Analyses confirm effect of women's groups on maternal and newborn deaths

Abstract

In their insightful Comment on our meta-analysis testing the impact of participatory women’s groups on maternal and neonatal mortality, Betty Kirkwood and Rajiv Bahl suggest that including stratifi ed results from the MaiMwana trial might have led to erroneous conclusions. We re-ran the meta-analyses using MaiMwana’s factorial results (maternal mortality odds ratio 0·48, 95% CI 0·26–0·91; neonatal mortality 0·85, 0·59–1·22). With all seven trials the results indicate a 34% reduction in maternal mortality (0·66, 95% CI 0·39–0·93), and a 20% reduction in neo natal mortality (0·80, 95% CI 0·68–0·93). These results lend support to the conclusion that exposure to groups is associated with signifi cant reductions in maternal and neonatal deaths. Studies on change mechanisms and the role of context are critical to understand the eff ect and generalisability of community inter ven tions. Much of the heterogeneity in the metaanalysis was explained by diff erences in the proportion of pregnant women participating in groups. The existence of a critical threshold for participation only became clear after most of the trials included in the meta-analysis were underway, but this has been applied to more recent initiatives. For example, recent work with groups in three districts of eastern India has achieved coverage of over 60% of pregnant women. With strong evidence from seven large, rigorous trials, the metaanalysis supports a review of safer motherhood and newborn care policy, especially in rural areas where mortality rates remain high. Participatory women’s groups have the potential to enhance the eff ect of home visits and essential obstetric care on mortality. We declare that we have no confl icts of interest.