Group-based programs to improve health outcomes in India: An accumulating body of evidence but questions remain

There is a strong global movement to scale-up various types of programs based on women's groups, both to empower women economically and to use the potential of women's groups to deliver other interventions. In India the movement has taken a strong hold, with the launch and scale-up of the National Rural Livelihoods Mission and several other associated programs. Many of these programs operate in poor states, or are focused on the poor even in richer states.

There is accumulating evidence that health/nutrition interventions delivered via the platform of women's groups are effective, as in the paper published in this issue of EClinicalMedicine. In addition, our own work and that of others, recognizes that pathways to achieving this impact are potentially long and complex [1]. The paper by Hazra and colleagues [2] in this issue certainly adds to that body of evidence, and demonstrates that benefits accrue to the poorest/most left behind. The interventions tested are ambitious in the anticipated outcomes they aim to shape – i.e., mortality. Given the real life nature of the program that was being rolled out, the quasi-experimental evaluation is designed in the context of this large-scale program, using sampling approaches to ensure that equity findings could be examined. The findings on equity are powerful.

The authors use sampling approaches to create a reasonable counterfactual, and gather data on several covariates to address confounding. However, some limitations in the published paper are worth keeping in mind. First, allocation bias remains a concern, especially given an emphasis on group maturity in the intervention areas. Given the scale at which this intervention took place, it seems like a missed opportunity to not have addressed this allocation bias using a randomized or stepped-wedge evaluation design as the program was rolling out. Often the onus of rigorous evaluation is placed on the researchers working on the evaluation; however, as noted by the authors, allocation of the intervention (random or not) was not a decision they made. The only solution to this challenge in other similar research would be to work closely with implementation teams early in the design process to establish random allocation or a better-matched counterfactual. A second potential challenge is that the potential pathways from the interventions to the outcomes are not explicitly laid out. It is not clear, therefore, whether the outcomes are because of increased knowledge, increased use of services, increased demand for better services, or other pathways. Finally, little is known about the supply-side of maternal and child health interventions, and the extent to which there were any changes in these that might have been induced by increasing accountability and demand.

Some lingering questions that remain for us in thinking about the application of these results in this context and beyond:

First, whose responsibility is it, ultimately, to help improve outcomes as serious as mortality? Surely, relying on behaviour change by the women themselves, is inadequate to solve the problem entirely. There are a range of supply side factors related to the health system that need serious attention, including the overall quality of maternal and child care in this context. In addition, several societal barriers related to gender, mobility and the position of women are likely not changeable by the women themselves. In situations where health services are poor and patriarchy is deep-seated, how much of the onus of improving outcomes should be on women?

Second, what are the opportunity costs for women of spending time in these groups? Are there others more effective and even more equitable ways to achieve the same outcomes, such as serious efforts at improving the supply-side services such that every woman and every child receive high quality services as a right? The costs of women's group programs, both from the point of view of implementation and from the point of view of opportunity costs for women, are important to consider in this, and other, contexts.

Third, in a rapidly urbanizing context, what is the long-term potential for group-based programs that are designed primarily on a rural development framework? Are there different ways to explore the concept of collective action and peer/social support in the urban context? There are promising studies examining the role of collectivization and community mobilization in improving outcomes among female sex workers in urban areas [3], [4], [5]. There are, however, only a handful examining maternal and newborn health outcomes, with modest outcomes [6]. Creating opportunities to learn from across urban and rural group-based interventions could potentially help catalyse better outcomes.

In closing, we congratulate the authors – this specific piece of work is a reminder that women's groups have the potential not just to improve outcomes in general, but when designed with a strong equity lens, they also have the potential to improve outcomes for those who are typically left behind. These findings have important implications for the plethora of women's group programs around the world, especially in poor and rural communities in Asia and Africa. At the same time, we also ask that implementers, funders and governments examine fundamental questions about why poor women should be in a position of having to improve their own lives instead of living in societies that value their health and invest in providing services that contribute to better outcomes.

Declaration of competing interest

Drs. Menon and Kumar have no conflicts of interest.