The effect of Self-Help Groups on access to maternal health services: evidence from rural India

The nature of SHG activities, where members meet regularly for transactions and training, creates solidarity and social capital. Social capital is built on features of social organization, such as trust, social norms and networks, that can improve the efficiency of society by facilitating coordinated actions[24]. The concept of social capital is further split into three connecting strands: bonding social capital (i.e. ties between immediate family members, neighbours and close friends); bridging social capital (i.e. ties between people from different ethnic, geographical, and occupational backgrounds); and linking social capital (i.e. ties between poor people and those in positions of influence in formal organizations such as banks and schools)[25]. SHGs, which bring village women together in a common organization for mutual support, are mobilized by existing bonding social capital, and then build linking social capital as the group members get involved in activities[26].

Several studies have found an association between social capital, generated from participation in microfinance, SHG, and participatory women’s groups on diverse health behaviours and health outcomes, as well as reducing inequity.

Globally there is emerging evidence to show that microfinance programmes have created non-financial benefits including improvements in health, hygiene and sanitation[19, 27, 28]. In post-tsunami Sri Lanka, a study using retrospective panel data from 350 randomly selected borrowers showed that microfinance loans provided after the disaster were instrumental in reducing the income gap between those who were hit and those who were not[29]. One analysis of a large dataset from three waves of the Indonesian Family Life Survey, showed a positive effect on changes in children’s health as MFI members were twice as likely to live in urban areas, have sewerage systems, regular garbage collection, electricity and better access to medical facilities[30]. Studies in India and Bangladesh have shown the positive effect of SHGs on reducing exclusion[31], improved childcare and contraceptive use[32, 33].

In Maharashtra state, a project that trained women SHG members as health workers, initiated literacy programmes and provided funds for household health emergencies showed in the two decades after 1970 a reduction in infant mortality from 176 to 19 per 1000, a birth-rate decline from 40 to 20 per 1000, nearly universal access to antenatal care, safe delivery, and immunization, and a decline in rates of malnutrition from 40% to less than 5%[34, 35].

A clustered randomized trial was conducted to assess the impact of a community mobilization programme through participatory women’s group among the indigenous communities of Jharkhand and Odisha states of India. The trial found newborn babies born to mothers associated with a women’s group significantly improved the likelihood of surviving within the first six weeks of their lives, compared to babies born to analogous households in control communities[36].

While available evidences, that includes rigorous randomised controlled trials and other rigorous methods, suggests the positive effect of social capital in reducing income gap, exclusion, saving newborn, and gender disparity in access to healthcare, Nayar[22] noted that most of the success stories from India are from large organization that incorporate self-help activities as part of other concurrent interventions. There has not previously been an assessment in India using nationally representative data.

The third round of the District Level Household Survey (DLHS-3), a national health survey conducted in 2007–08 in 601 districts of India included a question on the presence of a SHG in its village level questionnaire. Using this data, we are able to analyze the effect of SHGs on the knowledge and practices of women. This data provides the best available opportunity to analyze the influence of the presence of SHGs on women’s RCH knowledge and practices on a national scale. The results of our analysis are reported in this paper.

Pitt[38] identified three sources of bias in estimating cause-effect relationships: choice-based sampling, individual heterogeneity bias, and village heterogeneity bias. To address unmeasured individual and village attributes that affect both programme participation and health outcome, we instituted some controls. For individual heterogeneity we controlled for: respondent education (illiterate, primary, middle and higher secondary and above), wealth quintile, heard or seen health messages; and for village level heterogeneity we controlled: accessibility of Community Health Centre/Rural Hospital (CHC/RH), villages with any beneficiaries of JSY in last one year, and health and sanitation committee in village. Choice-based sampling is addressed by the sample size, and the national nature of the survey that can tease out the contribution of self-help groups independent of other concurrent activities or the organizational infrastructure. Table 1 shows the number and percentage distribution of responses by selected characteristics.

We computed forward stepwise logistic regressions adding different levels of control variables to a base model that regress our four outcome variables (institutional delivery, feeding colostrum, knowledge of family planning, and ever used family planning) on the availability of SHG in the village. We used a fixed effect unique to a district that captures the time-invariant differences across districts. For each of the four outcome variables, three models were estimated. In Model 1, the effect of presence or absence of a SHG in the village was modelled. This model represented the total variance in the four outcome variables with the presence or absence of a SHG. In Model 2 only individual level control factors (respondent education, work status, and heard or seen health messages) were included. In Model 3 individual and village background controls: village electrification, education facility available in the village, village connected through an all-weather road, distance from nearest hospital, beneficiaries of JSY in last one year, and health and sanitation committee were included. All models use survey weights to account for sample design and population weighting and standard errors are adjusted for clustering at the district level.

The focus of the analysis was the change in the coefficient of the presence of a SHG. The results are shown as odds ratios (ORs) with 95 per cent confidence intervals (CIs). The magnitude of the change was interpreted as the (exponentiated coefficient – 1.0) x 100. The small variance inflation factor of 1.09 (not reported) indicated the absence of any significant co-linearity between explanatory variables in the regression model.

As per DLHS-3 data, 57.9 per cent of Indian villages have a self-help group (Figure 1). The majority of these groups are located in southern and north-eastern India, followed by Maharashtra, Chhattisgarh, Rajasthan and Madhya Pradesh.

The descriptive statistics (Figure 2) show some interesting findings on the four measures of women and child health knowledge and practices. The overall use of family planning was found to be very low. The presence of a SHG has a positive and strong correlation with all four measures of knowledge and practices. Compared to households in villages without a SHG, households in villages with a SHG are more likely to go for institutional delivery, more likely to feed new-borns colostrum, and more likely to have knowledge of and use family planning products and services. Members engaged in self-help activity feel a sense of connectivity and discuss issues ranging from place of delivery to feeding the baby and family planning products and services.

As discussed in the methods sections earlier, we present our results on four measures of maternal and child health knowledge and practices: institutional delivery; feeding new-born colostrum; knowledge about family planning services; and ever used family planning.

Predictor of institutional delivery (Table 2): The presence of a SHG in a village is associated with 19 per cent higher odds of mother’s delivering in an institution (CI: 1.13 – 1.24), holding other variables constant. The reduction in odds from 1.30 to 1.19 in presence of individual and village level controls indicates the influence of other factors affecting the outcome. Model 2 adds individual control. The coefficients of individual control variables illustrate that mother’s education, wealth and having heard or seen health messages are important mediating pathways to influence institutional delivery. At the village level, the presence of a health and sanitation committee in the village, accessibility of CHC/RH (OR: 1.16), and beneficiary of JSY (OR: 1.30) are important mediating pathways that influence institutional delivery.

Predictor of feeding colostrum (Table 3): The presence of a SHG in a village is associated with 8 per cent higher odds of an increase in colostrum feeding. Mother’s education, wealth, and having heard or seen health message are important individual level mediating pathways, while being a beneficiary of JSY (OR: 1.32) is an important village level mediating factor predicting colostrum feeding. The presence of a health and sanitation committee in a village or accessibility of a CHC/RH does not appear to influence the outcome.

Knowledge about Family Planning ( Table 4): Households in villages with a SHG are at 48 per cent higher odds of knowing at least one modern family planning method. Model 1 produced an odds ratio of 2.13, indicating the strong influence of having heard or seen health messages in knowledge generation about family planning. More educated and wealthy women are more likely to have knowledge of family planning. Accessibility of a CHC/RH, and having been a beneficiary of JSY are village level variables influencing the outcome.

Ever-used family planning (Table 5): Presence of a SHG is associated with 19 per cent higher odds of ever using family planning. Women’s literacy does not show a positive association with use of family planning, suggesting the lack of empowerment and decision making on reproductive choice. Wealth status, and heard or seen health message are important individual level factors, while accessibility of CHC/RH, and beneficiary of JSY are important village level mediating pathways influencing use of family planning.

As information on women’s actual participation in SHG activities was not included in the DLHS-3 dataset, our analysis provides an instructive but partial picture of the impact of SHGs on health outcomes. There are a number of limitations to our study. First, we did the analysis at the aggregate country level. This masks variations in the spread and intensity of SHG activity across India, as depicted in Figure 2 above. Secondly, the presence of a SHG in a village could only partially explain the level of activity. The level of women’s participation in a SHG[39], the availability of credit[40] and the duration of association[31, 41] are other key predictors of health outcomes that the DLHS-3, due its limited scope and intent, did not address. Thirdly, we also did not find within the DLHS-3 manual an explicit definition of SHGs, or any distinction between the possible impact of SHGs and other women’s or community groups. The fourth limitation relates to the design and nature of the DLHS-3, including its reliance on self-reported information from respondents and the cross-sectional nature of the survey, as described by Jat et al.[42]. The survey collected the responses to the questionnaire only at their face value and had no opportunity to probe. Hence we could examine only the association between explanatory variables and four indicators of maternal health services uptake; we were not able to draw conclusions about causality. Nonetheless, ours is the first attempt to analyse these issues using a nationally representative dataset. Using this large national level dataset allowed us to address two important biases revealed in previous studies: choice based sampling[38] and teasing out the contribution of SHGs within the organizational infrastructure[22].