How big data can improve women’s health

At the Women Deliver conference in 2010, Melinda French Gates reflected on the challenge of designing solutions around women’s needs instead of areas of expertise. Like many health organizations, the Bill & Melinda Gates Foundation organized its strategies by disease. But “women aren’t preeclampsia or malnutrition or neonatal malaria, said French Gates. “They are human beings.” How could we better reflect that humanity in our work?

Almost 15 years later, the Gates Foundation and others focused on gender equality continue to wrestle with this issue. Organizing work into siloes can lead to inefficiencies, missed opportunities, and a whole that is less than the sum of its parts. For example, a safe childbirth initiative and a financial literacy initiative compete for the same pot of resources, then take two discrete approaches to achieving the same goal: more women and girls who are healthy and empowered. But because the work is done and measured in isolation, the interrelationships that hinder or drive gender equality work remain invisible. No one can pinpoint which of the countless number of interventions, or which combination, works best for whom.

However, it is not clear how to convert a nuanced understanding of the interrelationships in a woman’s life into a coherent set of policies or programs to support her. The more complexity factored into an analysis of the problem, the more daunting the solutions appear. The Gates Foundation and many others have experimented with more holistic, integrated strategies, but they are slower than working in well-established siloes.

Fortunately, thanks to recent advances in data collection, analysis and sharing, better tools are available to navigate these challenges in 2024 than in 2010. It is increasingly possible to generate data that conveys the richness of the human experience while suggesting concrete ways to improve it.

A shift like this will be a massive undertaking. Gender data are woefully underfunded, and the old agenda still isn’t finished. For example, as of June 2022, less than half of the data needed to monitor the gender-specific dimensions of the UN Sustainable Development Goals were available; meanwhile, in 2020, financing for gender data dropped by more than half. This underfunding may help to explain why the burden of many women’s health conditions (such as anemia and post-partum hemorrhage) has been dramatically underestimated. But while researchers continue to work on these basics, the future of gender data should embrace mixed methods, measure impact, and tailor to decision making.

Much of the data the field uses now, such as coverage and facility readiness, describe a type of work, not the lives of the women the work is for. Bringing in ethnographic research can provide a fuller picture. Since 2019, the Gates Foundation has been supporting Project Pathways, which combines quantitative and qualitative data to determine women’s vulnerability. Using variables such as the media that women consume and whether they are free to visit friends and family, Pathways helps governments to design programs and policies targeted at the biggest barriers faced by vulnerable women in their country. The data are still being used to generalize and simplify the complex totality of women’s experiences into a set of archetypes, but mixing data methods makes the generalizations more accurate, inclusive and useful.

If a program is designed, for example, to increase contraceptive use, it is relatively easy to determine with reasonable accuracy and precision whether it worked. But the point of family planning programs is not just to increase contraceptive use but also to make women’s lives better in every way. To move from single outcomes and fuzzy correlations to constellations of causal impact, the field needs to invest in more rigorous studies: longitudinal studies, randomized controlled trials, quasi-experimental studies, and causal models. BRAC’s Ultra-Poor Graduation Initiative in Bangladesh is a powerful example of what this transition can do. BRAC followed participants for several years to see whether they stayed above the poverty line as well as what difference the package of anti-poverty interventions made in other areas of their lives, such as health.

Policymakers want to make data-based decisions, but data generated from academic research do not always give them the information they need on costs and benefits. Governments want to be able to compare their policy and program options head-to-head, even when they don’t seem comparable. For example, mass administration of azithromycin to children in high-mortality settings has been shown to reduce all-cause mortality by up to 20%. The Ultra-Poor Graduation Initiative has a five-year success rate with 95% of participating women lifted out of poverty. Governments should be able to ascertain how much each program would cost to implement in their country, the effect of a child mortality decline on extreme poverty, and the effect of completing an Ultra-Poor Graduation program on child mortality. This kind of evidence can give countries the confidence to implement more integrated, tailored gender equality programs.

Breaking down siloes starts as a philosophy, a way of thinking about the work of gender equality. But to put the philosophy into practice, the field needs better tools. The ongoing data revolution is giving us one: information about women’s lives that is simultaneously complex enough to capture the messiness, and simple enough to act on.