9/24/2018

1

Data in International

Research:

Lessons from Design,

Implementation & Scaling in Low and Middle Income

Countries

Katherine Semrau, PhD, MPH

Ariadne Labs

September 24, 2018

Roadmap• Content

– Maternal & Newborn Health

• Collaboration

– Towards mul tidisciplinary teams

• Connectivity

– Transition of paper to electronic

data

• Culture

– Theory to on the ground reality

• Q&A

9/24/2018

2

Maternal & Neonatal Mortality

0

10

20

30

40

50

60

World LowerMiddleIncomeCountry

India UttarPradesh

USA

NM

R p

er

1,0

00

liv

eb

irth

s

1990 2013

OCCURRING AROUND THE GLOBE EACH YEAR

303,000maternal deaths

1.2 mil lion intrapartum-

related stillbirths

2.8 mil lion neonatal deaths

Moments of Greatest Risk

4

LABOR ONSET

MA

TE

RN

AL

& N

EO

NA

TA

L M

OR

TA

LIT

Y R

ISK

DELIVERY

ADMISSION TO FACILITY

DISCHARGE FROM FACILITY

ANTENATAL PERIOD

28 DAYS

42 DAYS

T I M E

9/24/2018

3

Global Maternal Mortality

303,000 deaths per year

Alekema et al. Lancet. 387 (10017); 462-474.

Methods to Prevent Mortality are Known

Skin-to-skin care post-partum

Newborn antibiotics if high or low temperature

Instructions on danger signs for mother and child

Presence of a birth companion

Use of handwashing and gloving

Informing when to call for help

Preparation of oxytocin, clean supplies, bag mask, sterile blade, suction

Assistant at the ready

Uterine massage and oxytocin within 1 minute

PRENATAL LABOR & DELIVERY POST-PARTUM

Early referral for at-risk women

Antibiotics for fever, ruptured membranes

Magnesium for hypertension, proteinuria

9/24/2018

4

We Don’t Change What We Don’t Measure

Measurement Settings Matter

9/24/2018

5

Collaboration: Towards Multidisciplinary Teams

10

9/24/2018

6

The WHO Safe Childbirth Checklist

ONADMISSION

AD MISSION

1234

Evaluations for feasibili ty and acceptabili ty were completed in nine

countries

The WHO established the Safe Childbirth Checklist Collaborative to

explore factors that influence the use of the Checklist in diverse settings around

the world

Following a successful pilot study in

Karnataka, India, Collaborative

members completed a series of implementation studies in Rajasthan,

Sri Lanka, and Bangladesh

Desire for RCT to provide evidence-

based impact; BetterBirth trial in Uttar Pradesh, India finished in early 2017

Safe Childbirth Checklist Collaboration and pilot testing

9/24/2018

7

The BetterBirth Trial

PARTNERS:

Gov ernment of India

Gov ernment of Uttar Pradesh

World Health Organization

Ariadne Labs: A joint center f or health sy stems innovat ion of Brigham and Women's Hospital and Harvard T.H. Chan School of Public Health

Population Serv ices International

Community Empowerment Lab, Lucknow

Jawaharlal Nehru Medical College, Belgaum

The Bill & Melinda Gates Foundat ion

Theory of Change

9/24/2018

8

The BetterBirth Trial

OUTCOMES OF INTEREST:

Adherence to birth pract ices by birth

attendants

Early (7-day ) maternal morbidity, mortality,

& perinatal mortality

PROJECT GOAL: The Bet terBirth Trial is a matched-pair,

cluster-randomized controlled trial seeking to

establish whether the BetterBirth interv ention

is ef f ective in reducing deaths and

complications in inst itutional childbirth in

resource-limited sett ings.

120 Faci li ties (60 control)

24 Dis tricts

Nov. 2014-Dec. 2016

157,145 births (mother-baby pairs)

149 maternal deathsMMR7-day : 94 per 100,000

7,445 perinatal deathsPMR: 47 per 1,000

BetterBirth Trial in Uttar Pradesh, India

9/24/2018

9

BetterBirth Safe Childbirth Checklist Implementation

Included in Govt. of India Maternal

Health Toolkit

EngageBuy-in at district and

facil ity level promotes systemic change

LaunchMotivational event

to introduce the Checklis t and assess facility gaps

CoachingPeer-to-peer model for

uptake of essential birth practices

Data feedback for changeFeedback from data and

observation shared in real time to foster change

Sustainability StrategyCapacity bui lding of

facil ity and dis trict champions to support change beyond program

Collaboration is Key

18

9/24/2018

10

…and Solutions Emerge

Connectivity:Transition from Paper to Electronic Data Capture

20

9/24/2018

11

Field Data Collection

Data Processing in Zambia

22

9/24/2018

12

Data Storage

23

24 http://www.viewsoftheworld.net/wp-content/uploads/2015/07/MobilePho nesInc reaseMaps2001t o201 1.jpg

Access: Mobile Telephone Subscriptions

9/24/2018

13

Pulse System: A Quality Improvement Data Feedback Loop

BetterBirth coaches work with birth attendants to improve adherence to the WHO Safe Childbirth Checklist

1

Data are automatically analyzed and available to the frontline within 24 hours.

3

Behavior monitoring heat maps shared at facility and district level

Reports are used by BetterBirth staff members to promote behavior change 4

Coaches capture behavioral data via smartphone

2

BetterBirth Intervention: Facility Level

Facility Level Heat Map

9/24/2018

14

Monitoring Implementation & Follow Up

Use of Data and Reports to Influence Change

• The Pulse System generates reports for use in data f eedback f or quality improv ement.

• When shared withbirth attendants, hospital leadershipand distr ict leaders, they prompt change to improv e conditions.

28

O A

M

A

9/24/2018

15

Call Center for Outcomes Data Collection

Contact information for mothers is input from clinic registers to smartphone app

These data are automatically sent to the call center, appearing on a call center staff’s desktop screen in Pulse

1 2

Operational and outcomes data can be accessed via automatic and rapidly generating

reports. These reports can be accessed in near to real-time via Pulse

4

Call center staff contacts mother to assess post-discharge

health outcomes (mortality and morbidities)

3

Data QualityAssurance & Protocols Created Success

Monitoring & Evaluation

TeamTraining

Standard Operating Procedures

Electronic Data Collection &

Reporting

Data Quality Assurance:

• Audits• Feedback• Supervision

Data Qual ity Assurance Protocol achieved 98.3%accuracy acrossal l research data-col lection activities in the trial

Fol low up of 99.7%out of >157,000 mothers-newborns

9/24/2018

16

Culture:Theory to on the ground reality

31

Call Center for Outcomes Data Collection

Contact information for mothers is input from clinic registers to smartphone app

These data are automatically sent to the call center, appearing on a call center staff’s desktop screen in Pulse

1 2

Operational and outcomes data can be accessed via automatic and rapidly generating

reports. These reports can be accessed in near to real-time via Pulse

4

Call center staff contacts mother to assess post-discharge

health outcomes (mortality and morbidities)

3

9/24/2018

17

When a Call was Not Possible, Home Visits Made

33

Call Center & Home Visit Consistency

34

93%

54%

92%

Call Center/Call Center Call Center/Home Visit Call Center/RemoteHome Visit

9/24/2018

18

Call center follow up for health outcomes

85%

13%

2%

Call Center Home Visit with Call

Home Visit Only Lost to follow Up

N=157,145

Collaboration

CultureConnectivity

36

9/24/2018

19

Acknowledgements

Thank you to all study participants, birth attendants, women, and their newborns.

Partners:• Government of India (GoI)

• Government of Uttar Pradesh (GoUP)

• World Health Organization (WHO)

• Population Services International (PSI)

• Community Empowerment Lab, Lucknow

• Jawaharlal Nehru Medical College, Belgaum

• The Bill and Melinda Gates Foundation

• The John D. and Catherine T. MacArthur Foundation

• Ariadne Labs at Brigham & Women's Hospital and Harvard T.H. Chan School of Public Health

• BetterBirth Scientific Advisory Board Members

9/24/2018

20

Thank you and Questions!

Contact information:

Katherine Semrau

Director, BetterBirth Program

Email: ksemrau@bwh.harvard.edu

Twitter:

@k_semrau

Intervention sites had a significantly higher adherence rate to practices after 2 months of coaching

40

0% 0% 1%

20%

4%

11%15%

1%

19%

30%

79%

70%

79%

84%

0%

20%

40%

60%

80%

100%

At admission:Partograph

started

Within 1 hour:Skin to skin at 1

hour*

Just beforepushing: Properhand hygiene*

Within 1 minute:Oxytocin

administered*

Within 1 hour:Breastfeeding

ini tiated*

Within 1 hour:Skin to skinini tiated*

Just beforepushing: Cleantowel available*

Perc

ent of

del

iverie

s w

here

pra

ctic

e perfo

rmed

Control Intervention

SMALLEST GREATEST

* p<0.001

9/24/2018

21

0

5

10

15

20

25

30

35

Mat

ern

al m

orb

idit

y (%

)

0

50

100

150

200

250

300

350

400

450

500

7-d

ay m

ate

rnal

mo

rtal

ity

pe

r 1

00

,00

0 b

irth

s

0

20

40

60

80

100

120

Pe

rin

atal

mo

rtal

ity

rate

pe

r 1

,00

0 b

irth

s

7-day health outcomes

41

Maternal Mortali tyPerinatal Mortali ty Maternal Morbidity

were not different across arms

Control Interv ention Control Interv ention Control Interv ention

New Analyses: Context Matters

Prov ider Skills

Community /Patient Factors

Facility Inf rast ructure

Supplies

Leadership and Culture

Surveillance

CommunicationProcesses

Motivation