effectiveness of antimalarial drugs during pregnancy: implications for prevention pro‐ grammes in Plasmodium falciparum‐endemic sub‐Saharan Africa. Trop Med Int Health 2003, 8:488‐506.

4. Nosten F, McGready R, d'Alessandro U, Bonell A, Verhoeff F, Menen‐ dez C, Mutabingwa T, Brabin B: Antimalarial Drugs in Preg‐ nancy: A Review. Current Drug Safety 2006, 1:1‐15.

5. WHO: Antimalarial drug combination therapy. In WHO/CDS/ RBM/2001 Geneva: World Health Organization; 2001.

6. WHO: Assessment of the safety of artemisinin compounds in pregnancy. Report of two informal consultations convened by WHO in 2002. In WHO/CDS/MAL/2003.1094 WHO/RBM/TDR/ Artemisinin/03.1 Geneva: WHO; 2003.

7. Clark RL, White TE, S AC, Gaunt I, Winstanley P, Ward SA: Devel‐ opmental toxicity of artesunate and an artesunate combina‐ tion in the rat and rabbit. Birth Defects Res B Dev Reprod Toxicol 2004, 71:380‐394.

8. Longo M, Zanoncelli S, Manera D, Brughera M, Colombo P, Lansen J, Mazue G, Gomes M, Taylor WR, Olliaro P: Effects of the antima‐ larial drug dihydroartemisinin (DHA) on rat embryos in vitro. Reprod Toxicol 2006, 21:83‐93.

9. Antimalarial treatment policies for P. falciparum and P. vivax by country in WHO Africa region [http://www.who.int/malaria/ amdp/amdp_afro.htm]

10. Bounyasong S: Randomized trial of artesunate and mefloquine in comparison with quinine sulfate to treat P. falciparum malaria pregnant women. J Med Assoc Thai 2001, 84:1289.

11. McGready R, Ashley EA, Moo E, Cho T, Barends M, Hutagalung R, Looareesuwan S, White NJ, Nosten F: A randomized comparison of artesunate‐atovaquone‐proguanil versus quinine in treat‐ ment for uncomplicated falciparum malaria during preg‐ nancy. J Infect Dis 2005, 192:846‐853.

12. McGready R, Brockman A, Cho T, Chq D, van Vugt M, Luxemburger C, Chongsuphajaisiddhi T, White N, Nosten F: Randomized com‐ parison of mefloquine‐artesunate combination versus qui‐ nine in treatment of multi‐drug resistant falciparum malaria in pregnancy. Trans R Soc Trop Med Hyg 2000, 94:689.

13. McGready R, Cho T, Leopold Villegas S, Brockman A, van M VI, Looareesuwan S, Whitezg NJ, Nosten F: Randomized comparison of quinine‐clindamycin versus artesunate in the treatment of falciparum malaria in pregnancy. Trans R Soc Trop Med Hyg 2001, 95:651.

14. Sowunmi A, Oduola AM, Ogundahunsi OA, Fehintola FA, Ilesanmi OA, Akinyinka OO, Arowojolu AO: Randomised trial of arte‐ mether versus artemether and mefloquine for the treat‐ ment of chloroquine/sufadoxine‐pyrimethamine‐resistant falciparum malaria during pregnancy. J Obstet Gynaecol 1998, 18:322‐327.

15. Wang TY: Follow‐up observation on the therapeutic effects and remote reactions of artemisinin (Qinghaosu) and arte‐ mether in treating malaria in pregnant woman. J Tradit Chin Med 1989, 9:28‐30.

16. McGready R, Cho T, Keo NK, Twai KL, Villegas L, Looareesuwan S, White N, Nosten F: Artemisinin antimalarials in pregnancy: a prospective treatment study of 539 episodes of multidrug‐resistant Plasmodium falciparum. Clin Infect Dis 2001, 33:2009.

17. McGready R, Cho T, Cho JJ, Simpson JA, Luxemburger C, Dubowitz L, Looareesuwan S, White NJ, Nosten F: Artemisinin derivatives in the treatment of falciparum malaria in pregnancy. Trans R Soc Trop Med Hyg 1998, 92:430‐433.

18. McGready R, Nosten F: The Thai‐Burmese border: drug studies of Plasmodium falciparum in pregnancy. Ann Trop Med Parasitol 1999, 93(Suppl 1):S19.

19. McGready R, Keo NK, Villegas L, White NJ, Looareesuwan S, Nosten F: Artesunate‐atovaquone‐proguanil rescue treatment of multidrug‐resistant Plasmodium falciparum malaria in pregnancy: a preliminary report. Trans R Soc Trop Med Hyg 2003, 97:592‐594.

20. McGready R, Stepniewska K, Ward SA, Cho T, Gilveray G, Looareesuwan S, White NJ, Nosten F: Pharmacokinetics of dihydroartemisinin following oral artesunate treatment of pregnant women with acute uncomplicated falciparum malaria. Eur J Clin Pharmacol 2006, 62:367‐371.

21. Deen JL, von Seidlein L, Pinder M, Walraven GE, Greenwood B: The safety of the combination artesunate and pyrimethamine sulfadoxine given during pregnancy. Trans R Soc Trop Med Hyg

Chapter 2

36

22. 2001, 95:424. 23. Adam I, Elwasila E, Mohammed Ali DA, Elansari E, Elbashir M: Artemether in the treatment of falciparum

malaria during pregnancy in eastern Sudan. Trans R Soc Trop Med Hyg 2004, 98:509. 24. Adam I, Ali DM, Abdalla MA: Artesunate plus sulfadoxine‐pyrimethamine in the treatment of

uncomplicated Plasmodium falciparum malaria during pregnancy in eastern Sudan. Trans R Soc Trop Med Hyg 2006, 100:632‐635.

25. Ashley EA, White NJ: Artemisinin‐based combinations. Curr Opin Infect Dis 2005, 18:531‐536. 26. Ashley EA, Krudsood S, Phaiphun L, Srivilairit S, McGready R, Leowat‐ tana W, Hutagalung R, Wilairatana P,

Brockman A, Looareesuwan S, Nosten F, White NJ: Randomized, controlled dose‐optimization studies of dihydroartemisinin‐piperaquine for the treatment of uncomplicated multidrug‐resistant falciparum malaria in Thailand. J Infect Dis 2004, 190:1773‐1782.

27. Ashley EA, McGready R, Hutagalung R, Phaiphun L, Slight T, Proux S, Thwai KL, Barends M, Looareesuwan S, White NJ, Nosten F: A ran‐ domized, controlled study of a simple, once‐daily regimen of dihydroartemisinin‐piperaquine for the treatment of uncomplicated, multidrug‐resistant falciparum malaria. Clin Infect Dis 2005, 41:425‐432.

28. Piola P, Fogg C, Bajunirwe F, Biraro S, Grandesso F, Ruzagira E, Babi‐ gumira J, Kigozi I, Kiguli J, Kyomuhendo J, Ferradini L, Taylor W, Checchi F, Guthmann JP: Supervised versus unsupervised intake of six‐dose artemether‐lumefantrine for treatment of acute, uncomplicated Plasmodium falciparum malaria in Mbarara, Uganda: a randomised trial. Lancet 2005, 365:1467‐1473.

29. Christianson A, Howson CP, Modell B: March of Dimes: Global report on birth defects. The hidden toll of dying and disabled children. New York: March of Dimes Birth Defects Foundation; 2006.

30. Adjei GO, Goka BQ, Kurtzhals JA: Neurotoxicity of artemisinin derivatives. Clin Infect Dis 2006, 43:1618. 31. McCall MB, Beynon AJ, Mylanus EA, van der Ven AJ, Sauerwein RW: No hearing loss associated with the use

of artemether‐lumefantrine to treat experimental human malaria. Trans R Soc Trop Med Hyg 2006, 100:1098‐1104.

32. Toovey S: A case‐control auditory evaluation of patients treated with artemether‐lumefantrine. Am J Trop Med Hyg 2006, 74:939‐940. author reply 940

33. Nosten F, Vincenti M, Simpson J, Yei P, Thwai KL, de Vries A, Chong‐ suphajaisiddhi T, White N: The effects of mefloquine treatment in pregnancy. Clin Inf Dis 1999, 28:808.

34. Adam I, Ali DA, Alwaseila A, Kheir MM, Elbashir MI: Mefloquine in the treatment of falciparum malaria during pregnancy in Eastern Sudan. Saudi Med J 2004, 25:1400‐1402.

35. Phillips‐Howard PA, Steffen R, Kerr L, Vanhauwere B, Schildknecht J, Fuchs E, Edwards R: Safety of mefloquine and other antimalarial agents in the first trimester of pregnancy. J Travel Med 1998, 5:121‐126.

36. Steketee RW, Wirima JJ, Slutsker L, Khoromana CO, Heymann DL, Breman JG: Malaria treatment and prevention in pregnancy: indications for use and adverse events associated with use of chloroquine or mefloquine. Am J Trop Med Hyg 1996, 55:50‐56.

37. Vanhauwere B, Maradit H, Kerr L: Post‐marketing surveillance of prophylactic mefloquine (Lariam) use in pregnancy. Am J Trop Med Hyg 1998, 58:17‐21.

38. Clark R, Kumemura M, Makori N, Nakata Y, Bernard F, Harrell A, White TEK, Arima A: Artesunate: Developmental Toxicity in Monkeys [Abstract]. 46th annual meeting of the Teratology Society; Tucson, Arizona 2006.

39. Laffan S, James A, Maleeff B, Pagana J, Bushdid P, Clark R, White T: Mitochondrial involvment of artesunate toxicity in rat embryonic erythroblasts. [Abstract]. 46th annual meeting of the Teratology Society; Tucson, Arizona 2006.

40. Lang T, Hughes D, Kanyok T, Kengeya‐Kayondo J, Marsh V, Haaland A, Pirmohamed M, Winstanley P: Beyond registration – measuring the public‐health potential of new treatments for malaria in Africa. Lancet Infect Dis 2006, 6:46‐52.

41. WHO: Antenatal care in developing countries: promises, achievements and missed opportunities: an analysis of trends, levels and differentials, 1990–2001. Geneva: WHO; 2003.

37

Chapter3

QuantifyingtheNumberofPregnanciesatRiskofMalariain2007:ADemographicStudy

StephanieDellicour1,AndrewJ.Tatem2,3,CarlosA.Guerra2,4,RobertW.Snow2,5,FeikoO.terKuile1,6

1 Child and Reproductive Health Group, Liverpool School of Tropical Medicine, Liverpool,

United Kingdom, 2 Malaria Public Health and Epidemiology Group, Centre for Geographic

Medicine, Kenyan Medical Research Institute–University of Oxford–Wellcome Trust

Collaborative Programme, Nairobi, Kenya, 3 Department of Geography and Emerging

Pathogens Institute, University of Florida, Gainesville, Florida, United States of America, 4

Spatial Ecology and Epidemiology Group, Department of Zoology, University of Oxford,

Oxford, United Kingdom, 5 Centre for Tropical Medicine, Nuffield Department of Clinical

Medicine, University of Oxford, CCVTM, Oxford, United Kingdom, 6 Department of

Infectious Diseases, Tropical Medicine & AIDS, Academic Medical Centre, University of

Amsterdam, Amsterdam, The Netherlands

PLoS Medicine 2010, 7:1 Copyright© This is an open‐access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited

Chapter 3

38

AbstractBackground: Comprehensive and contemporary estimates of the number of pregnancies at risk of

malaria are not currently available, particularly for endemic areas outside of Africa. We derived

global estimates of the number of women who became pregnant in 2007 in areas with Plasmodium

falciparum and P. vivax transmission.

Methods and Findings: A recently published map of the global limits of P. falciparum transmission

and an updated map of the limits of P. vivax transmission were combined with gridded population

data and growth rates to estimate total populations at risk of malaria in 2007. Country‐specific

demographic data from the United Nations on age, sex, and total fertility rates were used to

estimate the number of women of child‐bearing age and the annual rate of live births. Subregional

estimates of the number of induced abortions and country‐specific stillbirths rates were obtained

from recently published reviews. The number of miscarriages was estimated from the number of live

births and corrected for induced abortion rates. The number of clinically recognised pregnancies at

risk was then calculated as the sum of the number of live births, induced abortions, spontaneous

miscarriages, and stillbirths among the population at risk in 2007. In 2007, 125.2 million pregnancies

occurred in areas with P. falciparum and/or P. vivax transmission resulting in 82.6 million live births.

This included 77.4, 30.3, 13.1, and 4.3 million pregnancies in the countries falling under the World

Health Organization (WHO) regional offices for South‐East‐Asia (SEARO) and the Western‐Pacific

(WPRO) combined, Africa (AFRO), Europe and the Eastern Mediterranean (EURO/EMRO), and the

Americas (AMRO), respectively. Of 85.3 million pregnancies in areas with P. falciparum transmission,

54.7 million occurred in areas with stable transmission and 30.6 million in areas with unstable

transmission (clinical incidence,1 per 10,000 population/year); 92.9 million occurred in areas with P.

vivax transmission, 53.0 million of which occurred in areas in which P. falciparum and P. vivax co‐

exist and 39.9 million in temperate regions with P. vivax transmission only.

Conclusions: In 2007, 54.7 million pregnancies occurred in areas with stable P. falciparum malaria

and a further 70.5 million in areas with exceptionally low malaria transmission or with P. vivax only.

These represent the first contemporary estimates of the global distribution of the number of

pregnancies at risk of P. falciparum and P. vivax malaria and provide a first step towards a more

informed estimate of the geographical distribution of infection rates and the corresponding disease

burden of malaria in pregnancy.

Quantifying the number of pregnancies at risk of malaria

39

IntroductionMalaria in pregnancy can have devastating consequences to a pregnant woman and the developing

fetus, but comprehensive estimates of the annual number of women who become pregnant each

year in malaria endemic areas and are therefore at risk of malaria are not available, particularly for

Latin America and the Asia‐Pacific regions. These figures are an important first step towards

informing policy makers and for estimating the regional needs for therapeutic and disease

prevention tools for malaria in pregnancy. The most cited global estimate is from the Roll Back

Malaria Partnership, which states that ‘‘each year approximately 50 million women living in malaria

endemic countries throughout the world become pregnant’’ [1]. However, an explanation of the

methods used to derive these estimates is not provided. More comprehensive estimates exist for

Africa and are provided by the Africa Regional Office (AFRO) of the World Health Organization

(WHO) in their widely quoted strategic framework document for malaria prevention and control

during pregnancy in the African region [2]. Their estimate of 24.6 million pregnancies at risk of

malaria (predominantly P. falciparum), is based on the number of live born babies delivered in

malarious areas of Africa in the year 2000 using a combination of malaria risk maps [3] and estimates

of the number of live births from UNICEF [4]. A more recent estimate by the WHO states that ‘‘In

Africa, 30 million women living in malaria endemic areas become pregnant each year’’ [5]. Estimates

for outside of Africa are less clear, particularly for P. vivax. P. vivax is the most widely distributed

human malaria parasite and co‐occurs with P. falciparum in tropical areas but also occurs in

temperate regions outside the limits of P. falciparum transmission. It is the major cause of malaria in

much of Asia and Latin America [6,7], and recent evidence has shown that P. vivax infections are far

from benign and can result in significant morbidity in pregnant women with serious consequences

for maternal and infant health [8–10].

Here we define a global estimate of the number of pregnancies at risk of P. falciparum and P. vivax

malaria in 2007 by combining malaria spatial limits developed by the Malaria Atlas Project (MAP;

www.map.ox.ac.uk), which define the total population at risk of malaria [11], with country‐specific

demographic data on women of childbearing age provided by the United Nations and published data

on induced abortions and spontaneous pregnancy loss.

Methods

DataSourcesThegloballimitsofP.falciparummalaria.

The initial focus of the Malaria Atlas Project has been P. falciparum [12] due to its global

epidemiological significance [13] and better prospects for its control and local elimination [14]. The

global spatial limits of P. falciparum malaria transmission in 2007 have recently been mapped. This

was done by triangulating data on transmission exclusion using biological rules based on

temperature and aridity limits on the bionomics of locally dominant Anopheles vectors, data on

nationally reported case incidence rates, and other medical intelligence [15]. The resulting map

stratifies the malaria endemic world by stable and unstable transmission in 2007 [15]. Unstable

transmission refers to areas where transmission is plausible biologically, but limited, with a clinical

incidence of less than one case per 10,000 population per year. Stable transmission refers to areas

with a minimum of one clinical case per 10,000 population per year [15].

Chapter 3

40

ThegloballimitsofP.vivaxmalaria.

Initial attempts to map the limits of P. falciparum and P. vivax transmission were made by Guerra et

al. [16,17]. The resulting maps and ‘‘masks’’ (mapped areas that are filtered and excluded from

analyses) used were later tested against the Malaria Atlas Project parasite prevalence database to

assess their feasibility [18,19]. This testing revealed that the accuracy to define areas of zero

transmission risk due to very low population densities was limited because of the coarse spatial

resolution of the initial map. Moreover, in the initial mapping [16,17], a high density population

mask was used on the basis of the assumption that no transmission occurs in areas where the

population density is so high that conditions become unsuitable for transmission through the

process of urbanization. However, recent analyses [19] provide evidence suggesting that high

density population masks and urban extent maps should not be used to map zero risk because some

transmission can occur in high density urban areas, although this is significantly lower than in rural

areas [18,19]. Therefore, for the current analyses, the P. vivax limits were redefined using the same

methods as in Guerra et al. [16,17], but without applying the population‐based masks. Also,

previously excluded P. vivax endemic countries have now been added after a more extensive review

of the literature; these include Comoros, Djibouti, Madagascar, and Uzbekistan. This refinement of

the spatial limits of transmission for P. vivax accounted for an approximate 19% increase in the

population at risk (PAR) compared with previous estimates [16,17], principally (18%) due to the

inclusion of major cities.

Griddedpopulationdata.

The Global Rural‐Urban Mapping Project (GRUMP) alpha version provides gridded population counts

and population density estimates for the years 1990, 1995, and 2000, both adjusted and unadjusted

to the United Nations’ national population estimates [11,20]. The adjusted population counts for the

year 2000 were projected to 2007 by applying national, medium variant, intercensal growth rates by

country [21] using methods described previously [22].

Annualnumberofpregnanciespercountry.

The number of pregnancies was calculated as the sum of the number of live births, induced

abortions, and spontaneous pregnancy loss (including miscarriages and stillbirths) in 2007.

Livebirths.

The annual number of live births in 2007 was estimated per country using demographic data on the

proportion of women of childbearing age (WOCBAs) within a population and the total fertility rates.

The data were abstracted from the United Nations’ national population estimates, which provide

publicly accessible demographic information by year, age, sex, and country for Africa, Asia, and the

Americas [23]. The number of WOCBAs in each country, defined as the mid‐year resident number of

women aged between 15 and 49y, was obtained for the years 2005 and 2010 (interim years are not

available), and the number of WOCBAs for 2007 was calculated as the midpoint between the 2005

and 2010 estimates. The fraction of WOCBAs per country was then calculated as the number of

WOCBAs in 2007 divided by the mid‐year resident population at risk in 2007 (available by year).

The total fertility rate (TFR) is an age‐standardised measure of fertility and corresponds to

the total number of children that would be born alive to a woman entering her childbearing


41

years at age 15y if she lived to the end of her childbearing years (age 49y) and if her fertility

during these 35 reproductive years was the same as the average woman of childbearing age.

The total fertility rate divided by 35 is the average number of live births per WOCBA per year and

when multiplied by 1,000 this is expressed as the rate of live births per 1,000 WOCBAs per year.

Inducedabortions,miscarriages,andstillbirthrates.

Subregional data on induced abortion rates were obtained from a recently published review that

calculated the worldwide, regional, and subregional incidence of safe and unsafe abortions in

women of childbearing age in 2003 by use of reports from official national reporting systems,

nationally representative demographic health surveys, hospital data, other surveys, and published

studies [24].

Country‐specific information on stillbirth rates was abstracted from model‐based estimates

published by Stanton et al. [25] that derived data from vital registration, demographic and health

surveys (DHS), and data from study reports integrated into a regression model. Regional estimates

were used for three malaria endemic countries for which country‐specific estimates were not

available (French Guiana, Mayotte, and Timor‐Leste).

Country‐specific data on miscarriages (spontaneous abortions) are not available. To calculate the

proportion of pregnancies resulting in miscarriage, a method was applied that uses multipliers to

work backwards from the (known) number of live births and induced abortions to recover the

(unknown) underlying number of pregnancies that ‘‘produced’’ them, as described in detail

previously [24,26–28]. The method takes account of pregnancies that are terminated voluntarily

during the period of risk for miscarriage and estimates the number of spontaneous pregnancy loss

(stillbirths and miscarriages) as 10% of induced abortions plus 20% of live births. It is based on

clinical studies of rates of pregnancy loss by gestational age that indicate that for each 100 induced

abortions an additional ten clinically recognised pregnancies will have aborted spontaneously prior

to the average gestational age of induced abortions in that population, and that approximately 120

additional clinically recognised pregnancies are required to ‘‘produce’’ 100 live births [27,28]. For

example, in Afghanistan it was estimated that in 2007 1.182 million live births occurred among a

population of 27 million and a further 0.284 million induced abortions occurred. The number of

spontaneous pregnancy losses (the sum of the number of stillbirths and miscarriages) was therefore

estimated at 0.2 x 1,182 plus 0.1 x 0.284 = 0.265 million, and the total number of pregnancies as

1.731 million. The reported number of miscarriages used in this manuscript represents the number

of spontaneous pregnancy losses calculated through the multiplier method as described above,

minus the country‐specific number of stillbirths obtained from the review by Stanton et al. [25]. The

estimates provided in this study refer to clinically recognised pregnancies and do not take into

account the potentially large but unknown rates of embryonic loss that may occur in the first 4–6 wk

of gestation.

Estimatingtheannualnumberofpregnanciesexposedtomalaria.

To obtain the total population at risk, the limits of stable and unstable P. falciparum transmission

and the limits of P. vivax transmission described above were overlaid onto the Global Rural‐Urban

Mapping Project (GRUMP) alpha surface, projected to 2007. For every malaria endemic country of

the world, the population within each set of limits was extracted, following approaches described

Chapter 3

42

Total n of

aSource: United Nations Development Program (in millions). bThe total number of pregnancies is the sum of the number of live‐births, stillbirths, spontaneous, and induced abortions (in millions). cThe total pregnancy rate (TPR) and the annual pregnancy rate per 1,000 WOCBAs are weighted means per region and is for illustration purposes only. The number of pregnancies was derived directly as the sum of the national estimates within each region and globally.

previously [13]. The number of pregnancies at risk of malaria was then calculated from the total

annual number of pregnancies estimated to have occurred in 2007 in the entire country

multiplied by the fraction of the total resident population living within the spatial limits of

malaria transmission in that country.

ResultsTables 1 and 2 provide a summary of the total population living within the global spatial limits of

malaria transmission in 2007, and the corresponding number of total population, pregnancies,

and live births, stratified by species and transmission patterns (within areas of assumed unstable

and stable P. falciparum transmission), globally and by WHO region.

The compiled data showed that, globally, 125.2 million women living in areas with P. falciparum

and/or P. vivax transmission became pregnant in 2007: 77.4 million (61.8%) in the countries that fall

under the regional office of the WHO for the South East Asian (SEARO) and the Western Pacific

Region (WPRO); 30.3 million (24.2%) in AFRO; 13.1 million (10.5%) in the Eastern Mediterranean and

European Region (EMRO and EURO); and only 4.3 million (3.4%) in the American Region (AMRO)

(Table 3). Figures 1 and 2 display the same analysis by species, but depicted by continent rather than

by WHO region. Of the 125.2 million pregnancies, 82.6 million (66.0%) are estimated to result in live

births; 48.8 million (63.0%), 22.1 million (72.7%), 9.0 million (68.8%), and 2.7 million (63.1%) in the

SEARO/WPRO, AFRO, EMRO/ EURO, and AMRO regions, respectively (Table 4). It illustrates that the

proportional distribution of pregnancies at risk resulting in live births is slightly different from the

distribution of total pregnancies at risk, primarily reflecting the differences in the proportion of

pregnancies ending in induced abortions, which is much lower in the AFRO region (11.9%) compared

to the global average in the malaria endemic countries of 19.5% [24].

Table 1. Demographic data for malaria endemic countries

WHORO

Region

n of Total Population

WOCBAsa Total n of Pregnanciesb

TPRc Pregnancy Rate Per 1,000 WOCBAs

Live‐

births

Still‐

births

Spontaneous

Abortions

Induced

Abortions

AFRO 43 755 178 36 7.16 204 72.4% 2.3% 13.3% 11.9%

EMRO/EURO 19 544 142 19 4.76 136 68.8% 2.3% 13.1% 15.8%

AMRO 21 530 143 16 3.81 109 63.2% 0.9% 14.0% 22.0%

SEARO/WPRO 19 3,327 881 91 3.62 103 62.4% 1.6% 13.1% 22.8%

Global 102 5,157 1,343 162 4.23 121 65.5% 1.8% 13.3% 19.5%


43

Similar tables with risk estimates by continent and by pregnancy outcome (live‐birth, induced abortions, stillbirths, and miscarriages) are provided in Tables S1, S2, S3. aIncludes countries where P. falciparum and P. vivax co‐exist. bStable transmission, $1 autochthonous P. falciparum cases per 10,000 people per annum; unstable transmission, ,1 autochthonous P. falciparum cases per 10,000 people per annum [15]. MEC, malaria endemic countries; TPR, total pregnancy rate; WHORO, World Health Organization Regional Office.

Table 2. Total population at risk of P. falciparum and/or P. vivax malaria by WHO regional office in 2007 (in millions) (percent of the population in malaria endemic countries at risk).

P.falciparumMalariaOf the 125.2 million pregnancies defined above, 85.3 million occur in areas with P. falciparum

transmission, 51.8% of them (44.2 million) are in the combined SEARO‐WPRO regions and 35.1%

(30.0 million) in the AFRO region. The remainder live in the EMRO‐EURO (9.6%) and AMRO regions

(3.5%) (Figure 3; Table 3). As expected, the top five ranked countries with the highest number of

pregnancies at risk of P. falciparum malaria were the malaria endemic countries with the largest

overall populations: India (28.2 million), Nigeria (6.5 million), Indonesia (4.4 million), Pakistan (3.7

million), and the Democratic Republic of the Congo (3.3 million). Overall, 64.1% of 85.3 million

pregnancies at risk of P. falciparum malaria live in areas with assumed stable transmission (Figure 3).

However, this varies widely by region; from 98.7% in the AFRO region to none in the EURO region. As

depicted in Figure 3, 55.3% of the 44.2 million pregnancies at risk of P. falciparum in the

WPRO/SEARO region occur in areas of very low and unstable transmission.

P.vivaxMalariaGlobally, an estimated 92.9 million pregnancies occurred in areas endemic for P. vivax in 2007

(including in areas where both P. falciparum and P. vivax co‐exist) (Figure 2). The top five ranked

countries include: India (32.9 million), China (21.2 million), Indonesia (6.3 million), Pakistan (5.8

million), and Bangladesh (4.7 million). In the WPRO/SEARO region, where the majority of the

populations at risk of P. vivax live (Figure 4), approximately 98.2% of those pregnancies in malaria

endemic countries occur in areas with P. vivax transmission (alone or combined with P. falciparum).

By contrast this was only 11.9% for the AFRO region where P. vivax transmission is principally

restricted to the horn of Africa region, Madagascar, and the Comoros islands (Figure 4).

The country‐specific demographic data and population at risk estimates (Table S1), as well as

total pregnancies at risk and by specific pregnancy outcomes (live births, induced abortions,

stillbirths, and miscarriages; Table S2) and summary estimates by other regional categories

(continents instead of WHO regions; Table S3), are provided as supplemental information. In

addition, information is provided illustrating which countries are included in the different WHO

WHORO Region

P. falciparum Transmissiona P. vivax

Transmissiona

Any Species

Stable

Transmissionb

Unstable

Transmissionb

Overall Overall Overall

AFRO 599.9 (79.4) 8.4 (1.1) 607.8 (80.5) 73.2 (9.7) 615.4 (81.5)

EMRO/EURO 89.8 (16.5) 101.7 (18.7) 190.9 (35.1) 285.1 (52.4) 343.9 (63.2)

AMRO 41.2 (7.8) 50.2 (9.5) 91.4 (17.2) 96.2 (18.2) 138.2 (26.1)

SEARO/WPRO 654.9 (19.7) 824.9 (24.8) 1479.3 (44.5) 2,722.3 (81.8) 2,770.1 (83.3)

Global 1,385.8 (26.9) 985.1 (19.1) 2,369.4 (45.9) 3,176.9 (61.6) 3,867.6 (75.0)

Chapter 3

44


regions (also see Figure S1) [29]. In brief, all malaria endemic countries on the African continent

fall under the Africa Regional Office (AFRO), with the exception of Djibouti, Somalia, and Sudan,

which fall under the EMRO office.

Table 3. Number of pregnancies at risk of P. falciparum and/or P. vivax malaria by WHO regional office in 2007 (in millions) (column %).

WHORO Region

P. falciparum Transmissiona P. vivax

Transmissiona

Any Species

Stable

Transmissionb

Unstable

Transmissionb

Overall Overall Overall

AFRO 29.6 (54.1) 0.4 (1.2) 30.0 (35.1) 3.6 (3.9) 30.3 (24.2)

EMRO/EURO 4.0 (7.3) 4.2 (13.7) 8.2 (9.6) 10.4 (11.2) 13.1 (10.5)

AMRO 1.4 (2.5) 1.6 (5.2) 3.0 (3.5) 2.9 (3.1) 4.3 (3.4)

SEARO/WPRO 19.7 (36.1) 24.5 (79.9) 44.2 (51.8) 76.0 (81.8) 77.4 (61.8)

Global 54.7 30.6 85.3 92.9 125.2

DiscussionThis is the first time, to our knowledge, that contemporary species‐specific estimates of the annual

number of pregnancies at risk of malaria globally have been made. Our findings suggest that in

2007 approximately 125 million pregnancies occurred in areas with P. falciparum and/or P. vivax

transmission, resulting in 83 million live births; representing approximately 60% of all pregnancies

globally. Approximately 85 million pregnancies occurred in areas with P. falciparum transmission

and 93 million in areas with transmission of P. vivax transmission, of which about 53 million

occurred in areas where both species co‐exist. The pregnancies at risk estimates for P. falciparum

and P. vivax in Africa (32 million [30 million in the WHO‐AFRO region]) are consistent with the

previous estimates by WHO (25–30 million). By contrast, the numbers at risk outside Africa are

much higher (95 million) than previously estimated (25 million). Comparisons between the

estimates produced in this study and the previous WHO estimates are made difficult because

details of the methodology used by the WHO is not provided and it is not clear if they included all

transmission areas or only areas with stable malaria transmission. Inclusion of only those areas

with stable P. falciparum transmission in our study resulted in global risk estimates of just less

than 55 million pregnancies, 31 million in Africa and 23 million in the other regions, i.e., very

similar to the previous WHO estimates.

However, the numbers of pregnancies at risk outside Africa increase almost 4‐fold if areas with

unstable P. falciparum transmission are included (clinical incidence, 1 per 10,000

population/year) (30 million) and areas situated in the temperate regions outside the limits of P.

falciparum transmission that have P. vivax transmission only (40 million) are also included. It is also

not clear if the previous WHO estimates included pregnancies resulting in live births only or


45

included adjustments for induced abortions or spontaneous pregnancy loss. Since only

approximately two‐thirds of all pregnancies result in live births, estimates that include all

pregnancies are about one‐third higher than estimates based on live births only.

Figure 1. Malaria risk map for P. falciparum and corresponding number of pregnancies in each continent in 2007. doi:10.1371/journal.pmed.1000221.g001

Figure 2. Malaria risk map for P. vivax and corresponding number of pregnancies in each continent in 2007. doi:10.1371/journal.pmed.1000221.g002

Although risk estimates are widely quoted figures, it is important to place them in perspective.

The estimates provided here merely define the global distribution of pregnancies that occur

within the global spatial limits of malaria transmission. These estimates therefore represent ‘‘any

risk’’ of exposure to malaria during pregnancy, and do not represent the distribution of actual

Chapter 3

46


incidence or health burden on mothers and unborn babies, which is beyond the scope of this

paper. More than half (71 million) of the 125 million pregnancies occur in areas with unstable P.

falciparum transmission (31 million) or with transmission of P. vivax only (40 million), and the risk of

acquiring malaria in these areas is extremely low.

Table 4. Number of live‐births born to pregnancies at risk of at risk of P. falciparum and/or P. vivax malaria by WHO regional office in 2007 (in millions) (column %).

WHORO Region

P. falciparum Transmission a P. vivax

Transmissiona

Any Species

Stable b

Unstable Overall Overall Overall

AFRO 599.9 (79.4) 8.4 (1.1) 607.8 (80.5) 73.2 (9.7) 615.4 (81.5)

EMRO/EURO 89.8 (16.5) 101.7 (18.7) 190.9 (35.1) 285.1 (52.4) 343.9 (63.2)

AMRO 41.2 (7.8) 50.2 (9.5) 91.4 (17.2) 96.2 (18.2) 138.2 (26.1)

SEARO/WPRO 654.9 (19.7) 824.9 (24.8) 1479.3 (44.5) 2,722.3 (81.8) 2,770.1 (83.3)

Global 1,385.8 (26.9) 985.1 (19.1) 2,369.4 (45.9) 3,176.9 (61.6) 3,867.6 (75.0)

Thus, although these 71 million pregnancies represent more than 50% of the global number of

pregnancies at risk, they may only contribute a small proportion to the number of infections in

pregnancy. For example, if the actual incidence of malaria infection in these very low

transmission areas is 1 in 10,000 per person‐year (52 wk), and if the average pregnancy resulting

in a live birth takes 38 wk from fertilisation to term, then 71 million pregnancies at risk may result

in only 5,188 actual malaria infections, whereas in areas with infection rates of 1.36 or higher per

person‐year, all term pregnancies have been potentially exposed to malaria. Furthermore, the

definition of stable transmission for P. falciparum used included all areas with more than one

clinical case per 10,000 population per year. This included almost all pregnancies at risk in the

AFRO Region (99% of the 30 million pregnancies at risk) and 25 million of the 95 million (26%)

pregnancies in the other WHO regions. However, these stable transmission strata cover a very

wide range of transmission intensities and the actual risk of infection to the 55 million

individuals and the impact on maternal and infant health varies enormously within this range.

At the higher end of the transmission spectrum, the majority of malaria infections in pregnancy

remain asymptomatic or pauci‐symptomatic, yet are a major cause of severe maternal anaemia

and preventable low birth weight, especially in the first and second pregnancies. In areas with

stable, but low transmission, and certainly in areas with unstable and exceptionally low

transmission, infections can become severe in all gravidae groups because most women of

childbearing age in these regions have low levels of pre‐pregnancy and pregnancy‐specific

protective immunity to malaria [30]. The most recent version of the World Malaria Map [28]

from the Malaria Atlas Project shows that 89% of the population in stable P. falciparum areas

outside Africa live in areas characterised by low malaria endemicity (defined as P. falciparum

parasite rate in children 2–10 y of age of <5%). This total includes all of the stable P. falciparum


47

transmission areas in the Americas, and 88% of the populations at risk in the Central and South‐

East Asia‐Pacific region [31]. Our estimates do not take seasonality into account and include all

pregnancies occurring throughout the year, whereas those pregnancies that occur outside of the

transmission season may be at no risk, or very low risk of exposure. Our risk estimates for P. vivax

are likely to be less accurate than those for P. falciparum because of greater uncertainties about

the basic biology of transmission and clinical epidemiology. For example, the climatic constraints

on P. vivax transmission are less well defined, the accuracy of clinical reporting of P. vivax in areas

with coincidental P. falciparum is poor, and the untreated hypnozoite stage of P. vivax, which can

remain dormant in infected liver cells for months or years, provides an additional challenge to the

interpretation of prevalence and incidence data [15]. We used a refined P. vivax risk map that

resulted in a 19% increase over previous population at risk estimates (adjusted for population

growth) [18,19], principally resulting from the removal of the population density masks and

thereby the inclusion of many large cities. In most of these cities, pregnancies will be at low or

very low risk of autochthonous infections. Imported malaria associated with travel to rural areas

may be a greater risk factor in these cities. We did not consider infections with P. ovale or P.

malariae, as their distribution is not well described and the adverse effects on maternal health and

the newborn infant are unknown.

Figure 3. Distribution of the number of pregnancies in areas with P. falciparum malaria in 2007 by WHO regions and the corresponding proportion living under stable versus unstable transmission. Blue, SEARO and WPRO; green, AFRO; orange, EMRO; red, AMRO. doi:10.1371/journal.pmed.1000221.g003

Chapter 3

48

Figure 4. Distribution of the number of pregnancies in malaria endemic areas in 2007 by WHO regions and by species (P. vivax transmission only, P. falciparum transmission only or transmission of both species). Blue, SEARO and WPRO; green, AFRO; orange, EMRO; red, AMRO. Pv, P. vivax; Pf, P. falciparum. doi:10.1371/journal.pmed.1000221.g004

In the current analysis we used the map of the global spatial limits of P. falciparum malaria, which

stratifies the malaria endemic world by stable and unstable transmission published in 2008 [15].This

map uses a simple divide between very low risk and higher transmission intensities and a crude

proxy to account for the corresponding levels of acquired immunity in women of childbearing age.

As a next step, we will examine the burden of malaria in pregnancy in terms of health impact on the

pregnant women (e.g., febrile episodes, impact on maternal anaemia and maternal mortality), the

newborn baby (e.g., impact on the frequency of preterm births and low birth‐weight) and the infant

(e.g., susceptibility to malaria). For this project, we will use the more refined P. falciparum

transmission intensity model of risk within the defined stable limits which was developed recently by

the Malaria Atlas Project [31], allowing disease impact calculations across multiple transmission

strata to be made. It is also important to take the different pregnancy outcomes into account in

these further burden estimates. Of the 125 million pregnancies, one in five are estimated to be

terminated voluntarily during the period of risk for miscarriage, and only about two‐thirds (82.6

millions) are expected to result in live births. Although malaria in pregnancy is associated with

miscarriages and stillbirths [30], the majority of the health and economic burden is likely through the

impact on pregnancies that result in live births by increasing the risk of preterm births and low birth‐

weight [30] and by modifying the susceptibility to malaria in the infant [32–34].


49

Most of the existing research and policy guidance for malaria control in pregnancy has focused

on P. falciparum in the stable transmission regions of sub‐Saharan Africa. The results of this study

are consistent with the previous WHO‐RBM risk estimates for areas with stable P. falciparum

malaria in Africa, but our work offers advancement on the existing risk estimates for malaria

endemic countries outside Africa. In these regions, the burden of malaria in pregnancy is less

well defined, both in terms of the number of pregnancies and its actual impact on health. Policy

guidelines for malaria control in pregnancy are also less well developed for these regions.

These estimates of the number of pregnancies at risk of malaria provide a first step towards a

spatial map of the burden of malaria in pregnancy and a more informed platform with which to

estimate the associated disease and economic impact and its geographical distribution. Such

global estimates provide guidance in terms of priority setting for resource allocation for both

research and policy for the control of malaria in pregnancy. This project provides a dynamic

framework that allows risk estimates to be updated when new risk maps of P. falciparum and P.

vivax become available as the world attempts to move towards malaria elimination and

eradication.

Acknowledgments

We acknowledge the support of the Kenya Medical Research Institute. We would like to thank

Richard Steketee for helpful comments on the draft manuscript.

AuthorContributions

ICMJE criteria for authorship read and met: SD AJT CAG RWS FOtK. Agree with the manuscript’s

results and conclusions: SD AJT CAG RWS FOtK. Designed the experiments/the study: SD RWS FOtK.

Analyzed the data: SD AJT RWS FOtK. Collected data/did experiments for the study: AJT CAG FOtK.

Wrote the first draft of the paper: SD. Contributed to the writing of the paper: SD AJT CAG RWS

FOtK.

References

1. RBM Pregnant women and Infants, Available:http://malaria.who.int/ pregnantwomenandinfants.html Geneva: WHO, Accessed 5 January 2010.

2. World Health Organization (2004) A strategic framework for malaria prevention and control during pregnancy in the African region. Brazzaville: World Health Organization: Regional Office for Africa, AFR/MAL/04/01 AFR/MAL/04/01.

3. Snow RW, Craig M, Deichmann U, Marsh K (1999) Estimating mortality, morbidity and disability due to malaria among Africa’s non‐pregnant population. Bull World Health Organ 77: 624–640.

4. United Nations Children’s Fund (1998) The state of the world’s children. New York: Oxford University Press.

5. World Health Organization (2003) Available: http://www.who.int/features/ 2003/04b/en/ Geneva: WHO, Accessed 5 January 2010.

6. Baird JK (2007) Neglect of Plasmodium vivax malaria. Trends Parasitol 23: 533–539. 7. Hay SI, Guerra CA, Tatem AJ, Noor AM, Snow RW (2004) The global distribution and population at

risk of malaria: past, present, and future. Lancet Infect Dis 4: 327–336. 8. Singh N, Shukla MM, Sharma VP (1999) Epidemiology of malaria in pregnancy in central India. Bull

World Health Organ 77: 567–572. 9. Nosten F, McGready R, Simpson JA, Thwai KL, Balkan S, et al. (1999) Effects of Plasmodium vivax

malaria in pregnancy. Lancet 354: 546–549.

Chapter 3

50

10. Poespoprodjo JR, Fobia W, Kenangalem E, Lampah DA, Warikar N, et al. (2008) Adverse pregnancy outcomes in an area where multidrug‐resistant plasmodium vivax and Plasmodium falciparum infections are endemic. Clin Infect Dis 46: 1374–1381.

11. Balk DL, Deichmann U, Yetman G, Pozzi F, Hay SI, et al. (2006) Determining global population distribution: methods, applications and data. Adv Parasitol 62: 119–156.

12. Hay SI, Snow RW (2006) The Malaria Atlas Project: developing global maps of malaria risk. PLoS Med 3: e473. doi:10.1371/journal.pmed.0030473.

13. Snow RW, Guerra CA, Noor AM, Myint HY, Hay SI (2005) The global distribution of clinical episodes of Plasmodium falciparum malaria. Nature 434: 214–217.

14. Hay SI, Smith DL, Snow RW (2008) Measuring malaria endemicity from intense to interrupted transmission. Lancet Infect Dis 8: 369–378.

15. Guerra CA, Gikandi PW, Tatem AJ, Noor AM, Smith DL, et al. (2008) The limits and intensity of Plasmodium falciparum transmission: implications for malaria control and elimination worldwide. PLoS Med 5: e38. doi:10.1371/ journal.pmed.0050038.

16. Guerra CA, Snow RW, Hay SI (2006) Mapping the global extent of malaria in 2005. Trends Parasitol 22: 353–358.

17. Guerra CA, Snow RW, Hay SI (2006) Defining the global spatial limits of malaria transmission in 2005. Adv Parasitol 62: 157–179.

18. Guerra CA (2007) Mapping the contemporary global distribution limits of malaria using empirical data and expert opinion. Oxford: University of Oxford. 258 p.

19. Tatem AJ, Guerra CA, Kabaria CW, Noor AM, Hay SI (2008) Human population, urban settlement patterns and their impact on Plasmodium falciparum malaria endemicity. Malar J 7: 218.

20. Center for International Earth Science Information Network (CIESIN)/ Columbia University; International Food Policy Research Institute (IFPRI)/ the World Bank/and Centro Internacional de Agricultura Tropical (CIAT) (2007) Global Rural Urban Mapping Project (GRUMP) alpha: Gridded Population of the World; version 2 with urban reallocation (GPW‐UR). Palisades (New York): CIESIN; Columbia University, Available: http://sedac. ciesin.columbia.edu/gpw. Accessed 5 January 2010.

21. United Nations Population Division; Department of Economic and Social Affairs (2007) World population prospects; the 2006 revision; ST/ESA/SER.A/ 261/ES. New York: United Nations.

22. Hay SI, Noor AM, Nelson A, Tatem AJ (2005) The accuracy of human population maps for public health application. Trop Med Int Health 10: 1073–1086.

23. UNPD (2006) World population prospects: population database. Available: http://esa.un.org/unpp/. Accessed 5 January 2010.

24. Sedgh G, Henshaw S, Singh S, Ahman E, Shah IH (2007) Induced abortion: estimated rates and trends worldwide. Lancet 370: 1338–1345.

25. Stanton C, Lawn JE, Rahman H, Wilczynska‐Ketende K, Hill K (2006) Stillbirth rates: delivering estimates in 190 countries. Lancet 367: 1487–1494.

26. Leridon H (1977) Human fertility: the basic components (Table 4.20). Chicago: University of Chicago Press.

27. Hammerslough CR (1992) Estimating the probability of spontaneous abortion in the presence of induced abortion and vice versa. Public Health Rep 107: 269–277.

28. Henshaw SK, Binkin NJ, Blaine E, Smith JC (1985) A portrait of American women who obtain abortions. Fam Plann Perspect 17: 90–96.

29. World Health Organization (2009) WHO regional offices, Available: http:// www.who.int/about/regions/en/ Geneva: World Health Organisation, Accessed 5 January 2010.

30. Desai M, ter Kuile FO, Nosten F, McGready R, Asamoa K, et al. (2007) Epidemiology and burden of malaria in pregnancy. Lancet Infect Dis 7: 93–104.

31. Hay SI, Guerra CA, Gething PW, Patil AP, Tatem AJ, et al. (2009) A world malaria map: Plasmodium falciparum endemicity in 2007. PLoS Med 6: e1000048. doi:10.1371/journal.pmed.1000048.

32. Mutabingwa TK, Bolla MC, Li JL, Domingo GJ, Li X, et al. (2005) Maternal malaria and gravidity interact to modify infant susceptibility to malaria. PLoS Med 2: e407. doi:10.1371/journal.pmed.0020407.

33. Le Hesran JY, Cot M, Personne P, Fievet N, Dubois B, et al. (1997) Maternal placental infection with Plasmodium falciparum and malaria morbidity during the first 2 years of life. Am J Epidemiol 146: 826–831.

34. Schwarz NG, Adegnika AA, Breitling LP, Gabor J, Agnandji ST, et al. (2008) Placental malaria increases malaria risk in the first 30 months of life. Clin Infect Dis 47: 1017–1025.

51

Chapter3Appendix&SupportingInformation

Figure S1 Map of the WHO Regions (http://www.who.int/ about/regions/en/index.html).

Found at: doi:10.1371/journal.pmed.1000221.s001 (1.07 MB TIF)

Table S1 Demographic characteristics and total population at risk of P. falciparum and/or

P. vivax malaria by malaria endemic country and by WHO regional office in 2007 (in

millions).

Found at: doi:10.1371/journal.pmed.1000221.s002 (0.38 MB PDF)

Table S2 Total number of pregnancies by pregnancy outcome in areas with P. falciparum

and/or P. vivax transmission by continent in 2007 (in millions).


Table S3 Total population, number of pregnancies, and number of live‐births born to

pregnancies in malaria endemic countries by continent in 2007 (in millions).


Chapter 3 Appendix

52

Figure S1. Map of the WHO Regions (http://www.who.int/about/regions/en/index.html).

doi:10.1371/journal.pmed.1000221.s001


53

Table S1: Demograp

hic characteristics an

d total population at risk of P.falciparum and/or P.vivax malaria by malaria endemic country an

d by WHO regional office in

2007 (in m

illions)

Table S1

2007

Demograp

hic data

Total population at risk*

UN National Population Estim

ates

Total

Population

both

sexes*

WOCBAs*

Number of

pregnancies*†

TFR

TPR

Pregnancy

rate per

1000

WOCBAs

Percentage pregnan

cies ending in:

P. falciparum#

P. vivax#

Any

species

Countries

Live‐

births

Still‐

births

Induced

Abortions

Spontaneo

us

Abortions

Stable¶

Unstable¶

Overall

Overall

Overall

AFR

O 1

755

178

35

5.18

7.16

199

72.4%

2.3%

11.9%

13.3%

600

8

608

73

615

Angola

17.02

3.93

0.96

6.43

8.56

245

75.1%

2.5%

9.0%

13.4%

15.41

0.28

15.70

0.00

15.70

Ben

in

9.03

2.08

0.44

5.42

7.34

210

73.9%

2.3%

10.3%

13.5%

7.66

0.00

7.66

0.00

7.66

Botswana

1.88

0.50

0.06

2.90

4.25

121

68.3%

1.3%

16.4%

14.0%

0.89

0.00

0.89

0.00

0.89

Burkina Faso

14.78

3.39

0.79

6.00

8.12

232

73.9%

2.0%

10.3%

13.8%

14.23

0.00

14.23

0.00

14.23

Burundi

8.51

2.01

0.56

6.80

9.66

276

70.4%

2.5%

14.1%

13.0%

5.66

0.00

5.66

0.00

5.66

Cam

eroon

18.55

4.47

0.73

4.31

5.74

164

75.1%

2.1%

9.0%

13.8%

16.95

0.00

16.95

0.00

16.95

Cape Verde

0.53

0.14

0.02

3.37

4.56

130

73.9%

1.2%

10.3%

14.6%

0.00

0.25

0.25

0.00

0.25

Cen

tral African

Republic

4.34

1.03

0.18

4.58

6.10

174

75.1%

2.2%

9.0%

13.7%

4.15

0.00

4.15

0.00

4.15

Chad

10.78

2.41

0.57

6.20

8.26

236

75.1%

2.6%

9.0%

13.3%

9.76

0.16

9.92

0.00

9.92

Comoros

0.84

0.21

0.04

4.30

6.11

174

70.4%

1.9%

14.1%

13.6%

0.59

0.00

0.59

0.64

0.64

Congo

3.77

0.91

0.16

4.49

5.98

171

75.1%

2.2%

9.0%

13.8%

3.33

0.00

3.33

0.00

3.33

Dem

. Rep

. of the Congo

62.64

13.90

3.55

6.70

8.92

255

75.1%

2.6%

9.0%

13.3%

57.97

0.00

57.97

0.00

57.97

Côte d'Ivoire

19.26

4.58

0.79

4.46

6.04

173

73.9%

2.6%

10.3%

13.2%

17.80

0.00

17.80

0.00

17.80

Equatorial Guinea

0.51

0.12

0.02

5.36

7.14

204

75.1%

2.0%

9.0%

13.9%

0.51

0.00

0.51

0.00

0.51

Eritrea

4.85

1.20

0.25

5.05

7.17

205

70.4%

2.0%

14.1%

13.5%

3.33

0.95

4.28

4.80

4.82

Ethiopia

83.10

19.51

4.19

5.29

7.51

215

70.4%

2.6%

14.1%

12.9%

46.08

1.50

47.59

49.05

53.13

Gabon

1.33

0.35

0.04

3.06

4.08

116

75.1%

1.4%

9.0%

14.5%

1.33

0.00

1.33

0.00

1.33

Gam

bia

1.71

0.40

0.07

4.70

6.36

182

73.9%

2.1%

10.3%

13.7%

1.51

0.00

1.51

0.00

1.51

Ghana

23.48

5.84

0.87

3.84

5.20

149

73.9%

1.8%

10.3%

14.0%

22.21

0.00

22.21

0.00

22.21

Guinea

9.37

2.14

0.45

5.44

7.37

210

73.9%

2.2%

10.3%

13.6%

9.23

0.00

9.23

0.00

9.23

Guinea

‐Bissau

1.70

0.37

0.10

7.07

9.57

274

73.9%

2.7%

10.3%

13.1%

1.42

0.00

1.42

0.00

1.42

Ken

ya

37.54

9.14

1.84

4.96

7.04

201

70.4%

3.3%

14.1%

12.2%

25.62

0.18

25.80

0.00

25.80

Liberia

3.75

0.84

0.22

6.77

9.17

262

73.9%

2.4%

10.3%

13.4%

3.43

0.00

3.43

0.00

3.43

Chapter 3 Appen

dix

54

Table S1

2007

Demograp

hic data



ates

Total

Population

both

sexes*

WOCBAs*

Number of

pregnancies*†

TFR

TPR

Pregnancy

rate per

1000

WOCBAs

Percentage pregnan

cies ending in:

P. falciparum#

P. vivax#

Any

species

Countries

Live‐

births

Still‐

births

Induced

Abortions

Spontaneo

us

Abortions

Stable¶

Unstable¶

Overall

Overall

Overall

Madagascar

19.68

4.63

0.90

4.78

6.79

194

70.4%

2.1%

14.1%

13.4%

17.28

0.00

17.28

18.74

18.74

Malaw

i 13.93

3.11

0.70

5.59

7.94

227

70.4%

2.9%

14.1%

12.6%

13.45

0.00

13.45

0.00

13.45

Mali

12.34

2.80

0.71

6.52

8.83

252

73.9%

1.8%

10.3%

14.0%

12.34

0.47

12.34

0.00

12.34

Mauritania

3.12

0.76

0.13

4.37

5.92

169

73.9%

2.3%

10.3%

13.5%

0.93

0.40

1.33

0.00

1.33

Mayotte*

0.21

0.05

0.01

5.65

8.02

229

70.4%

2.3%

14.1%

13.1%

0.00

0.21

0.21

0.00

0.21

Mozambique

21.40

5.10

1.06

5.11

7.26

207

70.4%

2.3%

14.1%

13.2%

21.06

0.00

21.06

0.00

21.06

Nam

ibia

2.07

0.54

0.07

3.19

4.67

133

68.3%

1.3%

16.4%

14.0%

1.25

0.38

1.64

0.00

1.64

Niger

14.23

3.04

0.85

7.19

9.74

278

73.9%

2.9%

10.3%

12.9%

13.19

0.62

13.81

0.00

13.81

Nigeria

148.09

34.57

7.11

5.32

7.20

206

73.9%

2.3%

10.3%

13.5%

134.60

0.00

134.60

0.00

134.60

Rwanda

9.73

2.47

0.59

5.92

8.41

240

70.4%

2.2%

14.1%

13.3%

5.03

0.00

5.03

0.00

5.03

Sao Tome and Principe

0.16

0.04

0.01

3.85

5.13

147

75.1%

1.9%

9.0%

14.0%

0.13

0.00

0.13

0.00

0.13

Senegal

12.38

2.96

0.54

4.69

6.35

181

73.9%

2.0%

10.3%

13.8%

10.82

0.00

10.82

0.00

10.82

Sierra Leo

ne

5.87

1.36

0.34

6.47

8.76

250

73.9%

2.8%

10.3%

13.0%

5.50

0.00

5.50

0.00

5.50

South Africa

48.58

12.98

1.43

2.64

3.86

110

68.3%

1.2%

16.4%

14.1%

3.44

2.95

6.39

0.00

6.39

Swaziland

1.14

0.30

0.04

3.45

5.05

144

68.3%

1.5%

16.4%

13.8%

0.23

0.00

0.23

0.00

0.23

Tanzania

40.45

9.38

1.96

5.16

7.33

209

70.4%

2.1%

14.1%

13.4%

39.84

0.00

39.84

0.00

39.84

Togo

6.59

1.58

0.29

4.80

6.50

186

73.9%

2.0%

10.3%

13.8%

5.45

0.00

5.45

0.00

5.45

Uganda

30.88

6.65

1.74

6.46

9.17

262

70.4%

2.3%

14.1%

13.2%

27.03

0.00

27.03

0.00

27.03

Zambia

11.92

2.71

0.57

5.18

7.36

210

70.4%

2.2%

14.1%

13.3%

11.84

0.00

11.84

0.00

11.84

Zimbabwe

13.35

3.42

0.44

3.19

4.53

129

70.4%

1.6%

14.1%

13.9%

7.44

0.00

7.44

0.00

7.44


55

Table S1

2007

Demograp

hic data



ates

Total

Population

both

sexes*

WOCBAs*

Number of

pregnancies*†

TFR

TPR

Pregnancy

rate per

1000

WOCBAs

Percentage pregnan

cies ending in:

P. falciparum#

P.

vivax#

Any

species

Countries

Live‐

births

Still‐

births

Induced

Abortions

Spontaneo

us

Abortions

Stable¶

Unstable¶

Overall

Overall

Overall

EMRO

409

104

16

3.65

5.31

172

68.8%

2.5%

15.8%

12.8%

90

98

188

262

320

Afghanistan 2

27.15

5.85

1.73

7.07

10.35

296

68.3%

3.2%

16.4%

12.1%

4.56

12.53

17.10

16.60

20.43

Djibouti 1

0.83

0.22

0.03

3.95

5.61

160

70.4%

2.5%

14.1%

13.0%

0.02

0.41

0.43

0.71

0.71

Iran

2

71.21

20.69

1.77

2.04

2.99

85

68.3%

0.7%

16.4%

14.6%

0.15

2.72

2.87

47.46

47.67

Iraq

2

28.99

7.11

1.25

4.26

6.14

176

69.3%

1.9%

15.3%

13.5%

0.00

0.00

0.00

9.21

9.21

Oman

2

2.60

0.63

0.08

3.00

4.33

124

69.3%

0.7%

15.3%

14.7%

0.00

0.00

0.00

0.04

0.04

Pakistan 2

163.90

41.23

6.07

3.52

5.15

147

68.3%

2.9%

16.4%

12.4%

30.74

68.30

99.04

155.73

155.73

Saudi A

rabia 2

24.74

6.00

0.83

3.35

4.83

138

69.3%

0.8%

15.3%

14.6%

0.72

1.22

1.94

13.00

14.39

Somalia 1

8.70

2.04

0.50

6.04

8.58

245

70.4%

3.3%

14.1%

12.2%

8.70

0.55

8.70

8.70

8.70

Sudan

1

38.56

9.35

1.62

4.23

6.05

173

69.9%

3.9%

14.7%

11.5%

28.99

6.84

35.83

2.98

35.93

Syrian

Arab Rep

. 2

19.93

5.35

0.68

3.08

4.44

127

69.3%

0.8%

15.3%

14.6%

0.00

0.00

0.00

5.88

5.88

Yemen

2

22.39

5.18

1.17

5.50

7.93

227

69.3%

2.5%

15.3%

12.9%

15.93

5.72

21.65

1.88

21.74

EURO

135

37

3

2.23

3.24

91

68.9%

1.1%

15.7%

14.2%

0

3

3

23

23

Arm

enia 6

3.00

0.87

0.05

1.39

2.00

57

69.3%

0.8%

15.3%

14.6%

0.00

0.00

0.00

0.35

0.35

Azerbaijan 6

8.47

2.56

0.19

1.82

2.62

75

69.3%

0.4%

15.3%

15.0%

0.00

0.00

0.00

0.22

0.22

Geo

rgia 6

4.40

1.17

0.07

1.41

2.03

58

69.3%

1.0%

15.3%

14.4%

0.00

0.00

0.00

0.69

0.69

Kyrgyzstan 2

5.32

1.47

0.15

2.48

3.63

104

68.3%

1.3%

16.4%

14.0%

0.00

1.20

1.20

1.62

1.66

Tajikistan 2

6.74

1.78

0.25

3.35

4.90

140

68.3%

1.7%

16.4%

13.6%

0.00

2.16

2.16

4.41

4.87

Turkey 6

74.88

20.51

1.81

2.14

3.09

88

69.3%

1.0%

15.3%

14.4%

0.00

0.00

0.00

13.99

13.99

Turkmen

istan 2

4.97

1.42

0.15

2.50

3.66

105

68.3%

1.3%

16.4%

14.0%

0.00

0.00

0.00

1.35

1.35

Uzbekistan 2

27.37

0.01

0.79

2.49

3.65

104

68.3%

1.3%

16.4%

14.0%

0.00

0.00

0.00

0.29

0.29

Chapter 3 Appen

dix

56

Table S1

2007

Demograp

hic data



ates

Total

Population

both sexes*

WOCBAs*

Number of

pregnancies*†

TFR

TPR

Pregnancy

rate per

1000

WOCBAs

Percentage pregnan

cies ending in:

P. falciparum#

P.

vivax#

Any

species

Countries

Live‐

births

Still‐

births

Induced

Abortions

Spontaneo

us

Abortions

Stable¶

Unstable¶

Overall

Overall

Overall

AMRO

530

143

16

2.41

3.81

124

63.2%

0.9%

22.0%

14.0%

41

50

91

96

138

Argen

tina 5

39.53

9.95

1.03

2.25

3.64

104

61.8%

0.7%

23.5%

14.0%

0.00

0.00

0.00

3.02

3.02

Belize 4

0.29

0.07

0.01

2.93

4.35

124

67.3%

1.1%

17.5%

14.1%

0.00

0.17

0.17

0.25

0.26

Bolivia 5

9.53

2.37

0.38

3.50

5.66

162

61.8%

1.3%

23.5%

13.4%

0.22

2.61

2.83

4.05

4.06

Brazil 5

191.79

52.64

5.48

2.25

3.64

104

61.8%

0.8%

23.5%

13.9%

12.79

16.69

29.47

23.07

34.48

Colombia 5

46.16

12.71

1.30

2.22

3.59

103

61.8%

0.8%

23.5%

13.9%

5.26

7.74

13.00

17.32

21.44

Costa Rica 4

4.47

1.21

0.11

2.10

3.12

89

67.3%

0.7%

17.5%

14.5%

0.00

0.00

0.00

1.03

1.03

Dominican

Rep

ublic 4

9.76

2.52

0.34

2.81

4.67

133

60.2%

0.9%

25.3%

13.7%

1.41

2.83

4.24

0.00

4.24

Ecuador 5

13.34

3.44

0.41

2.58

4.17

119

61.8%

1.0%

23.5%

13.7%

4.12

1.65

5.77

3.73

5.85

El Salvador 4

6.86

1.82

0.21

2.68

3.98

114

67.3%

1.1%

17.5%

14.1%

0.00

0.00

0.00

1.10

1.10

Guatem

ala 4

13.35

3.24

0.01

3.27

5.29

151

61.8%

0.8%

23.5%

13.9%

0.14

0.00

0.14

0.20

0.20

Guyana 5

0.74

0.18

0.57

4.15

6.17

176

67.3%

2.3%

17.5%

12.9%

1.02

5.27

6.30

4.07

7.30

Haiti 4

9.60

2.47

0.02

2.33

3.77

108

61.8%

1.0%

23.5%

13.7%

0.14

0.52

0.66

0.74

0.74

Honduras 4

7.11

1.81

0.42

3.54

5.88

168

60.2%

1.5%

25.3%

13.0%

8.60

0.00

8.60

0.00

8.60

Mexico 4

106.54

29.58

0.25

3.31

4.92

141

67.3%

1.3%

17.5%

13.9%

0.88

2.63

3.51

5.46

6.47

Nicaragua 4

5.60

1.48

2.78

2.21

3.28

94

67.3%

0.5%

17.5%

14.7%

0.00

0.00

0.00

15.82

15.82

Panam

a 4

3.34

0.88

0.17

2.76

4.10

117

67.3%

1.3%

17.5%

13.9%

1.57

2.12

3.70

2.21

4.71

Paraguay 5

6.13

1.55

0.10

2.56

3.80

109

67.3%

0.9%

17.5%

14.3%

0.90

0.00

0.90

0.19

1.06

Peru 5

27.90

7.45

0.22

3.08

4.98

142

61.8%

1.0%

23.5%

13.7%

0.00

0.00

0.00

1.52

1.52

Surinam

e 5

0.46

0.12

0.86

2.51

4.06

116

61.8%

0.9%

23.5%

13.8%

3.87

1.70

5.57

8.14

8.33

Ven

ezuela 5

27.66

7.33

0.01

2.42

3.92

112

61.8%

0.9%

23.5%

13.8%

0.01

0.05

0.06

0.04

0.06

Fren

ch Guiana 5

0.20

0.05

0.86

2.55

4.13

118

61.8%

0.8%

23.5%

13.9%

0.22

6.24

6.46

4.29

7.89


57

Table S1

2007

Demograp

hic data



ates

Total

Population

both sexes* WOCBAs*

Number of

pregnancies*

† TFR

TPR

Pregnancy

rate per

1000

WOCBAs

Percentage pregnan

cies ending in:

P. falciparum#

P. vivax#

Any species

Countries

Live‐

births

Still‐

births

Induced

Abortions

Spontaneo

us

Abortions

Stable¶

Unstable¶

Overall

Overall

Overall

SEARO

1,770

458

52

2.62

3.94

122

66.3%

1.9%

18.5%

13.2%

564

712

1,276

1,632

1,639

Bangladesh 2

158.67

41.02

4.86

2.83

4.14

118

68.3%

2.7%

16.4%

12.6%

15.12

47.99

63.11

154.08

154.08

Bhutan 2

0.66

0.17

0.02

2.19

3.21

92

68.3%

1.7%

16.4%

13.6%

0.66

0.47

0.66

0.66

0.66

Burm

a 2

48.80

13.97

1.40

2.07

3.51

100

59.0%

1.3%

26.5%

13.2%

42.88

1.91

44.79

47.47

48.80

India 2

1,169.02

294.81

34.65

2.81

4.11

118

68.3%

2.1%

16.4%

13.2%

414.53

535.59

950.12

1109.03

1109.03

Indonesia 2

231.63

64.13

6.77

2.18

3.70

106

59.0%

0.7%

26.5%

13.7%

68.59

81.93

150.52

215.66

217.60

Rep

ublic of Korea 2

48.22

13.10

0.80

1.21

2.13

61

56.8%

1.0%

29.0%

13.2%

0.00

0.00

0.00

12.39

12.39

Nep

al 2

28.20

7.18

0.98

3.28

4.80

137

68.3%

3.9%

16.4%

11.4%

3.40

6.15

9.54

18.01

18.92

Sri Lanka 2

19.30

5.29

0.42

1.88

2.75

79

68.3%

0.7%

16.4%

14.6%

1.75

7.53

9.28

10.46

12.84

Thailand 2

63.88

17.85

1.60

1.85

3.14

90

59.0%

0.6%

26.5%

13.9%

16.53

30.53

47.06

63.88

63.88

Timor‐Leste 2

1.16

0.26

0.08

6.53

11.07

316

59.0%

0.8%

26.5%

13.7%

0.96

0.00

0.96

0.63

0.96

WPRO

1,558

423

40

1.88

3.27

139

57.3%

1.3%

28.4%

13.0%

91

113

203

1,090

1,131

Cam

bodia 2

14.44

3.87

0.60

3.18

5.39

154

59.0%

1.2%

26.5%

13.2%

10.77

2.55

13.33

12.65

14.44

China 2

1,328.63

361.61

31.48

1.73

3.05

87

56.8%

1.4%

29.0%

12.9%

17.13

20.32

37.45

896.81

915.87

Lao 2

5.86

1.52

0.24

3.21

5.44

155

59.0%

1.2%

26.5%

13.2%

5.30

0.01

5.31

2.32

5.32

Malaysia 2

26.57

7.06

0.89

2.60

4.41

126

59.0%

0.5%

26.5%

14.0%

6.29

16.17

22.46

10.15

22.66

Papua New

Guinea 3

6.33

1.60

0.25

3.78

5.45

156

69.3%

1.2%

15.3%

14.2%

4.11

0.00

4.11

3.90

4.46

Philippines 2

87.96

22.52

3.52

3.23

5.47

156

59.0%

0.7%

26.5%

13.7%

26.95

20.41

47.35

78.55

82.01

Solomon Islands 3

0.50

0.12

0.02

3.87

5.58

159

69.3%

0.9%

15.3%

14.4%

0.43

0.00

0.43

0.38

0.49

Vanuatu 3

0.23

0.06

0.01

3.74

5.39

154

69.3%

1.0%

15.3%

14.4%

0.22

0.00

0.22

0.20

0.22

Viet Nam

2

87.38

24.82

2.57

2.14

3.63

104

59.0%

0.7%

26.5%

13.8%

19.31

53.31

72.62

85.08

85.49

TOTA

L 5,157

1,343

162

2.77

4.23

159

65.5%

1.8%

19.5%

13.3%

1,386

985

2,369

3,177

3,868

* In m

illions; † The total number of pregnancies is the sum of the number of live‐births, stillbirths, spontaneous and induced abortions; # Includes countries where P.falciaprum

and P.vivaxco‐exist; ¶

Stable

transm

ission: >= 0.1 autochthonous P. falciparum cases per 1,000 peo

ple per annum; u

nstable transm

ission <0.1 autochthonous P. falciparum cases per 1,000 people per annum [15]; Abbreviation: TFR:

Total Fertility Rate; TPR: Total Pregnancy Rate; UN: U

nited

Nations; W

OCBA: W

omen

of Childbearing Age (15‐49 years of age); PWAR: Pregnant Women at Risk; Note: The regional and total estim

ates for

TFR, Stillbirth rate, TPR and Annual PR are weighted m

ean

and is for illustration purposes only. The number of pregnancies at risk was derived by adding national estim

ates within regions and should not be

calculated using regional or global estim

ates for pregnancy rates; C

ontinent inform

ation: A

frica is den

oted by 1; A

sia by 2; O

ceania by 3; N

orth America 4 , South America 5 and Europe 6.

Chapter 3 Appen

dix

58

Table S2: Number of pregn

ancies total and by pregn

ancy outcomes at risk of P.falciparum and/or P.vivax malaria by malaria endemic countries by WHO regional office in

2007 (in m

illions)

Table S2

2007

Number of pregn

ancies at risk of malaria*§

Number of live‐births born to

pregn

ancies at risk of malaria* §

Number of stillbirths born to

pregn


Number of miscarriages to

pregn


Number of induced abortions to

pregn


P. falciparum#

P.

vivax#

Any

species

P. falciparum#

P.

vivax#

Any

species

P. falciparum#

P.

vivax#

Any

species

P. falciparum#

P.

vivax#

Any

species

P. falciparum#

P.

vivax#

Any

species

Countries

Stable¶

Unstable¶

Overall

Overall

Overall

Stable¶

Unstable¶

Overall

Overall

Overall

Stable¶

Unstable¶

Overall

Overall

Overall

Stable¶

Unstable¶

Overall

Overall

Overall

Stable¶

Unstable¶

Overall

Overall

Overall

AFR

O 1

29.62

0.36

29.95

3.60

30.33

21.55

0.25

21.79

2.53

22.05

0.70

0.01

0.71

0.09

0.72

3.95

0.05

3.99

0.47

4.04

3.42

0.05

3.46

0.51

3.52

Angola

0.87

0.02

0.89

0.00

0.89

0.65

0.01

0.67

0.00

0.67

0.02

0.00

0.02

0.00

0.02

0.12

0.00

0.12

0.00

0.12

0.08

0.00

0.08

0.00

0.08

Ben

in

0.37

0.00

0.37

0.00

0.37

0.27

0.00

0.27

0.00

0.27

0.01

0.00

0.01

0.00

0.01

0.05

0.00

0.05

0.00

0.05

0.04

0.00

0.04

0.00

0.04

Botswana

0.03

0.00

0.03

0.00

0.03

0.02

0.00

0.02

0.00

0.02

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

Burkina Faso

0.76

0.00

0.76

0.00

0.76

0.56

0.00

0.56

0.00

0.56

0.01

0.00

0.01

0.00

0.01

0.10

0.00

0.10

0.00

0.10

0.08

0.00

0.08

0.00

0.08

Burundi

0.37

0.00

0.37

0.00

0.37

0.26

0.00

0.26

0.00

0.26

0.01

0.00

0.01

0.00

0.01

0.05

0.00

0.05

0.00

0.05

0.05

0.00

0.05

0.00

0.05

Cam

eroon

0.67

0.00

0.67

0.00

0.67

0.50

0.00

0.50

0.00

0.50

0.01

0.00

0.01

0.00

0.01

0.09

0.00

0.09

0.00

0.09

0.06

0.00

0.06

0.00

0.06

Cape Verde

0.00

0.01

0.01

0.00

0.01

0.00

0.01

0.01

0.00

0.01

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

Cen

tral African

Rep.

0.17

0.00

0.17

0.00

0.17

0.13

0.00

0.13

0.00

0.13

0.00

0.00

0.00

0.00

0.00

0.02

0.00

0.02

0.00

0.02

0.02

0.00

0.02

0.00

0.02

Chad

0.51

0.01

0.52

0.00

0.52

0.39

0.01

0.39

0.00

0.39

0.01

0.00

0.01

0.00

0.01

0.07

0.00

0.07

0.00

0.07

0.05

0.00

0.05

0.00

0.05

Comoros

0.03

0.00

0.03

0.03

0.03

0.02

0.00

0.02

0.02

0.02

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

Congo

0.14

0.00

0.14

0.00

0.14

0.10

0.00

0.10

0.00

0.10

0.00

0.00

0.00

0.00

0.00

0.02

0.00

0.02

0.00

0.02

0.01

0.00

0.01

0.00

0.01

Dem

. Rep

. of Congo

3.28

0.00

3.28

0.00

3.28

2.46

0.00

2.46

0.00

2.46

0.09

0.00

0.09

0.00

0.09

0.44

0.00

0.44

0.00

0.44

0.30

0.00

0.30

0.00

0.30

Côte d'Ivoire

0.73

0.00

0.73

0.00

0.73

0.54

0.00

0.54

0.00

0.54

0.02

0.00

0.02

0.00

0.02

0.10

0.00

0.10

0.00

0.10

0.08

0.00

0.08

0.00

0.08

Equatorial Guinea

0.02

0.00

0.02

0.00

0.02

0.02

0.00

0.02

0.00

0.02

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

Eritrea

0.17

0.05

0.22

0.24

0.24

0.12

0.03

0.15

0.17

0.17

0.00

0.00

0.00

0.00

0.00

0.02

0.01

0.03

0.03

0.03

0.02

0.01

0.03

0.03

0.03

Ethiopia

2.32

0.08

2.40

2.47

2.68

1.63

0.05

1.69

1.74

1.88

0.06

0.00

0.06

0.07

0.07

0.30

0.01

0.31

0.32

0.34

0.33

0.01

0.34

0.35

0.38

Gabon

0.04

0.00

0.04

0.00

0.04

0.03

0.00

0.03

0.00

0.03

0.00

0.00

0.00

0.00

0.00

0.01

0.00

0.01

0.00

0.01

0.00

0.00

0.00

0.00

0.00

Gam

bia

0.06

0.00

0.06

0.00

0.06

0.05

0.00

0.05

0.00

0.05

0.00

0.00

0.00

0.00

0.00

0.01

0.00

0.01

0.00

0.01

0.01

0.00

0.01

0.00

0.01

Ghana

0.82

0.00

0.82

0.00

0.82

0.61

0.00

0.61

0.00

0.61

0.01

0.00

0.01

0.00

0.01

0.12

0.00

0.12

0.00

0.12

0.08

0.00

0.08

0.00

0.08

Guinea

0.44

0.00

0.44

0.00

0.44

0.33

0.00

0.33

0.00

0.33

0.01

0.00

0.01

0.00

0.01

0.06

0.00

0.06

0.00

0.06

0.05

0.00

0.05

0.00

0.05

Guinea

‐Bissau

0.08

0.00

0.08

0.00

0.08

0.06

0.00

0.06

0.00

0.06

0.00

0.00

0.00

0.00

0.00

0.01

0.00

0.01

0.00

0.01

0.01

0.00

0.01

0.00

0.01

Ken

ya

1.26

0.01

1.26

0.00

1.26

0.88

0.01

0.89

0.00

0.89

0.04

0.00

0.04

0.00

0.04

0.15

0.00

0.15

0.00

0.15

0.18

0.00

0.18

0.00

0.18


59

Table S2

2007

Number of pregn



pregn



pregn



pregn



pregn


P. falciparum#

P.

vivax#

Any

species

P. falciparum#

P.

vivax#

Any

species

P. falciparum#

P.

vivax#

Any

species

P. falciparum#

P.

vivax#

Any

species

P. falciparum#

P.

vivax#

Any

species

Countries

Stable¶

Unstable¶

Overall

Overall

Overall

Stable¶

Unstable¶

Overall

Overall

Overall

Stable¶

Unstable¶

Overall

Overall

Overall

Stable¶

Unstable¶

Overall

Overall

Overall

Stable¶

Unstable¶

Overall

Overall

Overall

Liberia

0.20

0.00

0.20

0.00

0.20

0.15

0.00

0.15

0.00

0.15

0.00

0.00

0.00

0.00

0.00

0.03

0.00

0.03

0.00

0.03

0.02

0.00

0.02

0.00

0.02

Madagascar

0.79

0.00

0.79

0.86

0.86

0.56

0.00

0.56

0.60

0.60

0.02

0.00

0.02

0.02

0.02

0.11

0.00

0.11

0.11

0.11

0.11

0.00

0.11

0.12

0.12

Malaw

i 0.68

0.00

0.68

0.00

0.68

0.48

0.00

0.48

0.00

0.48

0.02

0.00

0.02

0.00

0.02

0.09

0.00

0.09

0.00

0.09

0.10

0.00

0.10

0.00

0.10

Mali

0.71

0.03

0.71

0.00

0.71

0.52

0.02

0.52

0.00

0.52

0.01

0.00

0.01

0.00

0.01

0.10

0.00

0.10

0.00

0.10

0.07

0.00

0.07

0.00

0.07

Mauritania

0.04

0.02

0.05

0.00

0.05

0.03

0.01

0.04

0.00

0.04

0.00

0.00

0.00

0.00

0.00

0.01

0.00

0.01

0.00

0.01

0.00

0.00

0.01

0.00

0.01

Mayotte*

0.00

0.01

0.01

0.00

0.01

0.00

0.01

0.01

0.00

0.01

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

Mozambique

1.04

0.00

1.04

0.00

1.04

0.73

0.00

0.73

0.00

0.73

0.02

0.00

0.02

0.00

0.02

0.14

0.00

0.14

0.00

0.14

0.15

0.00

0.15

0.00

0.15

Nam

ibia

0.04

0.01

0.06

0.00

0.06

0.03

0.01

0.04

0.00

0.04

0.00

0.00

0.00

0.00

0.00

0.01

0.00

0.01

0.00

0.01

0.01

0.00

0.01

0.00

0.01

Niger

0.78

0.04

0.82

0.00

0.82

0.58

0.03

0.61

0.00

0.61

0.02

0.00

0.02

0.00

0.02

0.10

0.00

0.11

0.00

0.11

0.08

0.00

0.08

0.00

0.08

Nigeria

6.47

0.00

6.47

0.00

6.47

4.78

0.00

4.78

0.00

4.78

0.15

0.00

0.15

0.00

0.15

0.88

0.00

0.88

0.00

0.88

0.67

0.00

0.67

0.00

0.67

Rwanda

0.31

0.00

0.31

0.00

0.31

0.22

0.00

0.22

0.00

0.22

0.01

0.00

0.01

0.00

0.01

0.04

0.00

0.04

0.00

0.04

0.04

0.00

0.04

0.00

0.04

S. Tome &Principe

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

Senegal

0.47

0.00

0.47

0.00

0.47

0.35

0.00

0.35

0.00

0.35

0.01

0.00

0.01

0.00

0.01

0.06

0.00

0.06

0.00

0.06

0.05

0.00

0.05

0.00

0.05

Sierra Leo

ne

0.32

0.00

0.32

0.00

0.32

0.24

0.00

0.24

0.00

0.24

0.01

0.00

0.01

0.00

0.01

0.04

0.00

0.04

0.00

0.04

0.03

0.00

0.03

0.00

0.03

South Africa

0.10

0.09

0.19

0.00

0.19

0.07

0.06

0.13

0.00

0.13

0.00

0.00

0.00

0.00

0.00

0.01

0.01

0.03

0.00

0.03

0.02

0.01

0.03

0.00

0.03

Swaziland

0.01

0.00

0.01

0.00

0.01

0.01

0.00

0.01

0.00

0.01

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

Tanzania

1.93

0.00

1.93

0.00

1.93

1.36

0.00

1.36

0.00

1.36

0.04

0.00

0.04

0.00

0.04

0.26

0.00

0.26

0.00

0.26

0.27

0.00

0.27

0.00

0.27

Togo

0.24

0.00

0.24

0.00

0.24

0.18

0.00

0.18

0.00

0.18

0.00

0.00

0.00

0.00

0.00

0.03

0.00

0.03

0.00

0.03

0.03

0.00

0.03

0.00

0.03

Uganda

1.53

0.00

1.53

0.00

1.53

1.07

0.00

1.07

0.00

1.07

0.04

0.00

0.04

0.00

0.04

0.20

0.00

0.20

0.00

0.20

0.21

0.00

0.21

0.00

0.21

Zambia

0.57

0.00

0.57

0.00

0.57

0.40

0.00

0.40

0.00

0.40

0.01

0.00

0.01

0.00

0.01

0.08

0.00

0.08

0.00

0.08

0.08

0.00

0.08

0.00

0.08

Zimbabwe

0.25

0.00

0.25

0.00

0.25

0.17

0.00

0.17

0.00

0.17

0.00

0.00

0.00

0.00

0.00

0.03

0.00

0.03

0.00

0.03

0.03

0.00

0.03

0.00

0.03

Chapter 3 Appen

dix

60

Table S2

2007

Number of pregn



pregn



pregn



pregn



pregn


P. falciparum#

P.

vivax#

Any

species

P. falciparum#

P.

vivax#

Any

species

P. falciparum#

P.

vivax#

Any

species

P. falciparum#

P.

vivax#

Any

species

P. falciparum#

P.

vivax#

Any

species

Countries

Stable¶

Unstable¶

Overall

Overall

Overall

Stable¶

Unstable¶

Overall

Overall

Overall

Stable¶

Unstable¶

Overall

Overall

Overall

Stable¶

Unstable¶

Overall

Overall

Overall

Stable¶

Unstable¶

Overall

Overall

Overall

EMRO

4.01

4.07

8.05

9.79

12.51

2.78

2.79

5.55

6.71

8.60

0.13

0.12

0.25

0.25

0.34

0.49

0.50

0.99

1.25

1.58

0.62

0.66

1.27

1.58

1.99

Afghanistan 2

0.29

0.80

1.09

1.06

1.30

0.20

0.55

0.74

0.72

0.89

0.01

0.03

0.03

0.03

0.04

0.04

0.10

0.13

0.13

0.16

0.05

0.13

0.18

0.17

0.21

Djibouti 1

0.00

0.02

0.02

0.03

0.03

0.00

0.01

0.01

0.02

0.02

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

Iran

2

0.00

0.07

0.07

1.18

1.18

0.00

0.05

0.05

0.80

0.81

0.00

0.00

0.00

0.01

0.01

0.00

0.01

0.01

0.17

0.17

0.00

0.01

0.01

0.19

0.19

Iraq

2

0.00

0.00

0.00

0.40

0.40

0.00

0.00

0.00

0.27

0.27

0.00

0.00

0.00

0.01

0.01

0.00

0.00

0.00

0.05

0.05

0.00

0.00

0.00

0.06

0.06

Oman

2

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

Pakistan 2

1.14

2.53

3.67

5.77

5.77

0.78

1.73

2.51

3.94

3.94

0.03

0.07

0.11

0.17

0.17

0.14

0.31

0.45

0.71

0.71

0.19

0.41

0.60

0.95

0.95

Saudi A

rabia 2

0.02

0.04

0.06

0.44

0.48

0.02

0.03

0.05

0.30

0.33

0.00

0.00

0.00

0.00

0.00

0.00

0.01

0.01

0.06

0.07

0.00

0.01

0.01

0.07

0.07

Somalia 1

0.50

0.03

0.50

0.50

0.50

0.35

0.02

0.35

0.35

0.35

0.02

0.00

0.02

0.02

0.02

0.06

0.00

0.06

0.06

0.06

0.07

0.00

0.07

0.07

0.07

Sudan

1

1.22

0.29

1.50

0.12

1.51

0.85

0.20

1.05

0.09

1.05

0.05

0.01

0.06

0.00

0.06

0.14

0.03

0.17

0.01

0.17

0.18

0.04

0.22

0.02

0.22

Syrian

Arab Rep

. 2

0.00

0.00

0.00

0.20

0.20

0.00

0.00

0.00

0.14

0.14

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.03

0.03

0.00

0.00

0.00

0.03

0.03

Yemen

2

0.83

0.30

1.13

0.10

1.14

0.58

0.21

0.79

0.07

0.79

0.02

0.01

0.03

0.00

0.03

0.11

0.04

0.15

0.01

0.15

0.13

0.05

0.17

0.02

0.17

EURO

0.00

0.11

0.11

0.62

0.64

0.00

0.08

0.08

0.43

0.44

0.00

0.00

0.00

0.01

0.01

0.00

0.02

0.02

0.09

0.09

0.00

0.02

0.02

0.10

0.10

Arm

enia 6

0.00

0.00

0.00

0.01

0.01

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

Azerbaijan 6

0.00

0.00

0.00

0.01

0.01

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

Geo

rgia 6

0.00

0.00

0.00

0.01

0.01

0.00

0.00

0.00

0.01

0.01

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

Kyrgyzstan 2

0.00

0.03

0.03

0.05

0.05

0.00

0.02

0.02

0.03

0.03

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.01

0.01

0.00

0.01

0.01

0.01

0.01

Tajikistan 2

0.00

0.08

0.08

0.16

0.18

0.00

0.05

0.05

0.11

0.12

0.00

0.00

0.00

0.00

0.00

0.00

0.01

0.01

0.02

0.02

0.00

0.01

0.01

0.03

0.03

Turkey 6

0.00

0.00

0.00

0.34

0.34

0.00

0.00

0.00

0.23

0.23

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.05

0.05

0.00

0.00

0.00

0.05

0.05

Turkmen

istan 2

0.00

0.00

0.00

0.04

0.04

0.00

0.00

0.00

0.03

0.03

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.01

0.01

0.00

0.00

0.00

0.01

0.01

Uzbekistan 2

0.00

0.00

0.00

0.01

0.01

0.00

0.00

0.00

0.01

0.01

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00


61

Table S2

2007

Number of pregn



pregn



pregn



pregn



pregn


P. falciparum#

P.

vivax#

Any

species

P. falciparum#

P.

vivax#

Any

species

P. falciparum#

P.

vivax#

Any

species

P. falciparum#

P.

vivax#

Any

species

P. falciparum#

P.

vivax#

Any

species

Countries

Stable¶

Unstable¶

Overall

Overall

Overall

Stable¶

Unstable¶

Overall

Overall

Overall

Stable¶

Unstable¶

Overall

Overall

Overall

Stable¶

Unstable¶

Overall

Overall

Overall

Stable¶

Unstable¶

Overall

Overall

Overall

AMRO

1.36

1.60

2.96

2.90

4.32

0.84

1.01

1.85

1.84

2.73

0.02

0.02

0.03

0.03

0.04

0.18

0.22

0.40

0.40

0.60

0.32

0.35

0.67

0.62

0.95

Argen

tina 5

0.00

0.00

0.00

0.08

0.08

0.00

0.00

0.00

0.05

0.05

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.01

0.01

0.00

0.00

0.00

0.02

0.02

Belize 4

0.00

0.01

0.01

0.01

0.01

0.00

0.00

0.00

0.01

0.01

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

Bolivia 5

0.01

0.10

0.11

0.16

0.16

0.01

0.06

0.07

0.10

0.10

0.00

0.00

0.00

0.00

0.00

0.00

0.01

0.02

0.02

0.02

0.00

0.02

0.03

0.04

0.04

Brazil 5

0.37

0.48

0.84

0.66

0.98

0.23

0.29

0.52

0.41

0.61

0.00

0.00

0.01

0.01

0.01

0.05

0.07

0.12

0.09

0.14

0.09

0.11

0.20

0.15

0.23

Colombia 5

0.15

0.22

0.37

0.49

0.61

0.09

0.14

0.23

0.30

0.37

0.00

0.00

0.00

0.00

0.00

0.02

0.03

0.05

0.07

0.08

0.03

0.05

0.09

0.11

0.14

Costa Rica 4

0.00

0.00

0.00

0.02

0.02

0.00

0.00

0.00

0.02

0.02

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

Dominican

Rep

ublic 4

0.05

0.10

0.15

0.00

0.15

0.03

0.06

0.09

0.00

0.09

0.00

0.00

0.00

0.00

0.00

0.01

0.01

0.02

0.00

0.02

0.01

0.02

0.04

0.00

0.04

Ecuador 5

0.13

0.05

0.18

0.11

0.18

0.08

0.03

0.11

0.07

0.11

0.00

0.00

0.00

0.00

0.00

0.02

0.01

0.02

0.02

0.02

0.03

0.01

0.04

0.03

0.04

El Salvador 4

0.00

0.00

0.00

0.03

0.03

0.00

0.00

0.00

0.02

0.02

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.01

0.01

Guatem

ala 4

0.01

0.00

0.01

0.01

0.01

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

Guyana 5

0.04

0.23

0.27

0.17

0.31

0.03

0.15

0.18

0.12

0.21

0.00

0.01

0.01

0.00

0.01

0.01

0.03

0.03

0.02

0.04

0.01

0.04

0.05

0.03

0.05

Haiti 4

0.00

0.01

0.02

0.02

0.02

0.00

0.01

0.01

0.01

0.01

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

Honduras 4

0.37

0.00

0.37

0.00

0.37

0.22

0.00

0.22

0.00

0.22

0.01

0.00

0.01

0.00

0.01

0.05

0.00

0.05

0.00

0.05

0.09

0.00

0.09

0.00

0.09

Mexico 4

0.03

0.09

0.13

0.20

0.23

0.02

0.06

0.08

0.13

0.16

0.00

0.00

0.00

0.00

0.00

0.00

0.01

0.02

0.03

0.03

0.01

0.02

0.02

0.03

0.04

Nicaragua 4

0.00

0.00

0.00

0.41

0.41

0.00

0.00

0.00

0.28

0.28

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.06

0.06

0.00

0.00

0.00

0.07

0.07

Panam

a 4

0.05

0.07

0.11

0.07

0.15

0.03

0.04

0.08

0.05

0.10

0.00

0.00

0.00

0.00

0.00

0.01

0.01

0.02

0.01

0.02

0.01

0.01

0.02

0.01

0.03

Paraguay 5

0.03

0.00

0.03

0.01

0.03

0.02

0.00

0.02

0.00

0.02

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.01

Peru 5

0.00

0.00

0.00

0.05

0.05

0.00

0.00

0.00

0.03

0.03

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.01

0.01

0.00

0.00

0.00

0.01

0.01

Surinam

e 5

0.12

0.05

0.17

0.25

0.26

0.07

0.03

0.11

0.16

0.16

0.00

0.00

0.00

0.00

0.00

0.02

0.01

0.02

0.03

0.04

0.03

0.01

0.04

0.06

0.06

Ven

ezuela 5

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

Fren

ch Guiana 5

0.01

0.20

0.20

0.13

0.25

0.00

0.12

0.12

0.08

0.15

0.00

0.00

0.00

0.00

0.00

0.00

0.03

0.03

0.02

0.03

0.00

0.05

0.05

0.03

0.06

Chapter 3 Appen

dix

62

Table S2

2007

Number of pregn



pregn



pregn



pregn



pregn


P. falciparum#

P.

vivax#

Any

species

P. falciparum#

P.

vivax#

Any

species

P. falciparum#

P.

vivax#

Any

species

P. falciparum#

P.

vivax#

Any

species

P. falciparum#

P.

vivax#

Any

species

Countries

Stable¶

Unstable¶

Overall

Overall

Overall

Stable¶

Unstable¶

Overall

Overall

Overall

Stable¶

Unstable¶

Overall

Overall

Overall

Stable¶

Unstable¶

Overall

Overall

Overall

Stable¶

Unstable¶

Overall

Overall

Overall

SEARO

16.64

20.95

37.57

47.97

48.17

11.02

14.01 25.02

31.88

32.00

0.31

0.40

0.71

0.91

0.91

2.21

2.78

4.98

6.35

6.38

3.11

3.76

6.86

8.84

8.88

Bangladesh 2

0.46

1.47

1.93

4.72

4.72

0.32

1.00

1.32

3.22

3.22

0.01

0.04

0.05

0.13

0.13

0.06

0.19

0.24

0.59

0.59

0.08

0.24

0.32

0.77

0.77

Bhutan 2

0.02

0.01

0.02

0.02

0.02

0.01

0.01

0.01

0.01

0.01

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

Burm

a 2

1.23

0.05

1.29

1.36

1.40

0.73

0.03

0.76

0.80

0.83

0.02

0.00

0.02

0.02

0.02

0.16

0.01

0.17

0.18

0.18

0.33

0.01

0.34

0.36

0.37

India 2

12.29

15.88

28.16

32.87

32.87

8.39

10.84 19.24

22.45

22.45

0.26

0.33

0.58

0.68

0.68

1.62

2.10

3.72

4.35

4.35

2.01

2.60

4.62

5.39

5.39

Indonesia 2

2.00

2.39

4.40

6.30

6.36

1.18

1.41

2.60

3.72

3.75

0.01

0.02

0.03

0.05

0.05

0.28

0.33

0.60

0.86

0.87

0.53

0.64

1.17

1.67

1.69

Rep

ublic of Korea 2

0.00

0.00

0.00

0.20

0.20

0.00

0.00

0.00

0.12

0.12

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.03

0.03

0.00

0.00

0.00

0.06

0.06

Nep

al 2

0.12

0.21

0.33

0.63

0.66

0.08

0.15

0.23

0.43

0.45

0.00

0.01

0.01

0.02

0.03

0.01

0.02

0.04

0.07

0.08

0.02

0.04

0.05

0.10

0.11

Sri Lanka 2

0.04

0.16

0.20

0.23

0.28

0.03

0.11

0.14

0.15

0.19

0.00

0.00

0.00

0.00

0.00

0.01

0.02

0.03

0.03

0.04

0.01

0.03

0.03

0.04

0.05

Thailand 2

0.41

0.76

1.18

1.60

1.60

0.24

0.45

0.70

0.94

0.94

0.00

0.00

0.01

0.01

0.01

0.06

0.11

0.16

0.22

0.22

0.11

0.20

0.31

0.42

0.42

Timor‐Leste 2

0.07

0.00

0.07

0.04

0.07

0.04

0.00

0.04

0.03

0.04

0.00

0.00

0.00

0.00

0.00

0.01

0.00

0.01

0.01

0.01

0.02

0.00

0.02

0.01

0.02

WPRO

3.11

3.52

6.62

28.03

29.27

1.84

2.06

3.91

16.08

16.81

0.03

0.03

0.06

0.35

0.36

0.42

0.48

0.90

3.66

3.83

0.81

0.94

1.76

7.93

8.27

Cam

bodia 2

0.44

0.11

0.55

0.52

0.60

0.26

0.06

0.32

0.31

0.35

0.01

0.00

0.01

0.01

0.01

0.06

0.01

0.07

0.07

0.08

0.12

0.03

0.15

0.14

0.16

China 2

0.41

0.48

0.89

21.25

21.70

0.23

0.27

0.50

12.06

12.32

0.01

0.01

0.01

0.30

0.30

0.05

0.06

0.11

2.73

2.79

0.12

0.14

0.26

6.15

6.28

Lao 2

0.21

0.00

0.21

0.09

0.21

0.13

0.00

0.13

0.06

0.13

0.00

0.00

0.00

0.00

0.00

0.03

0.00

0.03

0.01

0.03

0.06

0.00

0.06

0.02

0.06

Malaysia 2

0.21

0.54

0.75

0.34

0.76

0.12

0.32

0.44

0.20

0.45

0.00

0.00

0.00

0.00

0.00

0.03

0.08

0.11

0.05

0.11

0.06

0.14

0.20

0.09

0.20

Papua New

Guinea 3

0.16

0.00

0.16

0.15

0.17

0.11

0.00

0.11

0.11

0.12

0.00

0.00

0.00

0.00

0.00

0.02

0.00

0.02

0.02

0.02

0.02

0.00

0.02

0.02

0.03

Philippines 2

1.08

0.82

1.90

3.15

3.28

0.64

0.48

1.12

1.86

1.94

0.01

0.01

0.01

0.02

0.02

0.15

0.11

0.26

0.43

0.45

0.29

0.22

0.50

0.84

0.87

Solomon Islands 3

0.02

0.00

0.02

0.01

0.02

0.01

0.00

0.01

0.01

0.01

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

Vanuatu 3

0.01

0.00

0.01

0.01

0.01

0.01

0.00

0.01

0.01

0.01

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

Viet Nam

2

0.57

1.57

2.14

2.51

2.52

0.34

0.93

1.26

1.48

1.48

0.00

0.01

0.01

0.02

0.02

0.08

0.22

0.29

0.34

0.35

0.15

0.42

0.57

0.67

0.67

TOTA

L 54.74

30.61

85.28

92.90

125.24 38.03

20.20 58.19

59.47

82.63

1.18

0.58

1.76

1.63

2.38

7.25

4.04

11.29

12.22

16.51

8.27

5.78

14.04 19.58 23.71

*In m

illions; # Includes countries where P.falciaprum and P.vivax co‐exist; ¶

Stable transm

ission: >= 0.1 autochthonous P. falciparum cases per 1,000 peo

ple per annum; u

nstable transm

ission <0.1 autochthonous P. falciparum cases per

1,000 peo

ple per annum; §

The number of pregnant women

at risk was then

derived

as follow: P

regA

R= PARxFWOCBAxAnnualised

_PR, w

here: FWOCBA=fraction of W

OCBA in

2007; A

nnualised

_PR = the average number of pregnancies

per year; Note: The regional and total estim

ates for TFR, Stillbirth rate, TPR and Annual PR are weighted m

ean and is for illustration purposes only. The number of pregnancies at risk was derived

by adding national estim

ates within regions

and should not be calculated using regional or global estim

ates for pregnan

cy rates; C

ontinen

t inform

ation: A

frica is den

oted by 1; A

sia by 2; O

ceania by3; N

orth America 4 , South America 5 and Europe 6.


63

Table S3: Total population, n

umber of pregnan

cies and number of live‐births born to pregnan

cies in m

alaria endemic countries by continent in 2007 (in m

illions)

Continents

Demograp

hic data*

Total population at risk

(% of the population in

Malaria Endem

ic Countries at risk)

(United Nations National Population Estim

ates)

Malaria Endem

ic Countries (M

EC)

P. falciparum transm

ission#

P. vivax

transm

ission#

Any species

Number

of

Malaria

Endem

ic

Countrie

s

Total

Popul

ation

(both

sexes

)*

WOCBA

*

Total

number of

Pregnancies

*¥

TPR§

Pregnancy

rate per

1000

WOCBAs§

Percentage pregnan

cies ending in§:

Stable

transm

ission¶

Unstable

transm

ission

¶

Overall

Overall

Overall

Live‐

births

Still‐

births

Spontan

e‐ous

Abortion

s

Induced

Abortion

s

Africa1

46

803

190

38.5

7.12

198

72.1%

2.4%

13.3%

12.2%

637.6 (79.4)

16.2 (2.0)

652.8 (81.2)

85.6 (10.7)

660.8 (82.2)

Asia2

28

3,726

984

105.8

3.76

135

62.5%

1.6%

13.1%

22.8%

702.2 (18.8)

918.7 (24.7)

1620.5 (43.5)

2975.3 (79.9)

3,048.2 (81.8)

Oceania

3

3

7

2

0.3

5.46

156

69.3%

1.2%

14.2%

15.3%

4.8 (67.5)

0.0 (0.0)

4.8 (67.5)

4.5 (63.4)

5.2 (73.3)

North America4

10

167

45

4.9

3.84

127

66.5%

0.8%

14.3%

18.4%

14.4 (8.6)

13.0 (7.8)

27.4 (16.4)

30.1 (18.1)

50.6 (30.3)

South America

5

11

363

98

10.6

3.79

122

61.8%

0.8%

13.9%

23.5%

26.8 (7.4)

37.2 (10.2)

64.0 (17.6)

66.1 (18.2)

87.6 (24.1)

Europe6

4

91

25

2.1

2.95

70

69.3%

0.9%

14.5%

15.3%

0.0 (0.0)

0.0 (0.0)

0.0 (0.0)

15.3 (16.8)

15.3 (16.8)

Global

102

5,157

1,343

162.3

4.23

159

65.5%

1.8%

13.3%

19.5%

1385.8 (26.9)

985.1 (19.1)

2369.4 (45.9)

3176.9 (61.6)

3,867.6 (75.0)

Continen

ts

Number of pregn

ancies at risk f malaria (column %)

Number of live‐births born to pregn

ancies at risk of malaria (column %)


ission#

P. vivax

transm

ission#

Any species


ission#

P. vivax

transm

ission#

Any species

Stable

transm

ission¶

Unstable

transm

ission¶

Overall

Overall

Overall

Stable

transm

ission¶

Unstable

transm

ission¶

Overall

Overall

Overall

Africa1

31.3(57.3)

0.7 (2.3)

32.0(37.5)

4.3(4.6)

32.4(25.8)

22.8(59.8)

0.5(2.4)

23.2(39.9)

3.0(5.0)

23.5 (28.4)

Asia2

21.9(39.9)

28.3 (92.5)

50.2(58.8)

85.2(91.7)

88.0(70.3)

14.3(37.6)

18.7(92.6)

33.0(56.7)

54.3(91.2)

56.0 (67.8)

Oceania

3

0.2(0.3)

0.0 (0.0)

0.2(0.2)

0.2(0.2)

0.2(0.2)

0.1(0.3)

0.0(0.0)

0.1(0.2)

0.1(0.2)

0.1 (0.2)

North America4

0.6(1.0)

0.5 (1.6)

1.1(1.2)

0.9(1.0)

1.7(1.4)

0.4(0.9)

0.3(1.6)

0.7(1.2)

0.6(1.0)

1.1 (1.4)

South America

5

0.8(1.4)

1.1 (3.6)

1.9(2.2)

2.0(2.1)

2.6(2.1)

0.5(1.3)

0.7(3.4)

1.2(2.0)

1.2(2.1)

1.6 (1.9)

Europe6

0.0(0.0)

0.0 (0.0)

0.0(0.0)

0.4(0.4)

0.4(0.3)

0.0(0.0)

0.0(0.0)

0.0(0.0)

0.2(0.4)

0.2 (0.3)

Global

54.7

30.6

85.3

92.9

125.2

38.0

20.2

58.2

59.5

82.6

* Source: United

Nations Developmen

t Program

; ¥ The total number of pregnancies is the sum of the number of live‐births, stillbirths, spontaneo

us and induced abortions; § The total pregnancy rate (TPR) and the annual pregnancy

rate per 1000 W

OCBAs are weighted m

eans per region and is for illustration purposes only. The number of pregnancies at risk was derived

directly as the sum of the national estim

ates within each region and globally. They differ

slightly from sim

ilar estimates obtained

indirectly by use of the weighted regional or global estim

ates for pregnancy rates; # Includes countries where P. falciparum and P. vivax co‐exist; ¶

Stable transm

ission: >= 1 autochthonous P.

falciparum cases per 10,000 peo

ple per annum; Unstable transm

ission <1 autochthonous P. falciparum cases per 10,000 peo

ple per annum; A

bbreviation: N. A

merica: North America; S. A

merica: South America; M

EC: M

alaria

Endem

ic Countries; TPR: Total Pregnancy Rate; W

OCBA: W

omen

of Childbearing Age (15‐49 years of age). Continent inform

ation: Africa is den

oted by 1; A

sia by 2; O

ceania by3; N

orth America 4 , South America 5 and Europe 6

Regions inform

ation: Th

e Africa and Europe regions are defined

as the respective continen

ts; A

mericas is defined

as ‘North America’ plus 'South America' countries and Asia region defined

as Asia‐Pacific and Oceania combined

65

Chapter4

PregnancyExposureRegistriesforAssessingAntimalarialDrugSafetyinPregnancyinMalaria‐EndemicCountries

StephanieDellicour1,FeikoO.terKuile1,2,AndyStergachis3


United Kingdom, 2 Division of Infectious Diseases, Tropical Medicine & AIDS, Academic

Medical Centre Amsterdam, Amsterdam, The Netherlands, 3 Departments of

Epidemiology and Global Health, School of Public Health & Community Medicine,

University of Washington, Seattle, Washington, United States of America

PLoS Medicine 2008, 5:9

Copyright© This is an open‐access article distributed under the terms of the Creative

Commons Attribution License, which permits unrestricted use, distribution, and

reproduction in any medium, provided the original author and source are credited

Chapter 4

66

SummaryPoints There is an urgent need to develop targeted pharmacovigilance systems to assess the safety

of antimalarials in early pregnancy.

The artemisinins are effective antimalarials increasingly deployed in malaria‐endemic

countries; however, they have been shown to be embryo‐ toxic in animal models, and their

safety in early human pregnancies remains uncertain.

Modelling suggests that the probability an embryo will encounter artemisinins during the

critical six‐ week period (at week four to week ten of gestation) through accidental exposure is

12% for areas where adults receive on average one treatment with three days of artemisinin‐

based combination therapy per year.

Most of the approaches used in industrialised countries to evaluate a drug’s embryo‐foetal

toxicity have limited application in resource‐poor countries. Establishing an international

antimalarial pregnancy exposure registry would enable a targeted prospective

pharmacovigilance approach and timely assessment of the risk–benefit profile of

antimalarials.

Here we discuss methodological considerations for the systematic prospective assessment of

pregnancy outcomes and congenital malformations in women exposed to antimalarials early

in pregnancy, including approaches to capture drug exposure information in resource‐poor

settings, choice of comparison groups, and sample size considerations.

Antimalarial pregnancy exposure registries

67

Because pregnant women are routinely excluded from pre‐ licensure clinical trials for fear of harming

the mother or the developing foetus [1], most drugs are marketed with limited information on their

safety during pregnancy and therefore are not recommended for use by pregnant women. Yet drugs

are widely used by pregnant women, and medication often cannot be avoided in chronic diseases such

as epilepsy and HIV or other acute illness that harm the mother and the unborn child if left untreated.

Passive mechanisms of spontaneous reporting of adverse drug effects are inadequate for detecting

drug‐induced foetal risks or lack of such risks [2]. The US Food and Drug Administration and the

European Medicine Agency recommend active surveillance, such as the use of pregnancy exposure

registries (PERs), for products that are likely to be used during pregnancy or by women of

childbearing age (WOCBAs), particularly if there have been case reports of adverse pregnancy

outcome following exposure, if drugs in the same pharmacological class are known to pose risk

during pregnancy, or if pre‐clinical animal data suggest potential teratogenic risk [3,4]. In

industrialised countries, this information can be derived from medical records and automated

databases, including medical or pharmacy insurance claims. Such approaches are challenging in

developing countries where resources for routine pharmacovigilance are rare and automated data

sources generally do not exist [5–8]. Thus, nearly all developing countries rely on drug safety data

from industrialised countries. However, there are often no or limited safety data in pregnancy for

drugs targeting tropical diseases, as these are not widely used in the countries with more robust

pharmacovigilance systems [9].

Antimalarials are a good example [9]. Malaria can have devastating consequences for the mother and

foetus [10,11], and pregnant women require prompt treatment with safe and effective antimalarial

drugs when infected. The artemisinins are among the most effective and rapidly acting antimalarials

to date, providing life‐ saving benefits to children, adults, and pregnant women [12]. The limited

information regarding their safety is reassuring [13], and the World Health Organization (WHO) now

recommends the use of artemisinin combination therapies (ACTs) in the second and third but not yet

in the first trimester (unless alternatives are not available) [12], as uncertainty remains about their

safety in early pregnancy (Box 1) [14–16]. ACTs are rapidly being rolled out and may soon become

among the most widely used antimalarial drugs. Because there are no specific risk management

precautions to exclude WOCBAs from using ACTs, the potential for inadvertent exposure to

artemisinins early in pregnancy is high and in many cases unavoidable (Figure 1). Health care

providers, pregnant women, and policy makers urgently need valid information to make informed

decisions about the risks and benefits of ACTs for WOCBAs.

This paper describes the use of PERs as a targeted pharmacovigilance approach for assessing the safety

of antimalarial drugs used during early pregnancy in resource‐constrained malaria‐endemic countries.

AntimalarialPregnancyExposureRegistries

PERs are the most common approach used to monitor drug safety in pregnancy and provide

reassurance on the potential risk associated with certain drugs. They can serve both to generate

hypotheses and to evaluate suspected risks or risk factors that may have been identified during pre‐ or

post‐marketing phases [2]. In industrialised countries, 32 PERs are registered with the Food and Drug

Administration [17]. There is some variation in design, but they all use prospective approaches and

identify and follow exposed women until the end of pregnancy (i.e., before the outcome is known).

Chapter 4

68

The systematic prospective ascertainment of pregnancy outcomes has several major advantages over

case‐control designs and passive surveillance. This design reduces selection bias (for example, due to

self‐reporting) and recall bias, has the potential to use standardised methods to assess outcome,

and—because of the availability of both numerator and denominators—allows calculations of risk

estimates that can then be compared against comparison groups or background population rates [2].

One other attractive feature of PERs is that they can be time‐limited and terminated once the target

sample size to rule out a pre‐defined risk is reached.

AssessmentofDrugExposureandRecordLinkage

The design for reliably capturing the occurrence and timing of inadvertent drug exposure to ACTs in

early pregnancy requires special consideration. Firstly, the critical period occurs around the time

when many women may not yet be aware of their pregnancy (our current understanding from animal

models of the mechanism of embryotoxicity of the artemisinins suggests that in humans the sensitive

drug exposure time window is between week four to week ten of gestation [Box 1]). Secondly,

retrospective determination of the precise timing of exposure is challenging since a typical treatment

course is short (three days). Another difficulty is the accurate assessment of the gestational age at

the time of exposure. Lastly, malaria treatment is often home‐based or unsupervised and

antimalarials can be obtained from a variety of providers, often over‐the‐ counter. In contrast,

antiretroviral and anti‐tuberculosis drugs are typically provided by formal health services, which are

more likely to keep records. Furthermore, exposures are often long‐ term and continuous, making it

easier to determine if and when a woman was exposed to antiretroviral or anti‐tuberculosis

medication than with the short course of antimalarials.

Although most exposures to artemisinins in early pregnancy will be unintentional, deliberate

exposures can occur where the benefit is perceived to outweigh the potential risk, as recommended

by WHO (such as for severe life‐threatening malaria) [12]. Either way, reliable ascertainment of drug

exposure will require record linkage. This can be done using prospective approaches by linking

datasets containing drug dispensing information (e.g., malaria treatment records from out‐patient

Box1.MechanismofArtemisininToxicityinEarlyPregnancyAnimal reproductive toxicology studies show that artemisinin derivatives all have embryo‐toxic effects at

low‐dose ranges in all species studied (i.e., mice, rat, rabbit, frog, and primate models) [31–34].The

embryo‐toxic mechanism is thought to occur through depletion of embryonic erythroblasts (primitive

erythrocytes), which is associated with severe anaemia leading to cell damage and death due to hypoxia

[35]. In humans, the most sensitive time window may be between week four and week ten, when

erythroblasts circulate and have not yet been fully replaced by definitive erythrocytes [36]. In addition to

the window of sensitivity, the duration of exposure is also important. Rodents have a synchronous clonal

expansion of metabolically active erythroblasts, making them particularly vulnerable during a three‐ to

four‐day window early in pregnancy. In primates (and most likely also in humans), this may not be the case,

as different generations of erythroblasts co‐exist and are progressively replaced by definitive erythrocytes

over a period of weeks [35]. In cynomolgus monkeys, no embryo lethality or malformations were observed

with three‐day exposures (the typical duration of treatment with ACTs) or with seven‐day exposures

[31,32,35,36]. The predictive value of the animal models for humans is unclear, particularly because the

duration of daily exposure is likely to be short (hours) as the artemisinins are rapidly eliminated and limited

to three or at maximum seven days.


69

departments) to datasets that capture newly identified pregnancies (e.g., from antenatal clinics or

demographic surveillance systems). This can determine whether a WOCBA might have been pregnant

at the time of treatment. Alternatively, records of pregnant women can be linked retrospectively

with their earlier treatment records. A disadvantage of recruiting pregnant women rather than

WOCBAs is that miscarriages will be missed, as the pregnancy may not be sustained long enough for

women to attend antenatal care. The pregnant woman’s drug history should be taken to verify the

record linkage and to capture information on any drug use not dispensed through formal

pharmacies.

Figure 1. Probability that an Embryo Will Encounter Artemisinins Inadvertently During the Critical Six‐Week Period of Its Development (Week Four to Week Ten), According to the Average Number of ACT Treatments Received Per Year. doi:10.1371/journal.pmed.0050187.g001

In the figure, χ = number of treatments per year, t = embryo‐sensitive period in days (set as 42 days or six weeks), and p = period of treatment and persistence of drug (set as three days because ACTs are normally deployed as a three‐day regimen and artemisinins are eliminated within hours after each dose). The inadvertently exposed group will consist of women taking ACTs for confirmed malaria and for presumed malaria. It has been estimated that over 70% of malaria episodes in rural Africa and about 50% in urban areas are self‐treated without consulting trained professionals [39]. Thus, many of these will be presumptive treatments without involvement of the formal health services, diagnostic confirmation of malaria, or screening for potential pregnancy. Even if more women seek treatment at health facilities with the deployment of more expensive ACTs and rapid diagnostic tests, antimalarials are often administered disregarding any diagnostic test. Studies in Africa indicated that between 30% and 50% of patients with a negative diagnostic test (microscopy or rapid diagnostic test) were still prescribed antimalarial drugs [40,41]. These proportions are likely to increase further when successful malaria control reduces malaria exposure. (Adapted from [16].)

PregnancyOutcomeAssessment

The primary outcome of interest is a decisive factor for the choice of study design, study population,

11.6%

21.9%

31.0%

39.1%

46.2%

52.6%

58.2%

63.1%

67.5%

71.4%

0%

10%

20%

30%

40%

50%

60%

70%

80%

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 8 8.5 9 9.5 10

Ris

k of

Exp

osur

e (%

)

Number of ACT treatment per year

Probability of exposure during embryo-sensitive period: 1-(1-x/365) (t+p) *

1.2%

2.4%

3.6%

4.8%

6.0%

7.1%

8.3%

9.4%

10.5%11.6%

0%1%2%3%4%5%6%7%8%9%

10%11%12%

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Chapter 4

70

and target data sources for outcome ascertainment and needs to be defined a priori. Although pre‐

approval animal reproductive toxicology studies have ambiguous predictive value for human

embryo‐foetal toxicity, due to variations in species‐specific effects [18], the current data from animal

models suggest that the effects are not species‐specific and that exposure early in the first trimester

might cause birth defects and/or early embryo/foetal death with subsequent miscarriages or foetal

resorption. Most of the existing PERs monitor all pregnancy outcomes (i.e., live births, still births, and

miscarriages), but the design and sample size calculation focus on capturing birth defects [19]. Foetal

resorption and early miscarriages are very difficult to assess reliably; most will go unnoticed clinically

as they occur before eight to nine weeks, with the majority occurring before three weeks [20]. Only

repeated pregnancy testing with a switch from positive to negative tests may suggest objectively

early loss of pregnancy [21]. This is unlikely to be feasible or culturally acceptable in many malaria‐

endemic countries, and the frequent use of pregnancy testing itself reduces the probability of

inadvertent exposures in that population. We may thus have to accept that early pregnancy loss

cannot be captured reliably in sufficient numbers, in contrast to later miscarriages and stillbirths.

For birth defects, the duration of follow‐up of the infant needs to be considered carefully, since only

about half of major structural and functional defects in children can be detected or classified at birth

[22,23]. The prevalence also varies with the specific defect inclusion and exclusion criteria and

whether the case definition includes developmental, functional, or other types of congenital

disorders (e.g., non‐structural genetic disorders) [24]. Assessment of congenital malformation

requires careful examination by dedicated staff trained to examine newborns using a standard tool

and scoring system. Suspected birth defects could be reviewed by a centralised committee that

included dysmorphologists and other specialists (e.g., using digital photographs). Complimentary

visiting specialists could study additional outcomes such as cardiovascular and neuro‐developmental

defects and other potential long‐term effects in a selected sample later in infancy.

ComparisonGroups

Assessing the teratogenic potential of a drug requires comparison of the frequency of birth defects

against other groups to put a signal into context. These comparison groups can be external (i.e., from

peripheral sources) or internal (i.e., generated from within the same study or system). External

population data from national health statistics centres and/ or birth defects surveillance systems are

commonly used as sources to calculate background event rates. This type of external comparison data

is not currently available in most malaria‐ endemic countries. Furthermore, these comparisons need

to be interpreted with caution as many confounding factors or potential effect modifiers of risk may

differ from the exposed group of interest [2,25].

Internal comparison groups can consist of women with the same conditions who are unexposed to

drugs (in which case the possibility of confounding by indication should be taken into account), exposed

to a different drug with established safety, or exposed to the same investigational drug, but only outside

the critical period (e.g., in the second or third trimesters).


71

0

5,000

10,000

15,000

20,000

25,000

30,000

35,000

40,000

45,000

50,000

1 1.5 2 2.5 3 3.5 4 4.5 5

Num

ber E

xpos

ed

Miscarriage 15% Stillbirth 3% Major Malformations 2% Specific Malformations 0.1%

Relative Risk

0500

1,0001,5002,0002,5003,0003,5004,0004,5005,000

1 1.5 2 2.5 3 3.5 4 4.5 5

Figure 2. Sample Size Calculation for Pregnancy Exposure Registry by Defect Frequency and Detectable Difference. doi:10.1371/journal.pmed.0050187.g002

Exposed to comparison group ratio = 1:1, power = 80%, and one‐sided α = 0.05. Based on the formula for cohort design described

in Strom’s Pharmacoepidemiology [25]: N = 1/[p(1 − R)]2 ×[Z √((1 + 1/k)U(1 − U)) + Z√(pR(1 − Rp) + (P(1 − P))/k)]2 where p is the incidence of disease in unexposed; R is the minimum relative risk to detect; k is the ratio of unexposed controls to exposed, and U = (Kp + pR)/(k + 1).

SampleSizeCalculationThe main determinants of sample size are the degree of the teratogenic effect to be excluded

(relative risk) and the expected frequency of the endpoint of interest in the non‐exposed group

(Figure 2). A third factor is the type and number of potential controls. For example, with an

exposed/unexposed ratio of 1:4, approximately 522 exposed women and 2,090 unexposed women

are needed to exclude a 2‐fold increase in major malformations detectable at birth when the

predicted rate in the comparison group is 2% (power 80%, alpha 0.05). This would be 10,748/42,992

exposed/unexposed women for birth defects that occur at a frequency of one in 1,000 (such as cleft

lip/palate). Such numbers will only be achievable using several sentinel sites over several years. The

sample sizes will also need to account for loss to follow‐ up and the fact that not all births can be

examined for birth defects (e.g., foetal loss with discarding of expelled foetus prior to examination by

study staff). The rate of recruitment will depend on the likelihood of accidental exposure. This

depends on the fertility rate and frequency of drug exposure in the population. For example, in areas

where pregnancy testing is not available, the average number of ACT treatments is one per woman

per year, and the total fertility rate is 5.5, the probability is only 2.5% (or one in 40 women) (Box 2).

Chapter 4

72

ConcomitantMedicationAlthough the registry could be set up initially to address the specific question of the safety of

antimalarials in pregnancy, it is essential to capture concomitant diseases and medications such as

antiretrovirals because of potential drug interactions, confounding, and effect modification. As such,

these additional data could contribute to PERs for other diseases such as the Antiretroviral Pregnancy

Registry [26,27].

DataSourcesThere are many methodological challenges to designing PERs for antimalarials, including those

common to most pharmacovigilance methods in resource‐poor countries [28,29]. The specialised

nature of the reliable assessment of drug exposure and congenital malformations is not easily

achievable from routine pharmacovigilance surveillance systems (where they exist). Such an effort

will require dedicated sentinel sites that are capable of following WOCBAs and linking antenatal care

records with treatment records, such as sites with demographic health surveillance systems or sites

with captive populations where health care is provided centrally and well recorded (e.g., industrial

and agricultural estates or long‐term refugee camps).

While the primary source of information is prospective and observational, data from clinical trials

and other studies involving pregnant women, and retrospective case series (i.e., pregnancies with a

known outcome at the time of reporting) could be included as secondary data and analysed

separately, as is currently done with some of the existing PERs [30].

Box 2: Probability that a Woman of Childbearing Age Treated for Malaria at an Out‐Patient Clinic Had an

Undetected Pregnancy of Four to Ten Weeks Gestation

An approximation of the probability that a WOCBA attending an out‐patient clinic has an early pregnancy can be

indirectly estimated from published total fertility rates. For sub‐Saharan Africa, total fertility rate was 5.5 in

2004 [37] and is defined as the number of live‐born children an average women would have, assuming that

she lives her full reproductive lifetime of 35 years (1,820 weeks, from 15 to 49 years). The total pregnancy

rate (6.7) was then calculated as the total fertility rate (5.5) multiplied by a factor of 1.22 (1 / [1.0 − 0.15 −

0.03]) to take into account 15% pregnancy loss due to miscarriages (a conservative estimate) and 3% due to

stillbirths (the average rate of stillbirths observed in developing countries [38]). Thus, of 1,820 reproductive

weeks, a woman is pregnant for 268 weeks (6.7 × 40 weeks); of which 40.2 weeks (6.7 × 6 weeks) are during

the sensitive six‐week time window from week four to week ten of gestation. Under these conditions, 14.7%

(268 of 1,820) of WOCBAs are pregnant at any time (i.e., one in 6.8), and 2.2% (40.2 of 1,820) or one in 45

are pregnant between week four and week ten.

If accidental exposure is defined as unintentional treatment in early pregnancy only, than the risk of

accidental exposure is slightly higher than 2.2%, as later pregnancy weeks do not contribute to the

denominator. The average time for women in Africa to recognise and report a pregnancy is not well

described in settings where pregnancy testing is not readily available. If it is assumed that this is during the

first ten weeks of pregnancy, then the denominator is 1,619 weeks (the 1,552 weeks that she is not pregnant

[1,820 − 268] plus the 67 weeks of early pregnancy (6.7 pregnancies × 10 weeks), and the risk of accidental

exposure in the four‐ to ten‐week period is 40.2 out of 1,607 weeks or 2.5% (one in 40 women).

This assumes that the probability of getting clinical malaria is the same in these first ten weeks of pregnancy

as in non‐pregnant women.


73

ConclusionThe establishment of an international antimalarial pregnancy exposure registry, using specialised

sentinel sites to provide reliable exposure and outcome data for the primary data collection, is a

potentially cost‐effective targeted approach. Central collation of the information would enable

evaluation of the risk–benefit profile of antimalarials in a timely manner, and over time would allow

the detection of rare adverse drug reactions that could not be detected by any single study. New

levels of collaboration between pharmacovigilance programmes, antimalarial drug developers,

research groups, regulatory authorities, and WHO will be essential. This international multi‐product,

multi‐ sponsor approach will require good governance structures, such as those used by the

Antiretroviral Pregnancy Registry, and if successful could serve as a pathfinder for other PERs to

capture much‐needed safety information on other drugs used for tropical diseases [9].

References1. Mitchell A (1994) Special considerations in studies of drug‐induced birth defects. In: Strom B, editor.

Pharmacoepidemiology. Chichester: Wiley. 2. Kennedy DL, Uhl K, Kweder SL (2004) Pregnancy exposure registries. Drug Saf 27: 215‐228. 3. Center for Drug Evaluation and Research, Center for Biologics Evaluation and Research (2002) Guidance for

industry: Establishing pregnancy exposure registries. US Food and Drug Administration. Available: http://www. fda.gov/cber/gdlns/pregexp.htm. Accessed 10 August 2008.

4. European Medicine Agency (2005) Guideline on the exposure to medicinal products during pregnancy: Need for post‐authorisation data. Available: http://www.emea.europa.eu/pdfs/ human/phvwp/31366605en.pdf. Accessed 10 August 2008.

5. Lang T, Hughes D, Kanyok T, Kengeya‐ Kayondo J, Marsh V, et al. (2006) Beyond registration—Measuring the public‐health potential of new treatments for malaria in Africa. Lancet Infect Dis 6: 46‐52.

6. Staedke S (2005) Pharmacovigilance and antimalarial treatment in Africa: The practical challenges. The Newsletter of The East African Network for Monitoring Antimalarial Treatment 8: 2. Available: http://www.eanmat.org/Newsletters/Nwsltt21.doc. Accessed 10 August 2008.

7. Talisuna AO, Staedke SG, D’Alessandro U (2006) Pharmacovigilance of antimalarial treatment in Africa: Is it possible? Malar J 5: 50.

8. Pirmohamed M, Atuah KN, Dodoo AN, Winstanley P (2007) Pharmacovigilance in developing countries. BMJ 335: 462.

9. White NJ, McGready RM, Nosten FH (2008) New medicines for tropical diseases in pregnancy: Catch‐22. PLoS Med 5: e133. doi:10.1371/journal.pmed.0050133

10. Steketee RW, Nahlen BL, Parise ME, Menendez C (2001) The burden of malaria in pregnancy in malaria‐endemic areas. Am J Trop Med Hyg 64: 28‐35.


12. World Health Organization (2006) Guidelines for the treatment of malaria. WHO/HTM/MAL/2006.1108. Available: http://www.who.int/malaria/includes_en/ whomalariapublications19982004.htm. Accessed 10 August 2008.

13. McGready R, Cho T, Keo NK, Thwai KL, Villegas L, et al. (2001) Artemisinin antimalarials in pregnancy: A prospective treatment study of 539 episodes of multidrug‐ resistant Plasmodium falciparum. Clin Infect Dis 33: 2009‐2016.

14. Dellicour S, Hall S, Chandramohan D, Greenwood B (2007) The safety of artemisinins during pregnancy: A pressing question. Malar J 6: 15.

15. Nosten F, McGready R, Alessandro U, Bonell A, Verhoeff F, et al. (2006) Antimalarial Drugs in Pregnancy: A Review. Curr Drug Saf 1: 1‐15.

16. Ward SA, Sevene EJ, Hastings IM, Nosten F, McGready R (2007) Antimalarial drugs and pregnancy: Safety, pharmacokinetics, and pharmacovigilance. Lancet Infect Dis 7: 136‐144.

17. US Food and Drug Administration (2008) List of pregnancy exposure registries. Available: http://www.fda.gov/womens/registries/registries.html. Accessed 10 August 2008.

18. Warkary J (1974) Problems in applying teratologic observations in animals to man. Paediatrics 53: 820.

Chapter 4

74

19. Honein MA, Paulozzi LJ, Cragan JD, Correa A (1999) Evaluation of selected characteristics of pregnancy drug registries. Teratology 60: 356‐364.

20. Simpson JL (1990) Incidence and timing of pregnancy losses: Relevance to evaluating safety of early prenatal diagnosis. Am J Med Genet 35: 165‐173.

21. Andrews E, Stergachis A, Hecht J (1989) Evaluation of alternative methods of assessing pregnancy outcomes using automated indicators of pregnancy J Clin Res Drug Dev 3: 201.

22. Briggs GG (2002) Drug effects on the foetus and breast‐fed infant. Clin Obstet Gynecol 45: 6‐21. 23. Christianson RE, van den Berg BJ, Milkovich L, Oechsli FW (1981) Incidence of congenital anomalies

among white and black live births with long‐term follow‐up. Am J Public Health 71: 1333‐1341. 24. Siffel C, Correa A, Cragan J, Alverson CJ (2004) Prenatal diagnosis, pregnancy terminations and prevalence

of Down syndrome in Atlanta. Birth Defects Res A Clin Mol Teratol 70: 565‐571. 25. Strom BL (2000) Pharmacoepidemiology. Sussex: John Wiley & Sons. 26. Covington DL, Tilson H, Elder J, Doi P (2004) Assessing teratogenicity of antiretroviral drugs: Monitoring

and analysis plan of the Antiretroviral Pregnancy Registry. Pharmacoepidemiol Drug Saf 13: 537‐545. 27. Scheuerle A, Covington D (2004) Clinical review procedures for the Antiretroviral Pregnancy Registry.

Pharmacoepidemiol Drug Saf 13: 529‐536. 28. Joint CIOMS/WHO Working Group (2005) Drug development research and pharmacovigilance in resource‐

poor countries. Available: http://www.cioms.ch/frame_whats_ new.htm. Accessed 10 August 2008. 29. Huff‐Rouselle M, Simooya O, Kabwe V, Hollander I, Handema R, et al. (2007) Pharmacovigilance and new

essential drugs in Africa: Zambia draws lessons from its own experiences and beyond. Glob Pub Health 2: 184‐203.

30. Tilson J, Watts J, Covington D (2006) Effective use of supplemental data in pregnancy exposure registries [presentation]. 22nd International Conference on Pharmacoepidemiology & Risk Management; 24‐27 August 2006; Lisbon, Portugal.

31. Clark R, Kumemura M, Makori N, Nakata Y, Bernard F, et al. (2006) Artesunate: Developmental toxicity in monkeys [abstract]. 46th Annual Meeting of the Teratology Society; 24‐29 June 2006; Tucson, Arizona, United States.

32. Clark RL, White TE, A Clode S, Gaunt I, Winstanley P, et al. (2004) Developmental toxicity of artesunate and an artesunate combination in the rat and rabbit. Birth Defects Res B Dev Reprod Toxicol 71: 380‐394.

33. Longo M, Zanoncelli S, Manera D, Brughera M, Colombo P, et al. (2006) Effects of the antimalarial drug dihydroartemisinin (DHA) on rat embryos in vitro. Reprod Toxicol 21: 83‐93.

34. Longo M, Zanoncelli S, Della Torre P, Rosa F, Giusti A, et al. (2008) Investigations of the effects of the antimalarial drug dihydroartemisinin (DHA) using the Frog Embryo Teratogenesis Assay‐Xenopus (FETAX). Reprod Toxicol 25: 433‐441.

35. World Health Organization (2007) Assessment of the safety of artemisinin compounds in pregnancy: Report of two joint informal consultations convened in 2006. WHO/ CDS/MAL/2003.1094. Available: http:// www.who.int/tdr/publications/publications/ pdf/artemisinin_compounds_pregnancy.pdf. Accessed 10 August 2008.

36. White TEK, Clark RL (2008) Sensitive periods for developmental toxicity of orally administered artesunate in the rat. Birth Defects Res B Dev Reprod Toxicol. E‐pub 9 July 2008.

37. World Health Organization (2004) World health statistics. Basic demographic and socio‐economic indicators: Total fertility rate, female, for latest available data. WHO Global Health Atlas. Available: http://www.who.int/ GlobalAtlas/. Accessed 10 August 2008.

38. Say L, Donner A, Gulmezoglu AM, Taljaard M, Piaggio G (2006) The prevalence of stillbirths: A systematic review. Reprod Health 3: 1.

39. McCombie SC (1996) Treatment seeking for malaria: A review of recent research. Soc Sci Med 43: 933‐945. 40. Hamer DH, Ndhlovu M, Zurovac D, Fox M, Yeboah‐Antwi K, et al. (2007) Improved diagnostic testing and

malaria treatment practices in Zambia. JAMA 297: 2227‐2231. 41. Reyburn H, Mbakilwa H, Mwangi R, Mwerinde O, Olomi R, et al. (2007) Rapid diagnostic tests compared

with malaria microscopy for guiding outpatient treatment of febrile illness in Tanzania: Randomised trial. BMJ 334: 403.

75


Table S1. Number of pregnancies needed in each group to detect various relative risks for

pregnancy outcomes with different prevalence.

Figure S1. Pregnancy Exposure Registry Sample Size calculation according to the exposed to

unexposed ratio for major malformation.

Found at doi:10.1371/journal. pmed.0050187.sd001 (596 KB DOC).

Chapter 4 Appendix

76

Table S1. Number of pregnancies needed in each group to detect various relative risks for pregnancy

outcomes with different prevalence (Ratio of exposed to unexposed of 1:4, Power 80% and one‐

sided α=0.05). These sample size calculations are based on a one‐sided approach because pregnancy

exposure registries are designed to detect safety signals rather than to examine potential protective

effects. Based on the formula for Cohort design described in Strom’s Pharmacoepidemiology[31]:

N=1/[p(1‐R)]2x [Z 1‐α√((1+1/k)U(1‐U))+Z1‐β√(pR(1‐Rp)+(P(1‐P))/k)]2 where p is the incidence of disease

in unexposed; R is the minimum relative risk to detect; k is the ration of unexposed controls to

exposed and U=(Kp+pR)/(k+1). These estimates do not include loss to follow‐up and do not account

for the expected 16‐18% of pregnancies that may not result in a live birth. The latter needs to be

considered for outcomes such as major malformation and specific birth defects.

*This is the number of exposed pregnancies required against a known background rate that was

derived using PASS software for cohort studies [Reference: Hintze J (2006) NCSS, PASS and GESS. In:

NCSS, editor. Kaysville, Utah].

Relative Risk

Exposed to Unexposed Ratio of 1:1

Exposed to Unexposed Ratio of 1:4

Against Background Rate*

Number Exposed

Number Unexposed

Total Number Exposed

Number Unexposed

Total Number Exposed

Miscarriage (p=15%)

1.2 1,897 1,897 3,794 1,168 4,672 5,921 1,098

1.5 335 335 670 202 808 1,011 191

2 95 95 190 56 224 280 54

5 7 7 15 4 16 20 5

10 NA NA NA NA NA NA NA

Stillbirths (p=3%)

1.2 10,991 10,991 21,981 6,747 13,493 20,712 5,490

1.5 1,988 1,988 3,976 1,192 4,767 5,959 955

2 591 591 1,182 343 1,372 1,715 268

5 69 69 139 37 146 183 26

10 22 22 45 11 45 56 8

Major Malformations (p=2%)

1.2 16,674 16,674 33,348 10,233 40,934 51,884 8,236

1.5 3,022 3,022 6,043 1,810 7,241 9,052 1,432

2 901 901 1,802 522 2,090 2,612 402

5 108 108 216 57 227 284 39

10 36 36 73 18 72 90 11

Specific Birth Defect (p=0.1%)

1.2 340,629 340,629 681,259 208,978 835,913

1,059,540 164,712

1.5 61,922 61,922 123,845 37,067 148,270 185,337 28,636

2 18,571 18,571 37,143 10,748 42,992 53,740 8,038

5 2,317 2,317 4,634 1,206 4,824 6,030 777

10 836 836 1,673 409 1,638 2,047 229


77

Figure S1. Pregnancy Exposure Registry Sample Size calculation according to the exposed to

unexposed ratio for major malformation (assuming a background rate of 2%; Power 80% and one‐

sided α=0.05).

*This is the number of exposed pregnancies required against a known background rate that was derived using PASS

software for cohort studies [Reference Hintze J (2006) NCSS, PASS and GESS. In: NCSS, editor. Kaysville, Utah].

901 650 565 522 497 479 467 457 450 444 402

901 1300 16962090

24832875

32673659

40514443

1802 19512261

2612

2979

3354

3734

4117

4501

4887

0

1000

2000

3000

4000

5000

6000

7000N

umbe

r of P

regn

anci

es

Ratio of Unexposed-control per Exposed pregnancy

Unexposed Exposed

79

Chapter5

ProbabilisticRecordLinkageforMonitoringtheSafetyofArtemisinin‐BasedCombinationTherapyintheFirstTrimesterofPregnancyinSenegal

StephanieDellicour1,2,3,PhilippeBrasseur4,PerThorn5,OumarGaye6,PieroOlliaro7,8,MalikBadiane9,AndyStergachis10,FeikoO.terKuile1


United Kingdom, 2 Kenya Medical Research Institute (KEMRI), Center for Global Health

Research (CGHR), Kisumu, Kenya, 3 Graduate School, Academic Medical Centre,

Amsterdam, The Netherlands, 4 Institut de Recherche pour le Développement (IRD),

Dakar, Sénégal, 5 Thorn IT Services Limited, London, United Kingdom, 6 Service de

Parasitologie, Faculté de Médecine, Université Cheikh Anta Diop, Dakar, Senegal 7

UNICEF/UNDP/WB/WHO Special Programme for Research & Training in Tropical Diseases

(TDR), Geneva, Switzerland, 8 Centre for Tropical Medicine and Vaccinology, Nuffield

Department of Medicine, University of Oxford, Churchill Hospital, Oxford, United

Kingdom, 9 District Médical d'Oussouye, Oussouye, Sénégal, 10 Departments of

Epidemiology and Global Health, Global Medicines Program, School of Public Health,

University of Washington, United States of America

Drug Safety 2013, 36:505–513

Copyright© Adis

Chapter 5

80

AbstractBackground There are insufficient data on the safety in early pregnancy of the artemisinins,

a new class of antimalarials. Assessment of drug teratogenicity requires large sample sizes for

an adequate risk‐benefit assessment. There is currently limited pharmacovigilance

infrastructure in malaria‐endemic countries. Monitoring drug safety in early pregnancy is

especially challenging, as it requires early pregnancy detection to assess any potential

increased risk of miscarriage, prospective follow‐up to reduce recall and survival biases, and

accurate data on gestational age assessment. Record linkage approaches for pregnancy

pharmacovigilance using routinely generated health records could be a pragmatic and cost‐

effective approach for pharmacovigilance in early pregnancy, but has not been evaluated in

resource‐poor settings.

Objective Our objective was to assess the feasibility of record linkage using routinely

collected healthcare data as a pragmatic means of monitoring the safety in early pregnancy of

artemisinin‐based combination therapies (ACTs) in Senegal.

Methods Data (2004–2008) from paper‐based registers from outpatient clinics, antenatal care

services (ANC) and the delivery unit from the St Joseph dispensary in Mlomp, south‐western

Senegal, were entered into databases. Record linkage based on a probabilistic matching

approach was used to identify pregnancies exposed to ACTs in the first trimester of pregnancy.

Two record linkage software packages (Link‐Plus and FRIL) were compared and output data were

reviewed independently by two investigators.

Results Information on 685 pregnancies was extracted, 536 of which were from the geographic

catchment area and eligible for record linkage; 94.6 % of them resulted in live births, 2.6 % in

stillbirths and 2.8 % in miscarriages. Major congenital malformations were identified in 1.6 % of

births. Seventy‐three and 75 true matches between pregnancy outcome and the outpatient

treatment registers were identified by two different record linkage software packages,

respectively. Record linkage identified seven exposures to ACTs in the first trimester, all of which

resulted in normal live‐births.

Conclusion Probabilistic record linkage is a potentially cost‐effective method to assess the safety

of antimalarials in early pregnancy in resource‐constrained settings to assess increased risk of

overall birth defects, and stillbirths in settings with good existing health records and well defined

target populations.

Probabilistic Record linkage to assess antimalarials safety in pregnancy

81

BackgroundIn the last decade, Senegal and other malaria‐endemic countries changed their first‐line treatment

policy for malaria to artemisinin‐based combination therapy (ACT). The artemisinin class of

antimalarials all have embryotoxic effects at low dose ranges in all animal species studied [1, 2]

and the World Health Organisation (WHO) does not recommend their use for non‐severe

malaria in the first trimester as there is insufficient information about their safety in humans.

Because of their widespread use, many women in endemic countries risk inadvertent exposure

early in pregnancy when they are either unaware of their pregnancy or do not report being

pregnant. To date, the data from 359 well documented exposures in the first trimester suggests

that the benefits outweigh the potential safety concerns but more data from a wider range of

malaria‐endemic countries are required to provide adequate reassurance [4–7].

Passive mechanisms of spontaneous adverse drug effects reporting are inadequate to detect drug‐

induced fetal risks or lack of such risks [3]. Different prospective study designs, including

pregnancy registers, are used to monitor safety of medication used during pregnancy in the post‐

marketing phase [4, 5]; however, these require considerable resources and a well functioning

health system and records infrastructure, which are often unavailable in resource‐constrained

settings [3, 6].

In the last few years there have been an increasing number of pregnancy postmarketing studies

using record linkage approaches in developed countries [7–10], which may also have application

in resource‐limited countries to evaluate the teratogenicity of a drug. It enables rapid evidence

generation by using existing healthcare data collected prospectively, while eliminating potential

recall bias. Often data on drug exposure, prenatal and pregnancy outcomes are available from

multiple linkable data sources. In industrialised countries, this information can be derived from

medical records and automated databases, including insurance claims [11]. A recent study

showed the feasibility of using record linkage in resource‐constrained settings to assess adverse

reactions to antiretroviral therapy [12], but such approaches have not yet been applied to

assess drug safety in pregnancy in such settings.

Record linkage requires access to long‐term, comprehensive and stable population datasets

with personal identifiers. In situations where unique identifiers are available, a deterministic

record linkage technique, which involves exact matching, can be applied. Probabilistic record

linkage is used in situations where there is no universal unique personal identifier and is based on

the assessment of similarity between pairs of records, allowing for a level of error (such as

typographical or spelling differences) in matching variables [13–15]. The mathematical framework

underlying probabilistic record linkage was first formulated by Newcombe and Kennedy in 1962

[16] and further developed by Fellegi and Sunter [13] in 1969. A linkage probability score is

computed for each pair of records based on the sum of the probability of agreement for each

matching variable. Probability of agreement for each matching variable reflects the probability

that the values match by chance based on the frequency of that value occurring in the

datasets. Thresholds for the combined linkage probability score are set to determine which pairs

are a true match, those that are potential matches and those that are not a match.

We report the results of a study to determine the feasibility of record linkage as a pragmatic

approach to retrospectively examine the potential risk associated with inadvertent exposure to

Chapter 5

82

ACTs in early pregnancy using data generated through routine clinical practice in a rural mission

dispensary in southern Senegal.

Methodology

StudySettingThis study was conducted in 2009 in a mission dispensary based in Mlomp, a rural village of

approximately 8,000 inhabitants in the District of Oussouye, Casamance, south‐ western Senegal.

Malaria is meso‐endemic in this area, occurs year round and peaks during the rainy season (July

to December). A recent study showed that malaria transmission intensity in southern Senegal has

been decreasing significantly in the past 15 years [17].

Mlomp has had a private dispensary operated by French catholic nuns since 1961. The dispensary

offers outpatient services, antenatal care service (ANC) once a week, and has a delivery unit.

Nearly all pregnant women attend ANC, and health facility deliveries have increased from 50 %

in 1961 to 99 % in 1999 [18, 19].

The setting is potentially well suited for record linkage studies. The clinic serves a stable, well

defined population and the dispensary registers for ANC, delivery, child welfare clinic and general

outpatient visits have been meticulously kept since 1993 [20]. Almost all antimalarials used are

provided by the clinic because there is no external source of drugs within the study area; the closest

pharmacy or drug store is in the nearest town, Oussouye, 10 km away, with limited options for

public transport.

DataCollectionThe general outpatient, antenatal, pregnancy complications/miscarriage and delivery registers from

Mlomp dispensary covering the period 2004–2008 were digitalised using a digital camera (Canon

EOS 450D with external flash and tripod). The data from the digital images of the registers were

subsequently entered into Microsoft® Excel spreadsheets. For the purpose of this feasibility study,

only entries for women of childbearing age (15–49 years) with a prescription for an ACT between

January 2004 and December 2007 were extracted from the outpatient register. All treatment

information, including treatment given at the time of antenatal visits, is recorded in the outpatient

register. All recorded pregnancy outcomes between May 2004 and September 2008 were entered

from the pregnancy complication and the delivery registers (Table 1). Information on newborn

abnormalities was recorded in the delivery register based on observation by the midwife attending

the delivery. Data from the outpatient register were double entered and the entries for the delivery

register were compared with data extracted for a previous study [20].

RecordLinkageMethodProbabilistic record linkage based on first names, surname, address and year of birth was used to

link information from the different registers because they did not contain unique patient

identifiers (Table 2). Two stand‐alone software packages were used and compared: (i) Registry

PlusTM Link Plus (Emory University, v.2.0) [21] (henceforth referred to as ‘Link‐Plus’), a royalty‐free

probabilistic record linkage program developed at the Division of Cancer Prevention and Control

of the Centers for Disease Control and Prevention (CDC), USA; and (ii) Fine‐grained Record Linkage

software (FRIL v. 2.1.5), a free open‐ source tool developed by Emory University and the CDC

[22, 23].

Probabilistic R

ecord lin

kage to

assess antim

alarials safety in

pregn

ancy

83

Table 1 Su

mmary o

f data availab

le in each

health

registe

r from th

e Mlomp Disp

ensary in

Senegal

G

eneral outpatient register A

ntenatal register P

regnancy complication register

Delivery register

Identifiers and demographic

• First andlast

names

•First

andlast

names

•First

andlast

names

•First

andlast

names

• A

ge

• Sex • A

ge

• Address

• Date of birth/age

• Marital status

• Date of birth/age

• Address

• A

ddress • M

arital status

• Em

ployment

Clinical

informatio n

• Presenting sym

ptoms

SeeA

NC

row4

• Hospitalisation dates

• Drugs

providedduring

delivery and care provided

• Diagnosis

• Treatm

ent

• Com

ments

• Care and drugs provided

or post-partum to m

other and baby

Obstetric history

None

• No. of previous pregnancies

• No. of previous deliveries

• No. of pregnancies

• No. of live-births

• No. of stillbirths

• No. of m

iscarriage

• No. of children alive

• Date of delivery

• Nam

e of child

• Sex • P

lace of delivery

• Child status

AN

C

None

• Date of A

NC

visit

• LM

P

• Weight, H

b, fundal height, blood pressure for each visit

• Risk factors

• Tetanus vaccination

None

• Date of A

NC

visits • W

eight, Hb, fundal height,

blood pressure for each visit

• Tetanus vaccination

AN

C antenatal care, L

MP

last menstrual period

Chapter 5

84

Following agreement on the matching criteria, two investigators conducted the record linkage

independently. Any discrepancies in results between investigators were compared and discussed

until a consensus was reached.

Table 2 Description of matching variables from the delivery register

Matching Variables Discriminating power/ # possible values

Missing value (with

ANC info)a

Variable limitations

First Name 230 0 • Variation in spelling/ or spelling errors • Truncated/nicknames • Hyphenated /multiple names • Inconsistent and interchangeable use and recording of traditional and

Christian first and second names

Surname 73 0.1% • Variation in spelling/ or spelling errors • Commonly changes for women of child bearing age after

marriage/divorce

Address 29 33% (10%) • Changes over time • Commonly changes for women of child bearing age after

marriage/divorce • Address only consists of neighborhood name which is not very

discriminating

Age or Year of birth 33 10% (4%) • Inaccuracy in age or date of birth is common in the study population a Manual search of the corresponding records in the ANC register enabled completing certain missing values for age and

address.

A variety of matching methods are available for each matching variable to take into account

different spelling, recording or typographical errors. To identify the optimal parameters that

produce the best linkage results, different linkage setups were compared by varying the weights

given to the matching variables. The adequacy of the linkage result was assessed by the

distribution of matches, uncertain and non‐matches according to the linkage score, the number of

true matches identified and the positive predictive value (PPV) for the selected threshold. In Link‐

Plus, the matching score threshold was set to 0 (minimal threshold option) in order to derive

histograms from the linkage output for each matching set up (this was not an option in FRIL).

True matches were defined as records with matching first names, surname, and address, allowing

for some misspelling and year of birth within 5 years of each other. Pairs were assigned as

uncertain matches if first name(s) matched, year of birth was within 10 years of each other but the

address and/or surname did not match. Surname and address can change over time, and this is

especially common among women of childbearing age when they get married. All other pairs of

records were considered as non‐match records. Details of the selected linkage setups are

provided in Table 3. For information on the linkage optimisation procedures, see appendices I

and II. Deduplication of the delivery register was performed using the deduplication function of

Link‐Plus, as some women could have multiple pregnancies within the 4‐year study period, and

to enable one‐to‐many matching.

AnalysisDescriptive statistics were used to determine the proportion of women with inadvertent exposure

to ACTs during the first trimester of pregnancy using the data obtained by linking the records

from the outpatient and the delivery/ pregnancy complication registers. Characteristics of

women included and excluded from the record linkage were compared using the Pearson Chi‐

square statistic. A first trimester exposure was defined as an ACT prescription (either artesunate‐


85

amodiaquine or artemether‐lumefantrine) provided to any woman in 2–14 weeks (inclusive) of

pregnancy. Data on last menstrual period (LMP) were not collected in either the outpatient or the

delivery registers. Information on the estimated gestational age was only available in the delivery

register and based on the assessment by the midwife at birth. This was categorised as ‘term’ or

‘pre‐term’. For pre‐term deliveries, the gestational age at delivery was specified in months from

LMP in the delivery records, whereas for all ‘term’ deliveries, the specific gestational age was not

recorded other than the notation that they were ‘term’ births. It was therefore set at 40 weeks

from LMP for the purpose of the analysis. The gestational age at the time of exposure was

derived from the date of the ACT prescription in the outpatient register and the estimated date

of conception calculated from the gestational age assessment at birth. Analysis was done in SPSS

version 18.

Table 3 Description of linkage parameters in Link‐Plus and FRIL

Link‐Plus set up FRIL set upb

Linkage Variablesa

M‐ probabiltyc

Matching methodd Weight Matching methodd

First Name 0.97 First Name (J‐W) 32 Edit distance (0.1‐0.4)

Second Name 0.7 Generic String 10 Edit distance (0.1‐0.4)

Surname 0.85 Last Name (J‐W) 25 Edit distance (0.1‐0.4)

Address 0.85 Generic String 23 Edit distance (0.1‐0.4)

Year of birth 0.2 Generic String 10 Numeric distance (+/‐ 5 years) a all variables are standardized b m‐probability determines the reliability of the variable, it ranges from 0 (unreliable) to 1 (very reliable). cMatching Methods descriptions:

J‐W: Jaro‐Winkler Metric is a string comparator which measures the partial agreement between two strings accounting for random insertion, deletions, and transpositions.

Generic String and Edit‐distance: incorporates partial matching to account for typographical errors and calculates the number of operations (insertion, deletion, or substitution of a single character) needed to transform one string into the other. The approve and disapprove levels need to be specified in FRIL for edit distance function.

Numeric distance: allows users to specify a range of values that will have a non‐zero match score. dFRIL doesn’t calculate m‐probability as in Link‐Plus but enables the user to set weights for each matching variables summing to 100. To compute the total score for a pair, the weight of each matching variable is multiplied by the value returned by chosen edit distance function and all these values are added together.

Results

DescriptionBetween May 2004 and September 2008, 685 pregnancy outcomes were captured from the

pregnancy complication and delivery registers. Women who attended their ANCs outside of the

Mlomp catchment area (n = 149) were excluded from the record linkage because no outpatient

treatment records were available from these other clinics.

Their characteristics did not differ significantly (at 5 % significance level) from other women living in

the catchment area, except that they were more likely to be primiparous women and to be slightly

younger. Figure 1 depicts the number of records that contributed to the probabilistic matching. Of

the pregnancies included, 94.6 % resulted in live births, 2.6 % in stillbirths and 2.8 % in miscarriages

(Table 4). Overall, 11 cases of birth defects (1.6 %) were captured in the delivery register (Table 5).

Chapter 5

86

Fig. 1 Flow diagram for inclusion in record linkage, matching result and resulting ACT pregnancy exposure. Note that of the 451 women, 372 had only one pregnancy during the study period (2004–2008), 73 had two pregnancies and six had three pregnancies during the 4‐year period

Table 4 Characteristics of 536 eligible pregnancies from the delivery andpregnancy complication registers (2004–2008) St Joseph Dis‐ pensary Mlomp, Senegal

Table 5 Description of congenital abnormalities identified from the delivery registers (2004–2008) in St Joseph Dispensary Mlomp, Senegal

Characteristic N (%)a

Age in years

Mean (SD) 29.2 (6.6)

Range 15–47

Parity

0 94 (17.5)

1 87 (16.2) 2? 355 (66.2)

Range 0–13

Number reporting previous stillbirth 26 (4.9)

Number reporting previous miscarriage 55 (10.3)

Pregnancy outcomes Live births 507 (94.6)

Stillbirths 14 (2.6)

Miscarriages 15 (2.8)

Preterm at delivery 26 (4.9)

a Unless otherwise indicated

Congenital abnormalitiesa Number

Anophthalmia 1

Ambiguous genitalia 1

Anencephaly 3

Down syndrome 6

Club foot 2

Hydrocephalus 1

Imperforate anus 1

Unspecified malformations 1

aSome cases had more than one abnormality. There were 16abnormalities among 11 births: two cases had both Down syndrome and anencephaly, one case of Down syndrome also had a club foot, one case of anophthalmia also had a club foot and one case with Down syndrome also had hydrocephalus

ProbabilisticMatchingThe optimum set up with highest PPV (86 %) and highest number of matches was accomplished by

including second names. Figure 2 depicts the distribution of match, uncertain and non‐match for

this set up. The relatively high number of false negatives (n = 13) and false positives (n = 8)

suggests clerical review would still be required for pairs with a linkage probability score between 6

and 9.


87

Fig. 2 Histogram of the optimum set up depicting the distribution of matches, uncertain and non‐matches according to the probability score derived from Link‐Plus

Potential cut off

Seventy‐one and 75 matched pairs between the pregnancy outcome registers and the ACT

treatment data from the outpatient registers were detected using Link‐Plus and FRIL, respectively.

The four additional matches identified through FRIL were not detected through Link‐Plus as it

only allows one‐to‐many matching. After running another round of matching in Link‐Plus using

the non‐match records only and adjusting the matching variables weight, an additional two

matches were identified (total of 73 matched pairs). The other two treatment records were

matched by Link‐Plus with incorrect records from the delivery register and were not compared

again to the correct delivery records. The two cases missed in Link‐Plus were not exposed during

the pregnancy period.

ArtemisininExposureinPregnancyOut of the true matches, 11 of the 536 pregnancies (2 %) had evidence for ACT exposure during

pregnancy, seven during the first trimester (1 %), three of whom were 4–10 weeks (the

projected embryo‐sensitive period from animal models) pregnant (0.6 %) and four in the second or

third trimester. There was no exposure in the month prior to the estimated conception dates.

None of the women exposed during pregnancy had an adverse pregnancy out‐ come, preterm

birth or a baby with malformations.

DiscussionThe findings from this feasibility study suggest that record linkage using routine healthcare data is

feasible in resource‐constrained settings with a relatively well defined catchment population.

Among the 451 women with pregnancy outcomes, 75 records from the delivery register were

matched to an ACT treatment in the outpatient registers, although only 11 were ACT exposures

during the pregnancy period; the other matches represent ACT treatment received before or

after pregnancy. Matching was feasible despite the limited discriminating power resulting from the

lack of variance among two of the five matching variables due to clustering of commonly used

surnames, lack of address details and the lack of precise dates of birth for many individuals.

Chapter 5

88

Furthermore, the interchangeable use of traditional and Christian first names by single individuals

was very common and often one was recorded, but seldom both.

Although there was no other source of medication in Mlomp and the nearby surrounding areas,

the risk of exposure misclassification remains, as women could have obtained drugs from

relatives or friends, could have travelled during their pregnancy and obtained drugs elsewhere or

the drugs could have been prescribed and recorded in the register but never actually taken by the

patient. The limited detail on gestational age (categorised as term or pre‐term with gestational

month) could also cause misclassification for timing of exposure. Furthermore, there could have

been errors or omissions during recording of data in the health registers and at time of data entry.

We attempted to minimise the latter by using double data entry for the general outpatient register

and by comparing the entries for the delivery register with data extracted for a previous study [20].

The prevalence of birth defects was within the expected range for major malformations detectable

at birth by surface exams by non‐specialists without special training for newborn examination.

Although the background prevalence of birth defects in developing countries is unknown,

extrapolation from birth defect prevalence found in industrialised countries suggests that the

defect prevalence detectable at birth would be around 1 % after exclusion of defects of genetic

aetiology and heart defects, which require specialist assessment to be diagnosed. The background

rate of stillbirth (2.3 %) was also within the projected range for Senegal (2.7 %) and sub‐Saharan

Africa (3.2 %) [24]. As expected, the proportion of miscarriages (2.5 %) was much lower than the

risk of 12–15 % typically quoted for miscarriages because most pregnant women do not present for

their first ANC visit before the second trimester. A previous record linkage study from Mlomp

dispensary showed that half of all pregnant women presented for their first ANC visit after 20 weeks

gestation, thus only allowing the prospective recording of miscarriages relatively late in the second

trimester (20–28 weeks) [20].

The two record linkage software programs used had comparable performance. The main advantage

of FRIL was that it included a matching method for numeric variables, which allowed variation in

numeric values (i.e. year of birth could be within 5 years of each other and considered a match).

There is no function for numeric variables in Link‐Plus, and year of birth had to be treated as a

string variable. Many‐to‐many pairwise comparison is possible in FRIL, whereas only one‐to‐many is

enabled in Link‐Plus (many‐to‐many will be available in the next version, 3.0 [25]). This was

highlighted by the two missed true matches identified through FRIL but not Link‐Plus described

above. Limitations for FRIL included the more limited flexibility to export datasets after linkage

(only matched pairs can be exported). Link‐Plus, on the other hand, enables the export of all pairs

reviewed manually (matches, uncertain and non‐matches), which then allows more complete

secondary analysis and graphical depiction of results.

Ruling out the teratogenic risk of a drug requires a very large sample size, as teratogens usually

induce specific patterns of birth defects that occur rarely in the general population. We have

shown previously that 10,748 well characterised exposures and four times as many unexposed

controls would be required to exclude a doubling of risk of a specific birth defect that occurs at a

frequency of 0.1 % [11]. Obtaining reliable data on early pregnancy drug exposures is challenging.

McGready and colleagues recently reported 64 well documented first trimester exposures to

ACTs after reviewing 25 years of data on 48,426 pregnancies [26]. This was enough to exclude


89

a doubling of risk of miscarriage for first trimester exposures [27]. To obtain a similar level of

reassurance for major malformations, data from several sentinel sites over several years are

required. Collaborating pregnancy registry sites have been set up by WHO [28] and the Malaria

in Pregnancy consortium [29] for this purpose.

Record linkage studies require minimal staff (for data extraction, data entry and analysis) and

resources (i.e. digital camera and a few computers for data entry). In settings with electronic

medical records, only staff for data management and analysis would be required. However,

assessment of the risk of miscarriage and specific birth defects, such as congenital heart defects,

requires dedicated studies. A ‘miscarriage’ endpoint requires an observational cohort study

involving women of childbearing age to capture data as early as possible in their pregnancy, while

assessment of specific rare birth defects would probably require nested case control or case

cohort studies in settings where drug exposure information is captured routinely. Additional data

from sites where record linkage is a possibility would greatly contribute to the achievement of an

adequate sample size to guide policy makers. Requirements for such sites include availability of

reliable and comprehensive medical records for treatment, pregnancy and maternity services, in a

relatively stable population where healthcare is provided in a central location with a high level

of health facility deliveries and limited availability of the drug of interest outside the central health

facility. Settings such as private agricultural estates (e.g. tea, coffee, sugar or cotton plantations),

where employees and their families get healthcare centrally and routine healthcare data are

typically recorded, could be used for similar record linkage studies. Data from health insurance

schemes, where unique identifiers are available, could be used with a combination of deterministic

and probabilistic record linkage [30, 31].

ConclusionOur findings suggest that record linkage to assess drug safety in pregnancy in resource‐

constrained settings is feasible for assessment of stillbirths and major congenital malformations

detectable by surface examination. Tapping into readily available data sources of sites adequate

for record linkage would greatly contribute to the high numbers needed to provide adequate

reassurance for ACT use in the first trimester of pregnancy.

Acknowledgments

This work was made possible by the dedication of the healthcare personnel of St Joseph Dispensary

in Mlomp and the help of Sister Marie Joelle.

Author’s contributions SD, FtK, OG and PO contributed to the concept of the project. SD and FtK developed the protocol with contributions from PO, PB and AS. SD and PT conducted the data collection and record linkage analyses. PB supervised the data collection. SD and FtK wrote the first draft of the manuscript; all authors reviewed and revised the final version.

References1. WHO. Assessment of the safety of artemisinin compounds in pregnancy. Report of two informal

consultations convened by WHO in 2002. Geneva: WHO; 2003.

2. WHO, TDR. Assessment of the safety of artemisinin compounds in pregnancy: report of two joint informal

consultations convened in 2006 by The Special Programme for Research and Training in Tropical Diseases

(TDR) sponsored by UNICEF/UNDP/World Bank/WHO. Geneva: WHO; 2006.

3. Kennedy DL, Uhl K, Kweder SL. Pregnancy exposure registries. Drug Saf. 2004;27(4):215–28.

Chapter 5

90

4. Ka llen BA. Methodological issues in the epidemiological study of the teratogenicity of drugs. Congenit

Anom (Kyoto). 2005; 45(2):44–51.

5. Hernandez‐Diaz S, Hernan MA, Meyer K, Werler MM, Mitchell AA. Case‐crossover and case‐time‐control

designs in birth defects epidemiology. Am J Epidemiol. 2003;158(4):385–91.

6. Adam MP, Polifka JE, Friedman JM. Evolving knowledge of the teratogenicity of medications in human

pregnancy. Am J Med Genet C Semin Med Genet. 2011;157(3):175–82.

7. Andrews EB, Tennis P. Promise and pitfalls of administrative data in evaluating pregnancy outcomes.

Pharmacoepidemiol Drug Saf. 2007;16(11):1181–3.

8. Charlton RA, Weil JG, Cunnington MC, Ray S, de Vries CS. Comparing the General Practice Research

Database and the UK Epilepsy and Pregnancy Register as tools for postmarketing teratogen surveillance:

anticonvulsants and the risk of major congenital malformations. Drug Saf. 2011;34(2):157–71.

9. Colvin L, Slack‐Smith L, Stanley FJ, Bower C. Pharmacovigi‐ lance in pregnancy using population‐based

linked datasets. Pharmacoepidemiol Drug Saf. 2009;18(3):211–25.

10. Irvine L, Flynn RW, Libby G, Crombie IK, Evans JM. Drugs dispensed in primary care during pregnancy:

a record‐linkage analysis in Tayside, Scotland. Drug Saf. 2010;33(7):593–604.

11. Dellicour S, Ter Kuile FO, Stergachis A. Pregnancy exposure registries for assessing antimalarial drug

safety in pregnancy in malaria‐endemic countries. PLoS Med. 2008;5(9):e187.

12. Corbell C, Katjitae I, Mengistu A, Kalemeera F, Sagwa E, Mabirizi D, et al. Records linkage of electronic

databases for the assessment of adverse effects of antiretroviral therapy in sub‐ Saharan Africa.

Pharmacoepidemiol Drug Saf. 2012;21(4):407–14.

13. Fellegi IP, Sunter AB. A theory for record linkage. J Am Stat Assoc. 1969;64(328):1183–210.

14. Jaro MA. Probabilistic linkage of large public health data files. Stat Med. 1995;14(5–7):491–8.

15. Meray N, Reitsma JB, Ravelli AC, Bonsel GJ. Probabilistic record linkage is a valid and transparent tool to

combine dat‐ abases without a patient identification number. J Clin Epidemiol. 2007;60(9):883–91.

16. Newcombe H, Kennedy J. Record linkage: making maximum use of the discriminating power of identifying

information. Commun ACM. 1962;5(11):563–6.

17. Brasseur P, Badiane M, Cisse M, Agnamey P, Vaillant MT, Ol‐ liaro PL. Changing patterns of malaria

during 1996–2010 in an area of moderate transmission in southern Senegal. Malar J. 2011; 10:203.

18. Enel C, Pison G, Lefebvre M. De l’accouchement traditionnel a l’accouchement moderne au Senegal.

Cahiers Sante. 1993;3: 441–6.

19. Pison G, Kodio B, Guyavarch E, Etard J‐F. La mortalite ma‐ ternelle en milieu rural au Senegal.

Population. 2000;55(6): 1003–18.

20. Olliaro PL, Delenne H, Cisse M, Badiane M, Olliaro A, Vaillant M, et al. Implementation of intermittent

preventive treatment in pregnancy with sulphadoxine/pyrimethamine (IPTp‐SP) at a district health

centre in rural Senegal. Malar J. 2008;7:234.

21. US Centers for Disease Control and Prevention. Registry PlusTM

22. Link Plus technical information and installation. Atlanta (GA): CDC; 2006 [online].

http://www.cdc.gov/cancer/npcr/tools/registry plus/lp_tech_info.htm. Accessed 24 Apr 2013.

23. Jurczyk P. Fine‐grained record integration and linkage tool. Open source record linkage software. 2.1.4 edn.

Atlanta (GA), 2009.

24. Jurczyk P, Lu JJ, Xiong L, Cragan JD, Correa A. Fine‐grained record integration and linkage tool. Birth

Defects Res A Clin Mol Teratol. 2008;82(11):822–9.

25. Stanton C, Lawn JE, Rahman H, Wilczynska‐Ketende K, Hill K. Stillbirth rates: delivering estimates in 190

countries. Lancet. 2006;367(9521):1487–94.

26. NPCR. Link Plus Development Priority List (updated January 26, 2011). 2011.

http://www.cdc.gov/cancer/npcr/tools/registryplus/ lp_features.htm. Accessed 7 Feb 2012.

27. McGready R, Lee SJ, Wiladphaingern J, Ashley EA, Rijken MJ, Boel M, et al. Adverse effects of falciparum

and vivax malaria and the safety of antimalarial treatment in early pregnancy: a population based

study. Lancet Infect Dis. 2012;12(5):388–96.


91

28. Desai M, Dellicour S. Effects of malaria and its treatment in early pregnancy. Lancet Infect Dis.

2012;12(5):359–60.

29. Gomes M. New global pregnancy register established. 2009 [online].

http://www.who.int/maternal_child_adolescent/

news_events/news/2009/pregnancy_register/en/index.html. Acce‐ ssed 15 May 2012.

30. MiPc. Malaria in Pregnancy Consortium. 2012. http://www.mip‐ consortium.org/. Accessed 2012.

31. Romitti PA, Watanabe‐Galloway S, Budelier WT, Lynch CF, Puzhankara S, Wong‐Gibbons D, et al.

Identification of Iowa live births in the Agricultural Health Study. Arch Environ Occup Health.

2010;65(3):154–62.

32. Roos LL Jr, Wajda A, Nicol JP. The art and science of record linkage: methods that work with few

identifiers. Comput Biol Med. 1986;16(1):45–57.

92


Text S1. Description of the steps used for data preparation before record linkage.

Text S2. Description of record linkage optimisation using Link‐Plus.


93

TextS1.Descriptionofthestepsusedfordatapreparationbeforerecordlinkage.

Semi‐structuring:

Age was transformed to year of birth as the data covered a 4 year period and some records only included year of birth.

First names with more than one name or hyphenated were parsed into first name, second name, and third name. The order of names was standardized so that Christian name was set as first name and Senegalese names as second/third names. In the study population it is not uncommon for people to have multiple first names (one Christian/European name and a Senegalese name) and the order by which they were used vary by occasion or the preference of the person recording the information.

Standardisation:

Spelling of common names was standardised (e.g. Lucy and Lucie). Standardisation of names was particularly important as the available phonetic algorithms commonly used for record linkage (e.g. Soundex and NYSIIS) are based on English names and would therefore have limited application for French and Senegalese name with non‐English pronunciation.

Abbreviation and nicknames were replaced by full name were possible (e.g. Fatou and Fatoumata or Cons and Constance).

Missingvalues:

Missing values can considerably affect the matching success. Significant missing values were found for age/date of birth (33%) and address (10%) in the delivery registry (see table 2). As data on ANC visits is available in the delivery register (including date of each visit), records with missing information were manually crossed‐check with the ANC registers and missing data extracted were available.

To avoid exposure misclassification, women who attended ANC outside of the catchment areas were excluded (n=149), as these women could have sought treatment and care for malaria elsewhere as well.

Standardfileformat:

After performing the transformations outlined above the dataset files were saved to standard file format (CSV) for import into the record linkage software.

Deduplication

Deduplication of the delivery register was performed as some women could have multiple pregnancies within the 4 years covered and to enable one‐to‐many matching. This was performed using the deduplication function of Link‐Plus. Following identification of duplicates, the files were transformed in order to have one record per individual in SPSS (v18).

Chapter 5 Appendix

94

TextS2‐DescriptionofrecordlinkageoptimisationusingLink‐Plus.With the linking variables fixed, different linkage set ups with varying matching methods and weights for each matching variables were compared to identify the optimal parameters. The linkage variables, matching methods and associated linkage weights are presented in Table i.

Link‐Plus

For optimisation purposes the matching score threshold was set to 0 (minimal cut‐off allowed) in order to produce histograms for each matching set up. Histograms were derived from the distribution of matches, uncertain and non‐matches according to the matching scores for the different set ups. Pairs were assigned as uncertain matches if first name(s) matched, year of birth was within 10 years of each other but the address and/or surname did not match. Surname and address can change over time, the former in case of marriage (which is likely to occur during child bearing years) and address if the person moves home after marriage or for other reasons. Youden indexes were derived for each option by adding the sensitivity and specificity and subtracting 1, to provide an indication of suitability of the proposed cut off values.

Option A: matching variable weights were estimated by the expectation‐maximization (EM) algorithm through maximum likelihood estimation based on the current data. This resulted in 50 true matches, 127 non‐matches and 96 uncertain matches. The histogram (figure i) below shows that the distribution of matches is spread and overlaps with non‐matches between scores 3.2 to 10.7. There is no clear cut off point for assigning true match & non‐match. Setting a cutoff point at 7 (as per the other set ups, Youden index 0.69), the PPV was estimated at 78%.

Figure iv. Option A histogram‐ linkage weights based on expectation‐maximization algorithm

Option B: The matching approach was adjusted by increasing the weight assigned to first name, decreasing the weight assigned to surname, address and the year of birth. This resulted in 67 true matches, 197 non‐matches and 142 uncertain matches with PPV of 82% (Youden index 0.81). This option enables a better discrimination between matches and non‐matches, with potential cut‐off point around a score of 7 (see figure i). The high number of false negatives (n=13) and the high number of false positive (n=8) suggest clerical review would be required between score 5.2 and 8.

0

5

10

15

20

25

30

0

0.3

0.8

1.3

1.9

3.4

3.8

4.2

4.7

5.1

5.4

5.7 6

6.3

6.8

7.1

8.1

8.5

9.4

10.1

10.6

11.1

11.8

12.6

14.6

Number of Pairs

Linkage Probability Score

Match

Non‐Match

Uncertain


95

Figure v. Option B histogram‐ adjusted linkage weights and normalised linkage variables.

Option C: To assess the benefit of using normalised names and address, the same matching Option as B was run using raw data for first name, surname and address. This resulted in much lower number of true matches (n=36), 180 non‐matches and 150 uncertain matches for clerical review. This Option has the lowest PPV (71%, Youden Index 0.82) and the highest number of uncertain matches (41% of pairs reviewed). Figure iii below shows that discriminating power is lost, with a much wider spread of matches and overlapping non‐matches.

Figure vi. Option C histogram‐ adjusted linkage weights and raw linkage variables.

0

5

10

15

20

25

30

0

0.5

1.7

2.1

2.6

2.9

3.2

3.6

4.2

4.6 5

5.3

5.6 6

6.3

6.7

7.1

7.5

7.8

8.3

8.8

9.1

9.4

9.9

10.2

10.9

12.2

13.6

Number of Pairs


Match

Non‐Match

Uncertain

Potential cut off

0

5

10

15

20

25

30

0

0.3

0.8

1.3

1.9

3.4

3.8

4.2

4.7

5.1

5.4

5.7 6

6.3

6.8

7.1

8.1

8.5

9.4

10.1

10.6

11.1

11.8

12.6

14.6

Number of Pairs


Match

Non‐Match

Uncertain

Potential cut off

Chapter 5 Appendix

96

Option D: As the use of multiple/hyphenated names is common in the study population but their use is irregular, the inclusion of second name as a linkage variable was assessed. This resulted in 71 true matches, 219 non‐matches and 93 uncertain matches for clerical review with a PPV of 86% (Youden Index 0.78). Figure iv below shows improved discriminating power.

Figure vii. Option D histogram‐ adjusted linkage weights and normalised linkage variables including middle name.

Table i below summaries the outputs of the different linkage Options. The optimal option should have the least overlap for true and false matches to minimise the number of pairs which have to be reviewed manually. In this case set up D provided the highest number of matches and best positive predictive value (PPV) although set up C has the highest sensitivity.

Table i. Description of linkage parameters and matching result for each set up in Link‐Plus.

Linkage Variables Weight Option A Option B Option C Option D

First Name 0.96 (normalised) 0.97 (normalised) 0.97 (raw) 0.97 (normalised)

Second Name NA NA NA 0.7(normalised)

Surname 0.97 (normalised) 0.85 (normalised) 0.85 (raw) 0.85 (normalised)

Address 0.95 (normalised) 0.85 (normalised) 0.85 (raw) 0.85 (normalised)

Year of birth 0.95 0.2 0.2 0.2

Linkage Output (Score range)

Pairs for Review 269 366 366 383

Match 50 (score 3.2‐14.9) 67 (score 2.4‐13.6) 36 (score 1.7‐12.8) 71 (score 6.2‐14.7)

Uncertain 92 (score 0.1‐14.6) 142 (score 1.7‐10.5) 150 (score 1.4‐10.4) 93 (score 1.1‐11.4 )

Non‐match 127 (score 0‐10.7) 197 (score 0.3‐8.0) 180 (score 0.3‐8.0) 212 (score 0.2‐10.1)

Linkage quality Threshold=7 (YI 0.69) Threshold=7 (YI 0.71) Threshold=7(YI 0.82) Threshold=7 (YI 0.78)

Positive Predictive value 78% 82% 71% 86%

Sensitivity 78% 88% 89% 79%

Specificity 91% 93% 93% 96%

0

5

10

15

20

25

30

0.2

0.9

1.7

3.1

3.5

3.8

4.2

4.5

4.9

5.3

5.8

6.2

6.5

6.8

7.2

7.6

7.9

8.3

8.6

9.4

9.8

10.2

10.6 11

11.4 12

13.1 14

(bla…

Number of Pairs


Match

Non‐Match

Uncertain

Potential cut off

97

Chapter6

ExploringRiskPerceptionandAttitudestoMiscarriageandCongenitalAnomalyinRuralWesternKenya

StephanieDellicour1,2,MeghnaDesai1,3,LindaMason2,BeatriceOdidi1,GeorgeAol4,PenelopeA.Phillips‐Howard2,KaylaF.Laserson3,5,FeikoO.terKuile2

1 Malaria Branch, Kenya Medical Research Institute (KEMRI)/ Centers for Disease Control

and Prevention (CDC) Research and Public Health Collaboration, Kisumu, Nyanza Province,

Kenya, 2 Department of Clinical Sciences, Liverpool School of Tropical Medicine, Liverpool,

Merseyside, United Kingdom, 3 Center for Global Health, Centers for Disease Control and

Prevention, Atlanta, Georgia, United States of America, 4 International Emerging Infection

Branch, Kenya Medical Research Institute/ Centers for Disease Control and Prevention

Research and Public Health Collaboration, Kisumu, Nyanza Province, Kenya Research and

Public Health Collaboration, Kisumu, Nyanza Province, Kenya, 5 Health and Demographic

Surveillance System Branch, Kenya Medical Research Institute/ Centers for Disease

Control and Prevention Research and Public Health Collaboration, Kisumu, Nyanza

Province, Kenya

PLoS One 2013, 8:11

Copyright© This is an open‐access article distributed under the terms of the Creative

Commons Attribution License, which permits unrestricted use, distribution, and

reproduction in any medium, provided the original author and source are credited

Chapter 6

98

Abstract

BackgroundUnderstanding the socio‐cultural context and perceptions of adverse pregnancy outcomes is

important for informing the best approaches for public health programs. This article describes the

perceptions, beliefs and health‐seeking behaviours of women from rural western Kenya regarding

congenital anomalies and miscarriages.

MethodsTen focus group discussions (FGDs) were undertaken in a rural district in western Kenya in

September 2010. The FGDs included separate groups consisting of adult women of childbearing age,

adolescent girls, recently pregnant women, traditional birth attendants and mothers of children with

a birth defect. Participants were selected purposively. A deductive thematic framework approach

using the questions from the FGD guides was used to analyse the transcripts.

ResultsThere was substantial overlap between perceived causes of miscarriages and congenital anomalies

and these were broadly categorized into two groups: biomedical and cultural. The biomedical causes

included medications, illnesses, physical and emotional stresses, as well as hereditary causes.

Cultural beliefs mostly related to the breaking of a taboo or not following cultural norms. Mothers

were often stigmatised and blamed following miscarriage, or the birth of a child with a congenital

anomaly. Often, women did not seek care following miscarriage unless there was a complication.

Most reported that children with a congenital anomaly were neglected either because of lack of

knowledge of where care could be sought or because these children brought shame to the family

and were hidden from society.

ConclusionThe local explanatory model of miscarriage and congenital anomalies covered many perceived

causes within biomedical and cultural beliefs. Some of these fuelled stigmatisation and blame of the

mother. Understanding of these beliefs, improving access to information about the possible causes

of adverse outcomes, and greater collaboration between traditional healers and healthcare

providers may help to reduce stigma and increase access to formal healthcare providers.

Community perceptions of miscarriage and congenital anomalies

99

IntroductionIn Kenya, as in most countries of sub‐Saharan Africa, there is a lack of information on the risk of

adverse pregnancy outcomes. Reviews using regional estimates project the risk of miscarriages and

stillbirths in 2007 to be 12.2% and 3.3% per pregnancy, respectively [1,2]. Appropriate and prompt

management of adverse pregnancy outcomes can reduce maternal mortality. For example, the two‐

fold higher risk of maternal death following miscarriage compared to those who had a live‐birth

[3,4] mainly reflects poor management of retained products and subsequent infection.

Globally, it is estimated that every year over three million children die due to congenital

anomalies (defined as structural or functional anomalies which occur during the embryo‐fetal

development and are present at the time of birth). In addition, another three million babies born

with a congenital anomaly and do not access care at birth may be disabled for life [5]. The risk of

major congenital anomalies in industrialised countries is around 3‐5% [6,7]. It is unclear if this is

similar in low income countries or if it is higher due to a variety of factors such as different

health‐seeking behaviours, poor nutrition, a higher prevalence of infectious diseases, weak health

systems and a higher availability of a wide range of prescription‐only drugs over‐the‐counter

[5,8,9]. This availability of drugs, and the high prevalence of infectious disease, means that

pregnant women are more likely to be exposed to teratogenic drugs or drugs with limited

information on their safety for pregnant patients. Further, misuse of drugs for induced abortion

also has political ramifications, particularly in countries with strict abortion laws and limited

access to contraceptive and family planning services. Due to limited information and knowledge,

infants born with a congenital anomaly might miss the opportunity to have access to appropriate

care in time and suffer the social and economic burden of lifelong disabilities. Receipt of

appropriate care significantly impacts the prognosis of newborns with a congenital malformation;

for children born with club foot, for example, this can be the difference between a life with limited

mobility (and often loss of social and economic opportunities) and that of normal active life

with painless functional feet.

Understanding the local perception of health, stigma, and disclosure is essential to improve

access to health services and for patient uptake of new programmes [10,11]. However, women’s

perceptions, clients and service providers understanding and attitudes towards adverse

pregnancy outcomes, and how these affect health‐seeking behaviour, have received scant

research attention. This paper describes the results from a qualitative study that explored the

perceived causes of miscarriages and congenital anomalies, stigma associated with these adverse

outcomes, and issues around disclosure and health‐seeking behaviour.

MethodsThis study was part of a formative research program carried out to inform the best approach to

setting up a prospective pharmacovigilance pregnancy cohort to monitor the use of medications

during pregnancy, the safety of these drugs in pregnancy, and their impact on birth outcomes.

There was no teratogenic drug monitoring program at the time we conducted focus group

discussions in the study area. Topics around the socio‐cultural context of pregnancy and adverse

pregnancy outcomes were explored.

Chapter 6

100

StudysiteThe study took place in Rarieda District, Nyanza Province, in western Kenya (within Siaya County

since 2013). This area has been under continuous surveillance as part of the Kenya Medical

Research Institute and the US Centers for Disease Control and Prevention (KEMRI/CDC) Health

and Demographic Surveillance System (HDSS) since 2001 [12]. Subsistence farming is the main

occupation in this area and more than 95% of the population are from the Luo ethnic group [13].

The prevalence of malaria (KEMRI/CDC, unpublished data), HIV [14] and TB [15] are high. Malaria

transmission is intense and holo‐endemic with peaks after the two rainy seasons usually in June‐

July and November‐December. HIV prevalence in Nyanza province is 13.9% (16.0% among women

and 11.4% among men) compared to a national prevalence of 6.3% [14]. HIV prevalence among

pregnant women is around 27% [16]. The total fertility rate in this area calculated for 2009 was 4.3

live births per woman [17]. Only 17% of females have had a secondary school education, while

19% have neither primary nor secondary school education [17].

ParticipantsFemales, aged 15 years and older, purposively sampled by village based KEMRI/CDC fieldworkers

(village reporters; VRs) were invited to be part of the focus group discussions (FGDs). VRs are

community members involved with sensitisation activities for new initiatives or research projects

in their villages. Prior to participant selection, VRs were trained on the aims of the formative

research and the characteristics of the participants required for each FGD by the study moderator

(B.O.), a Kenyan social scientist and the note‐taker. The different FGD groups consisted of adult

women of childbearing age (WOCBAs; 18‐49 years), adolescent girls (15‐18 years), recently or

currently pregnant women, traditional birth attendants (TBAs; locally called Nyamrerwas) and one

group of mothers of children born with a congenital anomaly. Each group consisted of no more

than one woman from each village. The separate groups were selected to have fairly homogenous

demographic characteristics and avoided dominance from older women, enabling freer

discussions. A total of 10 focus groups (n=90 participants) were conducted between September

1st and September 22nd, 2010.

StudyproceduresThe FGDs took place in central locations such as schools, or churches, that were easily accessible

to all participants, and where privacy could be maintained. Information on the pharmacovigilance

study and the formative research was discussed at community meetings with village chiefs and

counsellors as well as with a community advisory board (CAB, comprising stakeholders and

members of the community who provide advice to researchers on proposed studies). Individual

participants gave consent after the information sheet had been verbally explained and discussed.

FGD guides were semi‐ structured, open‐ended and probing. The discussion guides were initially

written in English and then translated to Dholuo, the local dialect, by the Kenyan social scientist

and moderator. The FGD guides covered topics relating to pregnancy recognition, disclosure,

health‐seeking behaviour and pregnancy related behaviour change as well as use of medication

and herbal remedies during pregnancy, practices around delivery, perception of adverse

outcomes and unwanted pregnancies. Not all topics were covered by all groups due to time

limitations. The FGDs ranged from one to two hours. The moderator and note‐taker, both

females from the study area, had many years of fieldwork experience, including FGDs, and were

trained on the study protocol and tools before the first FGD. All FGDs were recorded using

audiotapes. Audio files were transcribed verbatim and translated by the moderator and two


101

independent transcribers; this activity was guided by the notes taken during the FGDs. Each

transcript was reviewed by the moderator to ensure consistency and accuracy was maintained.

DataanalysisA deductive thematic framework approach was used to provide a detailed account of a group

of themes using the questions from the FGD guides on perception, attitudes, and risk factors

associated with adverse pregnancy outcomes within the data. Each transcript was uploaded to

QSR Nvivo 9 software (QSR International Pty Ltd, Melbourne, Australia). The code structure was

developed according to the topic guide and the codes were applied to each transcript in a

systematic fashion across the entire dataset by collating data relevant to each code. Additional

codes were identified inductively from the transcripts. The coding was done independently by

S.D. and L.M. and comparisons made; any areas of disagreement were discussed and resolved on

revisiting the data. Thematic maps were generated to visualise the main themes and sub‐ themes

and how they connected. Tree maps were generated using Nvivo to compare coding references

between the different sub‐groups. Table S1 provides a list of the main themes and sub‐themes.

EthicsThe protocol and consent procedures were reviewed and approved by 1) the Kenya Medical

Research Institute (KEMRI, Nairobi, Kenya) National Ethics Review Committee; 2) the US Centers

for Disease Control and Prevention (CDC, Atlanta, GA, USA) IRB and 3)the Liverpool School of

Tropical Medicine (LSTM, UK) Research Ethics Committee. Informed consent was obtained verbally

in the local dialect (Dholuo) and tape recorded. This informed consent procedure was approved

by all 3 ethics committees. This is common procedure for FGDs as written consent can sometime

make participant wary or guarded, particularly in a setting were illiteracy is common and would

require finding an appropriate witness.

ResultsTable 1 provides a summary of the participants’ characteristics. The findings are presented as

three main themes relating to the specific research questions: 1) Perceived causes of miscarriages

and congenital anomalies, 2) stigma and the community perception of miscarriages and congenital

anomalies, and 3) health‐seeking behaviours for miscarriages and congenital anomalies. There was

no major difference observed in the beliefs and themes that emerged between the different

groups (i.e. WOCBAs compared with adolescents or TBAs or mothers of children born with a

congenital anomaly) and the findings are consequently presented jointly for all groups.

No definition of congenital anomaly was provided to the participants but a wide range of

anomalies, both structural and functional, were cited when participants were asked about the type

of congenital anomalies they had observed in the community. The most frequently mentioned

anomaly was deformity of the hand, followed by mental retardation, and paralysis. Deafness, cleft‐

lip and eye anomalies were also mentioned by different groups. Explicit anomalies were not

specified during the discussion on causes, stigma and health seeking behaviours; therefore we

refer to congenital anomalies in general in the text.

Chapter 6

102

Table 1. Summary of Focus Group Discussions Participant Characteristics.

FGD# Group N Average Age in years

Average Gravidity (range)

Marital status Education Single Married Widowed None Primary Secondary

FGD1 WOCBA 10 38 7 (3‐10) 0 9 (90%) 1 (10%) 0 7 (70%) 3 (30%)

FGD2 WOCBA 9 31 5 (1‐9) 0 9(100%) 0 0 8 (89%) 1 (11%)

FGD3 Recently Pregnant

8 32 6 (2‐10) 0 7 (88%) 1 (13%) 0 8 (100%) 0

FGD4 TBA/

Nyamrerwas 10 54 7 (4‐12) 0 4 (40%) 6 (60%) 3 (30%) 2 (20%) 5 (50%)

FGD5 Adolescent 9 16 0 9 (100%) 0 0 0 9 (100%) 0

FGD6 Adolescent 9 17 0 9 (100%) 0 0 0 9 (100%) 0

FGD7 Recently Pregnant

9 28 4 (2‐7) 0 9 (100%) 0 0 8 (89%) 1 (11%)

FGD8 TBA/

Nyamrerwas 8 41 6 (2‐10) 0 8 (100%) 0 1 (13%) 7 (88%) 0

FGD9 WOCBA 9 31 5 (1‐10) 1 (11%) 6 (67%) 2 (22%) 1 (11%) 6 (67%) 2 (22%)

FGD10

Mothers of children with congenital anomalies

9 31 5 (1‐10) 2 (22%) 6 (67%) 1 (11%) 1 (11%) 8 (89%) 0

Overall 90 32 5 (0‐12) 21 (23%) 58 (64%) 11 (12%) 6 (7%) 72 (80%) 12 (13%)

Acronyms: WOCBA‐ women of childbearing age; TBA‐ traditional birth attendant

PerceptionofcausesofmiscarriagesandcongenitalanomaliesMany different possible causes of miscarriages and congenital anomalies were reported with a

significant overlap between these two outcomes. We divided these into two broad groups: those

with a biomedical basis and those with a cultural basis.

1: Biomedical Causes. The role and potential dangers of biomedical causes were thematically

sub‐grouped into medications, illness, physical and emotional stress, and hereditary causes.

Most women reported that certain drugs could lead to miscarriages or congenital anomalies, but

only if they were not prescribed, or if the dose taken was above that prescribed by a healthcare

worker. There was an overarching theme of trust by the women that any treatment given or

prescribed by clinicians would be safe. One participant mentioned that a pregnant woman should

only take medicines if the illness is confirmed. This infers a natural understanding of the concept

of balancing the risks and benefits of treatment (i.e. it is only worth taking the risk of consuming

medicines if one is truly ill) by this participant. When asked about reservations over the ingestion

of drugs in pregnancy, two separate groups brought up the issue of nausea and heightened

sensitivity to smell (i.e. relating this to difficultly in taking drugs orally during pregnancy). Thus, it

was suggested that pregnant women should be given injections rather than tablets. This implies

that reluctance to take medication in pregnancy is associated with pregnancy related nausea,

rather than a fear of an adverse effect on the pregnancy or the fetus. Only one participant

mentioned fear of side‐effects (itching and drowsiness for the pregnant women) as a reason for

not taking the drugs given to her during antenatal care.

“I’m requesting that if one is pregnant you should avoid the drugs, it should just be

injection given when one is sick. They don’t take drugs. The drugs that I was given in 1997

when I gave birth just expired the other day. So it should be the injection which you come

and get.” (FGD2, P9)


103

“When some women go to the clinic and are given medicine because when you go to the

clinic you are given medicine for blood, you are given medicine that would give energy…when

they are given such medicine, they do not take saying that it smells bad.” (FGD1, P6)

“A pregnant woman should not just take medicine if it is not prescribed by the doctor.”

(FGD3, P1)

Family planning drugs were considered to be distinct types of drugs, separate from medications

used for illness, and all groups believed that these could lead to congenital anomalies.

“You find that when you give birth to a child with a defect then people would say it is

because of the family planning that has caused it.” (FGD10, P2)

Other drugs also deemed to be potentially harmful for the pregnancy were antimalarials,

including chloroquine (which is no longer recommended for treatment due to malaria parasite

resistance ), quinine (which is considered safe for use in all trimesters of pregnancy and used for

severe and complicated malaria which in itself can cause miscarriage), and artemether‐

lumefantrine (which is the first line treatment for malaria but not recommended for use in the first

trimester of pregnancy due to its unknown safety). Chloroquine and quinine were drugs reported

to be used by pregnant women within the community to induce abortions for unwanted

pregnancies. Anti‐retroviral drugs were mentioned by one participant as having the potential to

cause congenital anomalies. One participant noted antibiotics could hinder bone development of

the fetus.

“[M: Which other way would you use to do abortion?]In some cases I hear people overdose.

[What overdose?] For the tablets, there are some drugs if they take like the malariaquine

[chloroquine] if you overdose then you will abort.” (FGD7, p5)

When asked about the potential risk period for taking medication during pregnancy only one

participant mentioned that drugs should be avoided during the first three months of pregnancy.

The majority mentioned drugs should be avoided closer to the time of delivery.

Disease or infections were mentioned by all groups as a potential cause of miscarriage. Most

participants simply cited ‘disease’ without being specific. However, malaria was mentioned by a

few participants, and one person cited HIV. Other illnesses, particularly sexually transmitted

infections (gonorrhoea was specifically mentioned), and measles, were commonly cited causes of

congenital anomalies.

“Like gonorrhoea maybe you were sick and you don’t seek treatment, so if you gave birth to

a child he/she must be disabled in some way.” (FGD9, P1)

Participants considered that adverse pregnancy outcomes were the consequence of pregnant

women not attending antenatal care services, or not completing their vaccinations (supposedly

tetanus vaccine, which is provided routinely through ANC). This could be interpreted as a

perception that pregnant women are vulnerable to adverse effects from an illness, which

would have been prevented, had the woman attended ANC services.

Chapter 6

104

Carrying out strenuous work while pregnant was mentioned by all groups as a potential cause of

miscarriage. A few respondents also thought physical trauma (such as trauma near or around the

uterus) could lead to miscarriage. Women also considered that adverse effects were more likely

among women who had conceived many times, suggesting women understood the physical toll

of many pregnancies and short birth intervals between pregnancies. Similarly, the position of

the fetus in the uterus was reported by one TBA as a potential cause of deformity. Emotional distress

such as being shocked by bad news or arguing with a husband/partner was considered to be

dangerous for the pregnancy, and could lead to pregnancy loss. Being raped was also mentioned as a

cause of miscarriage.

“Miscarriage can come when you fight with someone and you are hit in the wrong place.”

(FGD10, P9)

“One can miscarry if you are shocked by bad news.” (FGD10, P4)

Hereditary transmission (referred to as “hereditary” or “inherited trait”) was a prominent sub‐

theme for congenital anomalies. Many women reported that the mother of a child with a

congenital anomaly is often thought to have conceived with a man other than the husband, since

the husband did not appear to have, or physically display, the same congenital anomaly. This

highlights an underlying theme of mistrust and blaming of mothers for any adverse pregnancy

outcome (explored more below under “cultural beliefs”). Incest was also cited as a cause for

congenital anomalies and miscarriages. Incest was reported as a reason to kill a baby born

with a congenital anomaly by one group.

“Sometimes they would say you conceived with a relative, so most people normally kill them

[child with congenital anomaly].” (FGD6, P7)

2: Cultural Beliefs. There were many beliefs around the causes of miscarriage and congenital

anomalies including extra‐marital sex, not respecting traditional ways, and being cursed or

possessed.

A strong theme for causing both miscarriage and congenital anomalies was infidelity, where the

woman conceived outside of wedlock. This includes being raped, although this could also be

associated with the physical strain and emotional distress caused, if the husband cheated on the

wife, and as a consequence to incest.

“I have heard one but I’m not sure about it, that if you spend [a night] with a man other than

your husband when pregnant then you can abort.” (FGD9, P9)

“For a baby born with a defect, people would think how possible it is especially if the mother

has no defect. Someone would think how she can give birth to a child with a defect like

having only one eye. She would think that perhaps this might have come as a result of going

out with other men.” (FGD8, P6)

Women in different groups reported that if their husband either had an extra marital affair or had

slept with an inherited wife, then the current wife would be more likely to miscarry or bear a child

with a congenital anomaly.


105

Not conforming to traditional rules was mentioned as a cause of miscarriage and congenital

anomalies. Mostly this related to not performing traditional rituals surrounding marriage,

becoming pregnant before the husband paid the bride‐price, or building a new house while the wife

is pregnant. For the latter it was not clear whether this was a consequence of physical

exhaustion of the pregnant woman, or the breaking of a specific traditional taboo. Other beliefs

that could all lead to adverse pregnancy outcomes included the breaking of the taboo around

sharing the cooking place with a grandmother, taking on the responsibilities of elders before

reaching this senior position, planting when not the owner of the field, or eating meat from a

pregnant cow. Many mentioned that adverse outcomes occurred because the woman or her

family broke a taboo that was prohibited by the clan and had been cursed. Being possessed by

demons was considered another possible cause of a miscarriage or a congenital anomaly.

“If she did something prohibited in her clan and then she gives birth to a deformed child then

the community would say it came as a result of committing a taboo.” (FGD4, P4)

“A curse can fall upon you depending on how you live in the home or according to some

cultural practices that you are supposed to do.” (FGD9, P5)

The above quotation indicates a very strong relationship between the breaking of taboos, against

cultural norms, and subsequent adverse pregnancy outcomes. These taboos appear to be deeply

rooted within the cultural belief systems, and across all age groups. Although the community

under study are largely Christian, and very religious, only a few suggested that adverse pregnancy

outcomes happen without explanation or have a spiritual/religious explanation as “God’s plan.”

“There is malformation that comes as a result of God’s creation; you can give birth to a child

without knowing that it will have a malformation. So that is the will of God.” (FGD8, P5)

StigmaassociatedwithmiscarriagesandcongenitalanomaliesA number of the causes given were associated with blame, either on the mother or her family,

such as a woman’s infidelity to her husband, or breaking a traditional taboo. This could lead to

distrust between husband, mother‐in‐law and the wife/ mother. Women reported that mothers

are thus stigmatised rather than supported after going through the distress of having an adverse

pregnancy outcome. Some reported that women who suffered from a miscarriage were

ostracised and not allowed to come out of their home until they had been cleansed by a spiritual

healer. Other pregnant women were also not able to come into contact with them as this might be

passed on and cause them also to miscarry.

“In our place, I normally see when one has miscarriage they are not allowed to go out. They

say that you stay in the house until maybe the church members come and clean you that is

when you can move out. So the reason why they don’t want people to go out and I don’t

know what would happen if you went out.” (FGD7, p8)

Neglect of children born with abnormalities was reported by a number of groups. Children with any

congenital anomaly/ disability were seen as not “useful” to the family and an extra burden, as

Chapter 6

106

well as stigmatized. This stigma might imply some taboo had been broken. Many respondents said

such children are often hidden in homes or even killed. Only three women mentioned that they

would love and provide the extra‐care the child needed.

“You would not love such kind of a child [with congenital anomaly]; I would not love the child

because I would feel that he will not be of any help to me. And then taking care of him would

be hard.” (FGD9, P5)

“I would love the child [born with a birth defect/disability] because he/she is mine; I’m the

one who gave birth to her/him.” (FGD9, P1)

“Sometimes the grandmother speaks a lot and she might say that the child is not of her

blood.” (FGD7, P7)

Health‐seekingbehaviourParticipants reported that women seek care either from a health facility or from a traditional

healer such as a TBA, herbalist, spiritual healer or “Jarwecho” (a special healer that deals with

skeleton issues and bone fracture). Women also may seek advice from family members such as a

grandmother, mother‐in‐law and/or the husband. Women reported that they wanted to seek care

from health facilities so that they could be treated in the event of complications (i.e. when there

is excessive bleeding), and also so that they could better understand the potential causes of the

adverse outcome. Many women reported they would go to the TBA to get herbs or go to see a

spiritual healer who would “pray for them” or cleanse them of the curse. In the case of a

miscarriage, it seems women would only seek care if there were perceived complications. Most

women would otherwise keep the miscarriage a secret. Many women reported that they, and

other women, would not seek care if their baby was born with a congenital anomaly.

M: “When they miscarry, do they seek for any care?”

P7: “They don’t seek for any care when they miscarry.”

P1: “Sometimes they go to the hospital.”

P7: “Some people would not like others to know if they have miscarried.” FGD8

“From what I hear, there is nothing one can do about it [child born with malformation] and

no one can ever give you a piece of advice on how they can be helped.” (FGD10, P7)

The decision about where to seek care was discussed by one participant, who noted that it

depended on the cost of care and resources available to the family.

DiscussionUsing FGD methodology, we explored the community explanatory models for adverse pregnancy

outcomes and how these explanatory models might influence health‐seeking behaviours. Although

various biologically plausible causes of miscarriage and congenital anomalies were known, cultural

beliefs seem to play a central role in the community perception of these adverse pregnancy

outcomes. Such beliefs can result in stigma and influence health‐seeking behaviours, as reported in

previous studies on relatively minor but disfiguring congenital anomalies [18]. Stigma seems to be

strongly related to the belief that the woman either cheated on her husband or she or her family


107

had broken some traditional taboo. Some of the cultural beliefs around miscarriage and

congenital anomalies lead to stigmatisation, with the mother largely held responsible for the cause

of the malformation; this may have a negative impact on health‐seeking behaviours and

disclosure of such pregnancy outcomes. Women who experienced adverse pregnancy outcomes

are thus often additionally burdened by negative attitudes and stigmatisation from the

community. In particular, mothers of children with congenital anomalies were often blamed for

cheating on their husbands or being cursed because they have apparently transgressed some

cultural taboos. Similar studies from rural India and Nigeria found that parents of disabled children

were socially marginalized because of widely held beliefs that they had broken a taboo [19], and

their children were neglected [20]. Improving access to information about the possible causes of

such adverse outcomes may help to reduce the stigma and shame that these women undergo and

increase access to formal healthcare providers.

These data suggest that mothers of children with congenital anomalies or disabilities do not know

where they can seek help or whether any help is available. This, together with social stigma, may

contribute to the neglected care of these children in this setting. Other factors such as

perception of treatment effectiveness, satisfaction with health care services, and external

barriers (e.g. financial constraints, accessibility of health services) also play an important role in

driving health‐ seeking behaviours. We note that limited disclosure of adverse pregnancy outcomes

could also impact on the accuracy of public health surveillance programmes and could lead to an

under‐estimation of the problem.

Fertility and successful childbearing remain very important factors for status and recognition of a

woman’s worth in society, in many countries, including those in sub‐Saharan Africa. Childbearing

confers social status and strengthens conjugal relationships, and children are seen as contributors

to the daily labour. The presence of children ensures the rights of property and maintains the

family lineage, as described more fully by Dyer [21]. Unsuccessful childbearing and adverse

pregnancy outcomes could thus be a potential cause or driver of gender‐based violence. Gender‐

based violence is widespread in Kenya, and particularly high in Nyanza Province where 60% of

women report having ever experienced emotional, physical, or sexual violence by their husbands

[14]. Although this was not reported in our FGDs, we noted that females in the focus groups

considered violence and rape to be causes of miscarriage. We also recognise that in some

countries, with strict abortion laws and poor access to contraception, and family planning,

women may be driven to take medicine perceived locally to induce abortion, but which may be

potentially teratogenic. This requires further study.

It is not clear how the overlap between biomedical knowledge and cultural beliefs is

incorporated into society’s perception, whether these beliefs are mutually exclusive, or blend

according to different societal pressures. Under the biomedical paradigm, illness and disease are

attributable to biologically plausible causes based on scientific theories such as factors with

biochemical effects. If understanding of ill health is based on a biomedical paradigm, the individual

is more likely to seek medical care from modern rather than traditional sources [22]. On the other

hand, cultural beliefs are based on supernatural phenomenon, traditional values or religious

beliefs. Within this explanatory model individuals would seek help or care from either a

spiritual or traditional healer [22]. However, adverse health outcomes are not always categorised

as one or another. For example Hausmann‐Muela et al [23], reported how in Tanzania there is

Chapter 6

108

a widespread belief that witchcraft can affect biomedical treatment of malaria and impede

detection of malaria parasites. Addressing such beliefs, and acknowledging the role of traditional

healers and birth attendants, should be part of public health and research programs. Creating

collaborative links between traditional and modern medicine, such as by empowering TBAs to

refer their clients to health facilities, is critical to increase access to care in developing countries

[24].

There was an underlying theme of trust among this rural population in western medicine and

healthcare providers. Medications were deemed safe when prescribed and when taken at the

correct dosage as prescribed, and adverse outcomes were considered to be due to a lack of

medical follow up for women who didn’t attend ANC. Similar to other studies, our research

findings suggest that women in rural African settings might suffer from a “white‐coat complex”

where they avoid asking questions to healthcare workers, and assume any prescribed drugs are

safe [25,26]. At the same time, however, pregnant women do not always take medicines given

at the health facilities, not because they fear the potential teratogenic risk, but because pregnancy

predisposes them to nausea. This has important implications for any intervention targeting

pregnant women, and consideration should be given to this to ensure compliance, including the

use of directly observed therapy where feasible. Furthermore, in a context where healthcare

clients do not ask questions and rely on the providers’ judgement, it is particularly important that

healthcare providers who see pregnant (or potentially pregnant) women should be familiar with

all drugs contraindicated during pregnancy and potential teratogens. Whilst women were aware of

the potential risk of using medication during pregnancy, most did not know the potential risk in

the first trimester. The few specific drugs that they did mention (such as quinine and chloroquine)

are safe for use in all trimesters of pregnancy, exemplifying this population’s lack of

knowledge. Others mentioned that antimalarials should be avoided towards the end of

pregnancy in the third trimester. This highlights the need for more information, education and

communication on the risk of medication used during the first trimester of pregnancy and the

importance for women to explicitly disclose their potential pregnancy to healthcare providers.

The widespread belief that modern family planning methods increases the risk of having a child

with congenital anomalies also urgently needs to be addressed, particularly within the context of

a country with a high fertility rate since the 1990s [27]. Family planning interventions that

promote healthy spacing of pregnancy are vital for development and it will be important to ensure

accurate information is available to offset myths which could negatively impact uptake of family

planning programmes. Only about 25% of women in the study area report using a family

planning method (unpublished data), compared to 73% nationally [14].

This study has some possible limitations. Despite efforts to provide a non‐threatening

environment, respondents may have withheld information and some members might have been

affected by a social desirability effect. FGD participants could have been inclined to give what

they thought would be acceptable answers, and may have feared telling the truth, particularly

around local myths and taboos. This is exemplified by the fact that most of the reported themes

under cultural beliefs were reported in the third person (i.e. “it is believed that” rather than “I

believe”). The FGDs were conducted as formative research for setting up a pregnancy

pharmacovigilance study, which may have influenced the moderator and the participant to focus

on issues around drug safety. Although one of the coders (S.D.) is the study coordinator for the

pharmacovigilance study and has a specific interest in perception of drug safety in pregnancy, the


109

second coder was not associated with the pharmacovigilance study. Many topics were included in

the FGDs which limited the depth and details of the information collected. Moreover, the findings

reflect the circumstances affecting a particular population, i.e., women in one rural area in

western Kenya, and might not be generalisable. Although validation of findings representing the

views elicited through 10 FGDs through data triangulation was not possible, investigator

triangulation (where two investigators independently coded the data and compared notes) was

used to enhance credibility of the emerging sub‐themes. This study did not include the perspective

of men or healthcare providers who often play an important role influencing healthcare seeking

behaviours. Future studies including such groups would contribute additional understanding to the

barriers to healthcare seeking for adverse pregnancy outcomes.

ConclusionTo our knowledge, this is the first qualitative study undertaken in Kenya exploring the

perceptions, attitudes and health‐seeking behaviours on adverse pregnancy outcomes. Better

understanding of these concepts can inform strategies to improve women’s health‐seeking

behaviour. Development of appropriate information, education and communication and outreach

materials should inform women of true causes of adverse outcomes. This will also help to reduce

the burden of guilt they feel, the stigmatisation from the community, and provide guidance on

caring for women who have adverse pregnancy outcomes and have children with disabilities. To

support this, there is a need for greater integration and collaboration between traditional healers

and modern medical practitioners, and to ensure culturally acceptable management to better

protect the unborn child and expectant mothers.

AcknowledgementsWe thank all the study participants, the late Beatrice Odidi (moderator), Jane Oiro (note‐taker)

and the village reporters for their diligent efforts in the field. We thank Florence Were and Walter

Mchembere for reviewing the FGD guides. This manuscript is approved by the KEMRI Director.

KEMRI/CDC is a member of The INDEPTH Network. The findings and conclusions in this paper are

those of the authors and do not necessarily represent the views of the Centers for Disease Control

and Prevention.

AuthorContributionsConceived and designed the experiments: SD. Performed the experiments: BO. Analyzed the data:

SD LM. Wrote the manuscript: SD MD LM BO GA PAP KFL FOtK.

References1. Dellicour S, Tatem AJ, Guerra CA, Snow RW, ter Kuile FO (2010) Quantifying the number of

pregnancies at risk of malaria in 2007: a demographic study. PLOS Med 7: e1000221. PubMed: 20126256.

2. Stanton C, Lawn JE, Rahman H, Wilczynska‐Ketende K, Hill K (2006) Stillbirth rates: delivering estimates in 190 countries. Lancet 367: 1487‐1494. doi:10.1016/S0140‐6736(06)68586‐3. PubMed: 16679161.

3. Rahman M, DaVanzo J, Razzaque A (2010) The role of pregnancy outcomes in the maternal mortality rates of two areas in Matlab, Bangladesh. Int Perspect Sex Reprod Health 36: 170‐177. doi: 10.1363/3617010.

4. Reardon DC, Ney PG, Scheuren F, Cougle J, Coleman PK et al. (2002) Deaths associated with pregnancy outcome: a record linkage study of low income women. South Med J 95: 834‐841. doi: 10.1097/00007611‐200208000‐00010. PubMed: 12190217.

Chapter 6

110

5. Howson C, Christianson A, Modell B (2008) Controlling Birth Defects: Reducing the Hidden Toll of Dying and Disabled Children in Lower‐ Income Countries. Washington, DC.

6. Baird PA (1999) Prenatal screening and the reduction of birth defects in populations. Community Genet 2: 9‐17. doi:10.1159/000016177. PubMed: 15178956.

7. Edmonds LD, James LM (1993) Temporal trends in the birth prevalence of selected congenital malformations in the Birth Defects Monitoring Program/Commission on Professional and Hospital Activities, 1979‐1989. Teratology 48: 647‐649. doi:10.1002/tera. 1420480606. PubMed: 8115973.

8. Bale JR, Stoll BJ, Lucas AO (2003) Reducing Birth Defects: Meeting the Challenge in the Developing World. National Academy of Sciences.

9. Penchaszadeh VB (2002) Preventing congenital anomalies in developing countries. Community Genet 5: 61‐69. doi: 10.1159/000064632. PubMed: 14960901.

10. Landrine H, Klonoff EA (1994) Cultural diversity in causal attributions for illness: the role of the supernatural. J Behav Med 17: 181‐193. doi: 10.1007/BF01858104. PubMed: 8035451.

11. Filippi V, Marshall T, Bulut A, Graham W, Yolsal N (1997) Asking questions about women's reproductive health: validity and reliability of survey findings from Istanbul. Trop Med Int Health 2: 47‐56. doi: 10.1046/j.1365‐3156.1997.d01‐126.x. PubMed: 9018302.

12. Odhiambo FO, Laserson KF, Sewe M, Hamel MJ, Feikin DR et al. (2012) Profile: the KEMRI/CDC Health and Demographic Surveillance System‐‐Western Kenya. Int J Epidemiol 41: 977‐987. doi:10.1093/ije/ dys108. PubMed: 22933646.

13. Adazu K, Lindblade KA, Rosen DH, Odhiambo F, Ofware P et al. (2005) Health and demographic surveillance in rural western Kenya: a platform for evaluating interventions to reduce morbidity and mortality from infectious diseases. Am J Trop Med Hyg 73: 1151‐1158. PubMed: 16354829.

14. Kenya National Bureau of Statistics (KNBS) (2010) Macro I. Kenya Demographic and Health Survey 2008‐09. Calverton, Maryland.

15. van't Hoog AH, Laserson KF, Githui WA, Meme HK, Agaya JA et al. (2011) High prevalence of pulmonary tuberculosis and inadequate case finding in rural western Kenya. Am J Respir Crit Care Med 183: 1245‐1253. doi:10.1164/rccm.201008‐1269OC. PubMed: 21239690.

16. Amornkul PN, Vandenhoudt H, Nasokho P, Odhiambo F, Mwaengo D et al. (2009) HIV prevalence and associated risk factors among individuals aged 13‐34 years in Rural Western Kenya. PLOS ONE 4: e6470. doi:10.1371/journal.pone.0006470. PubMed: 19649242.

17. KEMRI/CDC (2009) KEMRI/CDC Health and Demographic Surveillance System Report for 2009. Kisumu: KEMRI/CDC.

18. Olasoji HO, Ugboko VI, Arotiba GT (2007) Cultural and religious components in Nigerian parents' perceptions of the aetiology of cleft lip and palate: implications for treatment and rehabilitation. Br J Oral Maxillofac Surg 45: 302‐305. doi:10.1016/j.bjoms.2006.09.005. PubMed: 17056161.

19. Minhas RS, Hillard R (2007) Maternal Care Practices and Perceptions of Birth Defects in Central India. Univ Tor Med J 84.

20. Okunade AO (1981) Attitude of Yoruba of Western Nigeria to handicap in children. Child Care Health Dev 7: 187‐194. doi:10.1111/j. 1365‐2214.1981.tb00836.x. PubMed: 6456848.

21. Dyer SJ (2007) The value of children in African countries: insights from studies on infertility. J Psychosom Obstet Gynaecol 28: 69‐77. doi: 10.1080/01674820701409959. PubMed: 17538814.

22. Kgwatalala G (2009) Health Seeking Behaviour among the People of the Africa Gospel Church in Francistown. UNISA.

23. Hausmann‐Muela S, Muela Ribera J, Tanner M (1998) Fake malaria and hidden parasites—the ambiguity of malaria. Anthropol Med 5.

24. Leininger MM (1991) Culture Care theory and uses in nursing administration. NLN Publ: 373‐390. PubMed: 1795996.

25. Stokes E, Dumbaya I, Owens S, Brabin L (2008) The right to remain silent: a qualitative study of the medical and social ramifications of pregnancy disclosure for Gambian women. BJOG 115: 1641‐1647; discussion: 10.1111/j.1471‐0528.2008.01950.x. PubMed: 19035940.

26. Brabin L, Stokes E, Dumbaya I, Owens S (2009) Rural Gambian women's reliance on health workers to deliver sulphadoxine – pyrimethamine as recommended intermittent preventive treatment for malaria in pregnancy. Malar J 8.

27. UN (2010) World Population Prospects: The 2010 Revision. Population Division of the Department of Economic and Social Affairs of the United Nations Secretariat.


111


TableS1.Mainthemesandsub‐themesusedinthethematicanalysisforexploringriskperceptionandattitudestomiscarriageandcongenitalanomaly.

Themes Sub‐themes

Causes Biomedical

Illnesses

Medications

Stress

Hereditary

Cultural

Not conforming to traditional norms

Infidelity

Curse

None

Stigma Stigmatisation

Hide child with abnormality

Isolate woman who had a miscarriage

Blame/Gossip

No disclosure

No stigma

Accept child with a congenital malformation

Freely disclose adverse pregnancy outcome

Health seeking behaviours Health facility

Traditional

TBA

Spiritual healer

Herbal remedies

Family

Husband

Parents in law

Grand‐mother

None

113

Chapter7

Assessment of Knowledge and Adherence to the NationalGuidelines forMalariaCaseManagement inPregnancyamongHealthcare Providers and Drug Outlet Dispensers in rural,westernKenya

ChristinaRiley1,StephanieDellicour3,4,PeterOuma2,UrbanusKioko5,FeikoO.terKuile2,AhmeddinOmar5,SimonKariuki3,4,AnnM.Buff6,7,MeghnaDesai3,6,JulieGutman6

1 Rollins School of Public Health, Emory University, Atlanta, USA, 2 Liverpool School of Tropical Medicine, Liverpool, United Kingdom, 3 Kenya Medical Research Institute and Centres for Disease Control and Prevention (KEMRI/CDC)Research and Public Health Collaboration, Kisumu, Kenya, 4 KEMRI, Centre for Global Health Research, Kisumu, Kenya, 5 Malaria Control Unit, Ministry of Health, Nairobi, Kenya, 6 Malaria Branch, Division of Parasitic Diseases and Malaria, Center for Global Health, US Centers for Disease Control and Prevention, Atlanta, USA and Kenya, 7 US President’s Malaria Initiative, Nairobi, Kenya

To be submitted

Chapter 7

114

ABSTRACTAlthough prompt and effective treatment is a cornerstone of malaria control, information on

healthcare provider adherence to malaria treatment guidelines in pregnancy is lacking. Incorrect or

sub‐optimal treatment can cause adverse consequences to the mother and fetus.

We conducted a cross‐sectional study from September to November 2013, in all health facilities and

randomly selected drug outlets in the HDSS catchment area in Siaya County western Kenya, to assess

healthcare provider and drug dispenser adherence to and knowledge of national guidelines for

treatment of uncomplicated malaria in pregnancy. In health facilities, exit interviews of women of

childbearing age, inclusive of pregnant women, who had been assessed for fever were used to

assess adherence to malaria treatment guidelines. Simulated clients posing as 1st trimester pregnant

women or as relatives of women in 3rd trimester collected information from drug outlets.

Information on treatment was recorded from prescriptions or after reviewing medications in

patient’s possession. Standardized questionnaires were used to assess healthcare providers’ and

drug outlet dispensers’ knowledge of treatment guidelines.

Fifty‐six percent of health facility providers, versus none of the drug outlet dispensers, knew the

appropriate treatment for 1st trimester patients, while 39% and 87%, respectively, knew the

appropriate treatment for 2nd/3rd trimester. Prescription of the correct drug for pregnancy trimester

at the correct dosage was observed in 66% of cases in health facilities and 40% in drug outlets.

Prescribing was more often correct in 2nd/3rd trimester than in 1st (65% vs. 32%, p=0.004, and 38% vs.

0%, p<0.001, at health facilities and drug outlets, respectively). Sulfadoxine‐pyrimethamine, which is

no longer recommended for treatment of acute malaria, was prescribed in 3% of cases in health

facilities and 18% of simulations in drug outlets (p<0.001). Exposure to artemether‐lumefantrine in

1st trimester, which is contraindicated due to its unknown safety, occurred in 16% and 51% of cases

in health facilities and drug outlets, respectively (p=0.04); none were a result of quinine stock‐out.

This study highlights knowledge inadequacies and incorrect prescribing practices in the treatment of

malaria in pregnancy. These should be addressed through comprehensive trainings and adequate

supervision by the Kenya Ministry of Health to improve the quality of patient care and maximize

therapeutic outcomes.

Knowledge and adherence to malaria in pregnancy treatment guidelines

115

INTRODUCTIONIt is estimated that around 125 million pregnancies occur in areas at risk of P. falciparum and/or P.

vivax infections every year; an estimated 1.3 million of these are in Kenya [1]. Malaria in pregnancy

(MiP) can have devastating consequences for the woman and her unborn baby. Adverse effects of

MiP include maternal anemia, fetal loss, intrauterine growth retardation, premature delivery and

low birth weight (LBW); LBW associated with MiP results in an estimated 100,000 deaths each year

in Africa alone [2]. In order to prevent the adverse consequences associated with MiP, the World

Health Organization (WHO) and the Kenyan Ministry of Health recommend that pregnant women

living in malaria endemic area use long‐lasting insecticidal nets (LLINs), intermittent preventive

treatment in pregnancy (IPTp) with sulfadoxine‐pyrimethamine (SP), and receive prompt and

effective diagnosis and treatment of malaria infections with a safe drug.

In Kenya, following WHO recommendations, artemether‐lumefantrine (AL) is the 1st‐line treatment

for uncomplicated P. falciparum malaria in the general population and women in 2nd/3rd trimester of

pregnancy; due to insufficient safety data [3,4,5], this is not recommended in 1st trimester, and oral

quinine should be used instead [6,7]. In practice, this means that all women of childbearing age

(WOCBA) must be assessed for pregnancy inclusive of the trimester of pregnancy. In addition,

Kenyan guidelines, updated in 2010, stipulate that ACTs should only be provided for malaria cases

confirmed by diagnostic test. Antimalarial treatment on the basis of clinical suspicion of malaria

should only be considered in situations where a parasitological diagnosis is not accessible [6].

There is limited data on healthcare provider adherence to diagnostic and treatment guidelines for

MiP. A review of studies of antimalarial use in the general population found that only 51% of cases

were treated with recommended antimalarials during 2004‐2006 [8]. A 2008 study in Kenyan health

facilities reported limited health worker compliance with the recommended treatment guidelines in

patients over 5 years of age [9]. Although 99% of patients with a positive test received an

antimalarial, only 80% received AL, the recommended first line therapy, despite the fact that the

study was restricted to health facilities, which had both malaria diagnostics and AL available on the

survey day. A more recent 2010 study in Kenya found improved adherence, with 90% of test positive

patients receiving the recommended first‐line therapy [10]. Very few studies have looked at

adherence to treatment guidelines in pregnancy. In Uganda, 70% of women in the 1st trimester of

pregnancy received a drug that is contraindicated [11]; in Tanzania, 43% of drug dispensers in

registered pharmacies offered AL regardless of the pregnant client’s gestation [12]. In addition, the

few studies on provider knowledge and prescribing and dispensing practice of antimalarials for

WOCBA show limited evidence of knowledge regarding the potential for teratogenicity with ACTs

[12,13,14,15]; in one study only 20% of drug dispensers in private pharmacies knew that AL was

contraindicated in 1st trimester. Given that WOCBA represent about 25% of the total population, up

to 14% of whom could be pregnant at any time and one‐third of them in the 1st trimester, it is crucial

that providers recognize regimens that have the potential for teratogenicity and assess WOCBA for

pregnancy status and gestational age.

Understanding provider prescribing behaviour in pregnant patients can play a key role in improving

the prescribing, administration, and use of antimalarials while minimizing potential harmful

exposures. This cross‐sectional study assessed healthcare provider and drug dispenser prescribing

behaviors and knowledge of malaria treatment guidelines for pregnant clients in a malaria endemic

region of western Kenya.

Chapter 7

116

METHODSPrescribing practice was observed by a) use of simulated client approach within randomly sampled

drug outlets, b) exit interviews with WOCBA (18‐49 years) and pregnant clients being treated for

febrile illness at all health facilities (HF) within the study area, and c) provider surveys using

structured questionnaires conducted for healthcare providers and drug dispensers to assess

knowledge of malaria treatment guidelines and self‐reported prescribing behavior for case

management of MiP. The latter surveys were administered following completion of the provider

practice component so as to avoid any influence in provider behavior.

StudySite&Sampling

This study was carried out from September to November 2013, in and around the KEMRI and CDC

Health and Demographic Surveillance System (HDSS) catchment area in Siaya County, Nyanza

Province in western Kenya. The HDSS collects birth, death, and migration information quarterly from

a large, rural area of approximately 700 km2 with 220,000 inhabitants, 95% of whom are ethnically

Luo [16]. Malaria transmission in western Kenya is perennial and holo‐endemic with peaks following

the two rainy seasons, from March through May and October through December). In the study area,

approximately 20% of pregnant women coming for the first antenatal clinic visit are parasitemic,

70% are anemic [17], 26% are HIV positive, and 18% of women delivering in Siaya District Hospital

had placental malaria [18].

HealthFacilitySelection

All facilities in the HDSS study area and within a 5 km buffer zone were deemed eligible if they were

operational, stocked antimalarials, and were visited by WOCBA for treatment of potential febrile

illness. Nine facilities were excluded due to ongoing studies that could have influenced study results.

Fifty‐two health facilities, including hospitals, health centers, and dispensaries, were eligible for the

study; 50 consented to participate.

DrugOutletsSelection

Prior to the start of data collection, a census was conducted of all registered and unregistered

entities selling antimalarial drugs within the HDSS border (Kioko et al., unpublished). Forty‐one drug

outlets were randomly selected from the 179 (99%) outlets that consented to participate at the time

of census; an additional 18 were randomly selected for a reserve list in case of closure or stock‐out

at the time of the simulated client component. Assuming that approximately 45% of providers have

adequate knowledge and prescribing practice [12], sampling 40 drug outlets out of 200 allowed an

estimation of the proportion of providers with adequate knowledge with 14% precision at 80%

power.

DataCollection

Training

Fieldworkers underwent two weeks of training, including interviewing techniques, data recording,

and piloting of survey tools. A subset of four fieldworkers (two women and two men) was trained on

the methodology behind the simulated client approach; and the standardized, pre‐determined

scenarios were piloted for an additional week in outlets outside the study area prior to

implementation.


117

ExitInterviewsinHealthFacilities

Patients were approached for eligibility assessment after completion of provider consultation, from

either outpatient department (OPD) or antenatal care clinic (ANC), and receipt of all prescribed

medications. Eligible patients included any WOCBA, either pregnant or non‐pregnant, that presented

with febrile illness and consented to participate in the study. Fieldworkers tried to interview at least

one of each of the following categories of patients per facility: 1) WOCBA who could potentially be

pregnant, 2) women in early pregnancy (1st trimester, defined as up to 14 weeks inclusive), and 3)

women in late pregnancy (2nd/3rd trimester, defined as 15 weeks gestation or greater). Pregnancy

status was based on patient report; gestational age and trimester were later confirmed by

calculation from patient reported date of last menstrual period (LMP).

After obtaining informed consent, exit interviews were conducted following a standard format. In

cases where an antimalarial contraindicated for pregnancy had been prescribed, the patient was

informed of the national treatment guidelines for MiP. The field supervisor (a Kenyan clinician) and

study coordinator were immediately informed and the recommended treatment was given to the

patient with appropriate dosage instructions and information.

SimulatedClientsinDrugOutlets

The simulated client (also known as mystery clients or shoppers [19,20]) approach was used to

assess prescribing practice within drug outlets. Female fieldworkers presented themselves as either

WOCBA or in early pregnancy, and male fieldworkers presented as the husband of a WOCBA or

woman in 3rd trimester of pregnancy. All simulated clients exhibited care‐seeking behavior for

uncomplicated malaria by presenting with general malaise, and, if prompted, complained of fever,

headache, chills, joint or muscle pain and nausea. The simulated clients were trained not to disclose

pregnancy status unless it was asked about by the dispenser. If dispensers failed to assess pregnancy

status, following receipt of a prescription the simulated clients would then disclose pregnancy status

(using local language to convey either early or late pregnancy depending on the pre‐set scenario)

and note any changes in the prescribed treatment or advice given. Simulated clients were able to

purchase medications up to an allotted 250 KSh (3.00 USD), but were instructed not to take

pregnancy or malaria diagnostic tests or treatment, if offered. The study coordinator and/or field

supervisor were in the vicinity at the time of simulation in case the simulated client was uncovered.

Immediately following completion of the scenario, the checklist for simulated client interaction was

completed under guidance of the study coordinator and/or field supervisor.

ProviderSurveysinHealthFacilities&DrugOutlets

Following the completion of exit interviews or client simulations at each facility or outlet, a separate

fieldworker administered a structured questionnaire to the provider to assess knowledge and self‐

reported prescribing practice, including: training, knowledge of symptoms, diagnosis, availability of

the most recent treatment guidelines, and treatment and preventive regimens for different

scenarios including pregnant women and the general population.

DataManagement&Analysis

Information for the drug outlet census and mapping and the provider survey components was

collected via personal digital assistant (PDA), and data for the simulated client and exit interview

components were collected via scannable form.

Chapter 7

118

All datasets were cleaned and analyzed using SAS 9.3 (SAS Institute, Cary, NC, USA). Descriptive

statistics were performed on all data to identify the extent of adherence to and knowledge of

Kenyan National Treatment Guidelines as they pertain to MiP across the provider study population.

Exit interview and simulated client data were analyzed to describe provider prescribing and

dispensing behavior in reference to pregnancy status, malaria diagnosis prior to treatment, correct

treatment and dosage, and provision of appropriate information as pertaining to treatment advice.

Provider survey data were also analyzed across these categories pertaining to the malaria treatment

guidelines. Variables within these categories were coded to give a threshold for dichotomous correct

vs. incorrect practice or knowledge for each category.

Correct practice and adequate knowledge definitions (Table 1) were based on the 2010 Kenyan

National Malaria Treatment Guidelines (MTGs) [7]; where these were insufficient, the 2010 WHO

Malaria Treatment Guidelines [6] were used. For exit interviews and client simulations, treatment

was considered correct if either first‐ or second‐line treatment was prescribed.

Table 1. Definitions of Correct Practice & Adequate Knowledge

Correct Malaria Diagnosis

Utilization of microscopy or RDT

Clinical diagnosis when diagnostic test unavailable

Correct Pregnancy Assessment

Inquired about pregnancy and/ or offered pregnancy test

Inquiry on LMP or gestational age

Correct Treatment & Dosage

Acceptable Knowledge Answers Acceptable Prescriptions in Practice

Non‐pregnant Non‐pregnant

1st‐line: Artemether‐lumefantrine (4x2x3) Artemether‐lumefantrine (4x2x3)

2nd‐line: DHA‐piperaquine (3x1x3 or 4x1x3) DHA‐piperaquine (3x1x3 or 4x1x3)

Quinine (2x3x7)

1st Trimester 1st Trimester

Quinine (2x3x7) Quinine (2x3x7)

2nd/3rd Trimester 2nd/3rd Trimester

Quinine (2x3x7) Quinine (2x3x7)

Artemether‐lumefantrine (4x2x3) Artemether‐lumefantrine (4x2x3)

DHA‐piperaquine (3x1x3 or 4x1x3)

Treatment regimens:

Artemether‐lumefantrine tablets (20/120 mg): 4 tablets, 2 times daily for 3 days (4x2x3)

DHA‐piperaquine tablets (40/320 mg): 3 or 4 tablets, once daily for 3 days (3x1x3) (4x1x3)

Quinine: 2 tablets of 300 mg, 3 times daily for 7 days (2x3x7)

Acronyms: RDT, rapid diagnostic test; LMP, date of last menstruation period; DHA, dihydroartemisnin

Chi square test or Fisher exact test were used to assess statistical significance of comparisons

between categorical variables; p≤0.05 indicates statistical significance. The total proportion of

providers (clustering on facility) who met adequate pregnancy assessment standards, adequate

malaria diagnostic standards, and correctly prescribed drug and dosage was calculated; these

measures were used to define overall correct prescribing practice. Logistic regressions were

performed at the individual provider level (clustering on provider) to identify potential predictors of

knowledge and practice level. Bivariate analyses of all provider characteristics were regressed upon


119

binary ‘Adequate Knowledge Score;’ variables identified as significant predictors of adequate

provider knowledge (p<0.2) were included in the multivariate model and further analyzed for

interaction to determine which provider characteristics were the best predictors of provider

knowledge. Bivariate and multivariate analyses of provider characteristics and correct provider

practice were done similarly.

Ethics

This study was approved by the ethical and institutional review boards of the U.S. Centers for

Disease Control and Prevention (CDC), the Kenya Medical Research Institute (KEMRI), Liverpool

School of Tropical Medicine (LSTM), and Emory University prior to the start of study. In addition,

permission to operate within the study area was obtained via meetings with the Siaya County

Director of Health and the District Health Management Teams for Bondo, Gem, Rarieda and Siaya

districts. The community advisory boards for each district were notified of the study presence within

the area. Signed letters of approval were obtained from the Siaya County Director of Health and

each District Health Management Team. Verbal consent was obtained from health facility in‐charges

prior to any interview activities at the respective facility; informed consent regarding future

potential participation in a study for MiP treatment assessment was obtained from the dispenser

during the drug‐outlet mapping and census. Written, informed consent was obtained from all

providers and patients interviewed during the study period in the participants’ preferred language of

Dholuo, Kiswahili, or English. Study participants were free to withdraw from the study at any time.

RESULTS

PrescribingPracticeandDispensingBehaviours:ExitInterviewsinHealthFacilities

A total of 210 patients were interviewed across 51 health facilities: 108 non‐pregnant women, 19

women in the 1st trimester of pregnancy, 77 women in 2nd or 3rd trimester of pregnancy, and 6

women who were unsure of their pregnancy status; one interview with a woman in her 2nd trimester

of pregnancy with severe malaria was excluded from the analysis. The average age of the patients

was 26 years; 26% of respondents had completed secondary school (Tables 2a & 2b).

Table 2a. Exit interview: health facility characteristics

Total Facilities Total Interviews

N % N %

District 51 209

Bondo 6 11.8 18 8.6

Gem 20 39.2 89 42.6

Rarieda 9 17.6 28 13.4

Siaya 16 31.4 74 35.4

Facility Type

Hospital 4 7.8 18 8.6

Health Center 19 37.3 83 39.7

Dispensary 28 54.9 108 51.7

Facility Managing Authority

Government 44 86.3 188 90.0

Mission 2 3.9 4 1.9

Private 5 9.8 17 8.1

Chapter 7

120

Table 2b. Health facility exit interview: respondent characteristics

Overall Unsure Non‐Pregnant 1st trimester** 2nd/3rd trimester

N % N % N % N % N %

Education Level 209 6 108 19 76

No Education 16 7.7 1 16.7 13 12.0 1 5.3 1 1.3

Primary 138 66.0 5 83.3 68 63.0 13 68.4 52 68.4

Secondary 38 18.2 0 0 20 18.5 2 10.5 16 20.8

Higher than secondary

17 8.1 0 0 7 6.5 3 15.8 7 9.2

Age mean (range) & Standard Deviation

26.3 (17‐48)

7.2 26.8

(18‐45) 9.7

28.4 (18‐48)

7.8 24.5

(18‐35) 5.0

23.9 (17‐40)

5.6

Symptoms Reported to provider*

Fever 138 66.0 4 66.7 76 70.4 13 68.4 45 59.2

Headache 183 87.6 6 100.0 98 90.7 16 84.2 63 82.9

Pain 104 49.8 4 66.7 53 49.1 7 36.8 24 31.6

Nausea 72 34.4 2 33.3 25 23.1 7 36.8 31 40.8

Malaise 80 38.3 2 33.3 39 36.1 5 26.3 34 44.7

Chills 17 8.1 0 0.0 9 8.3 2 10.5 6 7.9

Stomach Pain 23 11.0 1 16.7 11 10.2 3 15.8 8 10.5

Cough 18 8.6 1 16.7 6 5.6 3 15.8 8 10.5

Dizziness 3 1.4 0 0.0 0 0.0 0 0.0 3 3.9

Diarrhea 2 1.0 0 0.0 0 0.0 1 5.3 1 1.3

Gravidity

0 33 21.2 1 33.3 11 17.5 3 15.8 18 25.4

1 31 19.9 0.0 11 17.5 7 36.8 14 18.3

2 33 21.2 0.0 12 19.0 3 15.8 18 25.4

3‐4 32 20.5 1 33.3 12 19.0 5 26.3 13 19.7

5‐6 23 14.7 1 33.3 13 20.6 1 5.3 8 11.3

7+ 4 2.6 0 0.0 4 6.3 0 0.0 0 0.0

Missing 53 3 45 0 5

*2 reported no symptoms to provider **Patients with gestational age of up to 14 weeks, 6 days were included in 1st trimester given that treatment guidelines use 'quickening' as a treatment indicator

MalariaDiagnosisinHealthFacilities

Of the 209 women interviewed, 160 (77%) women were tested for malaria (RDT or microscopy). Of

those women who were not tested, 28 women were appropriately clinically diagnosed in facilities

that did not have the capacity to perform malaria diagnostics. Taking this into account, 90% were

properly assessed for malaria according to the facility diagnostic capability (Table 3).


121

Table 3 Malaria diagnostics practice in health facilities as observed through exit interviews across facility type

Overall Hospital Health Center Dispensary

N % N % N % N %

Diagnostic Test Performed* 209 18 83 108

Yes 160 77.0 16 88.9 75 90.4 70 63.9

No 48 23.0 2 11.1 8 9.6 38 35.2

Don't Know 1 0.5 0 0.0 0 0.0 1 0.9

Malaria Test Results 160 16 75 69

Positive 151 94.4 14 87.5 70 93.3 68 97.1

Negative 3 1.9 2 12.5 1 1.3 0 0.0

Don't Know 6 3.8 0 0.0 4 5.3 2 2.9

Test Location 160 16 75 69

OPD 28 17.5 0 0.0 3 4.0 25 36.2

Laboratory 132 81.9 16 100.0 72 96.0 44 62.3

Pharmacy 1 0.6 0 0.0 0 0.0 1 1.4

No/Unknown Diagnostic Test 49 23.3 2 11.1 18 21.7 57 52.3

Correct Clinical Diagnosis** 28 57.1 NA NA 4 50.0 24 61.5

Incorrect Clinical Diagnosis*** 21 42.9 2 100.0 4 50.0 15 38.5

Correct Malaria Diagnosis 188 90.0 16 88.9 79 95.2 93 81.6

*Tested via Rapid Diagnostic Test (RDT) or microscopy **Correct Clinical Diagnosis indicates women presenting with fever, multiple symptoms, and/or were pregnant with symptom(s) at facilities without laboratory or RDT diagnostic capacity ***Incorrect Diagnosis indicates patients treated for malaria without diagnostic testing at facilities where it was available)

or without clinical presentation if at a facility with no diagnostic capacity

PregnancyAssessmentinHealthFacilities

Overall, 92 (44%) of 209 patients were asked about their potential pregnancy status; inquiry was

more common among pregnant patients (90% and 61% for 1st and 2nd/3rd trimesters, respectively)

than among non‐pregnant patients (26%, Table 4). Only 43% of women were asked about their LMP;

this occurred with much greater frequency in pregnant (70%) versus non‐pregnant patients (24%).

Only 20 (10%) women were offered a pregnancy test; 80% of these women were pregnant (9 in the

1st, 5 in the 2nd and 2 in the 3rd trimester). Three of the women who were unsure of their pregnancy

status and 23 women who reported to us that they were not pregnant had LMPs of greater than 6

weeks, thus should have been tested for pregnancy. Overall, 52% of patients were correctly assessed

for pregnancy status (Table 4); however this was significantly higher in pregnant women versus self‐

reported non‐pregnant women (84% vs. 24%, p<0.0001). There was also a statistically significant

difference across facility types (hospital 78%, health center 53% and dispensaries 45%; p=0.03).

Chapter 7

122

Table 4. Pregnancy assessment practice in Health Facilities as observed through exit interviews stratified across pregnancy status.

Overall Unknown Non‐Pregnant 1st Trimester

2nd/3rd Trimester

N % N % N % N % N %

All Patients 209 6 108 19 76

Pregnancy Status Inquiry 92 44.0 1 16.7 28 25.9 17 89.5 46 60.5

Pregnancy Test Offered 20 9.6 0 0 4 3.7 9 47.4 7 9.2

LMP Inquiry 89 42.6 1 16.7 26 24.1 16 84.2 46 60.5

Pregnancy Duration/Timing 67 70.5 NA NA 14 73.7 53 69.7

Additional Confirmation* 41 43.2 NA NA 5 26.3 36 47.4

Correct pregnancy assessment 108 51.7 1 16.7 26 24.1 18 94.7 63 82.9

*Additional confirmation included palpation in 1st trimester, and palpation or observation in 2nd/3rd trimester cases.

TreatmentPrescribedinHealthFacilities

An antimalarial medication was prescribed to 205 (98%) of the 209 women; the most frequently

prescribed was AL (84%), followed by quinine (14%), and sulfadoxine‐pyrimethamine (SP) (3%).

Overall, 66% of providers prescribed the correct treatment and dosage to the patient across all

pregnancy scenarios (Table 5). AL was incorrectly prescribed in 1st trimester to 3/19 (16%) women.

The majority of prescriptions for AL and SP were for the correct dosage (73% and 71%, respectively);

in contrast, the correct dose of quinine was prescribed only 31% of the time. The correct drug and

dosage was prescribed more frequently to non‐pregnant patients (70%) and those in the 2nd/3rd

trimester of pregnancy (66%) than to those in 1st trimester (32%, p=0.001). Hospital‐based providers

(78%) were the most likely to provide correct treatment to non‐pregnant patients (versus Health

Centre‐72%, Dispensary‐ 68%). However, they were least likely to provide correct treatment to

pregnant patients regardless of trimester (hospitals 33%, health centres 62% and dispensaries 58%).

The first dose of antimalarial was directly observed in 25% of cases. Very few patients (7%) were

informed of potential side effects.


123

Table 5. Malaria treatment practice in health facilities as observed through exit interviews stratified across pregnancy status.

Overall Unsure* Non‐Pregnant 1st Trimester 2nd/3rd Trimester

N % N % N % N % N %

Prescribed Antimalarials 205 6 107 18 74

Antimalarial prescribed Correct Dosage (tabs x doses x days)**

Artemether‐lumefantrine 173 83.9 6 102 94.4 3 15.8 61 82.4

(4x2x3) 125 72.7 2 33.3 74 72.5 3 100.0 46 75.4

DHA‐piperaquine 2 1.0 0 0.0 2 1.9 0 0.0 0 0.0

(3x1x3) 1 50.0 1 50.0

Quinine 29 14.1 0 0.0 4 3.7 13 68.4 12 16.2

(2x3x7) 8 27.6 0 0.0 6 46.2 2 16.7

(150mgxkg) 1 3.4 1 25.0 0 0.0 0 0.0

Sulfadoxine‐pyrimethamine 7 3.4 0 0.0 1 0.9 2 10.5 4 5.4

(3x1x1) 5 71.4 0 0.0 2 100.0 3 75.0

Artemether Injection 1 0.5 0 0.0 1 0.9 0 0.0 0 0.0

(60mg) 1 100.0 1 100.0

Correct Drug 195 98.0 NA 108 100.0 13 68.4 75 97.4

Correct Drug & Dosage 131 65.8 NA 76 70.4 6 31.6 50 66.2

Concomitant Medications

Analgesic 148 72.2 4 66.7 77 72.0 15 83.3 52 70.3

Antibiotic 75 36.6 1 16.7 37 34.6 6 33.3 31 41.9

Treatment Advice

Reason for Prescription 58 28.3 1 16.7 32 29.9 4 22.2 21 28.4

Side Effects 14 6.8 0 0.0 5 4.7 4 22.2 5 6.8

Any other advice 46 22.3 2 33.3 20 18.7 5 27.8 19 25.7

Any treatment Advice 83 40.5 2 33.3 41 38.3 10 55.6 30 40.5

*Unsure pregnancy status refers to women who self‐reported as not knowing if they were pregnant or not, correct prescribing practice was not assessable for those; these women are not included in the ‘Correct Drug’ or ‘Drug & Dosage’ denominator . **Percentage for correct dosage is based on the numbers receiving the specific antimalarial. ***Any other advice includes patient‐reported advice by the provider including emphasis of complete medication regimen, eating prior to taking medication, sleeping under ITNs, etc.

PredictorsofCorrectPracticeinHealthFacilities

Significant predictors of correct prescribing and diagnostic practice in health facilities were

respondent cadre, dispensing medication and working in the OPD (Table 6). Health facility‐based

pharmacists were more likely to correctly prescribe and diagnose patients then their clinical

counterparts [RNs, COs, MDs] (OR=1.8 [95% CI 1.4‐2.2]). Likewise, providers that reported

dispensing medicines were more adherent to correct practice than their counterparts who did not

dispense medicines (OR=1.2 [95% CI 1.1‐1.4]). Providers that reported sometimes working in the

OPD were less likely to correctly diagnose and treat patients than those who reported always

working in the OPD (OR=0.8 [95% CI 0.7‐0.9]). Age and gender were controlled for in the model;

however they had no significant effect on correct prescribing practice. Neither malaria diagnostic

training nor MiP training within the last five years were statistically significant predictors of correct

provider practice in health facilities.

Chapter 7

124

Table 6 Predictors of correct prescribing and diagnostic practice in health facilities

Provider Characteristic N % Crude OR

95% CI P

value Adjusted*

OR 95% CI P value

Respondent Cadre 184

Nurse (ref) 109 59.2 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐

Clinical Officer/M.D. 52 28.3 1.1 (0.9, 1.3) 0.44 1.1 (0.9, 1.3) 0.43

Pharmacist 5 2.7 1.6 (1.1, 2.3) 0.01 1.8 (1.4, 2.2) <0.001

Other 18 9.8 0.7 (0.7, 0.8) <0.001 0.8 (0.7, 0.9) <0.001

Dispenses Medicine

No (ref) 19 10.3 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐

Yes 165 89.7 1.2 (1.0, 1.3) 0.04 1.2 (1.1, 1.4) <0.001

Works in Outpatient Department

Always (ref) 95 51.6 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐

Sometimes 68 37.0 0.9 (0.7, 1.0) 0.09 0.8 (0.7, 0.9) <0.001

Never 21 11.4 0.7 (0.6, 0.8) <0.001 0.9 (0.7, 1.1) 0.18

Gender

Male (ref) 100 54.3 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐

Female 84 45.7 0.9 (0.8, 1.1) 0.23 1.1 (0.9, 1.3) 0.57

Age 1.0 (1.0, 1.0) 0.31 1.0 (1.0, 1.0) 0.42

Malaria Diagnostic training within 5 years

No (ref) 51 27.7 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐

Yes 133 72.3 0.9 (0.8, 1.1) 0.46 ‐‐ ‐‐ ‐‐

MiP training within 5 years

No (ref) 118 64.1 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐

Yes 66 35.9 0.9 (0.8, 1.1) 0.50 ‐‐ ‐‐ ‐‐

* Adjusted for respondent cadre, dispensing medicine, working in OPD, gender and age.

PrescribingPracticeandDispensingBehaviours:SimulatedClientsinDrugOutlets

Simulations were completed at 41 drug outlets; two facilities were homesteads (individuals selling

antimalarials from their residence) and were included as informal drug outlets. There were 81

simulated client‐drug dispenser interactions with 154 total scenarios simulated Dispensers were a

mean of 33 years old, and 59% were female. The majority had completed only primary (18%) or

secondary school (41%). Between 36 and 40 simulations per scenario were completed (Table 7).


125

Table 7 Drug outlet & simulation characteristics

Drug Outlets Overall Registered Pharmacy

Informal Drug Outlet*

General Shop

N %

[SD] N

%

N % N %

District/Sub‐county 41 9 15 17

Gem 13 31.7 3 33.3 6 40.0 4 23.5

Rarieda 10 24.4 2 22.2 5 33.3 3 17.6

Siaya 18 43.9 4 44.4 4 26.7 10 58.8

Provider Gender

Male 17 41.5 5 55.6 5 33.3 7 41.2

Female 24 58.5 4 44.4 10 66.7 10 58.8

Education Level**

Primary School 7 17.9 0 0.0 0 0.0 7 41.2

Secondary School 16 41.0 5 55.6 6 40.0 5 29.4

Higher Education 6 15.4 2 22.2 2 13.3 2 11.8

Clinical Officer/MD 1 2.6 0 0.0 1 6.7 0 0.0

Registered Midwife/Nurse 2 5.1 1 11.1 1 6.7 0 0.0

Enrolled Midwife/Nurse 1 2.6 0 0.0 1 6.7 0 0.0

Pharmacist 3 7.7 1 11.1 1 6.7 1 5.9

Other technical 3 7.7 0 0.0 3 20.0 0 0.0

Age mean (range) 32.5

(19‐60) [9.4]

28.0 (19‐46)

[8.5] 32.3

(20‐50) [8.2]

35.5 (21‐60)

[10.5]

Simulation Characteristics 154 35 56 63

WOCBA 39 25.3 9 25.7 15 26.8 15 23.8

1st Trimester Pregnancy 39 25.3 9 25.7 14 25.0 16 25.4

Relative of WOCBA 36 23.4 8 22.9 12 21.4 16 25.4

Relative of 3rd Trimester 40 26.0 9 25.7 15 26.8 16 25.4

*There were 2 homesteads visited; these were included as informal drug outlets. **Education level was obtained from matched provider surveys (39 of 41 dispensers were interviewed) and this is missing for 2 providers in the general shop

MalariaDiagnosisinDrugOutlets

Dispensers assessed for malaria in 37% of all interactions (Table 8). Assessment for malaria was

more common when the simulator posed as the patient as opposed to the patient’s husband. RDTs

were offered in 10% of interactions where the client was present; RDTs were offered in both of the

homestead client‐present scenarios. Simulators posing as husbands were asked if their febrile wife

had been previously given a diagnostic test in 7% of interactions. 36% of dispensers asked about

symptoms, with just less than half of these inquiring about specific symptoms. No drug outlet

dispenser offered to take the simulated client’s temperature. A prescription was requested by the

dispenser in only 5% of interactions.

Chapter 7

126

Table 8 Malaria diagnostics practice in drug outlets as observed through simulated clients across pregnancy status

Overall WOCBA / 1st Trimester Relative of WOCBA/

3rd Trimester N % N % N %

Symptoms 81 40 41

Any Inquiry 29 35.8 17 42.5 12 29.3

Specific 14 17.3 12 30.0 2 4.9

Fever 8 9.9 7 17.5 1 2.4

Chills 3 3.7 3 7.5 0 0.0

Headache 11 13.6 9 22.5 2 4.9

Nausea 6 7.4 4 10.0 2 4.9

Pain 4 4.9 3 7.5 1 2.4

Prescription 4 4.9 1 2.5 3 7.3

Temperature 0 0.0 0 0.0 NA

Diagnostic Test or Test Inquiry 7 8.6 4 10.0 3 7.3

Any Malaria Diagnostic 30 37.0 18 45.0 12 29.3

PregnancyAssessmentinDrugOutlets

There were only six unprompted pregnancy inquiries across 81 total interactions (7%) and a

pregnancy test was offered in two of these cases. Gestation was inquired in four of six interactions.

Dispensers were informed of positive pregnancy status in 73 interactions where there was no initial

inquiry on the part of the dispenser; in 58% of these interactions the dispenser followed up with

gestational age or LMP inquiry. Inquiry about gestational age was highest in registered pharmacies

(77%) and informal drug shops (74%), with general shops significantly lower (33%, p=0.003); this did

not differ between interactions where the client was the patient versus the patient relative (table 9).


127

Table 9 Pregnancy assessment practice in drug outlets as observed through simulated clients across pregnancy status

Overall 1st Trimester 3rd Trimester

N % N % N %

Pregnancy Inquiry 81 40 41

Unprompted Pregnancy Inquiry 6 7.4 2 5 4 9.8

Confirmation

Timing 4 66.7 2 100.0 2 50.0

Last menstruation period 1 25.0 1 50.0 0 0.0

Gestation 3 75.0 1 50.0 2 100.0

Pregnancy Test Offered 2 33.3 2 100.0 NA

None 2 33.3 0 0.0 2 50.0

Informed Provider of Pregnancy Status 73 90.1 37 92.5 36 87.8

Confirmation

Timing 42 57.5 22 59.5 20 55.6

Last menstruation period 2 4.8 2 9.1 0 0.0

Gestation 40 95.2 20 90.9 20 100

Pregnancy Test Offered 0 0.0 0 0.0 0 0.0

None 31 42.5 15 40.5 16 44.4

Correct Pregnancy Assessment* 48 59.3 23 57.5 25 61

*Correct Pregnancy Assessment indicates that the provider confirmed pregnancy status via LMP, gestational inquiry, or pregnancy test

TreatmentDispensedinDrugOutlets

Correct prescribing and diagnostic practice was only observed in 3% of the 154 interactions in drug

outlets, with no significant difference across outlet types (Table 11). There were highly significant

differences in correct treatment and dosage across pregnancy status, with 63% of non‐pregnant,

38% of 3rd trimester, and 0% of 1st trimester client simulations receiving the appropriate treatment

(p<0.0001). Antimalarials were dispensed in 83% of all interactions; AL was most commonly

dispensed (76%), followed by SP (22%). Quinine was not dispensed in the drug outlets. Dispensers

were 7.6 times more likely to prescribe SP for treatment of acute malaria to pregnant versus non‐

pregnant women (p<0.0001). About half (51%) of the 39 1st trimester clients were prescribed AL; all

but 1 of the remaining clients were prescribed SP. AL was initially prescribed to over 90% of

simulated client patients; in 27% of cases treatment was changed from AL to SP after finding out the

patient was pregnant, regardless of trimester, and in another 17% AL was withdrawn and the patient

was referred to a health facility.

Chapter 7

128

Table 10. Correct treatment and dosage characteristics by pregnancy status in drug outlets

Overall WOCBA 1st Trimester 3rd Trimester

N % N % N % N %

154 75 39 40

Dispensed Antimalarials 127 82.5 66 88.0 32 82.1 29 72.5

Antimalarial dispensed Correct Dosage (tabs x doses x days)*

Artemether lumefantrine 96 75.6 60 90.9 20 62.5 16 55.2

(4x2x3) 78 61.4 47 78.3 16 80.0 15 51.7

Artesunate amodiaquine 1 0.8 1 1.5 0 0.0 0 0.0

(4x1x3) 0 0.0 0 0.0

Amodiaquine 2 1.6 1 1.5 1 3.1 0 0.0

(3x1x1) 1 0.8 0 0.0 1 100.0

Sulfadoxine Pyrimethamine 28 22.0 4 6.1 11 34.4 13 44.8

(3x1x1) 22 17.3 4 100.0 7 63.6 11 84.6

Correct Drug 77 60.6 61 92.4 0 0.0 16 55.2

Correct Drug & Dosage 62 48.8 47 71.2 0 0.0 15 51.7

Concomitant Medications

Analgesic 102 80.3 53 80.3 28 87.5 21 72.4

Multivitamin 2 1.6 0 0.0 0 0.0 2 6.9

Treatment Advice

Dosage directions 110 86.6 56 84.8 26 81.3 28 96.6

Visual instructions 72 56.7 35 53.0 20 62.5 17 58.6

Emphasize need to finish dose 11 8.7 4 6.1 3 9.4 4 13.8

Side effects 2 1.6 1 1.5 0 0.0 1 3.4

Advice if symptoms persist 6 4.7 3 4.5 3 9.4 0 0.0

Any treatment advice 110 86.6 56 84.8 26 81.3 28 96.6

*Percentage for correct dosage is based on the numbers receiving the specific antimalarial

Of 27 clients not given an antimalarial, 17 (63%) were referred to the hospital, and 8 (30%) clients

did not receive a medication due to antimalarial stock out. Other reasons for not receiving treatment

included refusal to treat without a prescription, diagnostic test, or clinical evaluation.

Prior to dispensing, only 16% of dispensers questioned the simulated client if any previous treatment

had been given for the current illness and only 5% asked about potential allergies. Dosage directions

were given to 87% of simulated clients.

Table 11. Antimalarial dispensing practice in drug outlets as observed through simulated clients across pregnancy status.

Overall Registered Pharmacy Informal Drug Shop General Shop

N % N % N % N %

154 35 48 71

Malaria Diagnostics 55 35.7 13 37.1 21 43.8 21 29.6

Pregnancy Assessment 47 30.5 14 40.0 18 37.5 15 21.1

Treatment & Dosage 62 40.3 18 51.4 25 52.1 19 26.8

Non‐pregnant N=75 47 62.7 14 82.4 19 82.6 14 40.0

1st Trimester N=39 0 0.0 0 0.0 0 0.0 0 0.0

3rd Trimester N=40 15 37.5 4 44.4 6 46.2 5 27.8

Correct Practice 5 3.2 1 2.9 1 2.1 3 4.2


129

PredictorsofCorrectDispensingPracticeinDrugOutlets

Prior MiP training was the only predictor of correct practice in drug outlets in the adjusted model

(Table 12). Dispensers who had been trained on MiP in the last 5 years were more likely to have

correct prescribing and diagnostic practice than those who had not been trained (p=0.05). Age,

gender, and prescribing status were controlled for in the model; however they had no significant

effect on correct prescribing practice.

Table 12. Predictors of correct dispensing practice in drug outlets


95% CI P

value Adjusted OR*

95% CI P

value

Respondent Cadre 113

Pharmacist (ref) 45 39.8 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐

Clinical Officer/M.D./Nurse 7 6.2 1.2 (0.9, 1.4) 0.23 1.0 (0.9, 1.2) 0.56

Shopkeeper/Other 61 54.0 1.0 (1.0, 1.1) 0.14 1.0 (1.0, 1.1) 0.28

Prescribes Medicine

No (ref) 51 45.1 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐

Yes 62 54.9 1.1 (1.0, 1.1) 0.06 1.0 (1.0, 1.1) 0.15

Malaria diagnostic training within 5 years

No (ref) 94 83.2 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐

Yes 19 16.8 1.1 (1.0, 1.2) 0.13 ‐‐ ‐‐ ‐‐

MiP Training within 5 years

No (ref) 99 87.6 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐

Yes 14 12.4 1.1 (1.0, 1.3) 0.07 1.1 (1.0, 1.3) 0.05

Sources of Information (ref='No')

CME as a source of info 29 25.7 1.1 (1.0, 1.2) 0.04 ‐‐ ‐‐ ‐‐

Gender

Male (ref) 42 37.2 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐

Female 71 62.8 1.0 (0.9, 1.0) 0.34 1.0 (0.9, 1.0) 0.17

Age 1.0 (1.0, 1.0) 0.31 1.0 (1.0, 1.0) 0.15

* Adjusted for respondent cadre, prescribing medicine, MiP training, gender and age. Malaria diagnostic training was not included in the adjusted model due to collinearity with MiP training. Continuing Medical Education (CME) was not included in adjusted model as it was not statistically significant.

Chapter 7

130

KnowledgeoftheNationalMalariaTreatmentGuidelinesamongProviders

CharacteristicsofRespondents

We surveyed 114 providers across 88 locations; 75 in HF and 39 in drug outlets. 44% of respondents

were nursing staff, 15% were COs/MDs, 18% pharmacists, and 13% were shopkeepers. 68% of

providers stated that they both prescribed and dispensed medication (Table 13a & 13b).

Table 13a. Facility characteristics from the provider survey on national malaria treatment guidelines

Total Facilities Total Providers Surveyed

N % N %

District 88 114

Bondo 6 6.8 9 7.9

Gem 32 36.4 42 36.8

Rarieda 17 19.3 19 16.7

Siaya 33 37.5 44 38.6

Facility Type

Hospital 4 4.5 6 5.3

Health Center 19 21.6 28 24.6

Dispensary 26 29.5 41 36.0

Total Health Facilities 49 55.7 75 65.8

Drug Outlet Type N % N %

Registered Pharmacy 9 10.2 9 7.9

Informal Drug Shop 13 14.8 13 11.4

General Shop 15 17.0 15 13.2

Homestead 2 2.3 2 1.8

Total Drug Outlets 39 44.3 39 34.2

Facility Managing Authority

Government 42 47.7 67 58.8

Mission 2 2.3 3 2.6

Private 44 50.0 44 38.6


131

Table 13b. Provider characteristics from the provider survey on national malaria treatment guidelines

Overall Health facilities Drug Outlets

N % N % N %

114 75 39

Sex

Male 54 47.4 39 52.0 15 38.5

Female 60 52.6 36 48.0 24 61.5

Respondent Cadre 0.0

Registered Nurse 33 28.9 32 42.7 1 2.6

Enrolled Nurse 16 14.0 16 21.3 0 0.0

Clinical Officer/MD 18 15.8 17 22.7 1 2.6

Pharmacist 20 17.5 5 6.7 15 38.5

Shopkeeper 15 13.2 0 0.0 15 38.5

CHW/VR/other* 12 10.5 5 6.7 7 17.9

Professional Qualification 0.0

Primary School 9 7.9 2 2.7 7 17.9

Secondary School 23 20.2 7 9.3 16 41.0

Higher Education 19 16.7 13 17.3 6 15.4

Clinical Officer/MD 14 12.3 13 17.3 1 2.6

Registered Midwife/Nurse 22 19.3 20 26.7 2 5.1

Enrolled Midwife/Nurse 11 9.6 10 13.3 1 2.6

Pharmacist 4 3.5 1 1.3 3 7.7

Other technical 12 10.5 9 12.0 3 7.7

* Other included clerk, economist, statistical clerk, and support staff

NationalMalariaTreatmentGuidelinesAwareness

75% of all providers said they were aware of the National Malaria Treatment Guidelines (MTGs);

67% had read the MTGs, 55% were in possession of them, and 58% were aware of the government

initiative to disseminate them (Table 14). However, HF providers were much more likely to respond

in the affirmative to all MTG‐related questions compared to drug outlet providers. The majority of

all providers (54%) had attended a malaria management workshop (94% within the past 5 years),

however only 31% of all providers had attended a workshop specific to MiP. Over 86% of those that

reported attending any workshop within the past 5 years were HF providers.

Chapter 7

132

Table 14. Malaria treatment guideline awareness, comparing health facilities vs. drug outlets

Overall Health Facilities Drug Outlets P‐value*

N % N % N %

MTGs 114 75 39

Awareness of Government Initiative 66 57.9 62 82.7 4 10.3 p<0.001

Read the MTGs 76 66.7 67 89.3 9 23.1 p<0.001

In Possession 63 55.3 60 80.0 3 7.7 p<0.001

Additional Materials 73 64 71 94.7 2 5.1 p<0.001

Awareness of MTGs 85 74.6 74 98.7 11 28.2 p<0.001

Additional Sources of Information

Training/CME 64 56.1 52 69.3 12 30.8 p<0.001

DHMT/health facility memos 46 40.4 38 50.7 8 20.5 p<0.001

Colleagues 40 35.1 28 37.3 12 30.8 p=0.25

Media 57 50 33 44.0 24 61.5 p=0.04

Medical Journals 19 16.7 19 25.3 0 0.0 p<0.001

Medical Representatives 18 15.8 11 14.7 7 17.9 p=0.33

Other** 9 7.9 3 4.0 7 17.9 p=0.01

Training Workshops

Malaria Training 62 54.4 51 68.0 11 28.2 p<0.001

within past 5 years 58 93.5 50 98.0 8 72.7 p=0.01

training prior to 2008 4 6.5 1 2.0 3 27.3

MIP Training 35 30.7 30 40.0 5 12.8 p=0.001

within past 5 years 32 91.4 28 93.3 4 80.0 p=0.21

training prior to 2008 3 8.6 2 6.7 1 20.0

*P‐values from Chi‐square test **Other includes community meetings (Barazas), CDC staff, NGO staff, & Village Reporters Acronyms: MTGs, malaria treatment guidelines; CME, continuing medical education; DHMT, district health medical team; MiP, malaria in pregnancy

MalariainPregnancyConsequenceandDiagnosisKnowledge

98% of all providers surveyed knew that MiP can cause adverse effects; 90% were able to cite at

least one adverse effect. 90% of all providers suspected malaria in cases of fever; other clinical

symptoms cited included headache (84%), vomiting (82%), body ache (67%), and chills (65%).

Providers in health facilities had statistically significant greater knowledge of both MiP consequences

(maternal anemia, maternal death, LBW, and neonatal death) and clinical symptoms (fever,

headache, body ache, chills, vomiting) compared to drug outlet providers. 84% of providers in health

facilities reported utilizing laboratory diagnosis (81% RDT, 70% microscopy); however, of those that

did not use lab diagnostics, 25% reported always treating clinically (versus regularly, sometimes, and

never). In drug outlets, where diagnostics are not widely available, 33% reported utilizing RDTs or

microscopy while 33% reported that they always treat clinically. Correct malaria diagnostic usage

was reported more often by HF providers (97%) versus drug outlets (85%) (p=0.018). Of those

providers that reported always (59%) or sometimes performing a pregnancy assessment, 79%

reported asking for LMP and 48% reported offering a pregnancy test. Knowledge of correct

pregnancy assessment was higher among health facilities providers (93%) versus their drug outlet‐

based counterparts (49%, p<0.001) (Table 15).


133

MalariaTreatment&TreatmentContraindicationKnowledge

Thirty‐eight (51%) HF providers knew the correct 1st‐line treatment and dosage for all scenarios

(non‐pregnant, 1st trimester, and 2nd/3rd trimester women) compared to none of the 39 drug

dispensers. Correct knowledge specifically for 1st trimester patients was given by 56% of HF

providers and none of drug dispensers (p<0.0001). Correct treatment knowledge for 2nd/3rd

trimester patients was reported in 87% of HF and 39% of drug outlets (p<0.0001). Overall provider

knowledge was considerably higher for 1st‐line treatment versus 2nd‐line treatment (p<0.0001, Table

15).

Table 15. Adequate provider knowledge of malaria in pregnancy based on national malaria treatment guidelines comparing health facilities to drug outlets


N % N % N %

114 75 39

Consequences of MiP 112 98.2 74 98.7 38 97.4 p=0.34

Awareness of MTGs 85 74.6 74 98.7 11 28.2 <0.0001

Malaria Diagnostics 107 93.9 73 97.3 33 84.6 p=0.01

Pregnancy Assessment 89 78.1 70 93.3 19 48.7 p<0.0001

Treatment & Dosage 38 33.3 38 50.7 0 0.0 p<0.0001

NP‐1st Line 97 85.1 72 96.0 25 64.1 p<0.0001

NP‐2nd Line 43 37.7 37 49.3 6 15.4 p=0.0002

1st Tri‐ 1st Line 42 36.8 42 56.0 0 0.0 p<0.0001

2nd/3rd Tri‐ 1st Line 80 70.2 65 86.7 15 38.5 p<0.0001

Severe Malaria 72 63.2 63 84.0 9 23.1 p<0.0001

Adequate Knowledge 34 29.8 34 45.3 0 0.0 p<0.0001

*P‐values from Chi‐square test and Fisher Exact used for strata with <5 observations. Acronyms: MTG, malaria treatment guidelines, MiP, malaria in pregnancy, NP, non‐pregnant; Tri, trimester of pregnancy

SP was incorrectly cited as 1st and 2nd line treatment for adults by 4% of providers, in 1st trimester

pregnant patients by 19% of providers, in 2nd/3rd trimester by 7% of providers, and for treatment of

severe malaria in pregnancy by 5% of providers. Two‐thirds cited IPTp with SP as a preventive

measure for MiP; however, only 54% knew that SP could only be used as preventive therapy and not

as treatment. Additionally, 6% of all providers thought AL could be used as preventive therapy and

another 14% were not able to cite a drug for preventive treatment. An ACT was incorrectly cited as

the appropriate treatment in 1st trimester pregnancies by 5% of HF providers and 18% of drug outlet

providers.

The majority of providers stated the reason behind their chosen antimalarial for 1st and 2nd‐line

treatment was either due to the observed effectiveness of the drug in their practice (45% and 36%

,respectively) or due to national guidelines (35% and 30%, respectively). However, of the providers

that cited the national guidelines as the reason, 83% and 74% had chosen an antimalarial not

recommended by the national MTGs for 1st‐line and 2nd‐line, respectively. 71% of HF providers and

28% of drug outlet dispensers cited 1st trimester as a contraindication for AL treatment (p<0.0001).

37% of HF providers cited ‘allergy’ as a contraindication for both 1st and 2nd line treatments versus

13% (p=0.003) and 3% (p<0.0001) of drug outlet providers, respectively.

Chapter 7

134

Correct treatment knowledge for severe malaria in pregnancy was given by 84% of HF providers and

23% of drug outlet providers (p<0.0001). Overall adequate knowledge of MiP (inclusive of

diagnostics, pregnancy assessment, and treatment knowledge) was reported by 34 providers (30%),

all of which were HF based.

MalariainPregnancyComprehensiveCareKnowledge

85% of HF providers reported giving any type of comprehensive care practices to pregnant patients

with malaria versus 26% of drug outlet providers (p=0.8). Fetal monitoring (60% HF, 17% drug outlet)

and anemia treatment (61% HF, 10% drug outlet) were the most commonly cited practices, followed

by hypoglycaemia prevention (43% HF, 0% drug outlet) and guidance on antipyretic usage (32% HF,

3% drug outlet). 87% of all providers reported that they give pregnant patients instructions on

treatment; 56% reported informing a pregnant patient of potential side effects, 52% reported telling

the patient to return if symptoms continued, and 7% reported informing the patient of danger signs

to look out for. A greater proportion of HF providers than drug outlet‐based providers reported

giving such information to a pregnant patient (p<0.001), with the exception of danger signs (Table

16).

Table 16. Comprehensive care practices provided during pregnancy, comparing health facilities vs. drug outlets


N % N % N %

114 75 39

Care Practices in Pregnancy

Prevent Hypoglycemia 32 28.1 32 42.7 0 0.0 p<0.001

Fetal Monitoring 52 45.6 45 60.0 7 17.9 p<0.001

Anemia Treatment 50 43.9 46 61.3 4 10.3 p<0.001

Antipyretics 25 21.9 24 32.0 1 2.6 p<0.001

None 8 7.0 0 0.0 8 20.5 p<0.001

Other* 55 48.2 25 33.3 30 76.9 p<0.001

Information provided with Treatment

Instructions 99 86.8 71 94.7 28 71.8 p<0.001

Side Effects 64 56.1 57 76.0 7 17.9 p<0.001

Return if Symptoms Continue 59 51.8 47 62.7 12 30.8 p<0.001

Danger Signs** 8 7.0 5 6.7 3 7.7 p=0.41

Other 13 11.4 3 4.0 10 25.6 p<0.001

Any Information Given 104 91.2 74 98.7 30 76.9 p<0.001

*P‐values from Chi‐square test and Fisher Exact used for strata with <5 observations **Included nutritious diet, ITNs, IPTp, and medication compliance. *** Danger signs included death, convulsions, dizziness, spotting, & fetal movement

PredictorsofKnowledgeMalariainPregnancyTreatmentGuidelines

Important predictors of adequate knowledge of MiP diagnostic and prescribing practice for

providers in health facilities included facility type, respondent cadre, dispensing medication, gender,

and having received trainings on malaria diagnostics and MiP (Table 17). When compared to nurses,

pharmacists had higher odds to have adequate knowledge (OR= 1.6 95%CI [1.0‐2.6]) and providers

that identified as non‐clinically trained staff were less likely to possess adequate knowledge (OR=0.6


135

95% CI [0.5‐0.8]). Female providers had higher odds of adequate knowledge criteria than male

providers (OR=1.9 95% CI [1.5‐2.4]). Providers that had specifically attended workshops on malaria in

pregnancy were more likely to possess adequate knowledge than those who had not (OR=1.4 95% CI

[1.1‐1.8]).

Table 17. Health facility provider predictors of adequate knowledge of malaria in pregnancy treatment guidelines


95% CI P value Adjusted*

OR 95% CI P value

Facility Type 75

Health Center (ref) 28 37.3 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐

Hospital 6 8.0 0.6 (0.5, 0.9) 0.00 0.6 (0.4, 1.2) 0.15

Dispensary 41 54.7 0.8 (0.6, 1.0) 0.06 0.9 (0.7, 1.1) 0.21

Respondent Cadre

Nurse (ref) 48 64.0 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐

Clinical Officer/M.D. 17 22.7 0.9 (0.7, 1.2) 0.60 1.2 (0.9, 1.6) 0.14

Pharmacist 5 6.7 0.9 (0.6, 1.5) 0.69 1.6 (1.0, 2.6) 0.05

Other** 5 6.7 0.6 (0.5, 0.7) <0.001 0.6 (0.5, 0.8) <0.001

Dispenses Medicine

No (ref) 9 12.0 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐

Yes 66 88.0 1.3 (1.0, 1.8) 0.10 1.2 (0.9, 1.7) 0.15

Malaria diagnosis Training within 5 years

No (ref) 25 33.3 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐

Yes 50 66.7 1.3 (1.1, 1.7) 0.01 1.4 (1.0, 1.9) 0.06


No (ref) 47 62.7 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐

Yes 28 37.3 1.2 (1.0, 1.5) 0.09 1.4 (1.1, 1.8) <0.01

Sources of Information

CME as a source of info 52 69.3 1.3 (1.1, 1.7) 0.01 ‐‐ ‐‐ ‐‐

Gender

Male (ref) 39 52.0 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐

Female 36 48.0 1.2 (1.0, 1.5) 0.10 1.9 (1.5, 2.4) <0.001

Age 1.0 (1.0, 1.0) 0.60 1.0 (1.0, 1.0) 0.32

*Adjusted model included facility type, respondent cadre, dispensing medicine, Malaria diagnostic training and MiP training, gender, and age. CME dropped from multivariate model due to non‐significance. **Other includes community health workers, Village Reporters, or other technical staff at the health facility.

Correct knowledge of treatment and dosage in 1st trimester patients was removed from the

adequate knowledge definition for logistic regression with drug outlet providers because none of the

providers interviewed were able to provide the correct response. Facility type was a predictor of

adequate knowledge among drug outlet providers (Table 18). As compared to providers working in a

general shop, providers in registered pharmacies were more likely to know correct pregnancy

assessment, malaria diagnostic information and correct treatment and dosage in non‐pregnant and

2nd/3rd trimester patients, (OR= 1.1, 95%CI [1.1‐1.9]). Age, gender and respondent cadre were

controlled for, though they were not found to be significant predictors of knowledge.

Chapter 7

136

Table 18. Drug outlet dispenser predictors of adequate knowledge of malaria in pregnancy treatment guidelines


95% CI P value Adjusted

OR 95% CI P value

Facility Type 39

General Shop (ref) 15 38.5 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐

Registered Pharmacy 9 23.1 1.2 (1.0, 1.9) 0.03 1.1 (1.1, 1.9) <0.01

Informal Drug Outlet 15 38.5 1.2 (1.0, 1.5) 0.05 1.1 (0.9, 1.5) 0.25

Respondent Cadre

Pharmacist (ref) 15 38.5 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐

Clinical Officer/M.D./Nurse 2 5.1 1.5 (0.7, 2.8) 0.42 1.4 (0.6, 2.5) 0.52

Shopkeeper/Other** 22 56.4 1.1 (0.7, 1.1) 0.36 1.1 (0.7, 1.2) 0.38

Malaria diagnosis Training within 5 years

No (ref) 31 79.5 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐

Yes 8 20.5 1.2 (0.8, 1.6) 0.46 ‐‐ ‐‐ ‐‐


No (ref) 35 89.7 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐

Yes 4 10.3 1.3 (0.7, 1.7) 0.63 ‐‐ ‐‐ ‐‐

Gender

Male (ref) 15 38.5 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐

Female 24 61.5 1.1 (0.7, 1.1) 0.15 1.1 (0.8, 1.1) 0.39

Age 1.0 (1.0, 1.0) 0.24 1.0 (1.0, 1.0) 0.02

*Adjusted model included facility type, respondent cadre, gender and age. **Other included community health workers and Village Reporters

DISCUSSION

This study found that correct malaria case management in pregnant women with symptoms of

uncomplicated malaria in Siaya County is low overall, particularly in the 1st trimester. Only 43% of

WOCBA who were not visibly pregnant were assessed for a potential pregnancy in health facilities

and only 7% of the female simulated clients in drug outlets were assessed without being prompted.

Only 32% of women diagnosed with malaria in their 1st trimester attending health facilities and none

of the 1st trimester simulated clients attending drug outlets were correctly offered quinine as per

national guidelines.

The failure of providers to assess for possible pregnancy in a large proportion of women is

problematic and most often results in inadvertently exposing early pregnancies to ACTs. In addition,

this represents a missed opportunity to refer women for early antenatal care in an area where most

women initiate ANC late in pregnancy [21]. In our study, five of the women who were not assessed

for pregnancy by providers had an LMP consistent with a 1st trimester pregnancy, and all of these

women received AL. Likewise, all six women who reported being unsure of their pregnancy status

were prescribed AL; all these women warranted further assessment before prescribing AL.

The most common mistakes in prescribing practices were giving AL in early pregnancy, SP as

treatment, and the incorrect dosage of quinine. Prescribing of AL in known early pregnancy was very

common, consistent with our observation in the cohort studies of women of child bearing age

(Dellicour chapter 8). This is likely reflective of a lack of knowledge by healthcare providers regarding

the potential teratogenicity; when surveyed later, only 56% of providers reported they were aware

of contraindications to using ACTs in the 1st trimester. Overall, contraindicated regimens were

prescribed in 65% of 1st trimester pregnancies (there was no stock‐out of quinine reported); this is


137

similar to the 70% self‐reported exposure rate in early pregnancy Uganda in 2011 [11]. Correct

prescribing of AL in 61% of women who were not pregnant or in late pregnancy is similar to rates

observed previously [9,10]; however, only 75% of women received the correct dosage for AL in our

study.

The prescribing pattern in drug outlets uncovers a more troubling and complex knowledge issue.

Dispensers in drug outlets initially prescribed AL to over 90% of simulated client patients. Upon

being informed of pregnancy status, treatment was changed from AL to SP in 27% of cases due to

finding out the patient was pregnant, regardless of trimester. There was also a tendency among

dispensers to withdraw AL and refer the woman to a health facility upon learning of pregnancy

status. This occurred in only 13% of early pregnancy cases and occurred incorrectly in 21% of third‐

trimester pregnancy cases. In the provider survey, only 28% of drug outlet dispensers were aware of

contraindications to AL in 1st trimester.

The prescribing of SP in place of AL indicates that some dispensers still incorrectly believe that SP

can be used as treatment, which was supported by the provider survey in which 49% of drug outlets

providers and 33% of HF providers incorrectly reported that SP could be used for both treatment

and preventive purposes. This is alarming given that Kenya changed the national malaria treatment

policy to recommend artemisinin‐based combination therapy (ACT) as the first‐line treatment for

uncomplicated malaria in 2004, and by 2006, the new ACT guidelines were implemented

countrywide [3, 9]. Overall, SP was prescribed as treatment in 11% of all cases; 37% to women in

their 1st trimester. Using SP as treatment is likely to result in a high risk of treatment failure in this

area with high levels of SP resistance with potential for immediate consequences that affect both

the mother and fetus (Desai et al, personal communications & [22]).

Over 70% of all women that were prescribed quinine were given an insufficient supply (often 3‐5

days rather than the recommended 7) and incorrect instructions (incorrect number of tablets or

times per day), resulting in prescriptions ranging from 10‐70% of the full dose, increasing the risk of

treatment failure and malaria relapse, and the development of drug resistance [23,24]. This is

particularly troubling given that quinine is currently the only safe and effective treatment available

to women in early pregnancy. Poor knowledge of correct quinine dosage versus that of other

commonly prescribed antimalarials such as AL was likewise observed in the provider survey.

Although almost all HF providers reported a high rate of awareness of the national MTGs versus

only slightly more than a quarter of drug outlet dispensers, provider knowledge in both settings was

poor and were reflective of the low levels of correct case management observed in practice. The

greatest knowledge deficiencies were observed in pregnancy assessment, correct treatment and

drug regimen. Although a number of providers were aware of contraindications in 1st trimester,

knowledge of the correct treatment for 1st trimester patients was low for both HFs and especially

drug outlets, where not a single provider cited quinine as the correct drug of choice, consistent with

the observed practices.

The only consistently significant indicators for correct practice and/or knowledge were the

provider’s professional cadre in health facilities. MiP training in the last 5 years was only a predictor

of correct knowledge for health facility providers and correct practice in drug outlet dispensers. It is

likely that the number of dispensers in our study was underpowered for the purposes of MiP training

assessment. A push for dissemination of the MTGs in drug outlets may result in improved MiP

practice, however targeted trainings for MiP and overall capacity building via increased contact

Chapter 7

138

between non‐clinically trained dispensers and qualified healthcare professionals is likely to have

more of an impact in drug outlets; similar documented efforts have been shown to be effective in

the region [25,26]

Limitations&Challenges

The relatively short time‐frame of the overall study limited the number of exit interviews completed

at each facility. In particular, the identification of febrile patients in 1st trimester for interview was

challenging, possibly due to shortcomings in early pregnancy detection. Gestational age assessment

was based on reported LMP which could have lead to pregnancy trimesters misclassification for late

1st trimester pregnancies. Unless the provider had an alternative approach to assess gestation (such

as fundal height) it is unlikely that assessment of correct practice would have been affected.

It is likely that correct diagnostic practice was overestimated as the diagnostic capacity of health

facilities or drug outlets was not collected at the time of exit interviews nor simulated client

interactions. It was assumed that drug outlets, health centers and dispensaries didn’t have access to

diagnostic tests if none of the participants at these facilities received a diagnostic test and clinical

diagnosis in these facilities was considered correct.

Exit interviews and provider surveys were susceptible to courtesy bias. Information obtained from

patients via exit interviews may also be biased due to patient recall/information loss, although this

was minimized by conducting the interview immediately upon completion of the consultation. In

addition, errors may have been introduced if the patient did not understand the information given

or procedures done when in the presence of the provider.

CONCLUSION

Observed practice in health facility and drug outlet settings indicates a very low uptake of standard

clinical practice to assess all women of childbearing age for pregnancy and poor prescribing

practices, particularly for the case‐management of 1st trimester uncomplicated malaria. Timely,

significant efforts to increase pregnancy assessment and correct prescribing practices are required

to improve the safe and effective case management of malaria in pregnancy.

Disclaimer

The findings and conclusions presented in this manuscript are those of the authors and do not

necessarily reflect the official position of the U.S. President’s Malaria Initiative, United States Agency

for International Development, or U.S. Centers for Disease Control and Prevention.

AcknowledgementsWe are grateful to the communities of Asembo, Gem and Karemo for their participation in and

support of the HDSS. We also thank numerous field, clinical, data and administrative staff, without

whom, this work would not have been possible. We thank INDEPTH for their ongoing collaboration

to strengthen and support health and demographic surveillance systems; the KEMRI/CDC Research

and Public Health Collaboration is a member of the INDEPTH Network. This paper is published with

the permission of the Director, KEMRI.


139

Financialsupport This publication was made possible through support provided by the United States President’s

Malaria Initiative, U.S. Agency for International Development and U.S. Centers for Disease Control

and Prevention (CDC), under the terms of an Interagency Agreement with CDC and through a

Cooperative Agreement between the CDC and the Kenya Medical Research Institute (KEMRI). The

sponsor of the study had no role in study design, data collection, data analysis, data interpretation,

or writing of the report. The corresponding author had full access to all of the data in the study and

had final responsibility for the decision to submit for publication.

PotentialconflictsofinterestAll authors: No reported conflicts.

References1. Dellicour S, Tatem AJ, Guerra CA, Snow RW, ter Kuile FO (2010) Quantifying the number of pregnancies at

risk of malaria in 2007: a demographic study. PLoS Med 7: e1000221. 2. Desai M, ter Kuile FO, Nosten F, McGready R, Asamoa K, et al. (2007) Epidemiology and burden of malaria in

pregnancy. Lancet Infect Dis 7: 93‐104. 3. Ward SA, Sevene EJ, Hastings IM, Nosten F, McGready R (2007) Antimalarial drugs and pregnancy: safety,

pharmacokinetics, and pharmacovigilance. Lancet Infect Dis 7: 136‐144. 4. WHO (2003) Assessment of the safety of artemisinin compounds in pregnancy. Report of two informal

consultations convened by WHO in 2002. Geneva: WHO. 5. WHO (2007) Assessment of the safety of artemisinin compounds in pregnancy: report of two joint informal

consultations convened in 2006. Geneva: WHO. 6. WHO (2010) Guidelines for the treatment of malaria. Geneva: World Health Organisation. 7. DOMC (2010) Division of Malaria Control National guidelines for the diagnosis treatment and prevention of

malaria in Kenya.: Ministry of Public Health and Sanitation, Nairobi. 8. WHO (2009) Medicines use in primary care in developing and transitional countries: Fact Book summarizing

results from studies reported between 1990 and 2006 Geneva: World Health Organization. 9. Zurovac D, Njogu J, Akhwale W, Hamer DH, Larson BA, et al. (2008) Effects of revised diagnostic

recommendations on malaria treatment practices across age groups in Kenya. Trop Med Int Health 13: 784‐787.

10. Nyandigisi A, Memusi D, Mbithi A, Ang'wa N, Shieshia M, et al. (2011) Malaria case‐management following change of policy to universal parasitological diagnosis and targeted artemisinin‐based combination therapy in Kenya. PLoS One 6: e24781.

11. Sangare LR, Weiss NS, Brentlinger PE, Richardson BA, Staedke SG, et al. (2011) Patterns of anti‐malarial drug treatment among pregnant women in Uganda. Malar J 10: 152.

12. Kamuhabwa A, Jalal R (2011) Drug use in pregnancy: Knowledge of drug dispensers and pregnant women in Dar es Salaam, Tanzania. Indian Journal of Pharmacology 43: 345‐349.

13. Aviv RI, Chubb K, Lindow SW (1993) The prevalence of maternal medication ingestion in the antenatal period. S Afr Med J 83: 657‐660.

14. Kebede B, Gedif T, Getachew A (2009) Assessment of drug use among pregnant women in Addis Ababa, Ethiopia. Pharmacoepidemiol Drug Saf 18: 462‐468.

15. Rohra DK, Das N, Azam SI, Solangi NA, Memon Z, et al. (2008) Drug‐prescribing patterns during pregnancy in the tertiary care hospitals of Pakistan: a cross sectional study. BMC Pregnancy Childbirth 8: 24.

16. Odhiambo FO, Laserson KF, Sewe M, Hamel MJ, Feikin DR, et al. (2012) Profile: the KEMRI/CDC Health and Demographic Surveillance System‐‐Western Kenya. Int J Epidemiol 41: 977‐987.

17. Ouma P, van Eijk AM, Hamel MJ, Parise M, Ayisi JG, et al. (2007) Malaria and anaemia among pregnant women at first antenatal clinic visit in Kisumu, western Kenya. Trop Med Int Health 12: 1515‐1523.

18. Perrault SD, Hajek J, Zhong K, Owino SO, Sichangi M, et al. (2009) Human immunodeficiency virus co‐infection increases placental parasite density and transplacental malaria transmission in Western Kenya. Am J Trop Med Hyg 80: 119‐125.

19. Huntington D, Schuler SR (1993) The simulated client method: evaluating client‐provider interactions in family planning clinics. Stud Fam Plann 24: 187‐193.

Chapter 7

140

20. Madden JM, Quick JD, Ross‐Degnan D, Kafle KK (1997) Undercover careseekers: simulated clients in the study of health provider behavior in developing countries. Soc Sci Med 45: 1465‐1482.

21. Kenya National Bureau of Statistics (KNBS) and ICF Macro (2011) Kenya Demographic and Health Survey 2008‐09. Calverton, Maryland, USA. http://www.measuredhs.com/pubs/pdf/FR229/FR229.pdf.

22. McCollum AM, Schneider KA, Griffing SM, Zhou Z, Kariuki S, et al. (2012) Differences in selective pressure on dhps and dhfr drug resistant mutations in western Kenya. Malar J 11: 77.

23. Cheruiyot J, Ingasia LA, Omondi AA, Juma DW, Opot BH, et al. (2014) Polymorphisms in Pfmdr1, Pfcrt, and Pfnhe1 Genes Are Associated with Reduced In Vitro Activities of Quinine in Plasmodium falciparum Isolates from Western Kenya. Antimicrob Agents Chemother 58: 3737‐3743.

24. Tarning J, Kloprogge F, Dhorda M, Jullien V, Nosten F, et al. (2013) Pharmacokinetic properties of artemether, dihydroartemisinin, lumefantrine, and quinine in pregnant women with uncomplicated plasmodium falciparum malaria in Uganda. Antimicrob Agents Chemother 57: 5096‐5103.

25. Brantuo MN, Cristofalo E, Mehes MM, Ameh J, Brako NO, et al. (2014) Evidence‐based training and mentorship combined with enhanced outcomes surveillance to address the leading causes of neonatal mortality at the district hospital level in Ghana. Trop Med Int Health 19: 417‐426.

26. Hanson C, Waiswa P, Marchant T, Marx M, Manzi F, et al. (2014) Expanded Quality Management Using Information Power (EQUIP): protocol for a quasi‐experimental study to improve maternal and newborn health in Tanzania and Uganda. Implement Sci 9: 41.

141

Chapter8

Riskof inadvertentexposure toartemisininderivatives in thefirst trimester of pregnancy and its association withmiscarriage:aprospectivestudyinWesternKenya

StephanieDellicour1,2,MeghnaDesai1,3,GeorgeAol1,MartinaOneko1,PeterOuma1,GodfreyBigogo1,DeronBurton3,RobertBreiman3,MaryHamel3,LarrySlutsker3,DannyFeikin3,SimonKariuki1,FrankOdhiambo1,JayeshPandit4,KaylaF.Laserson3,5,N.Yan1,2, GregCalip,AndyStergachis5,FeikoO.terKuile1,2

1 Kenya Medical Research Institute and Centres for Disease Control and Prevention

(KEMRI/CDC)Research and Public Health Collaboration, Kisumu, Kenya, 2 Liverpool School

of Tropical Medicine, Liverpool, United Kingdom,3 Centres for Disease Control and

Prevention, Atlanta GA, United States of America, 4 Kenya Pharmacy and Poisons Board

(KPPB), Kenya, 5 Departments of Epidemiology and Global Health, University of

Washington, United States of America

To be submitted

Chapter 8

142

AbstractBackground: Artemisinins are widely deployed as artemisinin‐based combination therapy (ACT),

including as first‐line therapy for malaria in the 2nd and 3rd trimester of pregnancy. However, they are

not recommended for uncomplicated malaria during the 1st trimester because they are embryo‐toxic

in animal models and safety data from human early pregnancies are scarce. More safety data are

potentially accessible as inadvertent exposure in early pregnancy to ACTs is common, but methods

to capture this in resource poor settings have not been developed.

Methods: We conducted a prospective cohort study of women of childbearing age to document the

relationship between ACT exposure during the 1st trimester and the risk of miscarriage in a

population under health and demographic surveillance in western Kenya. Community‐based

surveillance was used to identify early pregnancies. To identify pregnancies exposed to ACTs and

other antimalarials in the 1st trimester, record linkage was used to combine study‐specific pregnancy

follow‐up questionnaire data with health surveillance data from out‐patient records and weekly or

twice‐monthly household‐level surveillance to ascertain recent history of symptoms and drug intake.

Cox proportional hazard models with left truncation were used accounting for potential survival bias

and controlling for confounders.

Results: In 2011‐2013, among 5,536 women of childbearing age, 1,453 pregnancies were detected;

1,126 of which were included in the analysis. The cumulative probability of miscarriage by 28 weeks

gestation was 19%. 296 pregnancies (26%) had data indicative of exposure to ACTs in the 1st

trimester, 12 to quinine (1%) and an additional 22 (2%) to both. There were 75 (7%) confirmed

exposures in the 1st trimester, defined as matching exposure data in 2 of 3 source databases, and 42

(4%) confirmed ACT exposures in the suggested artemisinin embryo‐sensitive period (6‐12 weeks

post last menstruation). Pregnancies with confirmed ACT exposures in the 1st trimester, overall, or

during the embryo‐sensitive period where not at a significant increased risk of miscarriage relative to

women not exposed to antimalarials [Hazard ratio (HR)=1.60 95% CI (0.70‐3.68) and HR=0.85 95% CI

(0.22‐3.33) respectively]. Similar findings were observed for the small numbers exposed to quinine

alone compared to pregnancies unexposed to antimalarials: 1st trimester HR=1.53 95%CI (0.19‐

12.40); or 6‐12 weeks gestation HR=2.02 95%CI (0.24‐17.10).

Conclusion: ACT exposure in early pregnancy was more common than quinine exposure. No

significant association between confirmed early artemisinin exposure and miscarriage was found.

The limited sample size rules out a 3.7 fold or greater increased risk of miscarriage associated with

artemisinin exposure in the 1st trimester. Further confirmatory studies will be needed. Targeting

malaria prevention strategies for women in the early stages of pregnancy should be a priority.

Association between artemisinin exposure in the first trimester and miscarriage

143

BackgroundThe artemisinin‐based combination therapy (ACT) antimalarials have been adopted as first‐line

treatment for P. falciparum in almost all endemic countries, providing life‐saving benefits to

children, adults and pregnant women globally.[1] However, their safety when used in early

pregnancy is uncertain. Artemisinins are embryotoxic in several animal species, including primate

models. Teratogenic effects observed in mice and rabbits included death of the fetus, malformations

of the heart, great vessels and limb defects. Primate models exposed to prolonged courses of ACTs

had high rates of fetal loss.[2] Animal models suggested that artemisinin embryotoxicity targets

primitive erythroblasts which are the primary form of red blood cells in circulation between weeks 4‐

10 post‐conception in humans. Therefore the suggested artemisinin embryo‐sensitive period in

humans, if any, is thought to occur at 6‐12 weeks gestation post‐last menstrual period

(LMP).[3,4,5,6]

There are limited data available to assess whether ACTs are similarly embryotoxic or teratogenic in

humans; fewer than 700 exposures in the 1st trimester have been well‐

documented.[7,8,9,10,11,12,13] After reviewing all existing evidence in 2003 and then in 2006, the

WHO recommended that artemisinins could be used during the second or third trimesters of

pregnancy and that, due to lack of sufficient safety data, treatment in the 1st trimester was not

recommended unless the life of the mother is at risk, or oral quinine is not available.[4,14] However,

as women may not be aware of their pregnancy or do not report an early pregnancy and because

clinicians often do no assess for pregnancy in women of childbearing age (WOCBA), the risk of

exposure to drugs not recommended in pregnancy, including to potential teratogens is high. We

have previously reported that in malaria endemic countries in Africa, up to 2 million pregnancies

could be inadvertently exposed to ACTs annually during a suggested embryo‐sensitive

time‐window.[15]

Malaria can have severe consequences to the health of the pregnant woman and her unborn baby

including maternal anaemia, fetal loss, preterm birth, low birth weight and perinatal mortality, and

in some cases maternal death. The impact of malaria infection in early pregnancy had been

identified as a major knowledge gap for estimating the burden of malaria in pregnancy.[16] Recent

studies provided insight into the potential adverse consequences of malaria infections early in

pregnancy with major impact on birth weight and maternal anaemia [17,18], and modelling by

Walker et al, suggest that as many as two thirds of placental malaria infections can be prevented by

avoiding exposure during the first 12 weeks of pregnancy.[19] Placental infections are thought to

play an important role in the adverse effect of malaria in pregnancy due to restricted nutrition flow

through the placenta.[20] McGready et al reported the findings from a retrospective analysis from

25 years of data from their research clinics on the Thai‐Burmese border and showed that malaria

infection in the 1st trimester of pregnancy (both symptomatic and asymptomatic) was a significant

risk factor for miscarriage.[11,21] They found no association between miscarriage and 1st trimester

ACT exposure. However more data from a wider range of malaria‐endemic countries are required to

provide an increased level of reassurance. We report the findings from a prospective cohort study of

WOCBA designed to examine whether ACT exposure in the 1st trimester is associated with

miscarriage.

Chapter 8

144

Methods

OverviewofstudydesignThis was a cohort study conducted among women aged 15‐49 years residing in the KEMRI/CDC

health and demographic surveillance system area (KEMRI/CDC HDSS) in a highly malarious area in

western Kenya. The study site also included facility‐ based and household level surveillance of recent

history of illness, symptom and drug intake [See Appendix for more details]. Pregnancies were

identified through health facility and community‐based strategies (described below), and followed

prospectively (i.e. before the pregnancy outcome was known) to document birth outcome. As this

was an observational study, the study staffs were not involved in treatment of study participants.

Participants were informed and reminded throughout the study that the purpose of the study was to

monitor safety of antimalarial medicine in early pregnancy and that ACTs are not recommended in

the 1st trimester. Participants received treatment through the usual channels including health

facilities and drug outlets.

RecruitmentofwomenofchildbearingageandpregnancydetectionBetween February 15, 2011 and February 15, 2013, WOCBA participating in an ongoing population‐

based infectious disease surveillance project (PBIDS) in rural western Kenya [22,23] were invited to

participate in the “Evaluation of Medications used in Early Pregnancy” (EMEP) cohort study. The

PBIDS catchment population included 33 villages within 5 km of Lwak Hospital, the designated

referral health facility in the area of the KEMRI/CDC HDSS located in Asembo, Bondo district.[24,25]

All PBIDS participants were visited weekly (from January 5, 2010 to May 26, 2011) and then twice

monthly (May 27, 2011 onwards) at home to collect information on any symptoms, where care was

sought and whether any medications were taken. PBIDS participants received free care at the

referral facility for all potential infectious illnesses and detailed information on diagnosis and

treatment was recorded. EMEP study staff visited all homes in the PBIDS and enrolled consenting

WOCBA, who met eligibility criteria for EMEP. WOCBA were eligible for EMEP if they were between

15 and 49 years of age and were active participants of PBIDS. EMEP exclusion criteria were the

following: refusal to participate or to be followed up at the end of pregnancy; and any condition

(physical, social or mental) that would interfere with the ability to give informed consent or to

provide an accurate medical history. WOCBA who consented to participate were asked if they could

be pregnant and offered a pregnancy test at the time of enrolment and again approximately every 3

months thereafter. Any participant with a detected pregnancy was referred to the antenatal clinic at

Lwak Hospital where trained EMEP study nurses would confirm the pregnancy and offer free

antenatal care (ANC). Additionally, all pregnant patients presenting at Lwak Hospital were assessed

for EMEP study eligibility by the study nurse and enrolled if all criteria were met.

GestationalageassessmentGestational age was assessed using multiple methods, including ultrasound scans at the 1st antenatal

visit at Lwak ANC (for participant presenting before 24 weeks); reported 1st day of LMP; reported

gestational age at the time of pregnancy loss; Ballard scoring for live‐births captured within 7 days of

delivery [26]; and fundal height measurements recorded at antenatal checkups. Not all methods

were available for all pregnancies since some were not seen at ANC (no fundal height or ultrasound

measurement available) or were seen at ANC but beyond 23 weeks. Ballard score was only available

for live‐births seen within 7 days of delivery. Furthermore some participants couldn’t recall their

LMP or in some instances hadn’t resumed their menses since their previous pregnancy. Therefore,


145

for this analysis, gestational age was determined using the most accurate measurement available for

each participant. Methods in order of decreasing accuracy were: ultrasound scan taken before 24

weeks gestation, Ballard estimates, LMP or reported gestation at time of pregnancy loss and lastly

gestational age derived from fundal height assessment [see Appendix].

PregnancyoutcomePregnancy outcomes were assessed using a combination of health facility‐based and home‐based

follow‐ups. The latter is particularly important for miscarriages, because the vast majority do not

occur at health facilities. Village‐based staff received monthly lists of participants with pending

estimated delivery dates in their respective catchment area. Study nurses were notified of

pregnancy outcomes by village‐based staff and follow ups were done either at home or at the health

facility. A detailed structured questionnaire about the delivery and outcome was administered in

addition to questions regarding any illnesses and medication used during pregnancy. Pregnancy

outcomes captured included pregnancy losses (miscarriages, induced abortions and stillbirths), live‐

births and major congenital malformations detectable at births by surface examination. The analysis

presented in this manuscript focuses on miscarriages defined as spontaneous pregnancy loss at or

before 28 completed weeks gestation (2‐28 weeks inclusive), which is considered the gestational age

of viability in resource constrained settings.[27] Risk of other adverse pregnancy outcomes

(stillbirths and major congenital malformations) associated with ACT exposures in early pregnancy

will be reported elsewhere.

AntimalarialdrugexposureascertainmentDrug exposure data were captured using 3 approaches: a) interviews with pregnant women visiting

the antenatal clinic in Lwak Hospital and at the time of pregnancy outcome follow up (from here on

referred to as EMEP data); b) record linkage of data on drugs prescribed to WOCBA at the outpatient

department in Lwak Hospital (from here on referred to as Lwak OPD data) and c) weekly to twice

monthly home visits by fieldworkers as part of PBIDS (Table 1).

Table 10. Description of drug information sources used to determine antimalarial and malaria exposure status

Approach Format Drug Information available

EMEP self‐report Retrospective self‐report of illness and medication used since the beginning of the pregnancy collected at every ANC visit and at pregnancy outcome follow up visit. A general open question about any drug use as well as a prompted question for specific antimalarials were included as using medication/indication‐specific questions have been shown to improve accuracy.[28,29,30] Visual aids depicting photographs of all antimalarial drugs found in the study area were used to facilitate recognition of drug names. Calendar marking public holidays and school closures was also used to enhance recall of dates.

Drug name

Drug start date Duration Number of tablets per day

Indication and indication diagnosis

Drug source

Lwak OPD records Prospective documentation by health facility clinician of diagnosis and treatment prescribed at out‐patient department (OPD) whenever a PBIDS participant sought care at Lwak Hospital for an infectious syndrome

Date of visit Diagnosis Prescribed treatment

PBIDS weekly & twice‐monthly home visits

Self‐report of symptoms, health‐seeking behaviour and medication. This information was collected continuously on a weekly (from January 5, 2010 to May 26, 2011) and then twice‐monthly basis (May 27, 2011 onwards). The same visual aids as described above were used for recall of drug intake.

Date of visit Symptoms in previous week/2 weeks

Treatment taken for the symptoms including drug name

If and where care was sought

Chapter 8

146

MalariainfectionInformation on malaria was captured through interviews by study nurses or through OPD records for

those who sought care from Lwak Hospital. Reliable information on malaria diagnosis was only

available from Lwak OPD which has the laboratory capacity to perform malaria slide microscopy

through PBIDS.[23]EMEP nurses asked about recent care‐seeking and diagnosis for each reported

illness including malaria episodes but reliable information on laboratory confirmed malaria was

mostly not available.

OtherCovariatesObstetric history and ANC laboratory information collected routinely at antenatal booking

(haemoglobin level, HIV and syphilis testing, and malaria microscopy) were abstracted by study

nurses from the ANC records at Lwak hospital or antenatal cards.

Dataanalysis

Miscarriagerateandsurvivalprobabilities Analyses were performed using Stata v12.1 (StataCorp LP, College Station, Texas). Survival analysis

with left truncation was used to estimate the miscarriage rate by gestational week to account for

delayed pregnancy detection and the range in gestational ages at the time of pregnancy detection.

Left truncation was used to account for survival bias as the average gestational age that pregnancies

were detected was only around 13 weeks and only pregnancies that survive the early weeks of

gestation (the highest risk of miscarriage) were followed prospectively.[31,32] The Kaplan–Meier

product‐limit method was used to calculate the cumulative probability of survival and cumulative

probability of miscarriage. Standard methods were used to calculate probability of miscarriage by

gestational week.[33] In brief, the miscarriage rate during the specific week of gestation was

converted to probability using the formula: (Miscarriage Rate)/(1+ (Miscarriage Rate x 0.5)). The

remaining risk of miscarriage by gestational week was calculated by subtracting the probability of

surviving the remaining weeks from 1. The probability of fetal survival during the remaining weeks is

the product of the probability of survival for week x and the probability of survival for week x+1.

ExposuredefinitionAnimal reprotoxicology studies suggest that the embryo‐sensitive period for artemisinins in humans

would be 6‐12 weeks post‐LMP (4‐10 weeks post conception).[6] A trend of increase in risk during

this artemisinin specific embryo‐sensitive period would corroborate the biological mechanism

observed in animal models and suggest a causal association with ACT exposures. Therefore it is

important to consider and compare the effect of these two exposure groups. The analysis focused

on two exposure definitions: 1) antimalarial drug reported/prescribed in the 1st trimester between

gestational weeks 2 and 13 inclusive (first trimester) and 2) antimalarial drug reported/prescribed

between weeks 6 and 12 post‐LMP (artemisinin embryo‐sensitive period as suggested by animal

reprotoxicology studies [6]). In the absence of a gold‐standard, we used agreement between at least

2 of the 3 data sources to define confirmed exposure. Probable exposures were defined as

antimalarial reported/prescribed according to at least 1 of the 3 data sources in the two periods

defined above. Probable exposures were considered in sensitivity analysis. Unconfirmed exposures

were defined as exposures identified by only 1 of the data sources.


147

CoxregressionmodelCox proportional hazard regression models with left truncation were fitted to estimate the effect on

miscarriage of ACT exposure during the 1st trimester and during the artemisinin embryo‐sensitive

period. Since women were exposed to ACTs at different gestational ages in the 1st trimester, ACT

exposure was treated as a time‐dependent variable. For example, if a woman took an ACT during

gestation week 9, she was considered unexposed during the 1st weeks and then exposed from week

9 onwards.[34] The reference group consisted of pregnancy‐weeks without exposure to

antimalarials or malaria during the 1st trimester according to any of the 3 exposure data sources. The

proportional hazard (PH) assumption was assessed for each covariate using a graphical approach

(the log‐log Kaplan‐Meier survival plots assessing if the curves for different strata are parallel) and

based on Schoenfeld residuals for the global test, and Schoenfeld scaled residuals for separate tests

with each covariate. The cluster function for robust standard errors in Stata was used to account for

the fact that some participants had multiple pregnancies during the study period. Pregnancies that

ended in induced abortions (9), withdrawal (4) or loss to follow up before 28 weeks (20) were

censored at the time of that event, or at the time of their last follow‐up visit in case of loss to follow‐

up.

We considered the following potential confounding factors based on known risk factors for

miscarriage as well as common socioeconomic variables: maternal age at the time of pregnancy

detection [35,36]; gravidity [37]; previous pregnancy loss [38]; socioeconomic characteristics

(education, marital status and occupation [39,40]). Although HIV status [41,42]; syphilis [43] and

anaemia (defined as Hb <11g/dl) could be potential confounders, we could not include these factors

due to missing data. Missing data were particularly frequent for miscarriage cases due to the fact

that most of these pregnancies didn’t persist long enough to attend ANC (mean gestational age for

1st ANC visit 20.8 weeks and mean gestational age for pregnancy loss 14.4 weeks). None reported to

smoke, while alcohol consumption (n=7), diabetes (n=2) and epilepsy (n=5) were infrequent and

these factors were not included in further analysis. Malaria infection was not considered in the

multivariate model since it is strongly correlated with the antimalarial drug exposure of interest. To

determine which variables remained in the final model, assessment of confounding was based on

the impact a variable had on the hazard ratio, followed by the consideration of its precision. If the

HR changed by >=10% the variable was retained in the model. Variables identified as non‐

confounders were dropped from the model provided their deletion led to a gain in precision for the

effect estimate from examining confidence intervals.[44,45]

The primary analysis focused on confirmed ACT exposures in the 1st trimester and in the suggested

artemisinin embryo‐sensitive period (6‐12 weeks post‐LMP, as defined above under Exposure

Definition). In order to assess potential exposure misclassification bias, alternative models were run

using the following exposure definitions:

1. Probable exposure defined as ACT exposure detected from at least one of the 3 data sources

regardless of whether they were confirmed by another data source.

2. ACT exposure excluding those with possible exposures within the error margin of gestation time.

That is, to address the uncertainty related to the assessment of gestational age, we excluded

pregnancies with exposure that could have been before conception or in the 2nd trimester due to

possible error in gestational age assessment (for example: an ACT exposure at 13 weeks

estimated using the Ballard Score method has an estimated error margin of 2 weeks therefore

Chapter 8

148

the true gestational age most probably lies between 11 and 15 weeks. Since the upper margin for

this exposure would fall outside the 1st trimester, such cases were excluded from the exposure

group).

To account for the possibility of confounding by indication (that is, the fact that malaria, for which

ACT was taken, itself may also cause miscarriage), a separate Cox regression model was run

comparing:

3. Pregnancies exposed to quinine (not known to cause miscarriage) in the 1st trimester to those

exposed to ACT in the 1st trimester.

4. Pregnancies with possible malaria infection (including both self‐report and OPD records of

confirmed malaria) in the 1st trimester compared to those without evidence of malaria or

antimalarial treatment. The PBIDS data were not included in this analysis as information on

malaria diagnosis was not captured through PBIDS system which only collected information on

symptoms, health seeking behaviour and treatment

EthicalreviewandconsentThe EMEP study protocol was reviewed and approved by the institutional review boards of CDC (No.

5889), KEMRI (No. 1752) and the Liverpool School of Tropical Medicine (No. 09.70). Written

informed consent or assent was obtained from each participant.

Results

ParticipantcharacteristicsOut of 5911 WOCBA approached, 5536 (94%) consented to participate and 1,453 pregnancies were

detected. Out of these, 1,126 (77.5%) were included in the data analysis; 327 were excluded because

pregnancy detection occurred beyond 28 weeks gestation (233) or at the time of pregnancy

outcome (33), lack of information on gestational age of exposure (15), loss to follow up immediately

after pregnancy detection (41), or inconsistent pregnancy end dates (5). The 1,126 pregnancies

involved a total of 951 women, 86 of whom had 2 pregnancies and 1 who had 3 pregnancies during

the study period. The mean gestational age at time of pregnancy detection was 13.3 weeks (SD=6.9).

Overall, 62% of deliveries took place at a health facility, and 25% of miscarriages took place at a

health facility. 67% of pregnancy outcomes were captured less than 1 week after the end of

pregnancy; however, for miscarriage this proportion was only 20%. The median number of days

between outcome and follow up was 3 overall (range: 0‐755) and for miscarriage was 24 (range: 0‐

602). This reflects the fact that follow ups were arranged at the convenience of participants and to

ensure suitable amount of time between miscarriages and follow up. Table 2 provides a description

of participants overall and by ACT exposure status.


149

Table 11 Characteristics of 1,126 pregnancies by ACT exposure status (n (%) otherwise stated).

Total

(N=1126)

No ACT 1st

trimester (N=830)

Unconfirmed ACT 1st trimester (N=221)

Confirmed ACT 1st trimester

(N=75)

P‐values*

Demographic and Obstetric Characteristics

Gestational week at pregnancy detection (mean (SD; range))**

13.3 (6.9; 0‐27.9)

13.3 (6.9; 0‐27.9)

13.0 (6.8; 0.3‐ 27)

13.4 (6.9; 2.4‐ 27.4)

0.794

Age in years (mean (SD; range)) 26.2 (6.8; 15‐47)

26.1 (6.7; 15‐45)

26.6 (7.2; 15‐47)

25.1 (6.4; 16‐41)

0.261

Age categories 0.743

15‐20 280 (24.9) 202 (24.3) 55 (24.9) 23 (30.7)

21‐25 286 (25.4) 218 (26.3) 49 (22.2) 19 (25.3)

26‐30 252 (22.4) 188 (22.7) 47 (21.3) 17 (22.7)

31‐35 182 (16.2) 130 (15.7) 42 (19) 10 (13.3)

>35 126 (11.2) 92 (11.1) 28 (12.7) 6 (8)

Gravidity Missing n= 17 Missing n= 14 Missing n= 2 Missing n= 1 0.075

Primigravidae 212 (19.6) 145 (17.8) 47 (21.5) 20 (27.0)

1‐3 pregnancies 511 (47.1) 405 (49.6) 90 (41.1) 30 (40.5)

4+ pregnancies 361 (33.3) 266 (32.6) 82 (37.4) 24 (32.4)

Previous pregnancy loss 150 (13,8), Missing n=37

109 (13.7), Missing n=34

30 (13.7), Missing n=2

11 (14.9), Missing n=1

0.961

Education level Missing n= 42 Missing n= 39 Missing n= 2 Missing n= 1 0.505

None/ Primary not completed 478 (44.1) 345 (43.6) 95 (43.4) 38 (51.4)

Primary completed 519 (47.9) 383 (48.4) 103 (47.0) 33 (44.6)

Secondary completed 87 (8.0) 63 (8.0) 21 (9.6) 3 (4.1)

Occupation Missing n= 54 Missing n= 50 Missing n= 2 Missing n= 2 0.191

Not working 361 (33.7) 265 (34.0) 67 (30.6) 23 (39.7)

Farming 361 (33.7) 261 (33.5) 79 (36.1) 21 (28.8)

Small business/ Skilled Labour 331 (30.9) 243 (31.2) 65 (29.7) 23 (31.5)

Other 19 (1.8) 11 (1.4) 8 (3.7) 0

Marital Status Missing n= 42 Missing n= 39 Missing n= 2 Missing n= 1 0.161

Single 227 (20.9) 156 (19.7) 50 (22.8) 21 (28.4)

Married 857 (79.1) 635 (80.3) 169 (77.2) 53 (71.6)

Antenatal Care Summary by the End of Pregnancy

Gestational week at 1st ANC study visit (mean (SD; range)

20.8 (7.8; 1.7‐41.0)

21.31 (7.8; 2.7‐37.1)

19.6 (7.6; 1.7‐41.0)

19.4 (7.8; 3.4‐37.0)

0.014

Number of ANC visit Missing n=39 Missing n=31 Missing n=5 Missing n=3 0.121

none 90 (8.3) 65 (8.1) 21 (9.7) 4 (5.6)

1 88 (8.1) 61 (7.6) 23 (10.7) 4 (5.6)

2 151 (13.9) 119 (14.9) 24 (11.1) 8 (11.1)

3 244 (22.5) 192 (24.0) 39 (18.1) 13 (18.1)

4+ 514 (47.3) 362 (45.3) 109 (50.5) 43 (59.7)

IPTp doses (in HIV negative) Missing n= 279 Missing n= 210 Missing n= 53 Missing n=16 0.579

none 243 (28.7) 184 (29.7) 43 (25.6) 16 (27.1)

1 93 (11.0) 59 (9.5) 27 (16.1) 7 (11.9)

2 170 (20.1) 126 (20.3) 32 (19.1) 12 (30.3)

3 222 (26.1) 163 (26.3) 42 (25.0) 17 (28.8)

4 119 (14.1) 88 (44.2) 24 (14.3) 7 (11.9)

Vaginal Bleeding 20 (2.4),

Missing n= 298 18 (3.0),

Missing n= 238 3 (1.7), Missing

n= 53 2 (2.9), Missing

n=7 0.680

Chapter 8

150

Total (N=1126)

No ACT 1st trimester (N=830)

Unconfirmed ACT 1st trimester (N=221)

Confirmed ACT 1st trimester

(N=75)

P‐values*

ANC Laboratory Profile from 1st visit

HIV positive 244 (24.5) Missing n=129

183 (24.9) Missing n=97

48 (24.7) Missing n=27

13 (18.6) Missing n=5

0.491

Hb (mean (SD; range)) 11.2 (1.9; 4.3‐17.2)

Missing n=309

11.1 (1.9; 4.3‐17.2)

Missing n=247

11.6 (2.0; 5.1‐16.6) Missing

n=55

11.6 (1.9; 6.9‐16.7)

Missing n=7

0.002

Anemia (Hb<11g/dl) 346 (42.4) Missing n=309

263 (45.1) Missing n=247

56 (33.7) Missing n=55

27 (39.7) Missing n=7

0.029

Malaria slide positive 114 (13.9) Missing n=306

72 (12.3) Missing n=244

32 (19.3) Missing n=55

10 (14.7) Missing n=7

0.070

Syphilis reactive test 71 (7.9) Missing n=225

49 (7.5) Missing n=176

17 (9.4) Missing n=41

5 (7.5) Missing n=8

0.684

*P values refer to Pearson Chi‐square test for categorical variables and ANOVA test for continuous variables ** Gestational age lowest estimate include 0 which reflects inaccuracy in the gestational age measurements Acronyms: ACT, artemisinin combination therapy; Hb, haemoglobin; SD, Standard deviation

Prevalenceof1sttrimesterACTexposuresOverall, 296 (26.3%) of the 1,126 pregnancies had evidence of an ACT exposure during the 1st

trimester as determined by at least 1 of the 3 exposure data sources (probable exposure) and 75 of

these (25.3% of exposures and 6.7% of all pregnancies) by at least 2 of the 3 sources (confirmed

exposure). The number reduced to 54 (18.4%, 4.8% of pregnancies) when pregnancies with

exposures within the potential gestational age error margin were excluded. Two‐hundred and one

(17.9%) pregnancies had evidence of probable exposure in the suggested artemisinin embryo‐

sensitive period but only 42 were confirmed exposures. Of these, 21 (2% of pregnancies) were within

the estimated gestational age confidence interval. Table 3 below depicts the different exposure

categories.

Table 12 Breakdown of antimalarial and ACT exposure status in the 1st trimester

Overall n (%)

N=1,126

Within gestational age confidence interval n (%)*

N=1,063

No antimalarial in 1st trimester 812 (72.1) 812 (76.4)

Probable non‐ACT antimalarial in 1st trimester**

18 (1.6) 13 (1.2)

Probable ACT exposure in 1st trimester

296 (26.3) 238 (23.0)

Confirmed ACT exposure in 1st trimester

75 (6.7) 54 (5.1)

Probable ACT exposure in 6‐12 weeks gestational age

201 (17.9) 116 (10.9)

Confirmed ACT exposure in 6‐12 weeks gestational age

42 (3.7) 21 (2.0)

* 63 cases have undefined exposure status as they fall in the potential error margin for gestational age.

** including 12 exposed to quinine, 5 to sulphadoxine‐pyrimethamine and 1 to amodiaquine.

RateofmiscarriagepergestationalweekThere were 89 (7.9%) miscarriages among the 1,126 pregnancies included in the analysis. The mean

gestational age at the time of miscarriage was 14.4 weeks (SD: 5.7) and the median was 13 weeks

(range: 4.3‐28); 75% of miscarriages occurred by 18 weeks. Accounting for late pregnancy detection,

the rate of miscarriage over the 1st 28 weeks of pregnancy calculated by survival analysis with left

truncation was 16.5 per 100 pregnancy‐weeks (95% CI: 13.4‐ 20.3) and the cumulative probability of


151

miscarriage was 19.2% (95% CI: 14.7‐ 24.9). The weekly miscarriage rate declined steadily with

increasing gestation (see Figure 1 and Table s1 in Appendix for miscarriage weekly rates and

probabilities) until approximately 16 to 20 weeks and then remained steady at approximately 3 per

1000 pregnancy‐weeks. Figure 2 shows the cumulative pregnancy survival probabilities per gestation

week.

Figure 1. Miscarriage rate by week of gestation with upper and lower estimates of 95% confidence interval.

Figure 2. Miscarriage Kaplan Meier survival curve by gestational week.

0

5

10

15

20

25

30

35

40

4 6 8 10 12 14 16 18 20 22 24 26

Rate per 1000 pregnan

cy‐w

eek

Gestational Week

Upper CI:117

6361

0.70

0.80

0.90

1

Per

cent

age

pre

gnan

cies

no

t yet

mis

carr

ied

46 297 514 675 809 916 1009 Number at risk

4 8 12 16 20 24 28Gestational Weeks

Chapter 8

152

RiskfactorsformiscarriageFactors associated with miscarriages included maternal age (increasing risk with increasing age);

gravidity (higher risk for primi and gravidity of 4 or more); previous pregnancy loss; occupation (with

highest risk for those doing physical labour/farming); reporting any vaginal bleed during the index

pregnancy; haemoglobin levels (1.2 g/dl higher Hb level observed in cases who miscarried compared

to other pregnancy outcomes which include stillbirths and live‐births); being HIV positive (with over

half of the miscarriage cases being HIV positive). Reported use of traditional remedies during

pregnancy was associated with a lower risk of miscarriage. Gestational age at first study ANC visit,

the number of ANC visits as well as the number of IPTp doses received had a significant protective

effect on miscarriage but this could reflect the fact that the majority of pregnancies ending in

miscarriage did not survive long enough to receive antenatal care since the average gestational age

for starting ANC in the overall cohort was 23 weeks, whereas the mean gestational age at the time of

miscarriage was 14 weeks.

Table 4. Distribution (n (%) or otherwise stated), rates of miscarriage and hazard ratios for covariates of interest.

Miscarriage

(N=89)

Other Pregnancy Outcomes (n=1037)

P‐values*

Rate of Miscarriage per 1000 pregnancy‐

weeks (95%CI)

Hazard Ratio (95%CI) §

Age in years (mean (SD)) 29.5 (7.9) 25.9 (6.6) P<0.001 NA 1.08 (1.04‐ 1.11)

Age categories P<0.001

15‐20 14 (15.7) 266 (25.7) 3.80 (2.29‐ 6.76) Ref

21‐25 14 (15.7) 272 (26.2) 3.50 (2.04‐ 6.51) 0.91 (0.43‐ 1.92)

26‐30 16 (18.0) 236 (22.8) 4.56 (2.83‐ 7.78) 1.15 (0.57‐ 2.31)

31‐35 21 (23.6) 161 (15.5) 8.85 (5.90‐ 13.85) 2.34 (1.22‐ 4.51)

>35 24 (26.8) 102 (9.8) 15.76 (10.45‐ 24.61) 4.14 (2.14‐ 7.98)

Gravidity Missing n= 1 Missing n= 16 P<0.001

Primigravidae 17 (19.3) 195 (19.1) 6.33 (3.98‐ 10.63) Ref

1‐3 pregnancies 23 (26.1) 502 (49.2) 3.12 (2.10‐ 4.83) 0.49 (0.26‐ 0.91)

4+ pregnancies 48 (54.6) 324 (31.7) 10.11 (7.64‐ 13.65) 1.61 (0.94‐ 2.77)

Previous pregnancy loss 19 (23.2),

Missing n= 7 131 (13.0) , Missing n= 30

P=0.004 10.17 (6.57‐ 16.46) 2.11 (1.28‐ 3.49)

Gestational week at detection (mean (SD))

8.0 (4.7) 13.8 (6.9) P=0.058 NA 0.94 (0.88‐ 1.01)

Education level Missing n= 10 Missing n= 32 P=0.541

Primary not completed 32 (40.5) 446 (44.4) 4.90 (3.50‐ 7.07) Ref

Primary completed 42 (53.2) 477 (47.5) 6.16 (4.59‐ 8.45) 1.21 (0.77‐ 1.91)

Secondary completed 5 (6.3) 82 (8.2) 4.16 (1.41‐ 17.36) 0.8 (0.27‐ 2.38)

Occupation Missing n= 11 Missing n= 43 P=0.002

Not working 22 (28.2) 339 (34.1) 4.90 (3.27‐ 7.66) Ref

Farming 33 (42.3) 328 (33.0) 6.35 (4.56‐ 9.10) 1.21 (0.71‐ 2.05)

Small business/Skilled Labour

18 (23.1) 313 (31.5)

4.03 (2.56‐ 6.67) 0.81 (0.44‐ 1.49)

Other 5 (6.4) 14 (1.4) 24.09 (9.43‐ 71.12) 4.54 (1.73‐ 11.91)

Marital Status Missing n= 10 Missing n= 32 P=0.140

Single 21 (26.6) 206 (20.5) 7.27 (4.81‐ 11.49) Ref

Married 58 (73.4) 799 (79.5) 4.97 (3.85‐ 6.54) 0.69 (0.42‐ 1.12)


153

Miscarriage (N=89)

Other Pregnancy Outcomes (n=1037)

P‐values*

Rate of Miscarriage per 1000 pregnancy‐

weeks (95%CI)

Hazard Ratio (95%CI) §

Antenatal Care Summary

Gestational week at 1st ANC study visit (mean (SD))

10.4 (4.9), Missing n=71

21.0 (7.7), Missing n=227

P<0.001 NA 0.85 (0.79‐ 0.91)

Number of ANC visit at end of pregnancy

Missing n= 0 Missing n=39 P<0.001

none 66 (74.2) 24 (2.4) 110 (84.06‐140) Ref

1 18 (20.2) 70 (7.0) 16.69 (10.54‐27.56) 0.17 (0.1‐ 0.3)

2 1 (1.1) 150 (15.0) 0.48 * 0 (0‐ 0.04)

3 3 (3.4) 241 (24.2) 0.92 (0.29‐4.45) 0.01 (0‐ 0.03)

4+ 1 (1.1) 513 (51.4) 0.13 * 0 (0‐ 0.01)

IPTp doses (HIV negative) at end of pregnancy

Missing n= 15 Missing n= 264 P<0.001

none 73 (98.7) 171 (22.0) 24.41 (19.27‐ 31.16) Ref

1 1 (1.4) 92 (11.9) 0.86* 0.04 (0.01‐ 0.32)

2 0 170 (22.0) 0 0 (0‐ 0)

3 0 222 (28.7) 0 0 (0‐ 0)

4 0 119 (15.4) 0 0 (0‐ 0)

Vaginal Bleeding 4 (22.2),

Missing n= 71 19 (2.4),

Missing n= 227 P<0.001 14.03(5.30‐46.92) 11.45 (3.96‐ 33.12)

ANC Laboratory Profile at 1st visit

BMI Missing n=73 Missing n=242 P=0.205

<18 3 (18.8) 47 (5.9) 3.74 (1.19‐17.48) Ref

18‐25 12 (75.0) 616 (77.5) 1.37 (0.80‐2.59) 0.4 (0.11‐ 1.4)

>25 1 (6.3) 132 (16.6) 0.60* 0.19 (0.02‐ 1.72)

HIV positive 15 (53.6),

Missing n=61 229 (23.6), Missing n=68

P<0.001 4.64 (2.85‐ 8.04) 3.83 (1.84‐ 7.96)

Hb (mean (sd)) 12.4 (1.9)

Missing n=72 11.2 (1.9)

Missing n=237 P=0.025 NA 1.31 (1.05‐ 1.63)

Anemia (Hb<11g/dl) 4 (23.5)

Missing n=72 342 (42.8)

Missing n=237 P=0.177 0.88 (0.33‐3.15) 0.47 (0.15‐ 1.46)

Malaria slide positive 2 (11.1)

Missing n=71 112 (14.0)

Missing n=235 P=0.737 1.25 (0.27‐12.32) 0.78 (0.18‐ 3.42)

Syphilis reactive test 1 (4.8)

Missing n=68 70 (8.0)

Missing n=157 P=0.635 1.05 * 0.62 (0.08‐ 4.51)

Traditional Remedies 6 (6.7),

Missing n=0 187 (18.3), Missing n=16

P=0.011 2.33 (1.06‐ 6.08) 0.36 (0.16‐ 0.81)

*P value refers to log‐rank test for categorical variables and to univariate Cox proportional hazard regression for continuous variables. **Confidence intervals are missing because of an insufficient number of failures.

Chapter 8

154

Associationbetweenmiscarriageand1sttrimesterexposuretoACTsorquinineNo difference in the risk of miscarriage was apparent when confirmed ACT exposures were

considered only (adjusted HRs for ACT exposures in 1st trimester 1.60 95% CI [0.70‐3.68] and

exposure in the embryo‐sensitive period 0.85 95% CI [0.22‐3.33]) compared to pregnancies without

evidence of malaria or antimalarial exposure in the 1st trimester. The risk of miscarriages was higher

among the pregnancies with probable exposure to ACTs in the 1st trimester (29/296, 7.8/1000

pregnancy‐weeks) compared to unexposed pregnancies (57/790, 5.2/1000 pregnancy‐weeks) by

crude (HR 1.68, 95% CI [1.07‐2.64]) and adjusted analysis (HR 1.79, 95% CI [1.10‐2.91]). Similar

findings were evident for probable ACT exposures restricted to the embryo‐sensitive period

although these were not statistically significant at the 5% level (table 5).

There was no statistically significant difference between pregnancies exposed to quinine (excluding

those with concomitant ACT exposure) during the 1st trimester compared to pregnancies not

exposed to antimalarials (crude HR 1.53 95% CI [0.19‐12.40] and adjusted 2.64 95% CI [0.32‐21.90]).

For comparison purposes, exposures during the suggested artemisinin embryo‐sensitive period (6‐12

weeks) are presented. Although the crude and adjusted HRs are slightly higher than for the whole 1st

trimester, these are not statistically significant. Comparison of pregnancies exposed to quinine in the

1st trimester to those with a confirmed exposure to ACT in the same period (HR=1.38 95%CI [0.19‐

9.95]) and for the embryo sensitive period (HR=0.79 95%CI [0.10‐ 6.54]) showed no statistical

difference although the number in the quinine comparison group were very small (n= 12 and 8

respectively).

The sensitivity analysis assessing the effect of using conservative estimates of gestational age

measurement for determination of 1st trimester showed no evidence of increase in the effect

estimates compared to exposures including those in the possible error margin for 1st trimester

gestational age measurement (crude HR 0.85 95%CI [0.28‐ 2.56] and adjusted HR 1.01 95%CI [0.32‐

3.15]).

Our study did not allow for an analysis of the risk of miscarriage by confirmed vs. unconfirmed

malaria as the use of diagnostic tests, or their results, were not systematically available. However,

pregnancies that reported malaria disease had a two‐fold higher hazard of miscarriage than

pregnancies without evidence of malaria disease or use of antimalarials, both in the crude and

adjusted models (HRs 2.06 95%CI [1.24‐ 3.41] and 2.32 95%CI [1.36‐ 3.94] respectively, table 6).

Asso

ciation betw

een artem

isinin exp

osure in

the first trim

ester and m

iscarriage

155

Table 5. M

iscarriage rate

, unad

juste

d an

d ad

juste

d hazard

rates fo

r the asso

ciation betw

een diffe

rent an

timalarial e

xposure cate

gorie

s and m

iscarriage.

a Th

e unexp

osed

pregn

ancies w

ere defin

ed as th

ose with

out h

istory o

f malaria d

isease or an

timalarial u

se in

the 1

st trimester acco

rding to

any o

f the 3 data so

urce. N

ote th

at the numerato

r and den

ominato

r for th

is unexp

osed

group re

main

the sam

e for all co

mpariso

ns, ye

t the rate

of m

iscarriage dep

icted in

the next co

lumn varies sligh

tly betw

een ro

ws. Th

is is becau

se antim

alarial exposure was fitted

as a time‐d

epen

dant variab

le an

d th

erefore th

e unexp

osed

group co

nsists o

f a combinatio

n of p

regnancies n

ever exp

osed

(e.g. 7

90) an

d of p

regnancies th

at were

eventually exp

osed

, but co

ntrib

uted

‘unexp

osed

pregn

ancy‐w

eeks’ to th

e unexp

osed

category u

ntil th

e day o

f exposure, after w

hich

they sw

itched

to th

e exp

osed

category.

b Covariates in

clude m

aternal age an

d occu

patio

n; c Exclu

ding th

ose also

exposed

to ACT in

1st trim

ester; d Confid

ence intervals co

uld not b

e calcu

lated becau

se of an

insufficien

t number o

f even

ts in th

e dataset; e Exp

osures are treated

here as tim

e‐indep

enden

t (a pregn

ancy is co

nsid

ered exp

osed fo

r the whole perio

d under o

f observatio

n disregard

ing th

e tim

e of exp

osure) an

d

the ACT gro

up exclu

des an

y quinine exp

osures. A

cronym

s: CI, C

onfid

ence Interval; A

CT, A

rtemisin

in Combinatio

n Th

erapy.

Exp

ose

d

Misca

rriage

ra

te p

er 1000

pre

gn

an

cy-we

eks

(95%

CI)

Une

xpo

sed a

Misca

rriage

rate

per 1

000

p

regn

ancy-w

eeks

(95

%C

I)

Exp

ose

d

#M

iscarria

ge/

#O

verall

Un

exp

ose

d a #M

iscarriag

e/

#O

verall

Un

ad

juste

d

Ha

zard

Ra

tio (9

5%

CI)

Adju

sted b

Ha

zard

R

atio

(95

% C

I) P

-valu

eP

-valu

e Antim

alarial Exposure Categories

Quinine c

1st trim

este

r (prob

able

)

1st trim

este

r (confirm

ed

)

6-1

2 w

eeks g

esta

tion (p

rob

able

)

6-1

2 w

eeks g

estatio

n (co

nfirm

ed)

1/ 1

2 0

/ 3

1/ 8

0

/ 1

57/7

90

5.3

7 (4

.13

, 7.1

0)

5.3

8 (4

.14

, 7.1

2)

5.3

7 (4

.13

, 7.1

1)

1.5

3 (0

.19

, 12

.40)

0.6

91

2.6

4 (0

.32

,21

.90)

0.3

68

6.3

1 d

0.0

0

57/7

90

57/7

90

57/7

90

2.0

2 (0

.24

, 17

.10)

3.8

7 (0

.44

, 33

.76)

0.5

20

0.2

21

9.5

6 d

0.0

0

5.3

8 (4

.14

, 7.1

2)

AC

T

1st trim

este

r (prob

able

)

1st trim

este

r (confirm

ed

)

1st trim

este

r (confirm

ed

& g

esta

tion C

I restricted

)

6-1

2 w

eeks g

esta

tion (p

rob

able

)

6-1

2 w

eeks g

estatio

n (co

nfirm

ed)

6-1

2 w

eeks g

esta

tion (co

nfirm

ed &

gesta

tion C

I restricte

d)

29/ 2

96

6/ 7

5

3/ 5

4

7.7

8 (5

.45

, 11

.47)

6.2

(2.8

6, 1

5.9

3)

57/7

90

57/7

90

57/7

90

57/7

90

57/7

90

5.2

4 (4

.03

, 6.9

4)

5.3

6 (4

.12

, 7.0

9)

1.6

8 (1

.07

, 2.6

4)

1.2

4 (0

.56

, 2.7

3)

0.8

5 (0

.28

, 2.5

6)

1.5

5 (0

.91

, 2.6

3)

0.8

0 (0

.21

, 3.0

8)

0.0

25

1.7

9 (1

.10

, 2.9

1)

1.6

0 (0

.7, 3

.68)

1.0

1 (0

.32

, 3.1

5)

1.6

3 (0

.93

, 2.8

5)

0.0

18

0.6

01

0.7

72

0.1

07

0.7

48

0.2

67

0

.985

0

.086

0

.818

4.1

6 (1

.34

, 19

.31)

7.2

4 (4

.62

, 11

.94)

3.7

5 (0

.84

, 34

.46)

0.0

0

5.3

6 (4

.13

, 7.1

0)

5.2

8 (4

.06

, 6.9

8)

5.3

6 (4

.13

, 7.1

0)

5.3

8 (4

.14

, 7.1

2)

18/ 2

01

2/ 4

2

0/ 2

1

0.8

5 (0

.22

, 3.3

3)

57/7

90

ACT vs. Quinine e

1st trim

este

r (prob

able

)

1st trim

este

r (confirm

ed

)

1st trim

este

r (confirm

ed

& g

esta

tion C

I restricted

)

6-1

2 w

eeks g

esta

tion (p

rob

able

)

6-1

2 w

eeks g

estatio

n (co

nfirm

ed)

6-1

2 w

eeks g

esta

tion (co

nfirm

ed &

gesta

tion C

I restricte

d)

27/ 2

74

5/ 6

8

2/ 4

9

16/ 1

81

1/ 3

6

7.2

3 (5

.01

, 10

.78)

5.4

6 (2

.33

, 15

.79)

2.8

9 (0

.65

, 26

.77)

6.5

4 (4

.08

, 11

.11)

1/ 1

2 1

/ 12

1.3

8 (0

.19

, 9.9

5)

1.0

2 (0

.12

, 8.5

4)

0.7

48

0.9

83

0.5

87

1.3

4 (0

.22

, 8.0

8)

1.3

6 (0

.18

, 10

.31)

0.7

49

0

.768

0

.521

5.3

6 d

5.3

6 d

5.3

6 d

8.4

3 d

8.4

3 d

8.4

3 d

1/ 1

2 1

/ 8

1/ 8

1

/ 8

0.5

4 (0

.06

, 4.9

8)

0.7

9 (0

.10

, 6.5

4)

0.2

5 (0

.02

, 2.8

2)

0.5

4 (0

.08

, 3.5

2)

0.7

2 (0

.10

, 5.1

7)

0.8

27

0.2

61

0.7

47

2.0

2 d

0.0

0

0/ 2

0

.05

15

Chapter 8

156

Table 6. Hazard ratio for the association between malaria in the 1st trimester and miscarriage.

#Miscarriages/ #Overall

Miscarriage rate per 1000

pregnancy‐week (95%CI)

Crude HR (95%CI) Adjusted§ HR (95%CI)

No antimalarial or reported malaria in 1st trimester

57/790 5.33 (4.10‐ 7.05) Ref Ref

Malaria in 1st trimester* 23/197 9.71 (6.48‐ 15.12) 2.04 (1.25‐ 3.35) 2.28 (1.36‐ 3.84)

§ Covariates include maternal age and occupation. * Malaria exposure in 1st trimester was derived from self report through interviews at ANC and pregnancy outcome study visits and from Lwak outpatient records on malaria diagnosis based on slide microscopy confirmation.

DiscussionThis study found that there was no statistically significant association between the risk of

miscarriage and confirmed ACT exposures in the 1st trimester of pregnancy or exposures in the

suggested artemisinin embryo‐sensitive period. Miscarriage is a relatively frequent endpoint and

although the sample size was limited to approximately 1000 pregnancies, our study had the power

to exclude a 3.7 fold or greater increased risk of miscarriage associated with artemisinin use in the

1st trimester as indicated by the upper bound of the 95% confidence interval (table 5). Sensitivity

analysis on the effect of using a less stringent definition of ACT exposure (i.e. probable exposure,

defined as exposure in at least one data source) in the first trimester showed a statistically

significant increase in the hazard of miscarriages of 1.8 compared to women who did not receive

antimalarials.

Several factors indicate that this increase in risk of miscarriage observed in the ACT exposed women

may be explained by the underlying malaria infection (i.e. potential confounding by indication).

Firstly, the ACT embryotoxic effect is expected to occur within 6‐12 weeks gestation when primitive

fetal erythroblasts, the target for ACT embryotoxicity, are in circulation.[2,6,46,47] The artemisinins

have very short half‐lives and are eliminated within hours. Therefore if ACTs were causing

miscarriage, the effect size would be expected to be highest for exposures restricted to that embryo‐

sensitive period compared to effect size of exposures encompassing the whole 1st trimester. We

observed no such trend. Secondly, although the effect estimate of quinine should be interpreted

with caution due to the small numbers of quinine‐only exposed women (12), the rates of miscarriage

in the quinine‐only and ACTs exposed pregnancies were similar. For example during the suggested

embryo‐sensitive period for the artemisinins the miscarriage rate was 7.2/1000 pregnancy weeks for

ACTs compared to 9.6/1000 pregnancy weeks for quinine. This is supported by the findings from

McGready et al in Thailand, indicating a higher risk of miscarriage associated with malaria in the 1st

trimester.[11] In this area of low malaria transmission, they found that asymptomatic malaria in 1st

trimester increased the odds of miscarriage by nearly 3 fold and symptomatic infections by 4 fold.

They found no difference in the proportions of pregnancies ending in miscarriages between women

treated with chloroquine (26%), quinine (27%) or artesunate (31%).

The small number of quinine exposures in the first trimester was surprising, but consistent with a

recent study on malaria in pregnancy treatment guidelines knowledge and prescribing practice

carried out in the same area of western Kenya (Riley et al, unpublished, chapter 7) and a study from

Jinja, Uganda.[48] The Kenya study found low adherence to treatment guidelines for women in their

1st trimester regardless of the provider’s malaria treatment guidelines knowledge (Riley et al,


157

unpublished). Very few women in their 1st trimester who had uncomplicated malaria (0% in drug

outlets and 68% in health facilities) were prescribed oral quinine (none as a result of stock out)

which draws attention to the need to assess reasons for poor adherence with quinine prescriptions

and malaria treatment guidelines. Quinine poor tolerability and poor compliance to its 7 days

regimen are known problems. [49]

This study provides the first description of the miscarriage rate in this rural Kenyan population in the

context of high malaria and HIV prevalence; there is very little data on miscarriage background rate

for sub‐Saharan Africa in general. The cumulative probability of miscarriages by 28 weeks gestation

accounting for staggered pregnancy detections in our study population was 19.2%, and the

probability by week declined from 16 weeks onward. The true rate is likely to be higher as

information from very early pregnancies (e.g. <6 weeks gestation) was not captured and the average

gestational age of pregnancy detection was 13.3 weeks, which meant that only 57% of pregnancies

were detected during the highest risk period for miscarriage (the first 12 weeks). However, the rate

of 19.2% is similar to that reported by McGready et al from the Thai‐Burmese border (20%)[11] and

consistent with that observed in other prospective studies in non malarious areas, which ranges

from 10% to 22%. [33,50,51,52]

This study had several limitations that should be considered. First, because only few women were

treated with quinine (the recommended 1st line malaria treatment in the 1st trimester) our ability to

compare ACT exposed pregnancies to a purportedly ‘control’ drug not known to be linked to

miscarriages was limited. Second, we could not control for confounding by indication (i.e. the

disease itself) because malaria was partly self‐reported (including self‐diagnosis, clinical diagnosis

and laboratory confirmed diagnosis) and partly laboratory confirmed. Controlling for laboratory

confirmed malaria diagnosis is important, as malaria itself has been suggested to reduce the

potential risk of embryotoxicity with artemisinin as was found in rat models. It was recently

suggested that in human pregnancies malaria may protect against artemisinin‐induced decreases in

reticulocyte count (a marker for embryotoxicity) by reducing the tissue levels of active drug and/or

ferrous iron which activates the drug to toxic free radicals, so that pregnant women without malaria

would be at greater risk of artemisinin‐induced embryotoxicity.[53] Third, there could have been

misclassification of miscarriage as an outcome since induced abortions are illegal in Kenya, they can

result in differential loss to follow‐up as well as misclassification of induced abortions as

miscarriages. However since inadvertent ACT exposure is very common and not perceived as an

indication for induced abortion, it is unlikely that such misclassification would differ according to

exposure status and therefore affect the ACT exposure effect estimate. Lastly, we found little

agreement between data sources regarding ACT exposures. Partly this is due to the fact that by

design, the PBIDS and Lwak OPD datasets could not capture all exposures. The Lwak OPD data

source was limited to participants receiving treatment from this facility; only 68 of the 169 (40%)

EMEP ACT exposures in the 1st trimester were reported to be from Lwak health facility and 48% of

cases reporting taking an antimalarial at a PBIDS home visit also sought care from Lwak health

facility. Further the antimalarial drug intake was not observed in most instances and it is unknown

what proportion of women complied with the prescribed regimen. The limitation of the PBDS (bi‐

)weekly home visits is in part missingness of data when no one was at home and the use of proxy

reporting where the participants herself was not at home on the day of interview and information

was provided by a relative responding on her behalf. The EMEP data source is limited by the

Chapter 8

158

participant recall, however pictorial aids and calendars were used to facilitate better recall. This data

source was also more prone to interviewer bias and reporter desirability bias.[54] Nevertheless the

group of ‘confirmed’ exposures is less likely to be misclassified as this was defined as evidence for

exposure in at least 2 of the 3 exposure data sources. Furthermore, the unexposed group was also

strictly defined as it excluded all possible antimalarial treatment according to any of the 3 data

sources. Another potential source of exposure misclassification is gestational age measurement

error. Our sensitivity analysis restricting the definition of ACT exposures to women whose exposure

fell within the potential error margins for gestational age measurement suggested this was not an

effect modifier. Also potential bias would only have been introduced if the error in gestational age

differed by exposure status.[55] We cannot account for exposure misclassification due to counterfeit

antimalarials, which is getting more common in sub‐Saharan Africa.[56]

Our study shows some of the challenges to conduct drug safety studies in resource poor settings. It

is very difficult to obtain reliable antimalarial exposure data outside a controlled environment such

as a clinical trial, as seen by the low level of agreement between the 3 data sources used for drug

ascertainment. In areas of high malaria transmission, having a febrile episode is not a particularly

memorable event and thus recalling the date is difficult. Furthermore, in rural settings where

illiteracy is common (estimated at 15% in Nyanza province [57]), reliable reports of drug name and

date are difficult to obtain. Antimalarials are widely available over‐the‐counter, and data from health

facility or clinic books do not capture history of drug intake from other sources or over‐the‐counter

exposures. Nevertheless, observational studies are essential to monitor drug safety in the post‐

marketing phase particularly for pregnant women and are often the only source of information

when administering a drug through a controlled clinical trial is considered unethical, or when the

outcome is rare or can occur a long time after the exposure. Such studies should be conducted in

sites where drug exposures can be confirmed, such as presented here, and be captured as

comprehensively as possible.

ConclusionThis prospective cohort study in women of childbearing age provides one of the first estimates of

weekly miscarriage rates in a rural African setting in the context of high malaria and HIV prevalence.

This information should be valuable to researchers and program managers for resource planning, to

monitor trends and impact of interventions as well as to clinicians in gauging miscarriage rate at a

given gestational week. Additionally, it adds to the limited information available on the use of

artemisinins in the 1st trimester of pregnancy. Our results are consistent with two previous

observational studies showing that the rate of miscarriage for pregnancies exposed to ACTs is similar

[11] or lower compared to quinine [12] and as such provides further reassurance regarding their

inadvertent or advertent use in early pregnancy for the treatment of acute malaria. Lastly, our

results suggest that ACT use in the 1st trimester is much more common than oral quinine, consistent

with our survey in clinics and drug outlets (Riley et al, unpublished). The lack of observed risk

associated with ACTs use in early pregnancy to date, and the limited compliance to treatment with

quinine suggests a trial comparing ACT vs. oral quinine for the treatment of uncomplicated malaria

in the 1st trimester may be merited. Before such a trial is considered, further safety data on the

association between ACT and congenital malformations that is forthcoming from studies conducted

by the Malaria in Pregnancy Consortium and WHO should be reviewed. An updated review to assess

the pooled evidence of the risk and benefits of artemisinin use in pregnancy is needed to potentially


159

revise the recommendations for the treatment of malaria in pregnancy by WHO. To prevent

exposure to malaria or antimalarial drug in the 1st trimester of pregnancy, malaria prevention

strategies should target women as early as possible in pregnancy.

AcknowledgementsThis research was conducted through the ongoing KEMRI and CDC collaboration. We are very

grateful to all participants and their babies for taking part in the study. We wish to thank the EMEP

study team for their perseverance and hard work. We are grateful to the International Emerging

Infection Program (IEIP) team for their help and collaboration. Furthermore we wish to thank the

Asembo District health and medical team and the Lwak Mission Hospital Board for their support. We

also wish to thank John Williamson and Jane Bruce for the statistical support and advice. KEMRI/CDC

HDSS is a member of the INDEPTH Network. The findings and conclusions in this paper are those of

the authors and do not necessarily represent the views of the US Centers for Disease Control and

Prevention. This paper is published with the permission of KEMRI Director.

References1. WHO (2013) World Malaria Report: 2013. Geneva: World Health Organisation. 2. Clark R, Kumemura M, Makori N, Nakata Y, Bernard F, et al. (2006) Artesunate: Developmental toxicity in

monkeys. . Bir Birth Defects Research (Part A) 76: 329. 3. Clark RL (2009) Embryotoxicity of the artemisinin antimalarials and potential consequences for use in

women in the first trimester. Reprod Toxicol 28: 285‐296. 4. WHO, TDR (2007) Assessment of the safety of artemisinin compounds in pregnancy: report of two joint

informal consultations convened in 2006. Geneva: WHO. WHO/GMP/TDR/Artemisinin/07.1. WHO/GMP/TDR/Artemisinin/07.1.

5. Kelemen E, Calvo W, Fliedner T (1979) Atlas of Human Hemopoietic Development; Springer‐Verlag, editor. Berlin.

6. White TE, Clark RL (2008) Sensitive periods for developmental toxicity of orally administered artesunate in the rat. Birth Defects Res B Dev Reprod Toxicol 83: 407‐417.

7. Adam I, Elhassan EM, Omer EM, Abdulla MA, Mahgoub HM, et al. (2009) Safety of artemisinins during early pregnancy, assessed in 62 Sudanese women. Ann Trop Med Parasitol 103: 205‐210.

8. Deen JL, von Seidlein L, Pinder M, Walraven GE, Greenwood BM (2001) The safety of the combination artesunate and pyrimethamine‐sulfadoxine given during pregnancy. Trans R Soc Trop Med Hyg 95: 424‐428.

9. Dellicour S, Hall S, Chandramohan D, Greenwood B (2007) The safety of artemisinins during pregnancy: a pressing question. Malar J 6: 15.

10. Manyando C, Mkandawire R, Puma L, Sinkala M, Mpabalwani E, et al. (2010) Safety of artemether‐lumefantrine in pregnant women with malaria: results of a prospective cohort study in Zambia. Malar J 9: 249.

11. McGready R, Lee SJ, Wiladphaingern J, Ashley EA, Rijken MJ, et al. (2012) Adverse effects of falciparum and vivax malaria and the safety of antimalarial treatment in early pregnancy: a population‐based study. Lancet Infect Dis 12: 388‐396.


13. Rulisa S, Kaligirwa N, Agaba S, Karema C, Mens PF, et al. (2012) Pharmacovigilance of artemether‐lumefantrine in pregnant women followed until delivery in Rwanda. Malar J 11: 225.

14. WHO (2003) Assessment of the safety of artemisinin compounds in pregnancy. Report of two informal consultations convened by WHO in 2002. Geneva: WHO.

15. Dellicour S, ter Kuile FO, Stergachis A (2008) Pregnancy exposure registries for assessing antimalarial drug safety in pregnancy in malaria‐endemic countries. PLoS Med 5: e187.


Chapter 8

160

17. Huynh BT, Fievet N, Gbaguidi G, Dechavanne S, Borgella S, et al. (2011) Influence of the timing of malaria infection during pregnancy on birth weight and on maternal anemia in Benin. Am J Trop Med Hyg 85: 214‐220.

18. Cottrell G, Mary JY, Barro D, Cot M (2007) The importance of the period of malarial infection during pregnancy on birth weight in tropical Africa. Am J Trop Med Hyg 76: 849‐854.


20. Brabin BJ, Romagosa C, Abdelgalil S, Menendez C, Verhoeff FH, et al. (2004) The sick placenta‐the role of malaria. Placenta 25: 359‐378.

21. Desai M, Dellicour S (2012) Effects of malaria and its treatment in early pregnancy. Lancet Infect Dis 12: 359‐360.

22. Bigogo G, Audi A, Aura B, Aol G, Breiman RF, et al. (2010) Health‐seeking patterns among participants of population‐based morbidity surveillance in rural western Kenya: implications for calculating disease rates. Int J Infect Dis 14: e967‐973.

23. Feikin DR, Audi A, Olack B, Bigogo GM, Polyak C, et al. (2010) Evaluation of the optimal recall period for disease symptoms in home‐based morbidity surveillance in rural and urban Kenya. Int J Epidemiol 39: 450‐458.

24. Adazu K, Lindblade KA, Rosen DH, Odhiambo F, Ofware P, et al. (2005) Health and demographic surveillance in rural western Kenya: a platform for evaluating interventions to reduce morbidity and mortality from infectious diseases. Am J Trop Med Hyg 73: 1151‐1158.


26. Sasidharan K, Dutta S, Narang A (2009) Validity of New Ballard Score until 7th day of postnatal life in moderately preterm neonates. Arch Dis Child Fetal Neonatal Ed 94: F39‐44.

27. WHO (1970) Spontaneous and Induced Abortion: Report of a WHO scientific group. Geneva. 28. Mitchell AA, Cottler LB, Shapiro S (1986) Effect of questionnaire design on recall of drug exposure in

pregnancy. Am J Epidemiol 123: 670‐676. 29. de Jong‐van den Berg LT, Waardenburg CM, Haaijer‐Ruskamp FM, Dukes MN, Wesseling H (1993) Drug use

in pregnancy: a comparative appraisal of data collecting methods. Eur J Clin Pharmacol 45: 9‐14. 30. Gama H, Correia S, Lunet N (2009) Questionnaire design and the recall of pharmacological treatments: a

systematic review. Pharmacoepidemiol Drug Saf 18: 175‐187. 31. Howards PP, Hertz‐Picciotto I, Poole C (2007) Conditions for bias from differential left truncation. Am J

Epidemiol 165: 444‐452. 32. Meister R, Schaefer C (2008) Statistical methods for estimating the probability of spontaneous abortion in

observational studies‐‐analyzing pregnancies exposed to coumarin derivatives. Reprod Toxicol 26: 31‐35. 33. Avalos LA, Galindo C, Li D‐K (2012) A Systematic Review to Calculate Background Miscarriage Rates using

Life Table Analysis. Birth Defects Research (Part A) 94. 34. Xu R, Luo Y, Chambers C (2012) Assessing the effect of vaccine on spontaneous abortion using time‐

dependent covariates Cox models. Pharmacoepidemiol Drug Saf 21: 844‐850. 35. Nybo Andersen AM, Wohlfahrt J, Christens P, Olsen J, Melbye M (2000) Maternal age and fetal loss:

population based register linkage study. BMJ 320: 1708‐1712. 36. Smith KE, Buyalos RP (1996) The profound impact of patient age on pregnancy outcome after early

detection of fetal cardiac activity. Fertil Steril 65: 35‐40. 37. Kline J (1978) I. An epidemiological review of the role of gravidity in spontaneous abortion. Early Hum Dev

1: 337‐344. 38. Regan L, Braude PR, Trembath PL (1989) Influence of past reproductive performance on risk of

spontaneous abortion. BMJ 299: 541‐545. 39. Banerjee B, Dey TK, Chatterjee P (2005) Work related physical exertion and spontaneous abortion. Indian J

Public Health 49: 248‐249. 40. Kline JK (1986) Maternal occupation: effects on spontaneous abortions and malformations. Occup Med 1:

381‐403. 41. D'Ubaldo C, Pezzotti P, Rezza G, Branca M, Ippolito G (1998) Association between HIV‐1 infection and

miscarriage: a retrospective study. DIANAIDS Collaborative Study Group. Diagnosi Iniziale Anomalie Neoplastiche AIDS. AIDS 12: 1087‐1093.

42. Brocklehurst P, French R (1998) The association between maternal HIV infection and perinatal outcome: a systematic review of the literature and meta‐analysis. Br J Obstet Gynaecol 105: 836‐848.


161

43. Holder WR, Knox JM (1972) Syphilis in pregnancy. Med Clin North Am 56: 1151‐1160. 44. Sonis J (1998) A closer look at confounding. Fam Med 30: 584‐588. 45. Kleinbaum DG, Klein M (2010) Chapter 6: Modeling Strategy Guidelines. Logistic Regression: A Self‐Learning

Text. 3rd edition ed: Springer. 46. Longo M, Zanoncelli S, Manera D, Brughera M, Colombo P, et al. (2006) Effects of the antimalarial drug

dihydroartemisinin (DHA) on rat embryos in vitro. Reprod Toxicol 21: 83‐93. 47. Longo M, Zanoncelli S, Torre PD, Riflettuto M, Cocco F, et al. (2006) In vivo and in vitro investigations of the

effects of the antimalarial drug dihydroartemisinin (DHA) on rat embryos. Reprod Toxicol 22: 797‐810. 48. Sangaré L, Weiss N, Brentlinger P, Richardson B, Staedke S, et al. (2011) Patterns of antimalarial drug

treatment among pregnant women in Uganda. Malaria Journal 10: 152. 49. Achan J, Talisuna AO, Erhart A, Yeka A, Tibenderana JK, et al. (2011) Quinine, an old anti‐malarial drug in a

modern world: role in the treatment of malaria. Malar J 10: 144. 50. Ellish NJ, Saboda K, O'Connor J, Nasca PC, Stanek EJ, et al. (1996) A prospective study of early pregnancy

loss. Hum Reprod 11: 406‐412. 51. Wang X, Chen C, Wang L, Chen D, Guang W, et al. (2003) Conception, early pregnancy loss, and time to

clinical pregnancy: a population‐based prospective study. Fertil Steril 79: 577‐584. 52. Wilcox AJ, Weinberg CR, O'Connor JF, Baird DD, Schlatterer JP, et al. (1988) Incidence of early loss of

pregnancy. N Engl J Med 319: 189‐194. 53. Clark RL (2012) Effects of artemisinins on reticulocyte count and relationship to possible embryotoxicity in

confirmed and unconfirmed malarial patients. Birth Defects Res A Clin Mol Teratol 94: 61‐75. 54. Delgado‐Rodriguez M, Llorca J (2004) Bias. J Epidemiol Community Health 58: 635‐641. 55. Howards PP, Hertz‐Picciotto I, Weinberg CR, Poole C (2006) Misclassification of gestational age in the study

of spontaneous abortion. Am J Epidemiol 164: 1126‐1136. 56. Newton PN, Green MD, Mildenhall DC, Plancon A, Nettey H, et al. (2011) Poor quality vital anti‐malarials in

Africa ‐ an urgent neglected public health priority. Malar J 10: 352. 57. Kenya National Bureau of Statistics (2013) Nyanza Province Multiple Indicator Cluster Survey 2011, Final

Report. Nairobi.

162


Text S1. Detailed description of study site and methodology.

Table S1. Weekly miscarriage rate, cumulative probabilities of survival and remaining risk of

miscarriage by gestational week.


163

TextS1:Detaileddescriptionofstudysiteandmethodology.

ThestudysiteandpopulationThe study area was located in Rarieda District in Siaya County, lying northeast of Lake Victoria in Nyanza Province, western Kenya. The cohort was carried out in 33 adjacent villages in the Asembo area (figure 1). The study area comprises one mission hospital, 2 government‐run health centres and 4 dispensaries. The provincial teaching hospital is located in Kisumu town approximately 50 km away and the closest district hospital is Bondo District Hospital 13km away. This area is under continuous health and demographic surveillance (HDSS) by Kenya Medical Research Institute (KEMRI) and the Centers for Disease Control and Prevention (CDC) Research and Public Health Collaboration since 2001.[1,2] Through quarterly household visits, information on births, deaths, and migrations are collected. Cause of death is established through verbal autopsy. In addition to the HDSS, the 33 villages are under enhanced morbidity surveillance since 2005 through the KEMRI/CDC International Emerging Infection program (IEIP) to investigate major infectious disease syndromes. This program has been described in details elsewhere.[3] As part of this population‐based infectious disease surveillance project (PBIDS), trained fieldworkers visit households regularly (on a weekly basis from January 5, 2010 to May 26, 2011 and then bi‐weekly basis from May 27, 2011 onwards) collecting information on all symptoms since the previous visit, the source of care and any medication taken, including specific antimalarial medication. In addition all visits for infectious symptoms Lwak Mission Hospital, the referral health facility for PBIDS, are recorded (including diagnosis made and prescriptions given) using the HDSS ID for record linkage. All care for acute illness is free at this health facility for PBIDS study participants, thereby enhancing health utilization. The morbidity data is linked to socio‐economic and demographic data collected by the HDSS. The HDSS and IEIP platform provided a unique opportunity to monitor drug use in pregnancy and develop/validate methodology for pharmacovigilance throughout the pregnancy period.

Rainfall is bimodal with long rains between March‐June and short rains in October‐December. Average temperature ranges between 18C and 36C at a mean altitude of approximately 1070 metres above sea level. Over 95 percent of the inhabitants are from the Luo ethnic group. The majority of the population is Christian and there is a small Muslim community. Polygamy is common with about 21% of women in Nyanza province in a polygamous relationship.[4] The main income generating activities are fishing, subsistence farming and cattle keeping and small scale businesses. Houses are typically made out of mud, brick or cement and roof are either made of thatched grass or iron sheets. The majority have completed primary school education but only 23% finished secondary school. The total population in the Asembo area of the HDSS was 64,491 with 22% women of childbearing age (15‐49 yrs) in 2010. The total fertility rate for 2009 was estimated at 4.3 in this area. Migrations are significant and the rates of out‐ and in‐migration for females in Asembo were 127 and 111 per 1000 person year respectively in 2009.[5]

Nyanza province has a high burden of disease and worst health indicators compared to the overall Kenyan national statistics.[6] Malaria transmission is perennial and holo‐endemic with peaks following the 2 rainy seasons. Annual cross‐sectional survey in this area showed parasitaemia of 42% in under‐5 years old, 60% in 5‐14 years old and 20% in over 14 year old (unpublished KEMRI/CDC data for 2010). Whereas the national HIV prevalence is 6% (4% for men and 8% for women), the prevalence for Nyanza province is close to double around 14% (11% for men and 16% for women).[4] In Asembo, infant mortality was 94 per 1000 live‐births and under five mortality ratio was 181 per 1000 live‐births (unpublished HDSS fact sheet for 2009).

Communitymobilizationandformativeresearch

The success and acceptability of any study is highly dependent on efforts made to involve the community and provide enough information in an accessible manner. Community mobilisation activities started in September 2010. A series of meetings were held with District Medical Officer for Health (DMOH), the village chiefs, district officers and counsellors, the community advisory board

Chapter 8 Appendix

164

(CAB) set up by KEMRI/CDC and community members to introduce and get feedback on the proposed study plans. “Barazas” (community meetings) where held in the pilot villages in November and in each IEIP study villages in February and March 2011. Study brochures were also distributed at the community meetings and at the central health facility. Formative research involving 10 focus group discussions was carried out in September 2010 with the aim to explore the socio‐cultural context around pregnancy and to investigate acceptability of proposed study procedures. A pilot study was subsequently carried out in 3 villages in November‐December to assess implementability and identify bottlenecks before wider implementation. Overall positive feedback was received through community meetings and acceptability during the pilot was high. Some concerns were raised regarding the need for the study to provide free ANC, delivery and referral services. Lessons learned in terms of field logistics were incorporated for the study implementation which started on the 15th of February 2011. Free ANC services were offered at the study health facility but covering delivery and full referral services was beyond the scope of this study and this was clarified at Barazas before implementation. Participants presenting to Lwak Mission Hospital for their ANC received a free delivery kit which entitled them to reduced costs of delivery at the same health facility (from KES 900 to KES 500 for a standard delivery).

Figure s8. Map of study area

StaffingandfieldinfrastructureThe core study team comprise of a study coordinator, part‐time field coordinator (shared with the IEIP surveillance activities), a VR supervisor, 2 field assistants and 3 study nurses. All went through 1 week training covering Good Clinical Practice (GCP), research ethics and consenting procedures, fieldwork etiquette and interview techniques, as well as protocol and study tools (including training


165

on the use of the netbook for data collection). In addition to the overall protocol training, study nurses were also trained on newborn examination for recognition of congenital malformations detectable by surface examination and Ballard score for gestational assessment (2 days) using training documents developed by the Malaria in Pregnancy Consortium (MIPc), use of ultrasound for gestational assessment (5 days) and a 2‐week motorbike training for home follow ups.

A team of 40 village based staff called “village reporters” or VRs were involved for the community mobilisation, enrolment and pregnancy detection activities. VRs usually act as community liaison officers for various KEMRI/CDC projects, some of them have been involved in research since the pivotal bednet trial which was carried out in Asembo and Gem in 1998‐2001.[7] All the 40 VRs involved in the study were females. They received a 3‐day training giving an overview of research ethics, the study (background, objective, procedure, expected output and role and responsibilities), consenting process and pregnancy test procedure and counselling. A 1 day refresher was provided before initiation of each village and each VR was accompanied by the supervisor or field assistants during the first few visits.

Datacollection

EnrolmentFollowing community mobilisation, door‐to‐door enrolment was carried out to provide information about the study to eligible women of childbearing age (15‐49 years). All female age 15‐49 years residents of households within the defined catchment area and participating in the population‐based infectious disease surveillance project (PIBDS) [3] were eligible for enrolment in the study. Women were excluded if they refused to participate; were unable to provide informed consent due to mental, physical or social inability or if they refused to be followed up at the end of the pregnancy. A list of all PIBDS female participants within the specified age group was provided to the VRs. Women above 18 years of age and mature minors (defined as an under 18 year old who is either married, head of a household or already a parent) could provide consent for themselves. Parental consent and assent were sought for minors (<18 years of age not qualifying as a mature minor). Information on screening and enrolment was collected on simple scannable forms in addition to the signed informed consent forms. Enrolment was active throughout the study period (from mid February 2011 to mid February 2013) whereby newly eligible women (turned 15 years of age or in‐migrant joining BIPDS) could be enrolled at any time. Of the women screened (not including those not at home after 3 attempts or migrated, N=1102), 92% (N= 5536) consented or assented to be part of the study. About 2% were not eligible to participate (68 were incapacitated, 17 were non‐PIBDS, 12 under 15 years of age, 1 was over 49 years of age and 1 was a male). Refusal rate was low at 6% (375) of screened participants. The majority of refusals (274/375) did not provide a reason as to why they dislike the study, of those who provided a reason 57% refused because they couldn’t become pregnant anymore (some felt too old and had already gone through menopause and some had done bilateral tubal ligation), for 1% the parents refused for their daughter to be part of the study and for less than 1% the husbands refused. Figure 2 depicts the participant flow from screening to inclusion in the data analysis. For girls under 18 years of age, 10% were mature minor and refusal rate was similar to the overall cohort (i.e. 6%). Participants were free to withdraw at anytime during the study, 1% (N=60) did so for various reason: the majority reported the study was not beneficial to them, 10% reported to dislike pregnancy test, 10% were later assessed as incapacitated and unable to answer study questions, 1 participants husband requested for his wife to be withdrawn, 2 participants felt too old and 4 refused to be followed up after identification of a pregnancy.

Chapter 8 Appendix

166

Figure s9. Study participant flow diagram from screening to inclusion in data analysis for miscarriage.

Analysis

Excluded from analysis (n=53) Detected at outcome (n=33) No GA information (n=15) Pregnancy end date error (n=5)

Assessed for eligibility N=6010

Excluded (n=474) Not meeting inclusion criteria (n=99) Declined to participate (n=375)

Consented (n=5536)

Pregnancies (n=1453)

Number of pregnancies per participants: Single pregnancy (n=1266) Two pregnancies (n=92) Three pregnancies (n=1)

Follow‐Up

Enrolment

Loss to follow up (n=85) Migrated (n=67) Withdrawal or refused follow up (n=13) Death (n=5)

Pregnancy Outcomes (n=1368)

Overall Pregnancy Outcomes (n=1315)

Miscarriage (n=1126)

Excluded from analysis (n=327) Detected at outcome (n=33) No GA information (n=15) Pregnancy end date error (n=5) No follow up (n=41) Entered after 28 weeks (n=233)


167

PregnancydetectionstrategyAt the time of enrolment, all consented participants were asked about their pregnancy status and offered to take a pregnancy test if they were not visibly pregnant. About 10% of participants were detected as pregnant at the time of enrolment. This was lower that the estimated proportion of WOCBAs pregnant 13% based on the fertility rate in this area (TFR 4.3)3.

Following the initial enrolment period, VRs (who were trained on pregnancy testing and counselling) visited all study participants every 3 months at their home to collect information on their pregnancy status and offer pregnancy testing4. Refusals remained very low at less than 2% of participants refusing the VRs visit every round. Although the number of women reporting to be menstruating at the time of the 3 monthly visit and in no need of taking a pregnancy test (7%) was higher than expected (about 2% (5/30 days) would be expected to be menstruating on any 1 day). This suggests some participant used menstruation as an excuse to avoid a pregnancy test. In between home visits the VRs also kept a stock of test strips, gloves and urine cups and participants were made aware that they could access pregnancy tests through the VRs or the study health facility any time they wished. These were accessed by 237 participants over the study period and only 12% (n=179) of pregnancy were detected that way. Although some pregnancies (n=30) were detected at delivery, about half (45%) were detected in their first trimester of pregnancy and 69% were detected using a pregnancy test before being visible. The average gestational age at the time of pregnancy detection varied according to the approach: pregnancies detected at the time of enrolment in the study had an average gestational age of 17 weeks (range 0‐42 weeks); pregnancies detected by pregnancy tests by VRs were detected around 16 weeks; whereas pregnancies detected at the ANC were on average 23 weeks (range 2‐41 weeks).

The graph below depicts the pregnancy detected by the different strategy and the line represent the average gestational age at detection. The latter decreased over the study period as the majority of pregnancies were detected in the first trimester. In the last rounds of active pregnancy detection gestational age was around 12 weeks gestation.

Figure s10 Graph of the number of pregnancies detected per round and average gestational age at detection over time.

3 This estimate of 13% was based on the formula below given a fertility rate of 4.3 and a pregnancy rate of 6.1 (adjusted for

pregnancy loss and with the proportion of live‐birth at 70.4% for Kenya):

Proportion pregnant at any year is 6.1/ (49‐15) = 0.18 and the proportion pregnant at any month in the year as 0.18 x (40/52) =0.13 4 The 3 monthly home visits started in October 2011

0

5

10

15

20

25

30

0

20

40

60

80

100

120

140

Feb

Mar

Apr

May Jun

Jul

Aug

Sep

Oct

Nov

Dec Jan

Feb

Mar

Apr

May Jun

Jul

Aug

Sep

Oct

Nov

Dec Jan

Feb

2011 2012 2013

Average Gestational Age (weeks)

Number of Pregnan

cies Detected

Outcome

ANC

Active detection

Passive detection

Enrolment

Chapter 8 Appendix

168

Characteristicsofthepregnanciesenrolledinthestudy

Around 72% of pregnancies were seen at least once at Lwak Mission Hospital before the end of their pregnancy. Overall 92% of women made at least 1 visit to ANC for the index pregnancy.

Table s1. Characteristics of pregnant participants.

Demographic characteristics N=1453

Age in years (mean (SD)) 26 (6.9)

Gravidity (n (%)) Missing n= 83

Primigravidae 298 (22%)

1‐3 pregnancies 627 (46%)

4+ pregnancies 445 (33%)

Gestational age at detection in weeks (mean (SD)) 17 (10.2) Missing n= 15

Education level (n (%)) § Missing n= 83

None 10 (1%)

Primary not completed 593 (43%)

Primary completed 652 (48%)

Secondary completed 94 (7%)

Higher education 21 (2%)

Occupation (n (%)) Missing n= 98

Not working 501 (37%)

Farming 415 (31%)

Salaried worker 37 (3%)

Skilled worker 55 (4%)

Small business 314 (24%)

Other 33 (3%)

Marital Status (n (%)) Missing n= 83

Single 298 (21%)

Married 1052 (72%)

Divorced 7 (1%)

Widowed 13 (1%)

PregnancyfollowupPregnant participants were referred for free antenatal care at Lwak Mission Hospital where trained study nurses were running the antenatal clinic. Participants were managed according to the Ministry of Health guidelines for ANC including a full ANC profile (covering blood group, syphilis, urinalysis, Hb and HIV) and in addition received ultrasound scan if they presented before 6 months of gestation. Following the ANC consultation the participant was asked a series of additional question for the study covering demographic characteristics, medical history and description of any illnesses and/or medication taken since the beginning of the index pregnancy. At each ANC revisits (scheduled according to MOH recommendation of minimum of 4 ANC visit per pregnancy), the participants were asked about any illnesses and treatment sought since their last ANC visits.


169

Table s2. Antenatal care summary for study participants.

Antenatal Care Summary

Gestational age at 1st study ANC visit in weeks (mean (SD)) 23 (8.7)

Number of ANC visit (n (%)) by end of pregnancy N= 1368, Missing n= 85

none 111/1368 (8%)

1 126/1368 (9%)

2 207/ 1368 (15%)

3 325/1368 (24%)

4+ 599/1368 (44%)

IPTp doses (HIV negative) (n (%))by end of pregnancy N= 1091, Missing n= 447

none 313 (29%)

1 136 (13%)

2 234 (16%)

3 270 (19%)

4 138 (10%)

ANC Profile

HIV positive (n (%)) 282/1235 (23%) missing n=218

Hb (mean (sd)) 11.1 (1.9) missing n=469

Anemia (Hb<11g/dl) 440/984 (45%)missing n=469

Malaria slide positive (n (%)) 133/987 (14%) missing n=466

Syphilis reactive test (n (%)) 79/1093 (7%) missing n=360

GestationalageassessmentGestational age was assessed using multiple methods and was determined using the most accurate measurement available for each participant. Ultrasound assessment of gestational age up to 24 weeks provides the most accurate prediction of the expected date of delivery, this was offered to all participants presenting at the study facility before 24weeks gestation. Out o f 995 pregnancies seen at the study health facility at least once for ANC, only 428 (43%) presented at or before 24 weeks and had a dating ultrasound completed. The data for 3 of those was unreliable and alternative method of gestational age assessment was used. Ballard score was performed on all live newborns seen by a study nurse within 7 days of birth (73% of 1221 live born babies). In case of a pregnancy loss, the participant was asked if she knew how many months pregnant she was at the time of the loss and 26% did and reported a gestational age. The first day of the last menstruation period (LMP) was asked to all participants at the time of pregnancy detection by trained VRs and was asked again during follow up visit by a study nurse either at ANC or at the time of outcome follow up (if the participant had not been seen at ANC at the study health facility). Only 19% of cases (which had both reported VR and ANC LMP) reported the same LMP date at the time of detection by village based staff and at the time of follow up by a study nurse. Forty four percent of the LMPs reported at these consecutive visits were within the same week. Eleven percent reported not to know the date of their LMP. Fundal height was also measured at each ANC visit at the study health facility and recorded by a trained study nurse. Only measurements done between 20 and 34 weeks gestation in singleton pregnancies were used and gestation estimated using the McDonald rule formula: # of cm x 8/7 = weeks gestation. There were 17 participants for whom no gestational age data was available as the participants didn’t know their LMP, had no ultrasound scan, fundal height or Ballard Score that could be used (including one which was lost to follow up).

The table below depicts the accuracy of each gestational age assessment method used and the corresponding expected error margin. The methods are listed in order of decreasing accuracy. In

Chapter 8 Appendix

170

addition if the LMP agreed within 10 days of ultrasound, fundal height or Ballard estimates the error margin was reduced to +/‐ 10 days. The last column represents the number of participants for which the specific method was used to assess gestational age in the data analysis.

Table s3. Accuracy correction for the gestational age assessment methods listed in order of decreasing accuracy.

Method of Gestational Age Assessment Accuracy Correction

Number of participants (% LMP confirmed)*

Mean gestational age at pregnancy detection in weeks

1. Ultrasound before 14 gestational weeks ± 5 days[8] 153 (68%) 8.4

2. Ultrasound between 14 and 24 gestational weeks

± 2 weeks[8] 273 (56%) 14.8

3. Ballard score within 7 days of birth ±2 week[9] 574 (41%) 21.8

4. LMP/Self report GA at end of pregnancy ± 4 weeks 431 (NA) 14.9

5. Fundal Height ± 6 weeks‐ when multiple measurements

[10]

5 (0%) 17.1

*17 had no gestational age information

SourceofinformationondrugexposureDrug exposure data was captured using 3 approaches: a) interview of pregnant participants seen at the study facility antenatal clinic and at the time of pregnancy outcome follow up (from here on referred to as EMEP data); b) record linkage of data on all drugs prescribed to WOCBAs at outpatient department in Lwak hospital (from here on referred to as Lwak OPD data) and lastly c) via an active approach involving regular home visits by field workers as part of the PBIDS.

a) EMEP self‐reported drug exposure: Pregnant participants attending ANC in Lwak were asked about drug exposure since the beginning of their pregnancy and at subsequent ANC visit about any drugs taken since their last visit. At the time of the pregnancy outcome follow up visit, participants were interviewed by a study nurse regarding drugs used in pregnancy and other medically relevant information. The limitation of drug histories is accuracy of recalled information. Visual aids depicting photographs of all antimalarial drugs found in the study area were used to facilitate recognition of drug names. Calendar marking public holidays and school closures was also used to enhance recall of dates.

b) Lwak OPD Health Facility records: The existing IEIP system in Lwak Mission Hospital is set up to document diagnosis and related treatments in all age‐groups at out‐patient department (OPD). All participants were already enrolled in the IEIP program and had access to free care at Lwak Mission Hospital for infectious syndrome.

c) PBDS (bi‐)weekly home visits: This active approach involved home visits by trained fieldworkers on a weekly (from January 5, 2010 to May 26, 2011) and then bi‐weekly basis (May 27, 2011 onwards), before a woman might becomes aware of her pregnancy. The interview component on drug exposure was also enhanced by introducing visual aids for recall (same as the one mentioned above under EMEP self‐reported drug exposure). This approach should limit drug exposure recall bias as the information is collected continuously (concurrently to when exposure occurs or close to this time) before a woman might become aware of her pregnancy. However a limitation of this approach is that interview data is incomplete for weeks when no one was found at home (54% of times) and on some occasion the interview was completed by a proxy (family member available on the day of the interview) instead of the participant herself.

Figures s4 and s5 below depict the number of ACT exposures during the 1st trimester of pregnancy and embryo‐sensitive period detected by the 3 different approaches as well as agreement between


171

the data. Overall there are 352 probable ACT exposure in the 1st trimester of pregnancy out of the 1453 pregnancies (24%), however only 88 (25% of the 352 possible exposures) were confirmed by at least 2 data sources. Similarly for exposures occurring in the embryo‐sensitive period, out of the 240 possible exposures only 49 (20% of these exposures) were confirmed by at least 2 of the data sources. After excluding exposures which fall within gestational age estimate error margin (for example: an ACT exposure at 13 weeks estimated using the Ballard Score method has an error margin of 2 weeks therefore the true gestational age most probably lies between 11 and 15 weeks. Since the upper margin for exposures falls outside the 1st trimester of pregnancy such cases are excluded from the exposure group), there are 284 probable 1st trimester exposure and 114 probable exposures within the embryo‐sensitive period of which 62 (22%) and 24 (21%) are confirmed by at least 2 data sources respectively. The 24 confirmed (by 2 data sources and within probable gestational age estimation margins) exposures during the embryo sensitive period represent 1.7% of the pregnancies which is lower than the anticipated 3% inadvertent exposure based on the assumption of 1 antimalarial treatment every 2 years for this population5.

Overall 34% of the EMEP ACT 1st trimester exposures were confirmed by one of the other data source and 87% (1028/1176) of EMEP unexposed were unexposed for all data sources. However, there is no gold standard to assess EMEP approach for detecting drug exposures. The Lwak OPD data source is limited to participants receiving treatment from this facility (only 68 of the 168 (40%) EMEP ACT exposures in the 1st trimester were reported to be from Lwak health facility and 48% of cases reporting taking an antimalarial at a 2 weekly home visit also sought care from Lwak health facility). Thus LWAK OPD data source is incomplete in terms of ACT received from other health facilities, drug outlets or relatives/friends and maybe misleading in case a woman was prescribed an antimalarial but did not take it. The limitation of the PBIDS home visits is in part the completeness of the data as 32% of participant did not have all home visit for the period of their 1st trimester of pregnancy (due to instances when no one was at home) and on the other hand 26% of first trimester visits were completed by proxy (i.e. a relative responding on behalf of the participant) which might not be as accurate and complete compare to interview completed by the woman herself. In order to estimate sensitivity of the EMEP interviews, EMEP ACT 1st trimester exposures were compared to exposures confirmed by both PBIDS home visits and Lwak OPD. Only 17 of the 46 PBIDS‐LWAK confirmed ACT 1st trimester exposures were detected by EMEP reflecting a low sensitivity of around 37%. The positive predictive value for this sub‐group shows than only 10% of cases defined as exposed according to EMEP data source are truly exposed. For specificity, EMEP ACT 1st trimester unexposed cases were compared to unexposed as defined by both PBIDS (only including participants with all visits during their 1st trimester of pregnancies 469/1453) and Lwak OPD. Specificity was also poor, with 66% (342 of the 519) unexposed pregnancies being correctly identified by EMEP as unexposed. Cohen's κappa statistic was run to determine if there was agreement between the different data sources and there was only slight (according to the commonly cited scale[12]) agreement between the EMEP and both PBIDS and Lwak OPD data sources, κ = 0.177 (p < 0.001) and κ = 0.103 (p < 0.001) respectively. Kappa agreement remained slight for comparison limited to pregnancies with complete PBIDS visit in their 1st trimester (κ = 0.138) and comparison limited to exposures reportedly obtained from LWAK OPD (κ = 0.208).

5 The average pregnancy is 266 days (38 weeks) from conception (280 days or 40 weeks from LMP). The average number of treatments with ACTs in adults in the study areas is approximately 0.5 treatments per year (1 every 2 years) and the total fertility rate is estimated at 5.5. Under these conditions and using the model developed by Dr. Ian Hastings, we estimate the probability that an embryo will encounter artemisinins inadvertently during the critical 42 day (6 weeks) period of its development (week 4 to week 9 inclusive, from conception) is about 6.0% [11. Ward SA, Sevene EJ, Hastings IM, Nosten F, McGready R (2007) Antimalarial drugs and pregnancy: safety, pharmacokinetics, and pharmacovigilance. Lancet Infect Dis 7: 136‐144.] Under these circumstances the potential ratio of exposed versus unexposed pregnant women would be 1:16. In Kenya, we anticipated that the probability of inadvertent exposure will reduce significantly and maybe halved to 3% following the introduction of pregnancy testing at the household level.

Chapter 8 Appendix

172

Table s4. Validity of EMEP approach for detection of ACT 1st trimester exposures in terms of sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), % agreement and % positive agreement with PBDS and Lwak OPD data sources.

ACT 1st Trimester Exposures

EMEP‐PBIDS

EMEP‐PBIDS (Full

1st Trimester1)

EMEP‐Lwak

EMEP‐Lwak Source Only2

EMEP‐PBIDS& Lwak CONF3

EMEP‐PBIDS(Full 1st

Trimester) OR LWAK4

Sensitivity 25% 19% 29% 25% 37% 65%

Specificity 91% 92% 89% 96% 89% 66%

PPV 33% 32% 12% 25% 10% 24%

NPV 87% 85% 96% 96% 98% 92%

% Agreement 81% 80% 86% 93% 87% 66%

% Positive Agreement 17% 14% 9% 14% 9% 21%

Kappa‐statistic 0.177 0.138 0.103 0.208 0.116 0.181

Kappa p‐values <0.001 0.002 <0.001 <0.001 <0.001 <0.001 1 Comparison between EMEP exposures and PBIDS exposures restricted to participants who had all PBIDS home visits for the period of their 1st trimester 2 Comparison between EMEP exposures and Lwak OPD exposures restricted to participants who reported that the ACT exposure was obtained from Lwak health facility at the time of EMEP interview 3 Comparison between EMEP exposures and PBIDS exposures confirmed in the Lwak OPD data source 4 Comparison between EMEP exposures and exposures defined by either PBIDS or Lwak OPD while unexposed are defined by PBIDS (only including participants with all visits during their 1st trimester of pregnancies 469/1453) and confirmed by Lwak OPD.

Table s4 summaries the relationship between potential predictors of EMEP data source ACT

exposure confirmation in the 1st trimester by another data sources (either PBIDS or Lwak OPD).

Overall demographic and obstetric characteristics were not significantly different between cases

with confirmed exposures detected by EMEP and those that weren’t detected. Factors associated

having a true positive ACT 1st trimester exposed according to EMEP data source, included having an

early 1st EMEP visit in the index pregnancy, the number of weeks between the exposure and the next

EMEP visit (cases with confirmed exposures having shorter time interval), the number of EMEP visits

(the higher the number the more likely to be detected), and having a follow up visit during

pregnancy as well as after the end of pregnancy follow up visit.

Asso

ciation betw

een artem

isinin exp

osure in

the first trim

ester and m

iscarriage

173

ACT Exp

osure in

1st

trimeste

r any d

ata source

N=353

Exposure co

nfirm

ed

by 2

data so

urce

s N=88 (2

5%)

EMEP

N=169

LWAK

N=70

PBIDS

N=219

30

3

17

38

111

134

20

A.

EMEP

N=131

LWAK

N=49

PBIDS

N=179

18

2

13

29

87

119

16

ACT Exp

osure in

1sttrim

este

r corre

cted fo

r gestatio

nal age

erro

r any d

ata

source

N=284

Exposure co

nfirm

ed

by 2

data so

urce

s N=62 (2

2%)

B.

Figure s1

1. V

enn diagram

of A

rtemisin

in Combinatio

n Th

erapy (A

CT) exp

osure in

1st trim

ester of p

regnancy acco

rding to

data so

urce. A

. All A

CT exp

osures. B

. ACT

exposures n

ot w

ithin possib

le error m

argins fo

r gestational age assessm

ent.

EMEP

N=101

LWAK

N=41

PBIDS

N=158

Exposure co

nfirm

ed

by 2

data so

urces

N=49 (2

0%)

19

210

18

70

111

11

ACT Exp

osure in

em

bryo

‐sensitive

perio

d an

y data

source N

=241

C.

ACT Exp

osure in

embryo

‐sen

sitive perio

d

corre

cted fo

r gestatio

nal age

erro

r any d

ata source

N=115

Exposure co

nfirm

ed

by 2

data so

urce

s N=24 (2

1%)

EMEP

N=48

LWAK

N=17

PBIDS

N=95

7

0

5

12

32

55

5

D.

Figure s1

2. V

enn diagram

of A

CT exp

osures in

embryo

‐sensitive p

eriod (6

‐12 weeks p

ost LM

P) acco

rding to

data so

urce. C

. All A

CT exp

osures. D

. ACT exp

osures n

ot w

ithin

possib

le erro

r margin

s for gestatio

nal age assessm

ent.

Chapter 8 Appendix

174

Table s5. Summary of factors associated with ACT exposure in 1st trimester of pregnancy confirmation by EMEP data source of exposures confirmed by prospective home visits (PBDS) and Lwak outpatient records.

Overall LWAK or PBIDS ACT 1st Trimester (N=233)

EMEP ACT 1st Trimester Unexposed

(N=177)

EMEP ACT 1st Trimester

Confirmed (N=56)

Chi‐Square P‐values

Age in years* 25.6 7.1 25.6 7.4 25 6.4 0.877

Gravidity 0.652

Primigravidae 62 27% 49 28% 13 23%

1‐3 pregnancies 88 38% 64 36% 24 43%

4+ pregnancies 82 35% 63 36% 19 34%

Previous pregnancy loss 33 14% 25 14% 8 14% 0.988

EMEP pregnancy number 0.703

1 220 94% 166 94% 54 96%

2 12 5% 10 6% 2 4%

3 1 0% 1 1% 0 0%

Any report of bleeding 3 1% 1 1% 2 4% 0.133

Gestational week at detection * 16.0 9.7 16.0 9.7 15.9 9.9 0.944

Education level 0.229

None or Primary not completed 114 49% 90 51% 24 43%

Primary completed 103 44% 73 41% 30 54%

Secondary completed 15 6% 13 7% 2 4%

Occupation 0.130

Not working 94 40% 68 38% 26 46%

Farming 71 30% 60 34% 11 20%

Small business or Salaried 62 27% 44 25% 18 32%

Other 4 2% 4 2% 0 0%

Marital Status 0.446

Single 63 27% 50 28% 13 23%

Married 169 73% 126 71% 43 77%

Pregnancy Outcome 0.601

Live birth 207 89% 155 88% 52 93%

Induced Abortion 2 1% 1 1% 1 2%

Miscarriage 16 7% 13 7% 3 5%

Stillbirth 1 0% 1 1% 0 0%

Neonatal death 7 3% 7 4% 0 0%

Gestational week at 1st EMEP visit* 21.7 8.6 22.4 8.6 20 8.5 0.077

Number of EMEP visit <0.001

1 51 22% 46 26% 5 9%

2 49 21% 42 24% 7 13%

3 34 15% 25 14% 9 16%

4+ 99 42% 64 36% 35 63%

EMEP visit at outcome only 51 22% 46 26% 5 9% 0.007

Weeks between EMEP interview & exposure*

19.3 14 20.8 14.9 15 9.7 0.003

*(mean (SD))


175

PregnancyoutcomeassessmentA combination of facility‐based and home‐based follow‐up systems were used to ascertain

pregnancy outcomes because the majority of deliveries occur in the home setting (only 44%

delivered in a health facility according to KDHS 2008/9). This is particularly important for

miscarriages which occur outside health facilities for the vast majority. Therefore, a home‐based

follow up system was developed, using VRs that alerted their supervisor by phone regarding any

pregnancy outcome. VRs were incentivised to report any pregnancy outcome as soon as possible

after delivery (they received Ksh120 for pregnancy outcomes reported within 4 days and Ksh 90

otherwise based on the rate paid by HDSS for birth notification at the time). Simple scannable forms

including the expected date of delivery (based on ultrasound EDD if available otherwise on LMP)

were distributed to the field team as a prompt that a delivery was expected. Participants were

encouraged and counselled on the benefits of delivering at a health facility although the study did

not cover delivery costs. At their last ANC visit, participants received a simple delivery kit (including

sterile gloves, gauze, cord clamp, soap, razor blade, maternity pads and clear plastic sheet) to bring

to the health facility which got them a discount on the delivery cost. Once notified about a

pregnancy outcome the VR supervisor would coordinate the follow up to be done by a study nurse

either at home or at the health facility. Overall 62% of deliveries took place at a health facility (70%

excluding pregnancies ending in miscarriages). A detailed questionnaire about the delivery, outcome

and newborn examination findings was administered in addition to the questions regarding

illnesses/medication used during pregnancy. This took approximately 60min. If the mother

consented, the baby was photographed. Any case of suspected malformation or abnormality was

reviewed by the study paediatrician either by going through the newborn exam data and the photos

or by full physical review if the latter information was inconclusive. Cases needing further

investigation or intervention were referred to the appropriate facilities/ specialist.

Sixty four percent of deliveries occurred at a health facility. It can be common for women to travel to

deliver close to the husband if he is working outside of the study area or to be with family. Overall

33% of pregnancy outcomes were captured over 1 week after the end of pregnancy and 16% of

outcomes were captured over 1 month later. The average number of days between outcome and

follow up was 28 range (0‐755).

Table s6. Summary of pregnancy outcomes

Pregnancy Outcome§ N (%)

Live‐Birth 1208 (88.3%)

Neonatal Death* 23 (1.6%)

Miscarriage (<=28 weeks)** 105 (7.7%)

Induced abortions 8 (0.6%)

Stillbirth (>28 weeks) 26 (1.9%)

§ Including 2 twin pregnancies which had 1 livebirth and 1 neonatal death/ 1 stillbirth. There were 85 (6%) loss to follow up.

* Babies were not followed up to 28 days, these are the numbers detected by the time of follow up.

**including 5 suspected induced abortions

Chapter 8 Appendix

176

Qualitycontrol

DatamanagementEnrolment and pregnancy status

All scannable forms were verified in the field by the study assistants when returned by the VRs

before being sent for scanning at the KEMRI/CDC offices in Kisian. Once the forms arrived in Kisian

they were scanned by the data management team who verified the entries on the computer screen

before uploading the data to the database. The forms were filled in secure office space for easy

retrieval in case of any query or data correction needed.

ANC visits and pregnancy outcome follow up

All data collected by the study nurses for pregnancy follow up at the ANC or at outcome follow up

were captured electronically on a Netbook computer with an in‐house program for data collection

developed using visual studio 2008 (VB.net programming language) and stored on MSSQL SERVER

2005/2008 on the Netbooks and Ms Access for the central database held in the KEMRI/CDC office.

The data entry system contained in‐built skip pattern depending on the entry, data checks for

inconsistencies (i.e. data out of expected range) and certain fields which could not be left blank.

Hard copies of the questionnaire were available in case the netbook had a problem. The data was

backed up daily on SD‐cards in the field and was downloaded once a week and brought to the

central database in Kisian.

EnrolmentandpregnancydetectionA sample of eligible participants from each village was selected for re‐visit by field assistants every 3

months following VRs visit. The data collected by the VR was compared to the data collected by the

field assistant to verify that 1) the participant had been visited, 2) had been explained the study

adequately and 3) the data collected on the form was correct.

NewbornexamOn a bi‐weekly basis 2 study nurses would independently carry out a newborn and Ballard score

examinations for the same baby and compare their findings on form provided. Any discrepancy was

discussed at the time and the baby re‐examined to identify the correct value. In case were there was

ambiguity about a correct value further explanation was sought from the study paediatrician and

refresher training on Ballard assessment was organised.

UltrasoundQuality control for ultrasound scans was implemented during the pre‐implementation phase

following the ultrasound training. All ultrasound images were downloaded from the machine to a

laptop and sent to ultrasound expert at the collaborating institution at the University of Washington

(after removing all identifiers from the images). The reviewers provided feedback on a formatted

excel spreadsheet and highlighted areas needing re‐training or special attention. If required,

additional training materials were sent for review with the study nurses.

EthicalreviewandconsentThe EMEP study protocol was reviewed and approved by the institutional review boards of CDC (No.

5889), KEMRI (No. 1752) and the Liverpool School of Tropical Medicine (No. 09.70). Informed written

consent or assent was obtained from each participant. For illiterate participants, an independent

witness had to be there during the consenting process and sign the consent form while the


177

participant marked the form with a thumbprint. Minors (girls under 18 years of age) needed parental

consent before being asked for assent. Mature minor (defined as an under 18 year old who is either

married, head of a household or already a parent) provided consent for themselves.

References1. Adazu K, Lindblade KA, Rosen DH, Odhiambo F, Ofware P, et al. (2005) Health and demographic surveillance

in rural western Kenya: a platform for evaluating interventions to reduce morbidity and mortality from infectious diseases. Am J Trop Med Hyg 73: 1151‐1158.


3. Feikin DR, Audi A, Olack B, Bigogo GM, Polyak C, et al. (2010) Evaluation of the optimal recall period for disease symptoms in home‐based morbidity surveillance in rural and urban Kenya. Int J Epidemiol 39: 450‐458.

4. Kenya National Bureau of Statistics (KNBS), Macro I ( 2010.) Kenya Demographic and Health Survey 2008‐09. Calverton, Maryland.

5. KEMRI/CDC (2010) Health and Demographic Surveillance System Report for 2009. Kisian: KEMRI/CDC. 6. Kenya National Bureau of Statistics (KNBS) and ICF Macro (2011) Kenya Demographic and Health Survey

2008‐09. Calverton, Maryland, USA. http://www.measuredhs.com/pubs/pdf/FR229/FR229.pdf. 7. Phillips‐Howard PA, Nahlen BL, Kolczak MS, Hightower AW, ter Kuile FO, et al. (2003) Efficacy of permethrin‐

treated bed nets in the prevention of mortality in young children in an area of high perennial malaria transmission in western Kenya. Am J Trop Med Hyg 68: 23‐29.

8. Seffah JD, Adanu RM (2009) Obstetric ultrasonography in low‐income countries. Clin Obstet Gynecol 52: 250‐255.

9. Sasidharan K, Dutta S, Narang A (2009) Validity of New Ballard Score until 7th day of postnatal life in moderately preterm neonates. Arch Dis Child Fetal Neonatal Ed 94: F39‐44.

10. White LJ, Lee SJ, Stepniewska K, Simpson JA, Dwell SL, et al. (2012) Estimation of gestational age from fundal height: a solution for resource‐poor settings. J R Soc Interface 9: 503‐510.

11. Ward SA, Sevene EJ, Hastings IM, Nosten F, McGready R (2007) Antimalarial drugs and pregnancy: safety, pharmacokinetics, and pharmacovigilance. Lancet Infect Dis 7: 136‐144.

12. Landis JR, Koch GG (1977) The measurement of observer agreement for categorical data. Biometrics 33: 159‐174.

Chapter 8 Appen

dix

178

TableS1Miscarriageweeklyrate,cumulativeprobabilitiesofsurvivalandfailureandrem

ainingriskofm

iscarriage.

Ges

tatio

nal

wee

k Pr

egna

ncie

s D

etec

ted

durin

g w

eek

Preg

nanc

y-W

eeks

at

Ris

k

Mis

carr

iage

Indu

ced

abor

tion

Loss

to

follo

w u

p &

w

ithdr

awal

s

Wee

kly

mis

carr

iage

rate

pe

r 100

0 pr

egna

ncy-

wee

ks

(95%

CI)

Cum

ulat

ive

Prob

abili

ty o

f Su

rviv

al

Cum

ulat

ive

Prob

abili

ty

of F

ailu

re

Rem

aini

ng

risk

of

mis

carr

iage

1

3 2.

7

0 0

0 0.

00

1.00

0

0.00

0

0.19

1

2 8

8.4

0

0 0

0.00

1.

000

0.

000

0.

191

3 15

18

.4

0 1

0 0.

00

1.00

0

0.00

0

0.19

1

4 22

32

.4

0 0

0 0.

00

1.00

0

0.00

0

0.19

1

5 43

68

.4

2 0

0 29

.23(

7.31

- 1

20)

0.

971

0.

029

0.

191

6 74

12

7.6

2

0 0

15.6

8(3.

92-

62.

69)

0.95

4

0.04

6

0.16

7

7 78

19

8.4

5

3 0

25.2

(10.

49-

60.

54)

0.93

0

0.07

0

0.15

3

8 69

27

3.1

2

1 1

7.32

(1.8

3- 2

9.2

8)

0.92

3

0.07

7

0.13

2

9 65

32

9.1

3

0 0

9.11

(2.9

4- 2

8.2

6)

0.91

5

0.08

5

0.12

5

10

64

388.

9

6 0

0 15

.4(6

.92-

34.

28)

0.

901

0.

099

0.

117

11

54

441.

4

7 1

1 15

.78(

7.52

- 3

3.1)

0.

886

0.

114

0.

104

12

59

490.

6

6 1

1 12

.2(5

.48-

27.

16)

0.

877

0.

123

0.

089

13

55

540.

4

12

1 0

22.2

3(12

.62-

39.

14)

0.85

8

0.14

2

0.07

8

14

42

575.

1

3 0

1 5.

21(1

.68-

16.

15)

0.

854

0.

147

0.

058

15

52

620.

3

4 0

0 6.

44(2

.42-

17.

15)

0.

848

0.

152

0.

053

16

41

661.

4

9 0

0 13

.59(

7.07

- 2

6.11

) 0.

838

0.

162

0.

046

17

43

699.

3

2 1

0 2.

86(0

.72-

11.

44)

0.

835

0.

165

0.

033

18

44

735.

3

5 0

0 6.

8(2.

83-

16.

34)

0.83

0

0.17

0

0.03

1

19

31

765.

3

5 0

0 6.

53(2

.72-

15.

7)

0.82

4

0.17

6

0.02

4

20

34

793.

6

2 0

1 2.

52(0

.63-

10.

08)

0.

822

0.

178

0.

018

21

29

823.

3

4 0

1 4.

86(1

.83-

12.

96)

0.

818

0.

182

0.

015

22

30

849.

6

1 0

1 1.

18(0

.17-

8.3

7)

0.81

6

0.18

4

0.01

0

23

24

870.

0

0 0

0 0.

00

0.81

6

0.18

4

0.00

9

24

35

903.

1

1 0

0 1.

11(0

.16-

7.8

7)

0.81

5

0.18

6

0.00

9

25

29

932.

6

0 0

0 0.

00

0.81

5

0.18

6

0.00

8

26

21

954.

0

2 0

0 2.

1(0.

52-

8.38

) 0.

813

0.

187

0.

008

27

33

984.

4

4 0

12

4.06

(1.5

3- 1

0.8

3)

0.81

0

0.19

1

0.00

6

28

29

100

0.0

2

0 5

1.99

(0.5

- 7.

94)

0.80

8

0.19

2

0.00

2

179

Chapter9

SummaryandDiscussion

Chapter 9

180

Post‐marketing surveillance of drugs used in pregnancy is challenging, especially in developing

countries where resources for pharmacovigilance are rare. There is a need to establish simple but

effective systems to monitor safety of drugs given during pregnancy in resource constrained

countries. Although ACTs are not recommended in early stages of pregnancy, inadvertent exposures

to ACTs are unavoidable; this will occur through different channels including outpatient clinics, the

informal sector, clinical trials or mass drug administration programmes because either the woman or

the prescriber is unaware of her pregnancy status. Deliberate exposures will occur in situations

where the benefit outweighs the potential risk, when no other effective antimalarial drugs are

available and particularly in the case of severe malaria. There are currently no established systems to

monitor drug safety in pregnancy in malaria endemic countries. The aim of this thesis was to develop

and evaluate pharmacovigilance systems in resource constrained settings to provide a better

estimate of the risk‐benefit profile of ACTs in pregnancy.

In Chapter 2 the published evidence with regard to the safety of artemisinin compounds when

administered during pregnancy was reviewed through systematic literature searches. At the time of

the review (November 2006), fourteen relevant studies (nine descriptive/case reports and five

controlled trials) were identified. Overall there were reports on 945 women exposed to an

artemisinin during pregnancy, 123 in the first trimester and 822 in second or third trimesters. The

limited data available suggested that artemisinins are effective and safe when used in late

pregnancy, although rare adverse events could not be ruled out. There was insufficient evidence to

effectively assess the risk–benefit profile of artemisinin compounds for pregnant women,

particularly with respect to exposure in the first trimester of pregnancy. This was in line with the

WHO recommendation following informal consultations convened in 2006 to assess the safety of

artemisinin compounds in pregnancy, highlighting the need to set up pharmacovigilance systems to

document first trimester exposure to ACTs.

In Chapter 3 we provided an estimate for the number of pregnancies living in areas with malaria

transmission in 2007. Updated maps of P. falciparum and P. vivax transmission were combined with

gridded population data and growth rates to estimate total populations at risk of malaria in 2007.

The numbers of pregnancies was derived from country‐specific demographic data from the United

Nations, combined with estimates for stillbirths and induced abortions derived from contemporary

published reviews and estimates of spontaneous abortions based on an established formula (using

the number of live‐births and induced abortions). We estimated that in 2007, 125 million

pregnancies occurred in areas with P. falciparum and/or P. vivax transmission of which

approximately 60% resulted in live births. Estimates from Africa (32 million) were similar to previous

estimates by WHO (25–30 million) but estimates for non‐African regions was much higher than

previously estimated (95 million vs. 25 million). These estimates of the number of pregnancies at

risk of malaria provided a first step towards a spatial map of the burden of malaria in pregnancy.

Recently Walker et al estimated the risk of placental infection and low birthweight attributable to

Plasmodium falciparum malaria in Africa taking into account heterogeneity in transmission and

parity‐dependent effects.[1] They report that 12 million pregnancies resulting in live‐births would

have been exposed to malaria infections without malaria protection (i.e. through insecticide treated

nets or IPTp) resulting in 900,000 low birthweight babies. They projected that the majority of

placental infections (65%) would occur early in the first trimester of pregnancy. Considering that

only about two thirds of pregnancies result in live‐births the number of pregnancies potentially

exposed to infection could be up to 18 million. It is difficult to predict what proportion of such

Summary and Discussion

181

infections would lead to treatment with an antimalarial as in areas of high transmission most malaria

infections remain asymptomatic particularly in multigravidae. Nevertheless even if only a small

proportion were to seek treatment, millions of pregnancies could be exposed to antimalarials each

year and many in the first trimester of pregnancy.

In Chapter 4, we estimated that the probability an embryo will encounter artemisinins during the

critical six‐ week period (weeks 6 to 12 post‐LMP) through accidental exposure is 12% for areas

where adults receive on average one treatment with three days of artemisinin‐based combination

therapy per year. This assumed that the likelihood of infections was similar throughout pregnancy. In

light of the modelling study by Walker et al, this probability is likely an underestimate. This

information is useful to grasp the scale of the problem related to malaria in pregnancy and the

potential for exposures to antimalarial in pregnancy. In this chapter we further described the

methodological considerations for the systematic assessment of pregnancy outcomes and congenital

malformations in women exposed to antimalarials early in pregnancy, as well as approaches to

capture drug exposure information, choice of comparison groups and sample size concerns. We

proposed a targeted prospective pharmacovigilance approach enabling timely assessment of the

risk‐benefit profile of antimalarials through the establishment of an international antimalarial

pregnancy exposure registry. The WHO recently published a description of methods for pregnancy

exposure registries in resource constrained settings which provides valuable tools for standard

methods of data collection facilitating pooled data analysis.[2]

A complementary approach to pregnancy exposure registries is illustrated in Chapter 5, which

describes the findings from a pilot study to assess the feasibility of record linkage using routinely

collected healthcare data as a means of monitoring the safety of ACTs in early pregnancy. This

study used data from paper‐based registers (2004–2008) from a mission‐run dispensary in Mlomp,

south‐western Senegal. Data from the outpatient clinics and delivery registers were linked based on

a probabilistic matching approach as no unique identifier was available. The findings from this

feasibility study suggest that record linkage using routine healthcare data is feasible in resource‐

constrained settings with a relatively well defined catchment population. Tapping into readily

available data sources of sites adequate for record linkage could greatly contribute to the high

numbers needed to provide adequate reassurance for ACT use in the first trimester of pregnancy,

in terms of stillbirths and major congenital malformations. However, in these settings assessment

of the risk of miscarriage and specific birth defects, such as congenital heart defects, would require

dedicated studies. Sites amenable for record linkage studies need to have reliable and

comprehensive medical records for treatment, pregnancy and maternity services. This requires a

high proportion of health facility deliveries and limited availability of the drug of interest outside the

central health facility record system (and therefore not captured in the health records). A prime

example of such sites is the Shoklo Malaria Research Unit (SMRU) clinic which has already

contributed significantly to what is known about ACT risk‐benefit profile in the first trimester of

pregnancy.[3] The unique set‐up of the SMRU allows it to serve a stable refugee population where

many pregnant women receive weekly antenatal care, pregnancies can be identified early, the

gestational age assessed at every contact, the health‐care provided, and drug exposure and birth

outcome recorded centrally. Other potential sites with comprehensive and reliable healthcare

records include agricultural estates providing free healthcare for their workers, refugee camps or use

of health insurance data where coverage is high and homogeneous.

Chapter 9

182

In Chapter 6 we investigated contextual information on perceptions of adverse pregnancy outcomes

in an area of high malaria transmission in western Kenya. Through ten focus group discussions we

explored the perceptions, beliefs and health‐seeking behaviours of women from rural western Kenya

regarding congenital anomalies and miscarriages. We found that lack of information regarding

causes of adverse pregnancy outcomes such as miscarriages and congenital anomalies could lead to

stigmatisation of the mother and further hamper health seeking behaviour. This could have

devastating consequences for children born with a malformation who are sometimes hidden from

society and don’t have access to care that could improve their quality of life. Although women

reported that care is usually sought in case of complications following a miscarriage, delaying care

until symptoms are pronounced could adversely affect recovery and the woman’s health. These

findings highlight the need for education about the potential cause of adverse pregnancy outcomes

and better information on where care is available. The qualitative data gathered through these focus

group discussions was informative to understand the socio‐cultural context around pregnancy and

pregnancy outcome. It was also interesting to understand that although some women were aware of

the potential danger of medicines used in pregnancy, drugs considered safe in medical practice were

being reported as potentially dangerous. Mostly a risk was perceived when drugs were used without

or not following clinician recommendations. This emphasised the need for education materials for

both community and healthcare providers regarding the consideration for the choice of treatment in

pregnancy.

In Chapter 7 we described healthcare provider and drug dispenser adherence to and knowledge of

national guidelines for treatment of uncomplicated malaria in pregnancy. We conducted a cross‐

sectional study from September to November 2013, in health facilities and randomly selected drug

outlets in an area of high malaria transmission in western Kenya. This study highlighted knowledge

inadequacies and incorrect prescribing practices in the treatment of malaria in pregnancy,

particularly in the first trimester. As the first line treatment for malaria, artemether‐lumefantrine, is

not recommended in the first trimester of pregnancy, all women of childbearing age should be

assessed for potential pregnancy. Such pregnancy enquiries were only observed in 44% of cases in

health facilities and 7% in drug outlets, although 93% and 49% reported routinely assessing for

potential pregnancy respectively. A reason for this could be that prescribers and dispensers do not

feel adequately prepared to enquire about potential pregnancy and/or the lack of access to

pregnancy tests. Prescription of the correct drug at the correct dosage was observed in 32% of all

cases in health facilities and 0% in drug outlets for first trimester cases. Knowledge was moderately

higher for health facility staff (56%) and remained nonexistent in drug outlets staff. The reason for

the discrepancy between what is reported and what was observed should be further explored.

Exposure to artemether‐lumefantrine in first trimester occurred in 16% and 51% of cases in health

facilities and drug outlets, respectively; none were a result of quinine stock‐out. Overall, SP was

prescribed as treatment in 11% of all cases and the vast majority of all women prescribed quinine

were given an insufficient supply. Provider knowledge in both settings was poor and was reflective

of the low levels of correct case management observed in practice. Such incorrect prescribing

practice can have serious consequence for the pregnant patients not only by the risk posed by using

a drug of undetermined safety for the unborn baby (ACTs) but also in terms of adverse

consequences due to the inability to clear a malaria infection when treated with an ineffective drug

(SP) or incomplete drug regimen (as seen with quinine). The latter also contributes to emerging

parasite resistance to these drugs. This study provided insights on the high potential for ACT


183

exposure in pregnancy in this setting but more broadly revealed a need for better dissemination of

guidelines for case management of malaria in women of childbearing potential with the need to

include both formal and informal sectors.

In Chapter 8 the findings from a prospective cohort study on the risk of inadvertent exposure to

ACTs in the first trimester of pregnancy and its association with miscarriage in Western Kenya were

presented. Community‐based surveillance was used to identify early pregnancies among women of

childbearing age under health and demographic surveillance in western Kenya. Multiple data

sources (including outpatient records, longitudinal household level morbidity surveillance data and

study specific pregnancy follow up questionnaires) were combined to ascertain ACT exposures in

pregnancy. Out of the 1,126 pregnancies included in the analysis, there were 42 (4%) confirmed ACT

exposures in the suggested artemisinin embryo‐sensitive period (6‐12 weeks post LMP) and 75 (7%)

confirmed exposures in the first trimester. This study showed that pregnancy detection in the first

trimester (before most women present for ANC) was feasible through community based testing in

this setting. We provided the first estimates for the rates of miscarriage stratified by gestational

week in this area and found that the cumulative probability of miscarriage was 19% in this

population during the period that women were under observation (up to 28 weeks of gestation).

Compared to pregnancies without reported malaria and antimalarial treatment, those with a

confirmed ACT exposure in the suggested artemisinin embryo‐sensitive period did not have a

statistically significant different hazard of miscarriage [HR=0.85 95% CI (0.22‐3.33)]. There was also

no statistical difference for the confirmed exposures in the first trimester as a whole, although the

effect estimate suggested a potential increase in risk [HR=1.60 95% CI (0.70‐3.68)]. As noted from

the upper bound of the confidence interval, we could not rule out a 3.7 fold increase in miscarriages

linked to ACT exposure in early pregnancy. In line with what was observed in the cross‐sectional

survey on prescribing behaviours (chapter 7), very few women in their first trimester of pregnancy

were treated with the recommended treatment, quinine (n=34 with 22 exposed to both quinine and

ACTs). This hampered our ability to compare pregnancies exposed to ACTs in the first trimester to

those exposed to a safe drug. Although an increased risk of miscarriage could not be discarded, the

results are consistent with recent findings reported from the Thai‐Burmese border and provide

further reassurance regarding the use of ACTs in early pregnancy.[3] Findings from pooled data

analysis combining data from this study and 2 other sites that were part of the Malaria in Pregnancy

Consortium, assessing risk of miscarriage, stillbirth and major congenital malformations are

forthcoming and will provide further insights on the risk‐benefit profile of ACTs.

ACT risk‐benefit profile in first trimester of pregnancy: where are we to date? Since the review presented in chapter 2, more data has been published, increasing the number of

documented first trimester exposures to 757 (see table 1). The evidence so far is reassuring and

rules out artemisinin derivatives as major teratogens (defined as a drug that would increase the risk

of major congenital malformations by 5 to 10 fold). The 757 documented first trimester exposures

confer enough statistical power to detect relative risk of 2.3 or greater for overall major

malformations and a RR of 1.3 for miscarriages (assuming the background rate of 0.9% for major

malformations detectable by surface examination at birth and 12% for miscarriage with ratio of

exposed to unexposed of 1:4 at 80% power). However for specific congenital anomalies which occur

at frequencies of 1/1000 for the most common, the 757 documented exposures could only detect a

relative risk of 6.6.

Chapter 9

184

Table 13. Description of documented ACT first trimester exposures.

Author Country Publication Year

Number of first trimester exposures

Reference

McGready TBB 2012 64 [3]

Deen Gambia 2001 77 [4]

Adam Sudan 2009 62 [5]

Mayando Zambia 2010 156 [6]

Rulisa* Rwanda 2012 96 [7]

Mosha Tanzania 2014 172 [8]

Poespoprodjo Indonesia 2014 11 [9]

Dellicour Kenya Not yet published 80 MiPc

Sevene Mozambique Not yet published 3 MiPc

Tinto Burkina Faso Not yet published 36 MiPc

Total 757

*Rulisa et al published results from an observational study following pregnant women exposed to artemether‐lumefantrine however pregnancy outcomes were not reported separately for first vs second/third trimester exposures

Interestingly, few studies reported the effect of exposures during the suggested artemisinin embryo‐

sensitive period (only McGready et al 2012 and the study presented in chapter 8). The suggested

period (6‐12 weeks post LMP) was derived from the observed causal mechanism in animal models

indicating that primitive erythroblasts are the primary target for embryotoxicity and these are the

weeks when primitive erythroblasts are the predominant form of erythrocyte in circulation in human

embryos.[10] As considering the exposure for the whole first trimester could bias effect estimates

towards the null, it is important to carry out analysis to explore the associations between ACT

exposures and outcomes during this period of maximum sensitivity in humans. This was performed

and presented in chapter 8 where we found no trend of increase in risk to exposure in the sensitive

period compared to the whole first trimester. Another point for consideration is the need to

investigate congenital cardiovascular defects as this was one of the safety signals identified from

preclinical reprotoxicology studies in rodents.[11,12] None of the studies were designed to assess

cardiovascular defects, which require review by a cardiologist and access to specialised equipment

(ultrasound for echocardiography, electrocardiogram and x‐ray). Furthermore, most of the studies in

table 1 were not designed to capture early pregnancy losses and adjustment for potential survival

bias will need to be considered when assessing miscarriages as an end point.

Meta‐analysis with individual patient data would be a useful next step for better evaluation of the

safety profile of ACTs in the first trimester of pregnancy. This will need to take into account

heterogeneity in data collection methods and different levels of certainty regarding potential

exposure misclassifications. Review of current evidence by the WHO Technical Expert Group (TEG)

on malaria chemotherapy will be a first step in deciding if more and what type of data is needed to

inform the Malaria Policy Advisory Committee (MPAC) which provides independent advice to the

WHO for the development of policy recommendations.

Safety is not an absolute property of a drug and it will depend on the appropriate use of the drug

considering indication, dosage, efficacy and suitability for the patient. A balance between risk and

benefit needs to be determined. In the case of ACTs, the deleterious effect of malaria infection in the

first trimester of pregnancy and the benefit of rapid and effective treatment needs to be taken into

consideration. Recent publications highlight first trimester malaria infections are more of a problem


185

than previously thought, with two‐thirds of all malaria placental infections occurring in the first 12

weeks of pregnancy (without preventive interventions), the high associated risk of maternal anaemia

and impact on birthweight as well as a 3‐4 fold increased odds of miscarriage with malaria infections

in early pregnancy in an area of low and seasonal malaria transmission.[1,3,13,14] We found that

treatment with ACTs was much more common than the recommended first line treatment quinine in

the first trimester of pregnancy in the study area in western Kenya. It is unclear whether this was

due to inadvertent exposures because of unknown or undisclosed pregnancies, because of lack of

knowledge regarding the national treatment guidelines or whether this was a deliberate choice

because quinine was out of stock or that an ACT was considered a better option for these women.

Poor tolerability to quinine and common low compliance to the long 7‐day regimen are known

problems.[15,16] There is a clear need for improved recommendations regarding treatment of

malaria in early pregnancy but also a need to investigate suitability of quinine for case management

of uncomplicated malaria in pregnancy.

Before a change in the treatment policy for malaria in the first trimester of pregnancy is

recommended, there will be a need to monitor and address poor compliance with current malaria

treatment guidelines for early pregnancy. We found poor awareness of contra‐indication of ACTs in

early pregnancy among healthcare providers, drug outlet dispensers and women in the community

(chapters 6‐7). Clearer guidelines emphasising the need for pregnancy assessments of all women of

childbearing age seeking treatment for malaria should be developed. Pregnancy testing in the

community was found acceptable in our study which suggests that pregnancy tests offered by

trained healthcare professionals could be feasible. There are at least 2 RDTs which combine malaria

and pregnancy in one test, and one has an acceptable performance profile as reviewed in the last

round of malaria diagnostic product testing reported by WHO.[17] These could be considered for use

in female patients of childbearing potential. Implementability (including training, costs and

availability) of pregnancy tests at point of care would need to be determined. In all situations a

woman’s privacy and the confidentiality of her pregnancy status should be preserved by handling

records carefully and choosing a suitable place for pregnancy assessments. This is particularly

delicate when dealing with minors and adolescent girls who might be reluctant to disclose their

pregnancy. This requires an objective, accepting and professional attitude from the healthcare

provider.[18] The purpose of the pregnancy assessment should be made clear to the client such that

it is to ensure the adequacy and safety of the treatment. Pregnancy assessments should be

considered in all antimalarial dispensing scenarios with women of childbearing potential. This

includes drug outlets, both formal and informal, and community based strategies for case

management of malaria. Furthermore, there is renewed interest in mass drug administration for

malaria elimination which will need to consider strategies for management of potential pregnancies.

The lack of pregnancy assessment in women of childbearing potential has important implications for

risk management programmes for other medications contraindicated in pregnancy in this setting.

What are the prospects for sustainable pharmacovigilance systems for pregnant women in resource constrained settings? No single method can capture all desired data needed to make appropriate risk benefit assessments

of a drug used in pregnancy. A combination of different methods and data sources is the only

feasible approach to gather the most complete picture of potential developmental toxicity of a drug.

The different approaches presented in this thesis have their own strengths and limitations but

provide complimentary information. Common challenges include the need to obtain reliable and

Chapter 9

186

accurate exposure data. In their retrospective analysis, McGready et al included fewer than 100 well‐

documented early exposures to artemisinins after review of 25 years of data including over 48,000

pregnancies.[3] We found only 25% of first trimester ACT exposures could be confirmed (chapter 8)

and the sensitivity of women’s self‐report at the time of pregnancy follow up interviews was

estimated at 37% (based on confirmed exposures by the 2 other data sources, see chapter 8

Appendix).This indicates that studies or surveillance systems cannot solely rely on women self‐report

for antimalarial drugs, particularly if the recall period stretches over several months. Secondly, it is

very challenging to account for confounding by indication (i.e. the fact that the disease for which the

drug is being indicated, itself causes adverse pregnancy outcomes) in settings where self‐medication

of febrile episodes is common and where few women are treated with quinine.[19,20] Furthermore,

due to the nature of these observational studies, it is not possible to account for factors influencing

antimalarial treatment choice by healthcare providers and patients. If more severe cases are usually

treated with one type of drug rather than other, this could confound the association between

exposure and adverse pregnancy outcome. Inclusion of internal unexposed controls is necessary (as

opposed to using background rates which is commonly done with pregnancy exposure registries in

high income countries) as there is a lack of data on background rates of most adverse pregnancy

outcomes (particularly miscarriages and birth defects) in most malaria endemic areas. Nevertheless,

the use of internal unexposed controls allows a more appropriate comparison and provides the

potential to control for the underlying disease provided there is enough variability in the different

class of antimalarials used. Randomised controlled trials are considered the “gold standard” for

generating reliable evidence in clinical research and could address these methodological challenges.

Depending on the forthcoming findings from pooled data analysis, a randomised controlled trial

comparing ACTs to quinine for confirmed malaria in the first trimester of pregnancy may be

warranted to provide the level of reassurance needed for policy change. Whereas, a clinical trial was

considered unsuitable during the last WHO consultations to review evidence on the safety of

artemisinin derivatives in pregnancy in 2006 due to the limited data on first trimester exposure in

humans, we may have reached the point of equipoise. Equipoise is defined as a state of genuine

uncertainty regarding the comparative merits of different therapeutic choices and considered

essential for the ethical acceptability of clinical trials.[21] WHO guidelines for the treatment of

uncomplicated malaria in the first trimester of pregnancy state that an ACT is indicated if there is

uncertainty of compliance with a 7 day quinine regimen.[22] In areas where treatment with ACTs

already predominates quinine in the first trimester, a randomised controlled trial should be

considered ethical.

Most teratogenic safety signals have been identified by astute clinicians through reporting of case

series of abnormal pregnancy outcomes to exposed mothers.[23] Although this is less than ideal and

often takes several years for the signal to be detected, the potential of such reports shouldn’t be

ignored. However this requires awareness of pharmacovigilance and special considerations for

treatment of women of childbearing potential. We found this to be low among healthcare providers

and drug dispensers in our study area and this probably is not unique to rural western Kenya.

Integration of pharmacovigilance topics into university and professional curricula for all health

disciplines has been proposed.[24] The recent establishment of UMC Africa (a WHO Collaborating

Centre for Advocacy and Training in Pharmacovigilance for Africa) in Ghana provides an opportunity

to share specific tools and education materials for pharmacovigilance methods for pregnant

women.[25]


187

Identification of sentinel sites able to capture reliable data on drug exposure and pregnancy

outcomes through a standard protocol as proposed in chapter 4 and recently by WHO would be

essential for safety signal detection and characterisation.[2] Such an approach could be used for the evaluation of a wide variety of medications that are used during pregnancy and estimation of the

risk of the spectrum of adverse pregnancy outcomes. However depending on the recruitment

strategy, assessment of miscarriages might not be feasible without introducing dedicated efforts to

detect these. Enrolling pregnant women attending for antenatal care is an interesting option, as

close to 90% of pregnant women in sub‐Saharan Africa attend antenatal clinics at least once.[26]

However as many women present for their first ANC visit after completion of the first trimester (the

average is around 22 weeks gestation), such an approach limits the possibility of studying early

miscarriages as an outcome. For this reason, prospective community‐based studies (as presented in

chapter 8) are needed for detection of early pregnancies or systems to incentivise women and/or

community health workers to refer pregnant women to present to ANC early should be explored.

Health and demographic surveillance sites make attractive candidates as they provide a framework

to identify women of childbearing age and usually have some level of health facility surveillance set

up. This is exemplified through the implementation of prospective observational cohorts to field‐test

active surveillance systems for identifying exposures to antimalarials during early pregnancy and for

monitoring pregnancy outcomes in three health and demographic surveillance sites in Africa

(Burkina Faso, Mozambique and Kenya as presented in chapter 8) as part of the Malaria in

Pregnancy consortium (MiPc).

Major limitations of traditional pregnancy exposure registries and prospective cohort studies include

costs and length of time required to achieve a desirable sample size. As an alternative, record

linkage studies using routine healthcare data provides a cost‐effective and relatively rapid means of

assessing the embryotoxic effect of a drug. Record linkage studies require minimal staff and

resources particularly in settings with electronic medical records. However, this necessitates sites

with comprehensive data on drug exposure and pregnancy outcomes. To minimize exposure

misclassification, it is essential to select sites within a relatively stable population where healthcare

is provided in a central location with a high proportion of health facility deliveries and limited

availability of the drug of interest outside the central health facility system. Sites where this

approach could be implemented effectively are few, and further study should assess suitability of

private agricultural and other industrial estates, mines and refugee camps as well as data from health

insurance schemes in malaria endemic countries.

How sustainable are these approaches? There are no obvious funding mechanisms for

pharmacovigilance activities. In resource constrained settings, priorities often lie with the delivery of

care and pharmacovigilance programmes currently have to compete for very scarce resources.[27]

However confidence in a public health programme can be seriously affected by adverse drug

reactions, particularly serious adverse events affecting pregnancy outcomes (such as pregnancy loss

or congenital anomalies).[28] Evidence on the safety of drugs implemented by public health

programmes has the potential to increase programme success by providing guidance on reducing

the risks of adverse drug reactions thereby increasing overall public trust.[29,30] The Roll Back

Malaria partnership issued guidance to countries applying for malaria funding for the inclusion of

pharmacovigilance component within donor‐supported programmes.[31] The Global Fund to fight

AIDS, tuberculosis and malaria has made provision for specific support towards pharmacovigilance

which could be used to fund such initiatives for antimalarials, antiretrovirals and tuberculosis

Chapter 9

188

treatments.[32] The United States' President's Malaria Initiative (PMI), is one of the major donors for

malaria control programmes in target countries in sub‐Saharan Africa and plays a key role in the

procurement of antimalarial treatment and assessment of drug safety. However, few proposals to

either funding bodies included a request for funding for pharmacovigilance activities.[33] In 2011,

the European Medicines Agency (EMA) approved Eurartesim (dihydroartemisinin‐piperaquine) for

marketing with the condition that a pregnancy registry be set up as part of the post‐marketing risk

management plan (among other recommendations).[34,35] This so far has resulted in the creation

of a European based pregnancy registry to monitor inadvertent exposures to Eurartesim in European

travellers, who rarely have access to Eurartesim, which is a treatment drug provided only to those

with clinical malaria. Little progress has been made with developing a platform to monitor the safety

of Eurartesim in pregnancy in malaria endemic countries where most of the exposures are likely to

occur.[36] Sustainability of the proposed approaches will need political commitment and key

stakeholders (including industry, governments and funders of public health programmes) buy‐in to

prioritise support for pharmacovigilance systems.

In conclusion, the reports presented in this thesis support the need and feasibility of implementing

various pharmacovigilance approaches for monitoring safety of medicines in pregnancy in resource

constrained settings. In light of the soaring number of first trimester pregnancies being exposed

inadvertently and deliberately to artemisinin derivatives, an update of the last 2006 WHO review of

the existing evidence is needed promptly to inform treatment guidelines for malaria in pregnancy.

Observational studies on ACT exposures in the first trimester of pregnancy have provided a degree

of reassurance and have not detected major teratogenicity signals in humans so far. However

numbers are still limited to less than a 1000 well documented first trimester exposures, and the rule

of 3 suggests that this can demonstrate a less than two‐fold increase in the overall incidence of

malformations.[37] Thus more research is needed, also to assess the risk of cardiovascular defects

associated with ACT exposures in early pregnancy as well as studies looking at congenital anomalies

detectable later in life including developmental delays. Furthermore, once enough data has accrued

to provide sufficient statistical power, it will be important to carry out a sensitivity analysis of the

suggested period of maximum artemisinin sensitivity in human embryos. With the accrued evidence

to date and the low adherence to the recommended treatment with quinine in first trimester of

pregnancy, a randomised controlled trial would be the fastest way to provide the level of evidence

needed. Until adequate evidence is available to comprehensively assess the risk‐benefit profile of

ACTs for early pregnancy, it will be important to minimise first trimester exposures. Current options

for treatment of pregnant women with malaria are few and appropriate prescribing and dispensing

is critical for rational drug use. As such healthcare providers and drug dispensers should have

relevant knowledge and skills regarding antimalarial drug use in women of childbearing potential.

Further research is needed to investigate the best approach to implement risk management to

minimise first trimester exposures including ways for appropriate dissemination of malaria

treatment guidelines, as well as pregnancy assessment strategies for healthcare professionals in

health facilities, drug dispensers working in the informal sector, and community health workers in

areas where their mandate includes treatment of malaria in women of childbearing age. This thesis

focused on treatment of malaria in pregnancy; however most of the findings are applicable to other

tropical diseases in pregnancy. The case of ACTs is not unique and most of the new medicines for

tropical diseases in pregnancy are not recommended for pregnancy due to lack of information.[38]

Given the high burden of disease in the tropics and the extensive overlap between malaria and HIV,


189

it will be important to consider all concomitant drug exposures and potential for drug‐drug

interactions. Joint efforts are needed to tackle this paradoxical situation to enable safe and effective

treatment of tropical diseases in pregnancy.

References


2. Mehta U, Clerk C, Allen E, Yore M, Sevene E, et al. (2012) Protocol for a drugs exposure pregnancy registry for implementation in resource‐limited settings. BMC Pregnancy Childbirth 12: 89.

3. McGready R, Lee SJ, Wiladphaingern J, Ashley EA, Rijken MJ, et al. (2012) Adverse effects of falciparum and vivax malaria and the safety of antimalarial treatment in early pregnancy: a population‐based study. Lancet Infect Dis 12: 388‐396.

4. Deen JL, von Seidlein L, Pinder M, Walraven GE, Greenwood BM (2001) The safety of the combination artesunate and pyrimethamine‐sulfadoxine given during pregnancy. Trans R Soc Trop Med Hyg 95: 424‐428.

5. Adam I, Elhassan EM, Omer EM, Abdulla MA, Mahgoub HM, et al. (2009) Safety of artemisinins during early pregnancy, assessed in 62 Sudanese women. Ann Trop Med Parasitol 103: 205‐210.

6. Manyando C, Mkandawire R, Puma L, Sinkala M, Mpabalwani E, et al. (2010) Safety of artemether‐lumefantrine in pregnant women with malaria: results of a prospective cohort study in Zambia. Malar J 9: 249.

7. Rulisa S, Kaligirwa N, Agaba S, Karema C, Mens PF, et al. (2012) Pharmacovigilance of artemether‐lumefantrine in pregnant women followed until delivery in Rwanda. Malar J 11: 225.


9. Poespoprodjo JR, Fobia W, Kenangalem E, Lampah DA, Sugiarto P, et al. (2014) Dihydroartemisinin‐piperaquine treatment of multidrug resistant falciparum and vivax malaria in pregnancy. PLoS One 9: e84976.

10. Clark RL (2009) Embryotoxicity of the artemisinin antimalarials and potential consequences for use in women in the first trimester. Reprod Toxicol 28: 285‐296.

11. WHO, TDR (2007) Assessment of the safety of artemisinin compounds in pregnancy: report of two joint informal consultations convened in 2006. Geneva: WHO. WHO/GMP/TDR/Artemisinin/07.1. WHO/GMP/TDR/Artemisinin/07.1.

12. Clark RL, White TE, S AC, Gaunt I, Winstanley P, et al. (2004) Developmental toxicity of artesunate and an artesunate combination in the rat and rabbit. Birth Defects Res B Dev Reprod Toxicol 71: 380‐394.

13. Cottrell G, Mary JY, Barro D, Cot M (2007) The importance of the period of malarial infection during pregnancy on birth weight in tropical Africa. Am J Trop Med Hyg 76: 849‐854.

14. Huynh BT, Fievet N, Gbaguidi G, Dechavanne S, Borgella S, et al. (2011) Influence of the timing of malaria infection during pregnancy on birth weight and on maternal anemia in Benin. Am J Trop Med Hyg 85: 214‐220.

15. Achan J, Talisuna AO, Erhart A, Yeka A, Tibenderana JK, et al. (2011) Quinine, an old anti‐malarial drug in a modern world: role in the treatment of malaria. Malar J 10: 144.

16. Taylor WR, White NJ (2004) Antimalarial drug toxicity: a review. Drug Saf 27: 25‐61. 17. WHO (2013) Malaria Rapid Diagnostic Test Performance – Summary results of WHO product testing of

malaria RDTs: Round 1‐5 (2008‐2013). Geneva: WHO. 18. Perlman SE, Postlethwaite D, Stump S, Bielan B, Rudy SJ (2001) Taking a sexual history from and counseling

women on teratogenic drugs. J Reprod Med 46: 163‐168. 19. Chipwaza B, Mugasa JP, Mayumana I, Amuri M, Makungu C, et al. (2014) Self‐medication with anti‐

malarials is a common practice in rural communities of Kilosa district in Tanzania despite the reported decline of malaria. Malar J 13: 252.

20. Geissler PW, Nokes K, Prince RJ, Odhiambo RA, Aagaard‐Hansen J, et al. (2000) Children and medicines: self‐treatment of common illnesses among Luo schoolchildren in western Kenya. Soc Sci Med 50: 1771‐1783.

21. Freedman B (1987) Equipoise and the ethics of clinical research. N Engl J Med 317: 141‐145. 22. WHO (2010) Guidelines for the treatment of malaria. Second Edition. Geneva.

Chapter 9

190

23. Carey JC, Martinez L, Balken E, Leen‐Mitchell M, Robertson J (2009) Determination of human teratogenicity by the astute clinician method: review of illustrative agents and a proposal of guidelines. Birth Defects Res A Clin Mol Teratol 85: 63‐68.

24. Isah AO, Pal SN, Olsson S, Dodoo A, Bencheikh RS (2012) Specific features of medicines safety and pharmacovigilance in Africa. Ther Adv Drug Saf 3: 25‐34.

25. UMC‐A the Uppsala Monitoring Centre Africa. 26. van Eijk AM, Hill J, Alegana VA, Kirui V, Gething PW, et al. (2011) Coverage of malaria protection in

pregnant women in sub‐Saharan Africa: a synthesis and analysis of national survey data. Lancet Infect Dis 11: 190‐207.

27. Bakare N, Edwards IR, Stergachis A, Pal S, Holmes CB, et al. (2011) Global pharmacovigilance for antiretroviral drugs: overcoming contrasting priorities. PLoS Med 8: e1001054.

28. Brabin BJ, Warsame M, Uddenfeldt‐Wort U, Dellicour S, Hill J, et al. (2008) Monitoring and evaluation of malaria in pregnancy ‐ developing a rational basis for control. Malar J 7 Suppl 1: S6.

29. Dodoo A, Adjei S, Couper M, Hugman B, Edwards R (2007) When rumours derail a mass deworming exercise. Lancet 370: 465‐466.

30. Pal SN, Duncombe C, Falzon D, Olsson S (2013) WHO strategy for collecting safety data in public health programmes: complementing spontaneous reporting systems. Drug Saf 36: 75‐81.

31. Roll Back Malaria (2008) Steps for Completing Global Fund’s Round 8 Proposal for Requesting funds for Pharmacovigilance Geneva.

32. Xueref S, Daviaud J, Pal S (2013) The Global Fund and pharmacovigilance systems in resource‐limited settings. Lancet 381: 1360.

33. Stergachis A, Bartlein RJ, Dodoo A, Nwokike J, Kachur SP (2010) A situational analysis of pharmacovigilance plans in the Global Fund Malaria and U.S. President's Malaria Initiative proposals. Malar J 9: 148.

34. European Medicines Agency (2011) Assessment Report Eurartesim. London. 35. Ubben D, Poll EM (2013) MMV in partnership: the Eurartesim(R) experience. Malar J 12: 211. 36. Sigma‐Tau (2014) Eurartesim Pregnancy Registry. 37. European Medicines Agency (2008) Guidelines on risk assessment of medicinal products on human

reproduction and lactation: from data to labelling. London: EMA. 38. White NJ, McGready RM, Nosten FH (2008) New medicines for tropical diseases in pregnancy: catch‐22.

PLoS Med 5: e133.

191

Samenvatting

Samenvatting

192

Nauwkeurige informatie over de veiligheid van geneesmiddelen in de zwangerschap na de afronding

van de formele klinische studies is moeilijk te verkrijgen, vooral in ontwikkelingslanden waar de

infrastructuur voor pharmacovigilantie niet goed ontwikkeld is. In deze landen is er behoefte aan

een simpel maar effectief systeem om de veiligheid van geneesmiddelen gedurende de

zwangerschap te volgen. Hoewel artemisinine gebaseerde combinatietherapie (ACT) niet is

aanbevolen voor gebruik tijdens het eerste trimester van de zwangerschap, is onbedoelde

blootstelling onvermijdelijk. Er zijn op dit moment geen gevestigde methodes om de veiligheid van

deze geneesmiddelen tijdens de zwangerschap te volgen in landen waar malaria endemisch is. Het

onderzoek beschreven in dit proefschrift evalueert surveillance systemen voor medicijnen in landen

met beperkte middelen om een beter beeld te krijgen van het risico‐baten profiel van het gebruik

van artemisinine derivaten tijdens de vroege zwangerschap.

In hoofdstuk 2 wordt het gepubliceerde bewijs met betrekking tot de veiligheid van artemisinine

derivaten gebruik tijdens de zwangerschap geëvalueerd door middel van systematisch literatuur

onderzoek. Ten tijde van de evaluatie (november 2006) werden 14 relevante studies gevonden met

in totaal 945 vrouwen die waren blootgesteld aan artemisinine derivaten gedurende de

zwangerschap, waarvan 123 in het eerste trimester. De beperkte hoeveelheid informatie

suggereerde dat artemisinine derivaten in het tweede en derde trimester effectief en waarschijnlijk

veilig waren. Zeldzame bijwerkingen konden echter niet worden uitgesloten. Er was onvoldoende

informatie voor een afweging van het risicoprofiel tijdens het eerste trimester, de meest gevoelige

periode voor eventuele ongewenste bijwerkingen op het ongeboren kind. De bevindingen van het

literatuur onderzoek waren overeenkomstig een aanbeveling van de Wereldgezondheidsorganisatie

(WHO) in 2006. De noodzaak om een systeem op te zetten om blootstelling aan, en gevolgen van

artemisinine derivaten tijdens het eerste trimester te documenteren werd benadrukt door de WHO

in 2006.

In hoofdstuk 3 geven wij een schatting van het jaarlijks aantal zwangerschappen in gebieden met

malaria transmissie. Geografische gegevens van de transmissie van Plasmodium falciparum en

Plasmodium vivax werden gecombineerd met demografische gegevens van de Verenigde Naties om

de totale populatie in risico gebieden voor malaria in 2007 te inventariseren. Wij schatten dat er in

2007 125 miljoen zwangerschappen waren in malaria gebieden, waarvan ongeveer 60%

resulteerden in een levend geboren kind. Schattingen voor Afrika (32 miljoen) waren vergelijkbaar

met eerdere berekeningen van de Wereldgezondheidsorganisatie (25‐30 miljoen), maar schattingen

voor regio’s buiten Afrika waren veel hoger dan voorheen. Deze berekening van het aantal

zwangerschappen in malaria endemische gebieden was een eerste stap om het aantal

zwangerschappen in te kunnen schatten dat jaarlijks wordt blootgesteld aan malaria en dus het

aantal eventuele behandelingen met antimalariamiddelen.

In hoofdstuk 4 berekenen we dat een embryo een kans heeft van 12 procent om onbedoeld aan

artemisinine derivaten blootgesteld te worden gedurende de 6 weken die het meest gevoelig zijn

voor mogelijke schadelijke effecten van artemisinine derivaten (6‐12 weken na de laatste

menstruatie). Deze informatie is nuttig om de omvang van de kans op blootstelling aan

antimalariamiddelen gedurende de zwangerschap te begrijpen. In dit hoofdstuk worden verder de

methodologische overwegingen beschreven voor een systematische beoordeling van de uitkomst

van zwangerschap en aangeboren afwijkingen bij vrouwen die vroeg tijdens de zwangerschap zijn

bloot gesteld aan antimalariamiddelen. Tevens worden verschillende methoden besproken hoe men

193

blootstelling aan medicijnen kan vast stellen, de keuze van mogelijke controle groepen en de

steekproefgrootte. Voor tijdige beoordeling van het risicoprofiel van antimalariamiddelen wordt een

doelgerichte prospectieve aanpak voorgesteld door middel van de oprichting van een internationaal

register voor blootstelling aan antimalariamiddelen. De Wereldgezondheidsorganisatie publiceerde

recentelijk een beschrijving voor het opzetten van landelijke registers om blootstelling aan

medicijnen tijdens de zwangerschap vast te leggen in gebieden met beperkte middelen. Dit biedt

mogelijkheden om gestandaardiseerde methodes te gebruiken bij het verzamelen van gegevens. [2]

In hoofdstuk 5, worden de bevindingen beschreven van een verkennende studie naar de

haalbaarheid van het gebruik van routinematig verzamelde gegevens over antimalariamiddelen in

registerboeken van algehele en verloskunde poliklinieken en over de uitkomst van bevallingen in

registers van verloskamers om de veiligheid van artemisinine derivaten tijdens het begin van de

zwangerschap te beoordelen. Deze studie werd verricht in een medische missie‐post in Mlomp, in

het zuidwesten van Senegal. De informatie van de verschillende registers werden gekoppeld op basis

van een statistische methode die gebruikt maakt van ‘probabilistische matching’ om

corresponderende gegevens van dezelfde patiënt te vinden in de verscheidene registers op basis van

een naam, leeftijd, woonplaats en dergelijke. De bevindingen lieten zien dat het koppelen van

routinematig verzamelde gezondheidszorggegevens via registers haalbaar is. Dit type gegevens zou

een belangrijke bijdrage kunnen leveren aan de aantallen zwangerschappen die nodig zijn om

zekerheid te geven dat het gebruik van artemisinine derivaten gedurende het eerste trimester van

de zwangerschap veilig is.

In hoofdstuk 6 worden de resultaten beschreven van een studie naar de belevingen, overtuigingen

en het gedrag met betrekking tot aangeboren afwijkingen en miskramen in het westen van Kenia. De

studie maakte gebruik van tien focusgroepdiscussies. Het gebrek aan informatie over de oorzaken

van miskramen en aangeboren afwijkingen kon leiden tot het stigmatiseren van de moeder en kon

verder het zoeken naar hulp en betere zorg belemmeren. Dit kon wrede gevolgen hebben voor

kinderen met aangeboren afwijkingen die vaak verborgen werden gehouden voor de gemeenschap

en geen toegang hadden tot zorg die de kwaliteit van hun leven zou kunnen verbeteren. Sommige

vrouwen die zich bewust waren van de potentiële gevaren van geneesmiddelen gebruik tijdens de

zwangerschap waren zich niet bewust van welke medicijnen veilig waren. Meestal werd een risico

gezien als een medicijn zonder voorschrift werd gebruikt of waarbij de aanbeveling van de clinicus

niet werd gevolgd. Dit benadrukt het belang van educatie materiaal voor zowel de gemeenschap als

zorgverleners met betrekking tot de keuze van medicijnen voor behandeling van aandoeningen

tijdens de zwangerschap. Deze bevindingen benadrukken tevens de noodzaak van voorlichting over

de potentiële oorzaak van slechte zwangerschapsuitkomsten en betere informatie over de

beschikbaarheid van zorg.

In hoofdstuk 7 beschrijven we de resultaten van een studie in west Kenia naar de kennis en naleving

van de nationale behandelingsrichtlijnen voor ongecompliceerde malaria tijdens de zwangerschap.

De studie werd uitgevoerd in klinieken en een willekeurige selectie van winkeltjes (‘Dukas’) die

geneesmiddelen verkochten. Vrouwen in de vruchtbare periode (15 tot 49 jaar) werden

geïnterviewd wanneer ze de kliniek verlieten. Informatie van de winkeltjes werd verzameld door

gebruik te maken van gesimuleerde patiënten die zich voordeden als vrouwen met een vroege

zwangerschap of als familie van vrouwen die in hun derde trimester waren. De studie bracht

onvoldoende kennis en onjuiste voorschrijfpraktijken aan het licht voor de behandeling van malaria

Samenvatting

194

tijdens de zwangerschap, vooral in het eerste trimester en in de informele sector. Bijvoorbeeld,

slechts 44% van de vrouwen in de klinieken en 7% van de simulatie patiënten werd gevraagd of ze

eventueel zwanger zouden kunnen zijn alvorens een geneesmiddel voor te schrijven. Dit was

tegenstrijdig met de bevinding dat 93% en 49% van de personen in de klinieken en winkeltjes

rapporteerden dat zij routinematig vroegen of een vrouw in de vruchtbare periode mogelijk zwanger

zou kunnen zijn. Het voorschrijven van het juiste medicijn, in de correcte dosering voor de

behandeling van malaria tijdens het eerste trimester werd waargenomen bij 32% van alle gevallen in

klinieken en bij 0% in de winkeltjes. Kennis was matig bij het personeel in de kliniek (56%) en afwezig

in de winkeltjes. Voorschrijven van artemether‐lumefantrine (een ACT) in het eerste trimester

gebeurde in 16% en 51% van de patiënten in de klinieken en winkels en dit was niet een gevolg van

een voorraadtekort aan kinine. Dergelijk onjuist voorschrijfgedrag kan ernstige gevolgen hebben

voor de zwangere patiënt, niet alleen omdat medicijnen werden gebruikt met een onduidelijk

veiligheidsprofiel (artemisinine derivaten), maar ook omdat ineffectieve medicijnen werden

voorgeschreven die niet geschikt zijn voor de behandeling van malaria of effectieve medicijnen,

maar in een te lage of te korte dosis (zoals bij kinine).

In hoofdstuk 8 worden de bevindingen van een prospectieve studie in het westen van Kenia

gepresenteerd om de associatie te bestuderen tussen miskramen en artemisinine derivaten gebruik

tijdens de vroege zwangerschap. De studie werd verricht in een gebied met een goed ontwikkeld

demografische surveillance informatiesysteem waar de gezondheidsstatus van de bevolking werd

gevolgd door wekelijkse (of twee keer per maand) huisbezoeken te doen en door surveillance in de

kliniek. Surveillance in de gemeenschap werd gebruikt om vroege zwangerschappen bij vrouwen in

de vruchtbare leeftijd te identificeren. Meerdere informatie bronnen werden gecombineerd om

blootstelling aan artemisinine derivaten tijdens de zwangerschap te identificeren en te verifiëren.

Deze studie liet zien dat vaststelling van zwangerschap in het eerste trimester (voordat de meeste

vrouwen een verloskundige bezoeken) met het gebruik van zwangerschapstesten in de

gemeenschap haalbaar was. Hieruit kon het percentage miskramen per zwangerschapsweek bepaald

worden. Het totale risico op een miskraam werd ingeschat op 19%. Van de 1,126 zwangerschappen

waren er 42 (4%) met bevestigde blootstelling aan artemisinine derivaten tijdens de embryo‐

gevoelige periode (6‐12 weken na de laatste menstruatie) en 75 (7%) tijdens het eerste trimester (0‐

13 weken). Zwangerschappen met malaria en blootstelling aan artemisinine derivaten in de embryo‐

gevoelige periode hadden geen statistisch significant verschillend risico op miskramen vergeleken

met zwangerschappen zonder blootstelling (risico ratio 0.85, 95% betrouwbaarheidsinterval 0.22‐

3.33). Een verhoogd risico door blootstelling in het eerste trimester kon niet worden uitgesloten

(risico ratio 1.60, 95% betrouwbaarheidsinterval 0.70‐3.68), maar de bovengrens van het 95%

betrouwbaarheidsinterval suggereert dat een meer dan 3,7‐voudige verhoogd risico uiterst

onwaarschijnlijk is. In lijn met hoofdstuk 7 werd gevonden dat kinine, het aanbevolen middel voor

behandeling tijdens het eerste trimester, weinig werd gebruikt (n=34 met 22 blootgesteld aan zowel

kinine als artemisinine derivaten). Dit belemmerde ons vermogen om het eventuele effect van

artemisinine derivaten in het eerste trimester te vergelijken met kinine. De resultaten waren echter

in overeenstemming met recente bevindingen van studies aan de Thai‐Birmese grens die ook geen

verhoogd risico vonden.[3] Binnenkort zullen tevens de bevindingen beschikbaar zijn van een

gemeenschappelijke data‐analyse waarbij de informatie uit deze en twee andere studies worden

gecombineerd. Dit zal verder inzicht geven in de risico’s en voordelen van het gebruik van

artemisinine derivaten in het eerste trimester.

195

Concluderend kan worden gesteld dat het onderzoek beschreven in dit proefschrift de noodzaak

benadrukt voor betere pharmacovigilantie van antimalariamiddelen tijdens de zwangerschap in

gebieden waar beperkte middelen voorhanden zijn zoals in Afrika. Geen enkele methode alleen is

toereikend om de gewenste informatie vast te leggen die nodig is voor een passende analyse van het

risicoprofiel van antimalariamiddelen. Een combinatie van verschillende methodes en bronnen van

informatie is nodig om een compleet beeld te krijgen voor het opsporen, evalueren en voorkomen

van eventuele ongewenste bijwerkingen van deze geneesmiddelen tijdens de zwangerschap. Een

geschikte studielocatie is essentieel waar betrouwbare gegevens zouden kunnen worden verkregen

over zowel de blootstelling aan geneesmiddelen als de uitkomst van de zwangerschap. Mogelijke

opties zijn gebieden waar gezondheidsstatus en demografische gegevens routinematig worden

bijgehouden zoals onderzoek locaties met demografische surveillancesystemen, agrarische of

industriële complexen, vluchtelingen kampen of via ziektekostenverzekering bedrijven die werkzaam

zijn in gebieden met malaria. Wij vonden een onverwacht hoog aantal zwangerschappen die

onbedoeld of opzettelijk aan artemisinine derivaten waren blootgesteld tijdens de vroege

zwangerschap in west Kenia. Conform het werk van andere groepen hebben deze en andere studies,

tot nu toe geen ernstige teratogene afwijkingen aan het licht gebracht. De aantallen zijn echter nog

beperkt, en in totaal zijn minder dan 1000 goed gedocumenteerde blootstellingen in het eerste

trimester beschreven wereldwijd. Dit geeft een zekere mate van vertrouwen in de veiligheid van de

artemisinine derivaten voor gebruik tijdens de vroege zwangerschap, maar de regel‐van‐drie

suggereert dat dit nog steeds een minder dan tweevoudige toename in the incidentie van

aangeboren afwijkingen kan inhouden. Meer onderzoek is dus nodig, ook om het risico op

cardiovasculaire afwijkingen in associatie met blootstelling aan artemisinine derivaten in de vroege

zwangerschap vast te stellen en om te kijken naar aangeboren afwijkingen die pas later in het leven

duidelijk zijn, zoals een ontwikkelingsachterstand. Ons onderzoek laat ook zien dat de aanbieders

van zorg en medicijnen in het westen van Kenia beperkte kennis hebben met betrekking tot het

veilig gebruik van antimalariamiddelen door vrouwen in de vruchtbare leeftijd. Met het tot nu toe

opgebouwde bewijs en de slechte naleving van de huidige aanbevolen behandeling met kinine in het

eerste trimester van de zwangerschap zou een gerandomiseerd klinisch onderzoek de snelste weg

zijn om het bewijs te verkrijgen dat noodzakelijk is. Deze studie was onderdeel van een multi‐landen

initiatief en het is belangrijk dat alle nieuwe gegevens nu zo snel mogelijk gedeeld kunnen worden

met de Wereldgezondheidsorganisatie voor het bijwerken van het risico‐baten profile van de

artemisinine derivaten tijdens de zwangerschap. Totdat er voldoende gegevens beschikbaar zijn om

een volledige kosten‐batenanalyse te maken van het gebruik van de artemisinine derivaten tijdens

de vroege zwangerschap, zal het belangrijk zijn om eerste trimester blootstellingen te minimaliseren.

Daarvoor is verder onderzoek nodig om de beste methode te bepalen ter verbetering van het

uitvoeren van de nationale malaria behandelingsrichtlijnen, vooral onder de

geneesmiddelenverkopers in de informele sector. De bevindingen in dit proefschrift zijn tevens van

toepassing voor de behandeling van andere tropische ziekten. De situatie met artemisinine

derivaten is niet uniek en het merendeel van de nieuwe medicijnen voor tropische ziekten die op de

markt komen worden niet aanbevolen voor gebruik tijdens de zwangerschap. Vaak is dit niet

gebaseerd op gegevens over onveiligheid, maar eerder over het gebrek aan informatie over

veiligheid in de zwangerschap, met als gevolg dat zwangere vrouwen vaak jarenlang moeten

toedoen met inferieure geneesmiddelen zoals kinine. Gezamenlijke inspanningen zijn nodig om deze

situatie aan te pakken zodat veilige en effectieve behandeling van tropische ziektes gedurende de

zwangerschap mogelijk is.

197

AcknowledgementsThe papers presented in this thesis are the result of joint efforts involving many people across

multiple institutions without whom the work could not have been successfully completed. I would

like to thank all study participants for their time and cooperation and all the staff involved in the

studies presented in this thesis.

Foremost, I would like to express my sincere gratitude to Feiko ter Kuile who has been an amazing

supervisor and played a pivotal role in the work that went into this thesis. I have learned a great deal

through his rigour, high scientific standards and integrity. He was always there if I needed him and all

his advices have been invaluable. I thank him for always driving me to go the extra mile, for making

time when he had very little, for considering “happiness levels” every step of the way and for making

work fun. To work with him over the past 7 years has been a real pleasure and a privilege.

I am grateful to Michael Boele van Hensbroek for accepting to be my second promotor and for his

help and guidance in the last few months of the production of this thesis.

I am most grateful to the Malaria in Pregnancy Consortium (MiPc) secretariat at the Liverpool School

of Tropical Medicine (LSTM), particularly Alison Reynolds for her friendship, great logistical support

and for proof reading part of this thesis. My deepest thanks to Jenny Hill and Jayne Webster for

creating the opportunity for me to work with them on the interesting MiPc implementation research

studies in Kenya and for the good times in Kweisos. Jenny also for the inspiring discussions while we

shared an office in Liverpool (which played an important role in my move to Kenya) and for keeping

me on track to meet PhD deadlines. Andy Stergachis and the MiPc ASAP team (Esperanca Sevene,

Tinto Halidou, Umberto d’Alessandro, Laura Sangare, Becky Bartlein, Gillian Levine and Greg Calip)

made important contributions to the work presented in chapter 8, I have greatly enjoyed our

collaboration in this stimulating work. My sincere thanks also to Andy for his support throughout the

MiPc years and the valuable input to study protocols and co‐authored papers. Special thanks to Rob

Nathan for his assistance with ultrasound quality control and for reviewing the numerous scans,

Stephen Rulisa and Mark Londema for training study nurses to do obstetric ultrasounds. At LSTM: I

thank Cheryl Pace and James Smedley for taking over the MiPc Central Safety Database and doing an

excellent job, and Helen Wong for her efficient administrative support. I am grateful to Bernard

Brabin for his encouragement and for the captivating malaria discussions.

I would like to thank the Malaria Branch, Centers of Disease Control and Prevention, Atlanta, for

financial and technical support for the studies in Kenya. Particularly, I am extremely grateful to

Meghna Desai, my co‐supervisor, who gave tremendous helpful advice on the fieldwork in Kenya

and for giving me the opportunity to be involved and learn from various, very interesting, projects

through the KEMRI/CDC collaboration. Her sense of humour and composure when things didn’t go

to plan were highly appreciated. I would like to thank Mary Hamel, without whom the EMEP study

(chapter 8) might not have seen the light of day. Her interest in this project, her hospitality during

my first visits in Kenya and her input to protocol development were invaluable. I thank Danny Feikin

and Deron Burton for making the collaboration with the IEIP program not only possible but

effortless. I also thank Meredith McMorrow for training the study nurses on Ballard score. My

sincere thanks go to Kayla Laserson, the KEMI/CDC director at the time of EMEP study (chapters 6 &

8), for her support throughout my time in Kisumu. Her enthusiasm, energy and dedication have been

198

truly inspiring. I am very grateful to Larry Slutsker for his support and constructive feedback on study

proposals, protocols and manuscripts.

I will be eternally grateful to Professors Brian Greenwood and Daniel Chandramohan at the London

School of Hygiene and Tropical Medicine and Jamie Robinson and Susan Hall at GlaxoSmithKline for

enabling a fellowship year whilst I was very new to malaria research and grant proposals.

To all my colleagues at KEMRI/CDC, Kisumu, Kenya: Peter Ouma, Abraham Katana, Florence Achieng,

Roseline Odera, George Aol, Godfrey Bigogo, Allan Audi, Frank Odhiambo and others from the

malaria and IEIP branches which are too many to cite ‐ thanks for making me feel welcome from the

day I arrived in Kisumu and for guiding me through local systems and situations. Special thanks go to

Dr Simon Kariuki, head of malaria branch, and Dr Vulule, director of the KEMRI research station in

Kisumu at the time, for their support throughout my time in Kisumu.

The work presented in chapters 6 & 8 was possible because of the dedication and hard work of the

EMEP study staff. I feel truly lucky to have worked with such a remarkable team, and I am

particularly grateful to Emily Ayanga, Everlyne Oteyo, Jane Oiro, Theresa Aluoch, Elizabeth Aballa,

Faith Samo, Eric Onyango and Joshua Auko as well as the 40 village reporters from Asembo and to

Tina Oneko for her careful review of newborns with suspected congenital anomalies‐ each team

member was invaluable. My deepest gratitude to George Aol for his guidance and wisdom

throughout the study which were instrumental to understand what would be culturally acceptable

and feasible. This study could not have happened without the support and collaboration of the IEIP,

MNHCU and STOPMIP teams‐ thank you all.

Friends and colleagues who have contributed positively to my life in Kisumu: Nelli and Matt

Westercamp, Alkesh Patel, Maria Oziemkowska, Tina Oneko, Penny Phillips‐Howard, Gemma Aellah,

Kurt and Amy Herman‐Roloff, Martin Mwangi, Ellen van Puffelen, Bobby Ochieng, Christina Polyak,

Titus Kwambai, Kat Tumlinson, Marieke Sassen, Erwan Piriou, Anja van’t Hoog, Josephine Walker,

Kelly Alexander, Jo Halliday, Daryn Knobbel and Nat Robinson‐ thank you for all the good memories.

My heartfelt thanks go to Professor Philippe Brasseur and Soeur Marie Joelle for their generous

support and for hosting us during our trip to Dakar and Mlomp, in Senegal (chapter 5).

Julie Gutman, Christina Riley and Ann Buff were critical to the successful completion of the cross‐

sectional survey presented in chapter 7. This has been a very conducive collaboration and I thank

them for putting up with sudden deadlines over the last few months.

Huge thanks to Annemieke van Eijk for doing the Dutch translation of the summary and to Olaya and

Eloisa Astudillo for designing the cover page of this thesis.

Last but not least, I would like to thank my family for their love and encouragement. My parents

Oswald and Dominique, who have supported me enthusiastically in all my undertakings and gave me

the thirst of travel. My sisters Savina and Ondine, for always being there for me. My daughter Lea,

for putting things in perspective and never failing to put a smile on my face. And especially to my

considerate and patient partner Per who has been an amazing support through all stages of this

PhD. I thank him for not only putting up with my moves and travels but also for joining me for part of

the adventures. Words cannot express how grateful I am for his help and encouragement,

particularly in the last few months, which carried me through to the end.

199

AbouttheauthorStephanie Dellicour was born in Belgium on December 4, 1980. In 1998, she finished her secondary

education at the European School in Brussels where she obtained a European Baccalaureate. She

subsequently moved to South Africa where she completed a BSc in Biochemistry and Genetics in

2001 and a Bsc Honours in Pharmacology in 2002 from the University of Cape Town. She obtained an

MSc in Epidemiology from the London School of Hygiene and Tropical Medicine (LSHTM) in 2004.

Her Msc dissertation was in pharmacoepidemiology and involved using the UK General Practice

Research Database (GPRD) to assess drug‐induced autoimmune haemolytic anaemia. She worked for

GlaxoSmithKline (GSK) from 2004 to 2005 in their pharmacoepidemiology department where she

first became interested in the issues surrounding the safety concerns for antimalarial drugs used in

pregnancy. At the time GSK was developing a risk management plan for potential exposures in early

pregnancy to CDA (chlorproguanil‐dapsone‐artesunate combination) which was in the pipeline. In

2005, she received a fellowship through GSK to conduct research at LSHTM with Professors

Greenwood and Chandramohan to investigate options for setting up studies assessing the safety of

antimalarials in the first trimester of pregnancy. In 2007, she joined the Liverpool School of Tropical

Medicine to work with the Malaria in Pregnancy consortium (MiPc) under Professor ter Kuile. There

she worked in the Safety Working Group coordinating pharmacovigilance cross‐cutting activities

within the MiPc and developing grant proposals for the work presented in this thesis. In 2009 she

moved to Kenya to coordinate studies for the MiPc at the Kenya Medical Research Institute (KEMRI)/

Centers for Disease Control and Prevention (CDC) Research and Public Health Collaboration. There

she coordinated implementation research studies focused on interventions for the control of malaria

in pregnancy. Concurrently she developed, implemented and coordinated the research project on

the pharmacovigilance system for monitoring the safety of antimalarial used in early pregnancy

throughout the 3.5 years of fieldwork. She moved back to the UK in 2013 to write up the work

coordinated in Kenya.

Additionalpublicationsbytheauthor:1. Dellicour S, Greenwood B (2007) Systematic review: Impact of meningococcal vaccination on pharyngeal carriage of meningococci. Trop Med Int Health 12: 1409‐1421.

2. Brabin BJ, Warsame M, Uddenfeldt‐Wort U, Dellicour S, Hill J, et al. (2008) Monitoring and evaluation of malaria in pregnancy ‐ developing a rational basis for control. Malar J 7 Suppl 1: S6.

3. Desai M, Dellicour S (2012) Effects of malaria and its treatment in early pregnancy. Lancet Infect Dis 12: 359‐360.

4. Hill J, Dellicour S, Bruce J, Ouma P, Smedley J, et al. (2013) Effectiveness of antenatal clinics to deliver intermittent preventive treatment and insecticide treated nets for the control of malaria in pregnancy in Kenya. PLoS One 8: e64913.

5. Kwambai TK, Dellicour S, Desai M, Ameh CA, Person B, et al. (2013) Perspectives of men on antenatal and delivery care service utilisation in rural western Kenya: a qualitative study. BMC Pregnancy Childbirth 13: 134.

UvA · ASSESSMENT OF PHARMACOVIGILANCE APPROACHES FOR MONITORING THE SAFETY OF ANTIMALARIAL DRUGS...

Documents

Transcript of UvA · ASSESSMENT OF PHARMACOVIGILANCE APPROACHES FOR MONITORING THE SAFETY OF ANTIMALARIAL DRUGS...