A promising method to estimate maternal mortality for Pakistan’s healthcare system
Reliable maternal mortality estimates are needed to plan and monitor interventions. Yet often births and deaths are under-recorded and regular surveys are too costly. Can a promising new mortality estimation technique help solve this problem for Khyber Pakhtunkhwa province and beyond?
A lack of good data makes it hard to assess whether maternal health is improving
All too often, when a woman in Pakistan dies due to pregnancy or childbirth it remains a private tragedy. Maternal deaths are often not recorded by the authorities and with each omission, a profound loss stays hidden from the institutions that might have prevented it.
Estimates from the 2006-07 Demographic and Health Survey suggest that maternal mortality is comparatively high. But these figures are now ten years old. Decision making in the Pakistani health system has recently been devolved from the federal level and there is little up-to-date information about patterns at the sub-national units where policies are now designed and implemented.
DFID Team Leader for Health and Nutrition, Chris Athayde, has been working together with the provincial governments and various partners to increase the availability of ante- and postnatal care in Punjab and Khyber Pakhtunkhwa. Although the collaboration has produced many positive outcomes, he points out that it is difficult to draw firm conclusions about the overall picture in relation to maternal health. “The lack of good mortality data means that we are not in a position to fully assess the impact of our collective efforts”, he explains. “At the very least we need some mechanism to understand whether the situation is changing.”
What experts like Chris really need, are regular and reliable maternal mortality data at the provincial and district levels. However, creating this evidence base is much easier said than done.
Survey estimates are of limited use to policy experts
Pakistan currently has no plans to make death registration mandatory and, like many countries where vital events are under-recorded, the best information about maternal mortality has tended to come from household surveys. However, large samples are required in order to create reliable mortality estimates and this means that survey data on maternal deaths are usually too costly and time-consuming to collect on a regular basis.
Other relatively cost-effective strategies exist to generate estimates from smaller samples. For example, the Sisterhood Method works by asking a sample of respondents to answer questions about the survival and pregnancy status of all adult sisters born to the same mother. Gathering information about all of the respondent’s adult sisters reduces the need for a large survey sample. However, approaches like this have other data collection weaknesses that can bias their estimates and limit their overall applicability.
The country’s maternal mortality statistics clearly need updating. But to provide timely and reliable data at the appropriate administrative levels, Pakistan needs a new measurement strategy every bit as much as it needs new estimates.
Using community informant networks to estimate maternal mortality
In response to the lack of reliable data in many developing countries, the Initiative for Maternal Mortality Programme Assessment at the University of Aberdeen has developed a promising new technique that uses community informant networks to collect information about maternal mortality.
The Maternal Death from Informants (MADE-IN) / Maternal Death Follow-On Review (MADE-FOR) begins by identifying suitable informant networks such as religious leaders, community health workers or traditional birth attendants. The informants are then asked to create a list of women of reproductive age who have died within in their communities and a “listing meeting” is held to review the results and to check for any inaccuracies or duplicates.
Having established the list, follow-up interviews are carried out with relatives of the deceased to check whether the death was from pregnancy-related causes. These interviews also establish the cause and circumstances of death; and collect information about socio-economic status and health-seeking behaviour. Where more than one informant network is used, or where there are other data sources available, the results are cross-checked with the alternative sources to increase accuracy. The final data are then used to estimate overall maternal mortality and to enable researchers to investigate the circumstances and causes of maternal death.
By relying on community members to identify deaths, MADE-IN/MADE-FOR is cheaper and less time-consuming to implement than a survey. The method aims to record all maternal deaths and, as a consequence, it also avoids the concerns about sampling error that are associated with other maternal mortality estimation techniques like the Sisterhood Method.
A viable and cost-effective method to identify and investigate maternal deaths
In early 2014 the Population Council received funding from the UK Department for International Development (DFID) and the Australian Department of Foreign Affairs and Trade to test the feasibility of the MADE-IN/MADE-FOR method in Pakistan. Focusing on the Chakwal district in Punjab province, the study successfully created district maternal mortality ratios and identified several informant networks at the community level. Data collection was then scaled-up to six more districts and used to estimate the provincial maternal mortality ratio for Punjab.
As with similar pilots in Indonesia and Somaliland, the Chakwal study found that MADE-IN/MADE-FOR is capable of generating precise estimates of maternal mortality down to the community level. Given the costs required to implement the method, the study also concluded that it would be feasible to carry out MADE-IN/MADE-FOR on a twice-yearly basis.
Piloting the method in two RMNHP districts
The Reproductive, Maternal and Newborn Health Project (RMNHP) implemented by Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) GmbH on behalf of the German Federal Ministry for Economic Cooperation and Development (BMZ) aims to strengthen the health system and improve the quality of healthcare for mothers and children. At the dissemination seminar for the Punjab Maternal Mortality Study in February 2015, GIZ committed to support the replication of MADE-IN MADE-FOR in Khyber Pakhtunkhwa province. This initiative complements other RMNHP activities (see, for example, ‘Implementing the WHO Safe Childbirth Checklist in Pakistan’) and contributes to the projects overall objective. RMNHP has partnered with the Population Council and the method was successfully piloted in the Nowshera and Haripur districts in the central KP region between August 2015 and January 2016.
This work identified a range of different community informant networks, including Lady Health Workers, Lady Health Visitors, religious leaders, Nikkah registrars and male and female councillors. To minimize recall bias, informants were asked only to report deaths that had occurred within the last two years. Under-reporting of maternal deaths was offset by comparing the results of two different informant networks in each district and the so-called “capture-recapture” technique was used to adjust the ratios to take account of any deaths that might have been overlooked. (“Capture-Recapture” matches the deaths captured by both informant networks and then uses this information to estimate the proportion of all deaths covered by each network. The number of deaths identified by the network are then adjusted to take account of its estimated coverage.)
Overall, the pilot study was successful in producing adjusted and unadjusted maternal mortality ratios for Nowshera and Haripur, as well as collecting a wealth of data on the causes and circumstances of these deaths. The results were then discussed with the local government authorities and presented at community meetings involving local health professionals and key informants including religious leaders and Nikkah registrars.
Creating a sustainable data collection mechanism for the whole province
Under a co-financing agreement with DFID data collection is now being scaled-up to include two districts in northern KP (Mansehra & Swabi) and two districts in southern KP (D.I. Khan & Kohat), so as to get representative maternal mortality ratio estimates for the entire province. MMR is the most important indicator of Reproductive Health and Dr Ali Ahmad, KP’s Director of Health Services, knows that the estimates are needed: “Reliable MMR estimates at provincial level are essential for planning, monitoring and evaluating of maternal healthcare interventions. The government has initiated many new projects for maternal and child health. New MMR data will guide us about the effectiveness of these programs and help us plan new interventions.”
MMR estimates will indeed help to characterize the current state of maternal health in KP. However, to design and monitor policies effectively it is necessary to establish a data collection mechanism that can also provide evidence about the trends in maternal mortality over time. To this end, the project in KP places a particular emphasis on working with District Health Officers and local statistical experts to enable the maternal mortality estimates to be generated routinely on a sustainable basis.
Dr Ali Mohammad Mir is Associate and Director Programs at the Population Council in Pakistan and Principal Investigator for the MADE-IN/MADE-FOR studies in Punjab and KP. He is particularly enthusiastic about this increased emphasis on capacity building: “It’s important that the provincial government has ownership”, he stresses. “Our ambition is that the method continues to be implemented using the knowledge that we have passed on”. But a sustainable system for obtaining reliable mortality estimates is not all that there is to gain from implementing MADE-IN/MADE-FOR technique in KP.
An estimation method and also an advocacy tool
Dr Mir points out that MADE-IN/MADE-FOR is a valuable advocacy tool as well as a cost-effective and precise method of generating subnational estimates. Indeed, presenting the results to the community creates an opportunity not only to raise awareness about the number of maternal deaths but also the reasons.
“One of the main advantages of this method is that it involves community members in discussions about issues they were not aware of” he remarks. “Local leaders are able to appreciate that many of the causes of mortality can be prevented through early diagnosis and simple interventions”.
The results from the pilot in Nowshera and Haripur have already sparked an animated debate about maternal health within the community. In particular, the data found that most pregnancy-related deaths occurred to women in low- or medium socio-economic status and that the families of the deceased considered the cost of accessing treatment to be prohibitively high.
In this way, implementing MADE-IN/MADE-FOR has not only established a sustainable mechanism to collect data on maternal deaths, it has also helped to draw the private suffering experienced by many families out into the public sphere where it can be acknowledged and addressed. And with data collection now extended to include four new districts, experts in KP will soon have both the information and the impetus necessary to design provincial-level policies that can save more women’s lives.