Health-system determinants of declines in maternal mortality in China between 1996 and 2013: a provincial econometric analysis

Abstract Background Maternal mortality in China has decreased strikingly over the past 20 years, and China is one of the few countries achieving Millennium Development Goal 5. Between 1996 and 2013, China substantially strengthened the maternal-health sector: a large number of funds were invested in maternal health, maternal health facilities were substantially improved, and maternal-health professionals received increased training. China also made substantial investments in the overall health-systems reform, including increases in overall health expenditure, universal health insurance coverage, and the number of health professionals per person. All of these health-systems improvements are thought to have contributed to reductions in maternal mortality in China, but which improvements carried the greatest weight were not known. The aim of this study was to examine health-systems determinates of declines in maternal mortality in China. Methods We used a panel dataset of 31 provinces from China between 1996 and 2013 to examine the effects of health-systems factors on maternal mortality. Our data on health systems were obtained from China Statistical Yearbooks, China Health Statistical Yearbooks, and other national and provincial publications. Data on maternal mortality were obtained from the Annual Report of Maternal and Child Health information. We ran fixed-effects regressions using the logarithm of maternal mortality as the dependent variable. The health-system determinants studied were health expenditure, number of hospital beds, human resources, and length of highways. We controlled for income, female literacy, birth rate, and local tax revenue capacity. All data used were publicly available, aggregate secondary data; therefore, this study was exempt from ethics approval. No individual patient data were used, so no informed consent from patients was necessary. Findings Maternal mortality fell from 66 to 14 deaths per 100 000 livebirths between 1996 and 2013 (32 to 8 deaths per 100 000 livebirths in eastern region, 52 to 13 deaths per 100 000 livebirths in central region, and 109 to 23 deaths per 100 000 in western region). Hospital beds per 1000 population (elasticity –0·78 [p=0·023]) and length of highways (–0·31 [p=0·010]) were significantly associated with maternal mortality. Health expenditure and the density of health workers had no effect on maternal mortality.” Interpretation Access to health care is important in accounting for variation in mater