Meta-analysis on maternal deaths and assessing significant factors for heterogeneity between states of India during 2016
Objective of the study was to provide a more precise estimate of maternal mortality (MM), maternal mortality ratio (MMR) and to identify significant factors contributing for heterogeneity between the states in India. “Metaprop” procedure in STATA software, which are specific to binomial data was applied on state wise MM data published by sample registration system (SRS) during 2014-16. An overall MM estimate and potential sources of heterogeneity could be identified using meta-regression. Corrected estimates of MMR by states were compared. SRS published the MM data by 17 Major states. Overall reported MM was 8.8 per 100, 000 women. Estimate obtained by random effect model was 8.3 (95% CI: 5.9-11.1) per 100,000 women. Heterogeneity between states was very high (I2-statistics =91.9%), and egger regression revealed no reporting bias (p=0.672). Meta-regression analysis indicated that the percent women attending full antenatal care (ANC) visits found to be highly significant (p<0.001) for MM with inverse relationship implying that the states with a higher percentage of women with full ANC visits are likely to have lesser MM. While the estimate of MMR by SRS was 130 per 100,000 live births, corrected MMR was 123 (95% CI: 87-164) accounting for 26% reduction from previous estimate 167 obtained in 2013. This paper provided a precious estimate of both MM and MMR adjusted for sampling weight. Further, the importance of either full ANC visits or four ANC visits could be demonstrated for reduction in MMR on achieving the Millennium development goal (MDG) in the country.