MSc Project Report 2014-2015library.lshtm.ac.uk/MSc_RSHR/2014-15/108664.pdf · 2016. 5. 19. · MSc...
Transcript of MSc Project Report 2014-2015library.lshtm.ac.uk/MSc_RSHR/2014-15/108664.pdf · 2016. 5. 19. · MSc...
-
Candidate no: 108664
The prevalence and risk factors of
puerperal sepsis in South Asia: a
systematic review
Supervisor: Laura Oakley
Candidate number: 108664
Word count: 9592
Project length: Standard
Submitted in part fulfilment of the requirements for the degree of MSc in
Reproductive and Sexual Health Research
September 2015
MSc Project Report
2014-2015
-
Candidate no: 108664
-
Candidate no: 108664
1
CONTENTS
ABSTRACT
ACKNOWLEDGEMENTS
1. INTRODUCTION
2. AIMS & OBJECTIVES
3. METHODS
3.1 Criteria for selecting studies
3.2 Search strategy
3.3 Data collection
3.4 Quality assessment
3.4.1 Quality of case diagnosis
3.4.2 Methodological quality
3.5 Data synthesis and analysis
4. RESULTS
4.1 Search results
4.2 Characteristics of included studies
4.3 Quality of case diagnosis of puerperal sepsis
4.4 Methodological quality of included studies
4.5 The prevalence of puerperal sepsis
4.5.1 Overall prevalence of puerperal sepsis
4.5.2 Prevalence of puerperal sepsis by country
4.5.3 Prevalence of puerperal sepsis by setting
4.5.4 Prevalence of puerperal sepsis by quality of case diagnosis
4.6 Meta-analysis of the prevalence of puerperal sepsis reported by high-quality
studies.
-
Candidate no: 108664
2
4.7 Risk factors for puerperal sepsis
5. DISCUSSION
5.1 The overall prevalence of puerperal sepsis in South Asia
5.2 The prevalence of puerperal sepsis by country
5.3 The prevalence of puerperal sepsis by setting
5.4 The prevalence of puerperal sepsis by the quality of case diagnosis
5.5 Risk factors for puerperal sepsis
5.6 Puerperal sepsis-related mortality
5.7 Strengths of systematic review
5.8 Limitations of systematic review
5.9 Implications & recommendations for future research
6. CONCLUSION
REFERENCES
APPENDIX 1. Search strategies
APPENDIX 2: Joanna Briggs Institute critical appraisal checklist used for assessing
methodological quality of included studies
APPENDIX 3. Data extraction form
APPENDIX 4: Included studies reporting additional types of puerperal infections that were
not used in analysis
-
Candidate no: 108664
3
LIST OF FIGURES
Figure 1. Identification and selection process of articles for inclusion in systematic review.
Figure 2. Prevalence of puerperal sepsis reported by all included studies.
Figure 3. Prevalence of puerperal sepsis by country.
Figure 4. Prevalence of puerperal sepsis by setting.
Figure 5. Prevalence of puerperal sepsis by quality of case diagnosis.
Figure 6. Meta-analysis of the prevalence of puerperal sepsis reported by high-quality
studies.
LIST OF TABLES
Table 1. International definitions of puerperal sepsis.
Table 2. Characteristics of included studies reporting puerperal sepsis prevalence.
Table 3. Critical appraisal results for included studies using the JBI Prevalence Critical
Appraisal Checklist.
Table 4. An overview of the characteristics of included studies. Table 5. Median prevalence and range of all included studies, overall and by strata.
Table 6. Identified risk factors for puerperal sepsis.
-
Candidate no: 108664
4
ABSTRACT
Background: Puerperal sepsis is a serious, yet highly preventable, cause of maternal
mortality and morbidity in South Asia. No estimates for the prevalence of this condition exist
for this region. The primary objective of this systematic review was to determine the
prevalence of puerperal sepsis in South Asia. A secondary objective was to review risk
factors leading to puerperal sepsis in this region.
Methods: Comprehensive electronic searches of five databases were conducted. Studies
reporting prevalence data for puerperal sepsis in South Asian countries were eligible for
inclusion. Prevalence data was stratified by country, by setting (facility-based or population-
based) and by quality of case diagnosis of puerperal sepsis. The methodological quality of
included studies was assessed and meta-analysis was performed on a sub-set of high
quality studies.
Results: 23 studies were included in this review. The median prevalence of puerperal sepsis
of all included studies was 3.84% of total deliveries (range 0% to 34.25%). The pooled
prevalence of a sub-set of seven studies of high methodological quality was 3.20% of total
deliveries (95% CI 2.03%, 5.02%). Clear trends in prevalence by country, setting and quality
of diagnosis were difficult to discern. Identified risk factors for puerperal sepsis maternal
factors such as pre-existing conditions, labour factors including peripartum bleeding,
stillbirth, perineal tears, prolonged delivery and unhygienic practices, health service factors
such as access to antenatal and obstetric care, and community factors such as lower
socioeconomic status and unhygienic postpartum traditions.
Conclusion: This systematic review is the first to attempt to estimate the prevalence of
puerperal sepsis in South Asia. The findings from this review indicate a prevalence between
3-4% of total deliveries. Although this estimate is relatively low, puerperal sepsis remains of
public health importance as it is a significant cause of obstetric morbidity and mortality in
South Asia, despite its highly preventable nature. Risk factors for puerperal sepsis are varied
and include maternal, labour, health service and community factors, which should be the
focus for prevention efforts.
-
Candidate no: 108664
5
ACKNOWLEDGEMENTS
I would like to thank my personal tutor and project supervisor, Laura Oakley, for her
invaluable guidance and feedback throughout every stage of completing this dissertation.
Many thanks to Veronique Filippi who helped me to form the focus of my dissertation in the
earlier stages, and to Clara Clavert for her advice on data analysis. I would like to also thank
the MARCH Centre staff for giving me the opportunity to present my dissertation and for their
useful feedback and suggestions.
As a personal acknowledgement, I would like to thank my parents for their continued support
and encouragement in pursuing my medical career and Masters. I would also like to express
my appreciation to my friends for their endless moral support and for always listening.
-
Candidate no: 108664
6
1. INTRODUCTION
Puerperal sepsis is a preventable postpartum complication that occurs when a recently
delivered mother contracts an infection related to giving birth. It can rapidly progress along a
spectrum of severity if untreated, leading to septicaemic shock and eventually death. In non-
fatal cases, it can cause significant long-term morbidity such as chronic pelvic inflammatory
disease and secondary infertility from tubal occlusion [1]. These serious consequences can
be easily avoided with better infection control, prompt diagnosis and appropriate treatment
[1]. Despite global progress in reducing maternal mortality, puerperal sepsis is a leading
cause of preventable maternal deaths, accounting for up to 11% of maternal deaths
worldwide [2]. The burden of maternal mortality is highest in the developing world, with
southern Asia and sub-Saharan Africa together accounting for 83.8% of all maternal deaths
[2]. South Asia has the highest proportion of maternal sepsis deaths globally (13.7%) [2], yet
little current information exists regarding the epidemiology of puerperal sepsis in this region.
Few reviews have attempted to produce estimates of the prevalence or incidence of
puerperal sepsis. AbouZahr (2003) estimated the global incidence of puerperal sepsis to be
4.4% of live births [3]. Dolea and Stein (2003) provide an incidence range of 0.07 to 9.3% of
live births from data collected in developing countries between 1986 and 1998, although no
studies from South Asian countries were identified or used in the review [4]. Seale et al.
(2009) investigate the burden of puerperal sepsis in sub-Saharan Africa, with prevalence
estimates reported by studies ranging from 0.1% to 17.9% [5]. Van Dillen et al. (2010)
discuss the global epidemiology of puerperal sepsis, although specific regional figures of
incidence were limited to high-income countries only [6]. Hussein and Walker (2010) suggest
that the global incidence of puerperal sepsis in developing countries ranges from 0.1 to 10%
of deliveries [7]. All of these reviews highlighted the difficulties in obtaining incidence
estimates for puerperal sepsis, attributed to marked differences in study design, the
diagnostic criteria used, and the setting, such as local conditions and whether the study is
population-based or facility-based.
Risk factors leading to puerperal sepsis are varied, including anaemia and other pre-existing
maternal conditions, obstetric issues such as obstructed labour, health service factors such
as poor hygiene and aseptic technique, and community factors such as socioeconomic
status [1]. A better understanding of the determinants of puerperal sepsis is crucial in
identifying factors that facilitate infection and in forming strategies for prevention.
Various definitions for puerperal sepsis exist, although none are used universally, which has
-
Candidate no: 108664
7
contributed to the difficulty in estimating its prevalence [7]. In addition, many articles refer to
the terms ‘puerperal sepsis’, ‘maternal sepsis’, ‘puerperal infection’, ‘puerperal pyrexia’ and
‘puerperal fever’ synonymously or without clarification of their definition [7]. Table 1 lists the
major international definitions of puerperal sepsis outlined in key documents by the World
Health Organization and the ICD-10 classification system. These definitions describe
puerperal sepsis as a clinical syndrome (a group of signs and symptoms) that may have
various primary aetiologies ranging from genito-urinary tract infections to obstetric wound
sepsis. The ICD-10 definition further incorporates extra-genitourinary complications
occurring in the puerperium under codes 087 – 092 [9], as does the WHO (1995) document
under the more general term ‘puerperal infection’ [8]. However, this review only addresses
puerperal sepsis associated with genito-urinary tract infections and obstetric wound
infections due to their direct association with the process of childbirth. The length of the
puerperium (postpartum period) is often cited as 42 days after delivery [8], although this may
vary between studies. The start of the puerperium is also disputed; the WHO (1995)
document [8] states that puerperal sepsis can occur from the onset of rupture of membranes
or labour, whereas the ICD-10 definition excludes sepsis during labour [9].
Identifying the prevalence of puerperal sepsis in South Asia is important in order to assess
the magnitude of this potentially life-threatening, yet preventable, condition in a region where
the burden of maternal mortality is already high. The aim of this systematic review is to
estimate the prevalence and review risk factors for puerperal sepsis in South Asia, using the
UN Millennium Development Goals region country groupings [10], taking into account the
study location, setting and the quality of case diagnosis of puerperal sepsis.
-
Candidate no: 108664
8
Table 1. International definitions of puerperal sepsis.
Document Definition of puerperal sepsis
The Prevention and Management of Puerperal Infections: Report of a Technical Working Group, Geneva, 20-22 May 1992 – World Health Organization (1995) [8]
Puerperal sepsis is defined as infection of the genital tract occurring at any time between the onset of the rupture of the membranes or labour and the 42nd day postpartum in which fever and one or more of the following are present: ▪ Pelvic pain ▪ Abnormal vaginal discharge e.g. presence of pus ▪ Abnormal smell/ foul odour of discharge ▪ Delay in the rate of reduction of size of uterus (
-
Candidate no: 108664
9
2. AIMS & OBJECTIVES
Overall aim
To perform a systematic review of the relevant literature, published from 2000
onwards, to estimate the prevalence of puerperal sepsis in South Asia, as well as
review risk factors leading to puerperal sepsis in this region.
Objectives
▪ Primary objective: to estimate the prevalence of puerperal sepsis in South Asia:
- by country;
- by setting (population-based vs. facility-based);
- by quality of case diagnosis of puerperal sepsis.
▪ Secondary objective: to review risk factors associated with puerperal sepsis in
South Asia.
-
Candidate no: 108664
10
3. METHODS
3.1 Criteria for selecting studies
Inclusion criteria
▪ Articles published from 2000 onwards were included.
▪ Grey literature or studies of any design reporting data on the prevalence of puerperal
sepsis or associated puerperal infections in South Asian countries were included. For the
purpose of this review, the country grouping of the Southern Asian region used by the UN
MDG initiative was used, comprising the following countries: Afghanistan, Bangladesh,
Bhutan, India, Iran, Maldives, Nepal, Pakistan and Sri Lanka [10].
▪ Only articles published in English were eligible for inclusion.
Exclusion criteria
▪ Articles reporting data on sepsis or fever within the puerperium associated with extra-
urogenital infections and incidental causes unrelated to childbirth were not included.
▪ As puerperal sepsis is a clinical syndrome, prevalence data of singular non-specific
symptoms such as ‘discharge’ or ‘pain’ were not included. However, an exception to this
criterion were reports of puerperal ‘fever’, which is identified by ICD-10 as a defining term of
puerperal sepsis under code 085 [9], and is acknowledged as the key defining symptom of
puerperal sepsis in the WHO (2008) definition [1] (see Table 1).
▪ Studies that identified cases of puerperal sepsis from study samples unrepresentative of all
deliveries within the study’s reported setting (facility or population-based) were excluded.
This included deliveries exclusive to women undergoing caesarean section, post-abortion
cases, grand multiparous women, women with obstetric-related acute renal failure, women
requiring emergency peripartum hysterectomy, and women with pre-term premature rupture
of membranes. Furthermore, deliveries to women admitted to intensive care units and ‘near
miss’ sepsis cases were excluded, as they would not include cases of less severe puerperal
infection.
▪ Studies reporting data of mortality from puerperal sepsis without prevalence data were
excluded.
3.2 Search strategy
Comprehensive electronic searches of the following databases were conducted: MEDLINE,
Embase, Global Health, Popline and WHO Reproductive Health Library. The main concepts
-
Candidate no: 108664
11
of ‘puerperal sepsis’ and ‘epidemiology’ were searched in MEDLINE, Embase and Global
Health databases using free text and MeSH headings. Region and country filters for South
Asia were applied in order to focus search results. For Popline and WHO Reproductive
Health Library databases, only the main concept of ‘puerperal sepsis’ was searched, with
manual review of study location and relevant epidemiological data. All searches were
conducted on 7th and 8th July 2015, with search results restricted to articles published from
2000 onwards. No language restrictions were applied, but only papers available in English
were included. Detailed search strategies for each database are provided in Appendix 1.
Manual review of reference lists and citation tracking of salient articles were also conducted.
-
Candidate no: 108664
12
Figure 1. Identification and selection process of articles for inclusion in systematic review.
3.3 Data collection
Data was extracted from included articles using a standardised data extraction form
comprising: the full title of study, author(s), year of publication, journal of publication, location
of study, study design, study dates (follow-up period or study duration), study setting, study
population, the case definition of puerperal sepsis/associated puerperal infections, the
number of puerperal sepsis cases, the sample size (denominator), prevalence data, data on
risk factors, and the method of case ascertainment. A template data extraction form can be
found in Appendix 3.
Databases searched: ▪ MEDLINE n =1,482 ▪ Embase n = 3,402 ▪ Global Health n = 1,831 ▪ WHO Reproductive Health Library n = 123 ▪ Popline n = 94 Total n = 6,932
Duplicates removed: n = 1,526
n = 5,406
Excluded at title & abstract screen:
n = 4,911
n = 495 Articles excluded at full text review: ▪ No relevant data n = 196 ▪ General: Full text unobtainable n = 74 Not written in English n = 2 Studies published before 2000 n = 3 ▪ Population: Data not from S. Asia/not possible to extract S. Asia-specific data n = 18 Inappropriate denominator n = 87 ▪ Outcome: Puerperal sepsis mortality data only n = 59 Sepsis not childbirth-related n = 25 Non-specific symptoms n = 10 ▪ Other: Secondary publication n = 1
n = 20
Manual citation tracking:
n = 3
Included articles: n = 23
-
Candidate no: 108664
13
Due to the nature of delivery outcomes, the terms ‘prevalence’ and ‘incidence’ have been
used with interchangeable definitions in studies reporting data on puerperal sepsis, with both
terms referring to the number of puerperal sepsis cases (n) in a specified population (N) over
a specified time period. Throughout this review, the term ‘prevalence’ has been used. ‘N’ is
represented by the total number of deliveries from which ‘n’ puerperal sepsis cases have
arisen, where one delivery may consist of one or more live-born or stillborn births [12]. In
addition, this review uses the term ‘delivery’ to also represent any termination of pregnancy,
including spontaneous or induced abortions.
If a study did not readily provide prevalence data, it was calculated by dividing the total
number of puerperal sepsis cases (n) by a suitable figure for the total number of deliveries
(N). Where studies reported stratified prevalence data (for example, cases of puerperal
sepsis by facility or by maternal body mass index), the total number of puerperal cases and
the total number of deliveries in each stratum were totalled and an overall prevalence
measure was calculated. Several studies reported prevalence data for more than one type of
puerperal infection. It there was no overlap between the different types, an overall
prevalence measure was calculated. If overlap existed between the prevalences of different
types of puerperal infection, the most generic type, such as ‘puerperal fever’ or ‘puerperal
sepsis’, was selected for analysis over more specific conditions such as ‘urinary tract
infection’ or ‘wound infection’. Details of other types of puerperal infection reported by
included studies that were not used in analysis can be found in Appendix 4.
If supplied, data on risk factors for puerperal sepsis demonstrating statistical significance
were also extracted from included studies.
3.4 Quality assessment
3.4.1 Quality of diagnosis
The quality of each study’s diagnosis of puerperal sepsis cases was assessed in order to
compare the prevalence of puerperal sepsis by high-, medium- and low-quality sub-groups.
The following classification was used:
▪ Low quality diagnosis:
- The study does not provide a case definition of puerperal sepsis and uses self-
reported symptoms to identify cases.
▪ Medium quality diagnosis:
- The study provides an acceptable case definition of puerperal sepsis, but uses self-
reported symptoms to identify cases;
-
Candidate no: 108664
14
- or the study does not provide a case definition of puerperal sepsis, but uses a
clinical source (including corroboration of self-reported symptoms by a physician/
trained staff/ medical records) to identify cases.
▪ High quality diagnosis:
- The study provides an acceptable case definition of puerperal sepsis, and uses a
clinical source (including corroboration of self-reported symptoms by a physician/
trained staff/ medical records) to identify cases.
3.4.2 Methodological quality
The methodological quality of each of the included studies was assessed using the Joanna
Briggs Institution (JBI) critical appraisal checklist for studies reporting prevalence data [11]
(see Appendix 2). A sub-set of high-quality studies could then be identified for meta-analysis.
For the purpose of this review, the JBI criteria deemed to be the most indicative of a high-
quality study were Question 1 (representative sample of the target population), Question 2
(appropriate recruitment of study participants), Question 6 (the use of objective, standard
criteria for measurement of the condition), and Question 7 (reliable measurement of the
condition). However, Question 1 was difficult to assess in included studies due to a lack of
reporting of whether or not abortions and stillbirths were included in the sample as a delivery
outcome, so was not used in this selection process. Studies were therefore classified as
high-quality if they demonstrated good overall quality by scoring a ‘yes’ across a minimum of
6 out of 10 questions, of which a ‘yes’ was specifically scored in Question 2, Question 6 and
Question 7 (see Table 3).
3.5 Data synthesis and analysis
Forest plots were used to present the prevalence of puerperal sepsis, expressed as a
percentage, with 95% confidence intervals calculated using the binomial exact method.
Prevalence data was stratified by country, by setting (facility-based or population-based) and
by quality of case diagnosis of puerperal sepsis (low, medium or high). Meta-analysis was
conducted on a sub-set of studies deemed to be of high methodological quality (see
methods section 3.4.2), in order to provide a pooled estimate of the prevalence of puerperal
sepsis in South Asia. The meta-analysis was conducted using the ‘metaprop’ command from
RStudio 3.2.0 with the package ‘meta’ using a random-effects model. Risk factors identified
as being statistically significant were extracted from included studies and tabulated.
-
Candidate no: 108664
15
4. RESULTS
4.1 Search results
Initial searches of the five databases produced a total of 6,932 results. After removing 1,526
duplicates, the remaining 5,406 articles underwent title and abstract screening. A further
4,911 articles were excluded and the full texts of the remaining 495 articles were assessed
for eligibility. 20 suitable articles were identified for inclusion, with a further three articles
identified by manual citation tracking. This produced a total of 23 studies included in this
systematic review. See Figure 1 for a flow diagram detailing the identification and selection
process of articles for inclusion in this review.
4.2 Characteristics of included studies
Geographical location
Of the 23 included studies, 11 studies were conducted in India [14, 19, 21, 23, 24, 26, 27,
29, 30, 32, 33], five in Bangladesh [13, 16, 25, 28, 35], five in Pakistan [15, 17, 18, 22, 31]
and two in Nepal [20, 34]. No studies from other South Asian countries met the inclusion
criteria for this review.
Year of publication & period of data collection
The included studies were published between 2000 and 2014, with data collection taking
place between 1992 and 2012. Study periods ranged from one month to 8 years, with a
median study period of 18 months.
Study design
Nine of the studies were cross-sectional in design, of which seven were prospective [13, 15,
17, 22, 25, 26, 29] and two were retrospective [21, 24]. 10 were prospective cohort studies
[14, 16, 19, 23, 27, 31-35], two were nested case-control studies within prospective cohorts
[18, 28], and two were randomised controlled trials [20, 30].
Study setting & participants
11 studies were facility-based, of which one study used data from primary health centres
[21], eight from tertiary facilities [18, 22-24, 28, 30-32], and two from a mix of secondary and
tertiary facilities [25, 33]. The remaining 12 studies were population-based [13-17, 19, 20,
26, 27, 29, 34, 35]. All of the studies used participants consisting of recently delivered
women or pregnant women recruited for follow-up after delivery (see Table 2 for further
details).
-
Candidate no: 108664
16
Case definition of puerperal sepsis & method of ascertainment
12 studies provided a case definition of puerperal sepsis or an associated puerperal infection
[13, 14, 16, 17, 20, 25, 26, 30-33, 35], whilst the remaining 11 of studies did not [15, 18, 19,
21-24, 27-29, 34]. Eight of the studies identified puerperal sepsis cases using self-reported
symptoms only [13, 15, 17, 19, 29, 30, 34, 35], 12 studies used a clinical diagnosis (i.e. by a
physician/trained healthcare worker or from medical records) [14, 18, 21-26, 28, 31-33] and
three studies used a combination of self-reported symptoms and/or clinical diagnosis [16, 20,
27]. In facility-based studies that did not explicitly specify the method of ascertainment of
sepsis cases, it was assumed that cases were identified by clinical diagnosis from a
physician or trained healthcare worker.
Risk factors for puerperal sepsis
Seven of the 23 included studies also identified statistically significant risk factors for
puerperal sepsis [13, 16, 17, 23, 28, 31, 35].
-
Candidate no: 108664
17
Table 2. Characteristics of included studies reporting puerperal sepsis prevalence.
No. Author(s), year
Region, Country
Study design & period of data collection
Setting Study Participants
Case definition of puerperal sepsis/ associated puerperal infections
Cases (n)
Total no. of deliveries (N)
Prevalence n/N (%)
Method of case
ascertainment
1 Goodburn et al., 2000 [13]
Bangladesh Prospective cross-sectional study, 1992-1993 (18 months)
Population-based, rural
Mothers delivered at home with a non-formal birth attendant
▪ ‘Postpartum maternal infection’ –Deemed to be present if 2 or more of the following symptoms were reported: lower abdominal pain; fever; foul smelling discharge - (reported at 2 and 6 weeks postpartum).
274 800 34.25 Self-reported symptoms
2 Bang et al.,
2004 [14]
Gadchiroli
district,
Maharashtra
– India
Prospective
cohort study,
1995-1998
(3 years)
Population-
based, rural
(39 villages)
All pregnant women
in selected study
villages
▪ ‘Puerperal fever’ – Temperature
>37.8oC on any day during 2–28 days
after delivery.
93
772
12.04
Clinical
observations by
trained health
workers and
physician
3
Fikree et
al.,
2004 [15]
Karachi
– Pakistan
Prospective
cross-
sectional
study,
Aug-Nov 2000
(3 months)
Population-
based
(5 low socio-
economic
settlements)
Muslim women
interviewed between
42 and 56 days
postpartum following
a live birth
▪ ‘High fever’
– No case definition given.
59
525
11.24
Self-reported
symptoms
-
Candidate no: 108664
18
Table 2. (continued)
No. Author(s), Year
Region, Country
Study design & period of data collection
Setting Study Participants
Case definition of puerperal sepsis/ associated puerperal infections
Cases (n)
Total no. of
deliveries (N)
Prevalence n/N (%)
Method of case
ascertainment
4 Fronczak et al., 2005 [16]
Dhaka – Bangladesh
Prospective cohort study, Nov 1993-May 1995 (19 months)
Population-based, (slum areas of 5 selected districts)
Pregnant women living in selected slum areas
▪ ‘Pelvic infection’ – Two of the symptoms: abdominal tenderness, fever, or foul vaginal discharge occurring 3 or more days postpartum (reported at 72 hours, 7 days & examined between 14-22 days postpartum).
14 1,506 0.93 Self-reported symptoms + corroboration by medical records for facility-based deliveries + physician examination 14 to 22 days postpartum
5 Ghani et
al.,
2007 [17]
Khyber
Agency
– Pakistan
Prospective
cross-
sectional
study,
July 2005
(1 month)
Population-
based
Pregnant or
postpartum married
women between 15
to 49 years
▪ ‘Vaginal infection’ – Perceived foul
smelling discharge, fever, lower
abdominal pain during postpartum
period (40 days).
162
1,000
16.20
Self-reported
symptoms
6 Jaleel &
Khan,
2008 [18]
Karachi
– Pakistan
Nested case
control study,
Jan-Dec 2007
(1 year)
Facility-
based,
urban
(tertiary
hospital)
All women admitted
for delivery and
during puerperium
▪ ‘Pelvis sepsis’ & ‘puerperal sepsis’ used
interchangeably (no postpartum period
specified)
– No case definition given.
5 735 0.68 Clinical
diagnosis
-
Candidate no: 108664
19
Table 2. (continued)
No. Author(s), Year
Region, Country
Study design & period of data collection
Setting Study participants
Case definition of puerperal sepsis/ associated puerperal infections
Cases (n)
Total no. of
deliveries (N)
Prevalence n/N (%)
Method of case
ascertainment
7 Patra et al., 2008 [19]
Chhainsa village, Haryana – India
Prospective cohort study, Aug 2002-July 2003 (1 year)
Population-based, rural
All women who experienced childbirth during the study period in Chhainsa village followed up to 6 weeks postpartum
▪ ‘High fever’ up to 42 days postpartum – No case definition given.
44
211
20.85
Self-reported symptoms
8 Christian
et al.,
2009 [20]
Sarlahi district
– Nepal
Cluster-
randomised
controlled
trial,
Jan 2000-Feb
2001
(1 year)
Population-
based,
rural
(30 villages)
Pregnant married
women living with
their husbands,
having at least 1 live
birth (control arm:
vitamin A
supplements; trial
arms: folic
acid/zinc/iron
supplement
combinations)
▪ ‘Puerperal sepsis’ – Measured
temperature ≥ 100.4 °F OR “hot to the
touch” on 2 or more days of the first
10 days postpartum exclusive of the first
day, and foul smelling vaginal discharge
on 2 or more days.
242
3,564
6.79
Self-reported
symptoms +/-
temperature
measurement
by trained
interviewer in
first 10 days
postpartum
9 Iyengar K.
& Iyengar
S.,
2009 [21]
Southern
Rajasthan
– India
Retrospective
cross-
sectional
study,
Jan 2000-Dec
2008
(8 years)
Facility-
based,
rural
(2 primary
health
centres)
Women in labour or
presenting with
antenatal, post-
partum or post-
abortion
complications
▪ ‘Puerperal fever’ (no postpartum
period specified)
– No case definition given.
6
3,171
0.19
Medical records
-
Candidate no: 108664
20
Table 2. (continued)
No. Author(s), Year
Region, Country
Study design & period of data collection
Setting Study participants
Case definition of puerperal sepsis/ associated puerperal infections
Cases (n)
Total no. of
deliveries (N)
Prevalence n/N (%)
Method of case
ascertainment
10 Shamshad et al., 2010 [22]
Abbottabad – Pakistan
Prospective cross-sectional study, Jan 2005-Dec 2007 (2 years)
Facility-based (Ayub Teaching Hospital; tertiary)
Obstetrical patients admitted over study period, including those presenting with post-abortion complications
▪ ‘Puerperal sepsis’ (within 42 days of termination of pregnancy) – No case definition given.
92
5,268
1.75
Medical records
11 Joshi et al.,
2011 [23]
Pune
– India
Prospective
cohort study,
Aug 2005-Sept
2007
(2 years)
Facility-
based,
urban
(tertiary
hospital)
Women receiving
prenatal care at the
study hospital
▪ ‘Postpartum fever’ (no postpartum
period specified)
– No case definition given.
38 1,200 3.17 Clinical
diagnosis
12 David et
al.,
2012 [24]
Vellore, Tamil
Nadu
– India
Retrospective
cross-
sectional
study,
2005-2010
(5 years)
Facility-
based,
(1 urban
health
centre, 1
tertiary
health
centre, other
government
& private
hospitals)
All women admitted
in labour to the study
facilities, including
patients transferred
intrapartum or in the
1st week postpartum
▪ ‘Puerperal sepsis’ (in the first week
postpartum)
– No case definition given.
0
1,873
0
Medical records
-
Candidate no: 108664
21
Table 2. (continued)
No. Author(s), Year
Region, Country
Study design & period of data collection
Setting Study participants
Case definition of puerperal sepsis/ associated puerperal infections
Cases (n)
Total no. of
deliveries (N)
Prevalence n/N (%)
Method of case
ascertainment
13 Huda et al., 2012 [25]
Matlab & Chandpur – Bangladesh
Prospective cross-sectional study, 2007-2008 (1 year)
Facility-based, rural (4 public hospitals & 26 private hospitals; mixed secondary & tertiary)
Women who gave birth during 2007-2008, excluding spontaneous and induced abortion cases
Puerperal infection up to 42 days postpartum, including: ▪ ‘Severe: septic shock or septicaemia’ – Genital source of infection and hyperthermia (fever 38.3°C and above for >48 hours) or hypothermia and low blood pressure (systolic
-
Candidate no: 108664
22
Table 2. (continued)
No. Author(s), year
Region, Country
Study design & period of data collection
Setting Study participants
Case definition of puerperal sepsis/ associated puerperal infections
Cases (n)
Total no. of
deliveries (N)
Prevalence n/N (%)
Method of case
ascertainment
14
Iyengar, 2012 [26]
Rajasthan – India
Prospective cross-sectional study, Jan 2007-Dec 2010 (3 years)
Population-based, rural (49 villages)
Recently delivered women
Puerperal infection in the first week postpartum, including: ▪ ‘Uterine infection’ – Temperature above 38 °C and one of the following symptoms: pain in lower abdomen, abnormal vaginal discharge, uterus not contracted well, history of heavy vaginal bleeding. ▪ ‘Upper urinary tract infection’ – Temperature above 38oC and one of the following symptoms: burning in micturition, flank pain. ▪ ‘Only fever’ – Temperature above 38 °C TOTAL:
64 28 73 165
4,975 4,975 4,975
4,975
1.29 0.56 1.47 3.32
Clinical examination by nurse-midwives
15 Iyengar et
al.,
2012 [27]
Rajasthan
– India
Prospective
cohort study,
June 2008-July
2010
(2 years)
Population-
based,
rural
Women who
delivered during the
study period and
were contacted
during the early
postpartum period
▪ ‘Fever’ – Temperature >38 °C reported
by nurse-midwife/doctor (in the first
week postpartum);
or ‘infected episiotomy/sepsis’ (in the
first week postpartum) – No case
definition given.
9
1,542
0.58
Home visit
examinations by
nurse-midwives
+ self-reported
symptoms
-
Candidate no: 108664
23
Table 2. (continued)
No. Author(s), year
Region, Country
Study design & period of data collection
Setting Study participants
Case definition of puerperal sepsis/
associated puerperal infections
Cases (n)
Total no. of
deliveries (N)
Prevalence n/N (%)
Method of case
ascertainment
16 Mannan et al., 2012 [28]
Dhaka – Bangladesh
Nested case-control study, July 2006-June 2007 (1 year)
Facility-based, urban (Sir Salimullah Medical College Hospital, Azimpur Maternity Center, some selected private hospitals; tertiary)
Pregnant women screened for gestational diabetes mellitus (GDM)
▪ ‘Puerperal sepsis’ (up to 4 weeks postpartum) – No case definition given.
17 144 11.81 Clinical diagnosis
17 Shriraam
et al.,
2012 [29]
Mugalivakkam,
Kancheepuram
district, Tamil
Nadu
– India
Prospective
cross-
sectional
study,
Nov 2008-Feb
2009
(3 months)
Population-
based,
rural
All women who
delivered in the last 6
months and
completed 42 days
puerperium
▪ ‘Puerperal sepsis’ (within 42 days
postpartum)
– No case definition given.
14
365
3.84
Self-reported
symptoms
18 Gupta et
al.,
2013 [30]
New Delhi
– India
Prospective
randomised
controlled
trial,
Oct 2005-
March 2007
(18 months)
Facility-based,
urban
(Lok Nayak
Hospital;
tertiary)
Asymptomatic
women in antenatal
clinic screened for
abnormal vaginal
flora (control arm: no
treatment, trial arm:
antibiotics)
▪ ‘Postpartum fever’ / ‘puerperal
pyrexia’ (used interchangeably) –
An oral temperature of ≥38.0°C on
any 2 of the first 10 days after
delivery, exclusive of the first 24
hours.
13
743
1.75
Self-measured
& self-reported
symptoms
-
Candidate no: 108664
24
Table 2. (continued)
No. Author(s), Year
Region, Country
Study design & period of data collection
Setting Study participants
Case definition of puerperal sepsis/ associated puerperal infections
Cases (n)
Total no.of deliveries
(N)
Prevalence n/N (%)
Method of
case
ascertainment
19 Khaskheli et al., 2013 [31]
Jamshoro, Hyderabad, Sindh - Pakistan
Prospective cohort study, Jan-Dec 2011 (1 year)
Facility-based (tertiary hospital)
Women delivering in this hospital or referred within 42 days after delivery
▪ ‘Puerperal pyrexia/ sepsis’ – Elevated body temperatures, abdominal distention, dehydration, foul smelling lochia, rise in total leukocyte count, intra-peritoneal collection, uterine collection, retained products of conception (within 42 days postpartum).
129 3,316 3.89 Clinical diagnosis (detailed history, examination & investigations)
20
Dasgupta
et al.,
2014 [32]
Pondicherry
– India
Prospective cohort study, 2 year period beginning Oct 2010
Facility-
based
(JIPMER;
tertiary)
Pregnant women up to 16 gestation wks on 1st antenatal visit, including abortion cases
▪ ‘Puerperal wound sepsis’ – Includes post- lower segment caesarean section wound discharge and episiotomy wound gape (postpartum period unspecified).
10
199
5.03
Clinical diagnosis
21 Hussein et al., 2014 [33]
Gujarat state – India
Prospective
cohort study,
Nov 2010-Feb
2012
(15 months)
Facility-
based
(6 hospitals;
mixed
secondary
and tertiary)
Women
delivering in the
study hospitals,
excluding
spontaneous and
induced abortion
cases
▪ ‘Puerperal sepsis and other puerperal
infections’ – As specified in ICD-10 codes 085
and 086, including infections of obstetric
surgical wounds, genital tract and urinary tract
infections following delivery (up to 42nd day
postpartum).
319
8,124
3.93
Clinical
diagnosis
22 Karkee et
al.,
2014 [34]
Kaski district
– Nepal
Prospective
cohort study,
Dec 2011-Oct
2012
(10 months)
Population-
based,
mixed
urban/rural
Women at least 5
months pregnant
completing
follow-up
interview
▪ ‘Postpartum severe fever’ (within 45 days
after delivery)
– No case definition given.
17
639
2.66
Self-reported
symptoms
-
Candidate no: 108664
25
Table 2. (continued)
No. Author(s), year
Region, Country
Study design & period of data collection
Setting Study participants
Case definition of puerperal sepsis/ associated puerperal infections
Cases (n)
Total no.of deliveries
(N)
Prevalence n/N (%)
Method of
case
ascertainment
23 Sikder et
al.,
2014 [35]
Bangladesh
(northwest)
Prospective
cohort study,
Dec 2007-June
2011
(4 months)
Population-
based,
rural
Pregnant women living with their husbands completing postpartum maternal morbidity interviews
▪ ‘Sepsis’ - High fever and foul-smelling vaginal discharge OR high fever and lower abdominal pain (30 days postpartum). If the woman had an induced abortion or miscarriage, these symptoms must have occurred within 7 days of the termination of pregnancy. ‘Near miss’ sepsis cases are defined as reports of nearly dying plus high fever.
3,259
42,124 7.74 Self-reported symptoms
-
Candidate no: 108664
26
4.3 Quality of case diagnosis of puerperal sepsis
All 12 studies that provided a case definition of puerperal sepsis or associated puerperal
infection used a diagnostic criteria acceptable under either the WHO or ICD-10 definitions
(see Table 1). However, the quality of a study’s diagnosis of puerperal sepsis also depends
on the method by which cases were ascertained. Clinical sources, including physicians,
trained healthcare workers/interviewers and medical records, are likely to provide a more
valid diagnosis of puerperal sepsis than self-reported symptoms [36].
Using the classification described in the methods section 3.4.1, four studies provided a low
quality diagnosis of puerperal sepsis cases [15, 19, 29, 34], 12 studies provided a medium
quality diagnosis [13, 17, 18, 20-24, 27, 28, 30, 35], and seven studies provided a high
quality diagnosis [14, 16, 25, 26, 31-33] (see Table 4).
-
Candidate no: 108664
27
4.4 Methodological quality of included studies
Each of the included studies were critically appraised using the JBI Prevalence Critical
Appraisal Checklist, consisting of ten questions addressing key methodological study
elements (see Appendix 2) [11].
Table 3. Critical appraisal results for included studies using the JBI Prevalence Critical Appraisal Checklist.
Y = yes, N = no, U = unclear, N/A = not applicable
Study author(s) & year
Q1 Q2 Q3 Q4 Q5 Q6 Q7 Q8 Q9 Q10 Included in meta-analysis
Goodburn et al., 2000 [13]
U Y U Y N Y N Y Y N/A -
Bang et al., 2004 [14] U Y U Y Y Y Y Y U N/A ✓
Fikree et al., 2004 [15] N Y U Y N N N Y U N/A -
Fronczak et al., 2005 [16] U Y U Y Y Y Y Y U N/A ✓
Ghani et al., 2007 [17] U Y U Y Y Y N Y U N/A -
Jaleel & Khan, 2008 [18] U Y Y Y N N Y Y N Y -
Patra et al., 2008 [19] U Y Y Y N N N Y U N/A -
Christian et al., 2009 [20] N Y U N Y Y N Y Y Y -
Iyengar K. & Iyengar S.,
2009 [21] U Y Y N N N Y U U Y -
Shamshad et al., 2010 [22]
U Y Y Y N N Y U N N/A -
Joshi et al., 2011 [23] U Y U Y N N Y Y U Y -
David et al., 2012 [24] U Y Y N N N Y U U N/A -
Huda et al., 2012 [25] N Y Y Y Y Y Y Y Y N/A ✓
Iyengar, 2012 [26] U Y Y Y Y Y Y Y Y N/A ✓
Iyengar et al., 2012 [27] Y Y Y N N N Y Y Y Y -
Mannan et al., 2012 [28] U Y U N N N Y Y U Y -
Shriraam et al., 2012
[29] U Y U Y N N N Y Y N/A -
Gupta et al., 2013 [30] U Y Y Y Y Y N Y N Y -
Khaskheli et al., 2013
[31] U Y Y Y Y Y Y Y N N/A ✓
Dasgupta et al., 2014 [32]
U Y Y N Y Y Y Y Y Y ✓
Hussein et al., 2014 [33] N Y Y Y Y Y Y Y Y Y ✓
Karkee et al., 2014 [34] U Y U Y N N N Y Y N/A -
Sikder et al., 2014 [35] Y Y Y Y Y Y N Y Y N/A -
-
Candidate no: 108664
28
All included studies used a sample consisting of recently delivered women or pregnant
women recruited for postpartum follow-up, representing the total number of deliveries within
that particular study setting. However, the studies varied as to whether they included
stillbirths and/or abortions as a delivery outcome. Two studies limited delivery outcomes to
live births only [15, 20] and two studies explicitly excluded abortions [25, 33], therefore
potentially underestimating the total number of deliveries and the number of sepsis cases as
a result of abortion or stillbirth events. Studies marked unclear under Q1 in Table 3 did not
mention whether both stillbirths and abortions were included in the study [13, 14, 16-19, 21-
24, 26-32, 34, 35].
All studies reported an appropriate method of recruitment of study participants, with most
providing sufficient detail regarding the characteristics of the setting and participants. 13
studies reported conducting a calculation for an adequate sample size; those which did not
were given an unclear status for Q3 (see Table 3). Most studies reported the reasons for
non-response, although few studies additionally described how non-responders differed from
responders with regards to their socio-demographic characteristics. All studies which
identified subpopulations amongst participants used objective criteria to classify subgroups.
12 studies provided an objective criterion for the diagnosis of puerperal sepsis cases [13, 14,
16, 17, 20, 25, 26, 30-33, 35], all of which were valid under the WHO or ICD-10 definitions
(see Table 1). However, the time period assigned to the puerperium in which sepsis cases
were identified differed between studies, ranging from the first seven days to the first 56
days postpartum, with a median of 40 days. Four studies did not specify a postpartum period
[18, 21, 23, 32]. Studies with earlier postpartum cut-offs may have missed sepsis cases
arising at later time periods, and those which did not provide any case definition may be
subject to measurement or classification bias. Cases were ascertained by different methods
across the studies, including self-reported symptoms and clinical sources (see Table 2).
Those studies relying on self-reported symptoms [13, 15, 17, 19, 20, 29, 30, 34, 35] may be
prone to information bias due to the subjectivity of self-perceived illness and the potential of
recall bias in retrospective study designs.
-
Candidate no: 108664
29
Table 4. An overview of the characteristics of included studies.
*BNG = Bangladesh, NPL = Nepal, IND = India, PAK = Pakistan
No. Author(s), year Location* Setting Quality of diagnosis of puerperal sepsis
BNG NPL IND PAK Facility- based
Pop- based
High Med Low
1 Goodburn et al., 2000 [13] ✓ ✓ ✓
2 Bang et al., 2004 [14] ✓ ✓ ✓
3 Fikree et al., 2004 [15] ✓ ✓ ✓
4 Fronczak et al., 2005 [16] ✓ ✓ ✓
5 Ghani et al., 2007 [17] ✓ ✓ ✓
6 Jaleel & Khan, 2008 [18] ✓ ✓ ✓
7 Patra et al., 2008 [19] ✓ ✓ ✓
8 Christian et al., 2009 [20] ✓ ✓ ✓
9 Iyengar K. & Iyengar S., 2009 [21]
✓ ✓ ✓
10 Shamshad et al., 2010 [22] ✓ ✓ ✓
11 Joshi et al., 2011 [23] ✓ ✓ ✓
12 David et al., 2012 [24] ✓ ✓ ✓
13 Huda et al., 2012 [25] ✓ ✓ ✓
14 Iyengar, 2012 [26] ✓ ✓ ✓
15 Iyengar et al., 2012 [27] ✓ ✓ ✓
16 Mannan et al., 2012 [28] ✓ ✓ ✓
17 Shriraam et al., 2012 [29] ✓ ✓ ✓
18 Gupta et al., 2013 [30] ✓ ✓ ✓
19 Khaskheli et al., 2013 [31] ✓ ✓ ✓
20 Dasgupta et al., 2014 [32] ✓ ✓ ✓
21 Hussein et al., 2014 [33] ✓ ✓ ✓
22 Karkee et al., 2014 [34] ✓ ✓ ✓
23 Sikder et al., 2014 [35] ✓ ✓ ✓
-
Candidate no: 108664
30
4.4 Selection of studies for meta-analysis
After visual inspection of data, performing a statistical meta-analysis for all included studies
was deemed to be inappropriate due to considerable heterogeneity. Therefore, a sub-set of
studies of high methodological quality was instead selected for meta-analysis using key
criteria from the JBI Critical Appraisal Checklist (see Appendix 2), as described in the
methods section 3.4.2. Using this criteria, seven studies were identified as high-quality [14,
16, 25, 26, 31-33] and were included in formal meta-analysis.
4.5 The prevalence of puerperal sepsis
4.5.1 Overall prevalence of puerperal sepsis
The reported prevalence of puerperal sepsis varied across all included studies (see Figure
2), with a median of 3.84% of total deliveries and a large range of 0% to 34.25%.
4.5.2 Prevalence of puerperal sepsis by country
Large variations in prevalence also existed within countries (see Figure 3). Nepal had the
highest median prevalence of 11.76%; however this is calculated from two studies only with
a wide range from 2.66% to 20.85%. The median prevalence in Bangladesh was 7.74%, with
a range from 0.88% to 34.25%. The prevalence tended to be lower in Pakistan (median
3.89%, range 0.19% to 16.20%) and India (median 3.32%, range 0% to 12.05%).
4.5.2 Prevalence of puerperal sepsis by setting
When stratified by setting (see Figure 4), the prevalence tended to be lower and less varied
in facility-based studies (median 1.75%, range 0.19% to 11.81%) compared to population-
based studies (median 7.26%, range 0.58% to 34.25%,).
4.5.4 Prevalence of puerperal sepsis by quality of case diagnosis
Figure 5 shows the prevalence of puerperal sepsis stratified by the quality of case diagnosis.
In studies with a high quality case diagnosis, the prevalence ranged from 0.58% to 12.05%,
with a median of 3.89%. Studies with a medium quality case diagnosis demonstrated a lower
median prevalence of 2.46%, with a range from 0% to 34.25%. Prevalence data from studies
with a low quality case diagnosis tended to be more varied, with a median of 7.54% and a
range from 2.66% to 20.85%.
-
Candidate no: 108664
31
4.6 Meta-analysis of the prevalence of puerperal sepsis reported by high-quality
studies
Meta-analysis was performed on a subset of seven studies deemed to be of high
methodological quality (see Figure 6). The pooled prevalence of puerperal sepsis was 3.20%
of total deliveries (95% CI 2.03%, 5.02%). The I-squared and tau-squared test indicate
substantial statistical heterogeneity among this sub-set of studies (p
-
Candidate no: 108664
32
Figure 2. Prevalence of puerperal sepsis reported by all included studies.
1 David et al. identified 0 puerperal sepsis cases, which is unable to be displayed.
Study identifier
Goodburn et al., 2000
Bang et al., 2004
Fikree et al., 2004
Fronczak et al., 2005
Ghani et al., 2007
Jaleel & Khan, 2008
Patra et al., 2008
Christian et al., 2009
Iyengar K. & Iyengar S., 2009
Shamshad et al., 2010
Joshi et al., 2011
David et al., 20121
Huda et al., 2012
Iyengar, 2012
Iyengar et al., 2012
Mannan et al., 2012
Shriraam et al., 2012
Gupta et al., 2013
Hussein et al., 2014
Dasgupta et al., 2014
Karkee et al., 2014
Sikder et al., 2014
Khaskheli et al., 2013
Prevalence % (95% CI)
34.25 (30.96, 37.65)
12.05 (9.83, 14.55)
11.24 (8.67, 14.26)
0.93 (0.51, 1.55)
16.20 (13.97, 18.63)
0.68 (0.22, 1.58)
20.85 (15.58, 26.97)
6.79 (5.98, 7.67)
0.19 (0.07, 0.41)
1.75 (1.41, 2.14)
3.17 (2.25, 4.32)
n/a
0.88 (0.51, 1.41)
3.32 (2.84, 3.85)
0.58 (0.27, 1.10)
11.81 (7.03, 18.2)
3.84 (2.11, 6.35)
1.75 (0.93, 2.97)
3.93 (3.51, 4.37)
5.03 (2.43,1, 9.05)
2.66 (1.56, 4.22)
7.74 (7.48, 8.00)
3.89 (3.26, 4.60)
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40
Prevalence of puerperal sepsis %
-
Candidate no: 108664
33
Figure 3. Prevalence of puerperal sepsis by country.
1 David et al. identified 0 puerperal sepsis cases, which is unable to be displayed.
Country
Bangladesh
India
Pakistan
Nepal
Goodburn et al., 2000
Bang et al., 2004
Fikree et al., 2004
Fronczak et al., 2005
Ghani et al., 2007
Jaleel & Khan, 2008
Patra et al., 2008
Christian et al., 2009
Iyengar K. & Iyengar S., 2009
Shamshad et al., 2010
Joshi et al., 2011
David et al., 20121
Huda et al., 2012
Iyengar, 2012Iyengar et al., 2012
Mannan et al., 2012
Shriraam et al., 2012
Gupta et al., 2013Hussein et al., 2014
Dasgupta et al., 2014
Karkee et al., 2014
Sikder et al., 2014
Khaskheli et al., 2013
Prevalence % (95% CI)
34.25 (30.96, 37.65)
12.05 (9.83, 14.55)
11.24 (8.67, 14.26)
0.93 (0.51, 1.55)
16.20 (13.97, 18.63)
0.68 (0.22, 1.58)
20.85 (15.58, 26.97)
6.79 (5.98, 7.67)
0.19 (0.07, 0.41)
1.75 (1.41, 2.14)
3.17 (2.25, 4.32)
n/a
0.88 (0.51, 1.41)
3.32 (2.84, 3.85)0.58 (0.27, 1.10)
11.81 (7.03, 18.2)
3.84 (2.11, 6.35)
1.75 (0.93, 2.97)3.93 (3.51, 4.37)
5.03 (2.43,1, 9.05)
2.66 (1.56, 4.22)
7.74 (7.48, 8.00)
3.89 (3.26, 4.60)
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40
Prevalence of puerperal sepsis %
-
Candidate no: 108664
34
Figure 4. Prevalence of puerperal sepsis by setting.
1 David et al. identified 0 puerperal sepsis cases, which is unable to be displayed.
Study setting
Facility-based
Population-based
Goodburn et al., 2000
Bang et al., 2004
Fikree et al., 2004
Fronczak et al., 2005
Ghani et al., 2007
Jaleel & Khan, 2008
Patra et al., 2008
Christian et al., 2009
Iyengar K. & Iyengar S., 2009
Shamshad et al., 2010
Joshi et al., 2011
David et al., 20121
Huda et al., 2012
Iyengar, 2012
Iyengar et al., 2012
Mannan et al., 2012
Shriraam et al., 2012
Gupta et al., 2013
Hussein et al., 2014
Dasgupta et al., 2014
Karkee et al., 2014
Sikder et al., 2014
Khaskheli et al., 2013
Prevalence % (95% CI)
34.25 (30.96, 37.65)
12.05 (9.83, 14.55)
11.24 (8.67, 14.26)
0.93 (0.51, 1.55)
16.20 (13.97, 18.63)
0.68 (0.22, 1.58)
20.85 (15.58, 26.97)
6.79 (5.98, 7.67)
0.19 (0.07, 0.41)
1.75 (1.41, 2.14)
3.17 (2.25, 4.32)
n/a
0.88 (0.51, 1.41)
3.32 (2.84, 3.85)
0.58 (0.27, 1.10)
11.81 (7.03, 18.2)
3.84 (2.11, 6.35)
1.75 (0.93, 2.97)
3.93 (3.51, 4.37)
5.03 (2.43,1, 9.05)
2.66 (1.56, 4.22)
7.74 (7.48, 8.00)
3.89 (3.26, 4.60)
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40
Prevalence of puerperal sepsis %
-
Candidate no: 108664
35
Figure 5. Prevalence of puerperal sepsis by quality of case diagnosis.
1 David et al. identified 0 puerperal sepsis cases, which is unable to be displayed.
Quality of diagnosis
High
Medium
Low
Goodburn et al., 2000
Bang et al., 2004
Fikree et al., 2004
Fronczak et al., 2005
Ghani et al., 2007
Jaleel & Khan, 2008
Patra et al., 2008
Christian et al., 2009
Iyengar K. & Iyengar S., 2009
Shamshad et al., 2010
Joshi et al., 2011
David et al., 20121
Huda et al., 2012
Iyengar, 2012
Iyengar et al., 2012
Mannan et al., 2012
Shriraam et al., 2012
Gupta et al., 2013
Hussein et al., 2014
Dasgupta et al., 2014
Karkee et al., 2014
Sikder et al., 2014
Khaskheli et al., 2013
Prevalence % (95% CI)
34.25 (30.96, 37.65)
12.05 (9.83, 14.55)
11.24 (8.67, 14.26)
0.93 (0.51, 1.55)
16.20 (13.97, 18.63)
0.68 (0.22, 1.58)
20.85 (15.58, 26.97)
6.79 (5.98, 7.67)
0.19 (0.07, 0.41)
1.75 (1.41, 2.14)
3.17 (2.25, 4.32)
n/a
0.88 (0.51, 1.41)
3.32 (2.84, 3.85)
0.58 (0.27, 1.10)
11.81 (7.03, 18.2)
3.84 (2.11, 6.35)
1.75 (0.93, 2.97)
3.93 (3.51, 4.37)
5.03 (2.43,1, 9.05)
2.66 (1.56, 4.22)
7.74 (7.48, 8.00)
3.89 (3.26, 4.60)
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40
Prevalence of puerperal sepsis %
-
Candidate no: 108664
36
Table 5. Median prevalence and ranges of all included studies, overall and by strata.
Figure 6. Meta-analysis of the prevalence of puerperal sepsis reported by high-quality studies.
Strata Median prevalence (%) Range (%)
All included studies
3.84 0 – 34.25
By country Bangladesh India Nepal Pakistan
7.74 3.32 11.76 3.89
0.88 – 34.25 0 – 12.05 2.66 – 20.85 0.19 – 16.20
By setting Facility-based Population-based
1.75 7.26
0 – 11.81 0.58 – 34.25
By quality of case diagnosis High quality Medium quality Low quality
3.89 2.46 7.54
0.58 – 12.05 0 – 34.25 3.54 – 13.64
-
Candidate no: 108664
37
4.7 Risk factors for puerperal sepsis
Of the 23 studies included in this review, eight studies identified a number of statistically
significant risk factors for puerperal sepsis, which can be categorised as maternal factors,
labour factors, health service factors and community factors (see Table 6.)
Table 6. Identified risk factors for puerperal sepsis.
Type of risk factor Risk factors for puerperal sepsis
Maternal factors Age 35 yrs [35] Nulliparity [35] Higher parity [16] Mid upper arm circumference
-
Candidate no: 108664
38
5. DISCUSSION
5.1 The overall prevalence of puerperal sepsis in South Asia
Findings from this systematic review suggest that the prevalence of puerperal sepsis in
South Asia falls between 3-4% of all deliveries, as identified from the median prevalence of
23 included studies (3.84%) and the pooled prevalence of seven high-quality studies
(3.20%). However, it is difficult to establish a single reliable estimate due wide variations in
reported prevalence by individual studies (ranging from 0% to 34.25%), as well as significant
statistical heterogeneity that is likely to be attributed to clinical and methodological variation
between studies.
No previous reviews have provided an estimate of the overall prevalence of puerperal sepsis
in South Asia specifically. Nonetheless, the median and pooled prevalence estimates found
by this review were comparable to the global estimate of 4.4% of live births produced by
AbouZahr (2003) [3]. As was the case with this review, wide prevalence ranges were also
identified for Sub-Saharan Africa by Seale et al. (2009) (range 0.1% to 17.9%) [5], and for
developing countries by Dolea and Stein (2003) (range 0.07% to 9.3%) [4] and Hussein and
Walker (2010) (range 0.1% to 10%) [7]. All of these reviews highlighted the difficulty in
estimating prevalence due to marked differences in study characteristics. This systematic
review therefore stratified prevalence data by the following key study features: country of
study, setting (facility-based or population-based), and the quality of case diagnosis of
puerperal sepsis.
5.2 The prevalence of puerperal sepsis by country
Despite stratification by country, wide variations in the prevalence of puerperal sepsis
persisted, making it difficult to discern any country-level trends. It should be noted that only
studies from India, Bangladesh, Pakistan and Nepal met the inclusion criteria for this
systematic review, therefore generalisability of results to the remaining South Asian
countries is limited.
5.3 The prevalence of puerperal sepsis by setting
The prevalence of puerperal sepsis tended to be lower and less varied in facility-based
studies (median 1.75%) compared to population-based studies (median 7.26%). It is likely
that facility-based studies identified fewer puerperal sepsis cases than population-based
studies due to the admission of patients with more severe infection only [20], as well as
missed sepsis cases among women with limited access to facilities due to factors such as
distance, financial constraints and cultural beliefs [4]. Furthermore, postpartum infections
-
Candidate no: 108664
39
can manifest several days after delivery, therefore cases that occur after being discharged
without re-admission often go undiagnosed and under-reported, particularly in settings with
no postnatal follow-up [7] [37].
Regardless of whether the recruitment of study participants was facility-based or population-
based, the proportions of women giving birth in a facility or at home varied within studies,
accounting for some of the variation in prevalence estimates. It is suggested that more
puerperal sepsis cases arise from home births than facility births due to higher levels of
unclean practices [25], such as traditions of inserting home-made substances into the vagina
[4] and restricted bathing postpartum [17]. The lack of skilled birth attendance at home births
is also associated with less hygienic delivery practices [13] [25]. On the other hand, facilities
may face higher rates of nosocomial puerperal infection as a result of suboptimal hygiene
standards and increased risk from multiple pelvic examinations, caesarean sections and
other invasive procedures [6] [26] [39]. The need for prevention and control of infections in
facilities is becoming a more pressing issue as the utilisation of already over-burdened
health facilities increases [31]. Despite the significant role that the place of delivery plays in
the contraction of puerperal infection, few studies specify the number of puerperal sepsis
cases by facility births and home births. Future research should not only identify puerperal
sepsis cases by study setting, but also consider the place of delivery and the presence of
skilled birth attendants.
5.4 The prevalence of puerperal sepsis by the quality of case diagnosis
Studies with a high-quality case diagnosis of puerperal sepsis reported a median prevalence
of 3.89%. This is likely to be a more reliable prevalence estimate than those ascertained
from studies with a medium-or low- quality diagnosis because of the use of an appropriate
definition and a clinical source of case ascertainment rather than self-report. However, the
comparability of prevalence estimates remains limited due to variations in the definitions and
types of puerperal infection included by individual studies. This may have led to an
overestimation or underestimation of prevalence, as some cases identified by some studies
will have been missed by others. For example, several studies include cases of puerperal
fever alone, whereas other study definitions would consider this insufficient for a diagnosis of
puerperal sepsis without other symptoms present. AbouZahr (2003) suggests that puerperal
fever is a reliable index of puerperal sepsis, as most causes of temperature elevation in the
puerperium are associated with pelvic infection [3]. However, it would also encompass cases
of fever associated with non-genitourinary infection and other incidental causes [7],
overestimating the prevalence of puerperal sepsis. The postpartum time period in which
cases were identified also varied between studies, so studies with earlier cut-offs may have
-
Candidate no: 108664
40
missed infections that manifested in later days or weeks. Furthermore, several studies
reported overlapping data for different types of puerperal infection, of which only the most
generic type was selected for analysis, underestimating the total number of puerperal sepsis
cases. It is therefore clear that reliable prevalence estimates of puerperal sepsis are difficult
to obtain without a universal diagnostic criterion.
The quality of a study’s case diagnosis is also dependent on the method of ascertainment of
puerperal sepsis cases. Clinical diagnosis by trained healthcare workers are considered to
be a more valid method of determining puerperal sepsis than self-reported symptoms [42]. It
has been noted that self-reported symptoms tend to over-estimate the incidence of clinically
present morbidities [4] [16], which may be due to self-perception of symptoms such as ‘fever’
without accurate measurement, and over-reporting by participants who feel obliged to report
something [19]. These may have been contributing factors to the anomalously high
prevalence estimates of 34.25%, 16.20% and 20.85% reported by Goodburn et al. (2000)
[13], Ghani et al. (2007) [17] and Patra et al. (2008) [19] respectively. Self-reported
symptoms may also underestimate prevalence due to limitations in recall and the perception
of postpartum morbidities as insignificant or normal [19]. Nonetheless, self-reported
symptoms remain important in determining health behaviour patterns and the perceived
discomfort and interference to women’s daily lives [15] [19]. It may also be the only feasible
method to collect data on maternal morbidities in resource-poor settings [4].
Clinical diagnoses ascertained by healthcare workers or from medical records are not
without fault, as they may be subject to misclassification by facilities with with poor record-
keeping [25] and limited awareness of the clinical presentation of puerperal sepsis. Signs
and symptoms vary depending on the source of infection [37] as well as the degree of
severity of sepsis [40] [41], which can make specific diagnosis ambiguous to healthcare
workers. For greater accuracy in diagnosing sepsis, Seale et al. (2009) [5] recommend the
use of adequate microbiological investigations. No included studies in this review used
microbiological tests to confirm a diagnosis of puerperal sepsis, although one study
(Khaskheli et al., 2013 [31]) performed blood tests as supplementary investigations. The lack
of microbiological and blood testing in studies is likely to reflect the technical difficulties in
obtaining uncontaminated samples [42], as well as the limited capacity of utilising such tests
in resource-poor settings.
5.5 Risk factors for puerperal sepsis
Risk factors leading to the development of puerperal sepsis are varied. Maternal factors
identified by studies include malnutrition in pregnant women, indicated by anaemia [16] and
-
Candidate no: 108664
41
a small mid-upper arm circumference [35], which increases the risk of infection due to a
weakened immune system response [35]. The risk pertained by overweight and obese
mothers [23] [30] is of increasing concern as the prevalence of obesity is rising in developing
countries, particularly amongst women [43]. Pre-existing reproductive tract and urinary tract
infections [13] [16] can develop into sepsis following delivery, therefore detection and
treatment during pregnancy may reduce the prevalence of puerperal sepsis [13]. Poor
glycaemic control in gestational diabetes is suggested to increase the risk of postpartum
infection [28] [44], which would also require screening and effective management. Sikder et
al. (2014) identify unwanted pregnancy as a risk factor for sepsis, which may be attributed to
decreased pregnancy care and attention and reduced attendance to antenatal care [35].
Other maternal risk factors identified include the extremes of ages (35 years)
[35], parity (nulliparity [35] or high parity [16]) and adverse obstetric history [35].
Risk factors that facilitate infection around the time of delivery include prolonged rupture of
membranes [31] and a long duration of labour [13], as the cervix remains open for a
lenghtened time and natural barriers to ascending infections from the vagina are impaired
[22]. Infection control measures are needed to prevent unhygienic practices around the time
of delivery, such as putting a hand into the vagina [13] or using unclean material to staunch
the flow of lochia [17]. Other determinants during labour include intrapartum bleeding,
perineal tears and stillbirth [16].
Health service factors associated with puerperal sepsis and other postpartum morbidities
include poor utilisation of antenatal care resulting in an unbooked status at delivery [31], and
limited access to health facilities for delivery [17]. Antenatal care is beneficial in screening
and treating complications such as anaemia, vaginal infections and urinary tract infections,
as well as raising awareness of the need for care at delivery [17]. It is suggested that
facilities are more attentive to aseptic measures that may not be performed in home births,
such as handwashing, use of a clean delivery service, clean cord cutting, perineal hygiene
and providing antibiotics after delivery [22]. This highlights the important role of
comprehensive good-quality antenatal and delivery services in reducing the prevalence of
puerperal sepsis.
Community risk factors for puerperal sepsis include lower socioeconomic status of mothers
[17], which may be associated with illiteracy, a lack of knowledge of signs and symptoms,
limited availability of obstetric care, and poorer hygiene and general health [4] [22].
Traditional practices resulting in poor hygiene, such as not taking a bath postpartum [17],
highlight the need to develop an understanding of the socio-cultural aspects that may
-
Candidate no: 108664
42
facilitate puerperal infection.
The risk factors discussed above are consistent with findings by the reviews undertaken for
developing countries [4] [7], as well as those identified in sub-Saharan Africa [5].
Nonetheless, risk factors are likely to differ in relative importance depending on the region
and setting. A key determinant of puerperal infection that is additionally mentioned in several
reviews is caesarean section [4] [5] [6] [7], which may increase rates of puerperal sepsis in
facilities due to increasing trends in surgical delivery. It should be noted that in this
systematic review, risk factors for puerperal sepsis were identified from included studies as a
secondary objective, and therefore were not investigated comprehensively.
5.6 Puerperal sepsis-related mortality
Mortality from puerperal sepsis was not investigated as an outcome in this review, but is an
important indicator of the magnitude of this condition. Without early diagnosis and
management, puerperal sepsis can rapidly progress along a clinical spectrum of morbidity
[31], leading to severe septicaemia and death. South Asia is identified to have the highest
proportion of global maternal deaths from sepsis at 13.7%, with 107,000 maternal sepsis
deaths occurring between 2003 and 2009 [2], and approximately 9,400 deaths occuring in
2013 [45]. Therefore, despite a relatively low estimated prevalence, puerperal sepsis
remains an important problem in South Asia due to potentially severe sequalae and its
impact on maternal mortality.
5.7 Strengths of systematic review
This systematic review is the first to estimate the prevalence of puerperal sepsis in South
Asia. It also provides a summary of risk factors for puerperal infection identified from
included studies. A comprehensive search strategy of several databases was used (see
Appendix 1), including a manual review of reference lists and forward citation tracking to
identify any studies missed by formal database searching.
5.8 Limitations of systematic review
This systematic review is limited by the identification of a small number of studies
reporting the prevalence of puerperal sepsis, of which only seven were deemed to be of
high methodological quality. Furthermore, included studies were located in four countries
only, so the generalisability of findings to the remaining South Asian countries is
restricted. Risk factors for puerperal sepsis were only identified from included studies
reporting prevalence data, therefore a comprehensive list was not provided.
-
Candidate no: 108664
43
The validity of the pooled prevalence estimate obtained by meta-analysis is limited due to
the analysis of a very small number of studies and significant statistical heterogeneity in
study outcomes. This is likely to be a result of differences in study location, setting, the
definition of puerperal sepsis used and the method of case ascertainment, as discussed
previously. Other differences in included studies that may have contributed to the large
discrepancies in prevalence estimates include variations in study design, quality, sample
size, participant characteristics (including socioeconomic status, living standards and
access to healthcare), and whether abortions and stillbirths were included as delivery
outcomes. Furthermore, most included studies investigated study participants from very
select populations or facilities, limiting the generalisability of findings to the whole region
or country.
5.9 Implications & recommendations for future research
In order to obtain more reliable prevalence estimates, it is important that studies use an
established standard definition of puerperal sepsis, as well as objective clinical methods
of ascertaining cases. Studies should also endeavour to collect population-representative
data of good quality that is inclusive of all delivery outcomes. Key sub-groups, such as
setting and place of delivery, should be identified between and within studies to account
for differences in prevalence and improve the comparability of prevalence estimates.
The risk factors for puerperal sepsis identified in this review have implications for
improving infection control and other prevention efforts. These include the detection and
management of antepartum maternal conditions such as anaemia, gestational diabetes
and pre-existing infections, ensuring good hygienic practice during and after delivery, and
improving access to good-quality obstetric care, including antenatal services and
postnatal follow-up. A more comprehensive review of risk factors should be conducted to
detect other determinants of puerperal sepsis that may have been missed.
In order to truly assess the magnitude of puerperal sepsis in South Asia, more data is
needed regarding sepsis-related mortality, as well as the burden of long-term morbidities
such as chronic pelvic inflammatory disease and secondary infertility, which appear to be
neglected in literature. Additionally, effective and feasible interventions for prevention and
treatment need to be identified in order to address and reduce the prevalence of
puerperal sepsis.
-
Candidate no: 108664
44
6. CONCLUSION
This systematic review is the first to attempt to estimate the prevalence of puerperal
sepsis in South Asia. Although a single reliable estimate of puerperal sepsis in South
Asia is difficult to establish, the findings from this systematic review indicate a relatively
low prevalence between 3-4%. Nonetheless, this should not detract from the public health
importance of this condition, as it remains a significant cause of obstetric morbidity and
mortality in South Asia despite its highly preventable nature. Risk factors for puerperal
sepsis are varied and include maternal, labour, health service and community factors,
which should be the focus for prevention efforts.
Producing better estimates for the prevalence of puerperal sepsis is important in
accurately measuring the magnitude of this condition and assessing progress in
improving maternal outcomes, particularly in South Asia where the global burden of
maternal sepsis deaths is highest [2]. This can only be achieved when good-quality data
is regularly collected using a standard definition of puerperal sepsis and a representative
population inclusive of all delivery outcomes.
-
Candidate no: 108664
45
References: [1] World Health Organization (WHO). 2008. Managing puerperal sepsis. Geneva: WHO Press.
[2] Say L., Chou D., Gemmill A., Tunçalp Ö, Moller A., Daniels J., Gülmezoglu A. M.,
Temmerman M., Alkema L. 2014. Global causes of maternal death: a WHO systematic
analysis. The Lancet, 2(6), pp.323-333.
[3] AbouZahr C., 2003. Global burden of maternal death and disability. British Medical
Bulletin, 67, pp.1-11.
[4] Dolea C. and Stein C., 2003. The global burden of maternal sepsis in the year 2000.
WHO.
[5] Seale A., Mwaniki M., Newton C., Berkley J., 2009. Maternal and early onset neonatal
bacterial sepsis: burden and strategies for prevention in sub-Saharan Africa. The Lancet
Infectious Diseases, 9(7), pp.428–438.
[6] van Dillen J., Zwart J., Schutte J., van Roosmalen J. 2010. Maternal sepsis:
epidemiology, eitiology and outcome. Current Opinion in Infectious Diseases, 23(3), pp.249-
54.
[7] Hussein J. and Walker L. 2010. Puerperal Sepsis in low- and middle- income settings:
past, present and future. Cambridge University Press: Royal College of Ostetricians and
Gynaecologists Study Group.
[8] World Health Organization (WHO). 1995. The prevention and management of puerperal
infections: report of a technical working group, Geneva, 20-22 May 1992. Geneva: WHO.
[9] International Statistical Classification of Diseases and Related Health Problems 10th
Revision (ICD-10)-2015-WHO Version, 2015. Complications predominantly related to the
puerperium (085-092). [online] Available at:
http://apps.who.int/classifications/icd10/browse/2015/en#/O85
[10] United Nations Statistics Division, 2013. Composition of macro geographical
(continental) regions, geographical sub-regions, and selected economic and other
groupings. [online] Available at: http://unstats.un.org/unsd/methods/m49/m49regin.htm#asia
[11] The Joanna Briggs Institute, 2014. The Joanna Briggs Institute Reviewers’ Manual 2014
– The systematic review of prevalence and incidence data. Australia: Joanna Briggs
Institute.
[12] Statistics Canada, 2012. Data quality, concepts and methodology: Definitions. [online]
Available at: http://www.statcan.gc.ca/pub/84f0210x/2008000/technote-notetech1-eng.htm
http://www.ncbi.nlm.nih.gov/pubmed/?term=Mwaniki%20M%5Bauth%5Dhttp://www.ncbi.nlm.nih.gov/pubmed/?term=Newton%20CR%5Bauth%5Dhttp://www.ncbi.nlm.nih.gov/entrez/eutils/elink.fcgi?dbfrom=pubmed&retmode=ref&cmd=prlinks&id=19555902http://www.ncbi.nlm.nih.gov/entrez/eutils/elink.fcgi?dbfrom=pubmed&retmode=ref&cmd=prlinks&id=19555902http://apps.who.int/classifications/icd10/browse/2015/en#/O85http://unstats.un.org/unsd/methods/m49/m49regin.htm#asiahttp://www.statcan.gc.ca/pub/84f0210x/2008000/technote-notetech1-eng.htm
-
Candidate no: 108664
46
[13] Goodburn E., Chowdhury M., Gazi R., Marshall T., Graham W., 2000. Training traditional birth attendants in clean delivery does not prevent postpartum infection. Health Policy and Planning, 15(4), pp.394-399. [14] Bang R., Bang A., Reddy M., Deshmukh M., Baitule S., Filippi, V., 2004. Maternal morbidity during labour and the puerperium in rural homes and the need for medical attention: A prospective observational study in Gadchiroli, India. BJOG: An International Journal of Obstetrics & Gynaecology, 111(3), pp.231-238. [15] Fikree F., Ali T., Durocher J., Rahbar M., 2004. Health service utilization for perceived postpartum morbidity among poor women living in Karachi. Social Science and Medicine, 59(4), pp.681-694. [16] Fronczak N., Antelman G., Moran A., Caulfield L., Baqui A., 2005. Delivery-related complications and early postpartum morbidity in Dhaka, Bangladesh. International Journal of Gynaecology & Obstetrics, 91(3), pp.271-278. [17] Ghani N., Rukanuddin R., Ali T., 2007. Prevalence and factors associated with postpartum vaginal infection in the Khyber Agency Federally Administered Tribal Areas, Pakistan. JPMA, Journal of the Pakistan Medical Association, 57(7), pp.363-367. [18] Jaleel R. and Khan A., 2008. Obstetric morbidity in the booked versus non-booked patients – a comparative study at Lyari General Hospital. Pakistan Journal of Surgery, 24(3), pp.196-202. [19] Patra S., Singh B., Reddaiah V., 2008. Maternal morbidity during postpartum period in a village of north India: A prospective study. Tropical Doctor, 38(4), pp.204-208. [20] Christian P., Khatry S., LeClerq S., Dali S., 2009. Effects of prenatal micronutrient supplementation on complications of labor and delivery and puerperal morbidity in rural Nepal. International Journal of Gynaecology & Obstetrics, 106(1), pp.3-7. [21] Iyengar K. and Iyengar S., 2009. Emergency obstetric care and referral: experience of two midwife-led health centres in rural Rajasthan, India. Reproductive Health Matters, 17(33), pp.9-20. [22] Shamshad, Shamsher S., Rauf B., 2010. Puerperal sepsis - still a major threat for parturient. Journal of Ayub Medical College Abbottabad, 22(3), pp.18-22. [23] Joshi S., Unni J., Vijay S., K