Women's groups that meet monthly to discuss, identify and address factors that contribute to poor maternal and neonatal health may improve birth outcomes in rural, low- resource settings—at least in some circumstances. A pair of randomized trials conducted in India and Bangladesh that examined the potential benefits of establishing such groups yielded mixed results.^1,2 In the Indian study, the odds of neonatal mortality were lower in areas with participatory women's groups than in control areas without groups (odds ratio, 0.68), particularly during the final two years of the three-year study (0.55). No decrease in maternal depression—an outcome assessed only in the Indian trial—occurred for the study as a whole, although the odds of moderate depression were reduced in the women's group areas during the final year (0.43). In the Bangladeshi trial, however, no improvements in neonatal or maternal outcomes occurred in areas with women's groups, a finding that may reflect the lower density of groups in the Bangladeshi program.

The trials were inspired by an earlier study in which a women's group intervention in rural Nepal produced neonatal mortality rates 30% lower than those in comparison areas. To examine whether this approach could be replicated in other low-resource settings, one follow-up trial was organized in three contiguous districts of Bangladesh (Bogra, Faridpur and Moulavibazar), and another in two Indian states (Jharkhand and Orissa). The trials, which were launched in 2005 and lasted about three years, were broadly similar in format. In each country, the study areas were divided into clusters, half of which were randomly chosen to serve as intervention areas. In Bangladesh, researchers met with leaders of 451 villages to obtain permission to establish women's groups in the intervention areas, and trained facilitators visited every 10th household to invite married women of reproductive age to join; in India, 172 existing women's groups (involved in savings and credit activities) were invited to participate in the intervention, and 72 additional groups were established. In both countries, groups met monthly and began by discussing the reasons for neonatal and maternal deaths and the problems that women face before, during and after delivery; at subsequent meetings, strategies for addressing these problems were discussed and, if possible, implemented. Groups used a variety of approaches, both in discussing problems (storytelling and role playing were frequently used) and in implementing solutions. Participatory group meetings were not held in the control clusters, although in these areas (as in the intervention areas) efforts were made to improve medical referral systems and to refresh providers, knowledge of neonatal and maternal care.

The studies, primary outcomes were neonatal mortality (death of a live-born infant within 28 days of birth) and, in the Indian trial, maternal depression; secondary outcomes included early and late neonatal mortality (deaths occurring within 1–6 and 7–28 days of birth, respectively), stillbirths, maternal mortality (pregnancy-related deaths within 42 days of the end of pregnancy) and home-care practices during and after delivery. To monitor outcomes and collect demographic information, trained informants (generally traditional birth attendants) identified all births, neonatal deaths and maternal deaths in the region, and women or family members were interviewed about six weeks after delivery; supervisors conducted "verbal autopsies" with mothers in cases of neonatal death and with family members in cases of maternal death. In India, maternal depression was assessed using a 10-item screening scale and classified as mild, moderate or severe. Analyses were by intention-to-treat.

Baseline surveys in the intervention and control areas of Bangladeshi women who had given birth in the year before the study revealed that most were Muslim (81–88%), aged 20–29 (60–65%) and had no more than a primary school education (76–82%). Fewer than half had use of a sanitary latrine (32–46%). Only 34–36% of the women had received any formal antenatal care during their last pregnancy, and relatively small proportions of deliveries had occurred in a medical facility (7–9%) or been supervised by a trained birth attendant (15–18% of home deliveries). The Indian study areas were impoverished as well: About three-quarters of women who had had births in the past year belonged to scheduled castes or scheduled tribes (73–78%), and similar proportions were illiterate (70–78%). Most women had obtained some prenatal care during their pregnancy (59–69%), but the majority of the deliveries had occurred at home (83–86%), generally without a birth attendant (61– 63%).

During the ensuing three years, data were collected on 18,775 births in India and 36,113 births in Bangladesh. Because pregnant women in Bangladesh often travel to their mother's home for the delivery, more than 10% of the interviewed women in that country were temporary residents of the relevant districts; they were included in the mortality analyses but not the analyses of secondary outcomes, which were restricted to the 30,952 births to permanent residents.

In India, the neonatal mortality rate was lower in intervention areas than in control areas (42 vs. 59 per 1,000 live births). After adjustment for clustering and baseline differences between regions, the odds of neonatal death were lower in the areas with women's groups than in the control areas (odds ratio, 0.68), especially during the last two years of the study (0.55). Similar reductions were apparent in early neonatal mortality (0.62) and perinatal mortality (0.79), but not in late neonatal mortality or stillbirth.

Overall, women in the intervention and control areas did not differ in rates of postpartum depression. However, during the third year—by which time more than half of the mothers had joined a women's group—the proportion of mothers with moderate depression was lower in the intervention areas than in the comparison areas (5% vs. 10%; odds ratio, 0.43). Rates of mild and severe depression did not differ between areas.

Finally, the researchers found changes in rates of beneficial home care practices. During the study's final two years, several such practices—including use of soap by the birth attendant, use of a safe delivery kit and use of a boiled thread to tie the umbilical cord—were utilized more often in intervention areas than in control areas (odds ratios, 2.3–4.3).

Unfortunately, the results were not as encouraging in Bangladesh, where only 9% of women aged 15–44 were group members by 2007. The neonatal mortality rate was lower than in India (34–38 per 1,000 live births), but it did not differ between the intervention and control areas, even in the later stages of the trial. Moreover, no differences were apparent in other neonatal outcomes, and maternal mortality was elevated in the intervention areas—a result the researchers suspect was a chance finding, in part because none of the deceased women had belonged to a women's group.

The authors of the Indian study suspect that the reduction in neonatal mortality in the intervention areas was due to the higher rates of hygienic practices in those regions; forthcoming analyses of the verbal autopsies may yield a more definitive explanation. Social support and group problem-solving may have led to the reduction in moderate maternal depression seen in intervention areas toward the end of the study. Overall, women's groups such as the ones in this study "have the potential to create improved capability in communities to deal with the health and developmental difficulties arising from poverty and social inequalities," the researchers state.

In the Bangladeshi trial, on the other hand, the intensity of the intervention may have been too low to provide benefits: One women's group was established for every 1,414 residents, compared with one per 468 residents in the Indian trial and one per 756 residents in the earlier Nepalese trial. Moreover, group facilitators in Bangladesh had more groups to supervise than did their counterparts in the other two trials. The researchers note that future studies should shed light on the factors, including population coverage, that influence the effectiveness of using women's groups to influence maternal and neonatal outcomes; in fact, an "intensive scale-up" of the intervention is under way in Bangladesh.

—P. Doskoch

REFERENCES

1. Tripathy P et al., Effect of a participatory intervention with women's groups on birth outcomes and maternal depression in Jharkhand and Orissa, India: a cluster-randomised controlled trial, Lancet, 2010, 375(9721): 1182–1192.

2. Azad K et al., Effect of scaling up women's groups on birth outcomes in three rural districts in Bangladesh: a cluster-randomised controlled trial, Lancet, 2010, 375(9721):1193–1202.