Background: Between 1990 and 2006, China reduced its under-five mortality rate (U5MR) from 64.6 to 20.6 per 1000 live births and achieved the fourth United Nation's Millennium Development Goal nine years ahead of target. This study explores the contribution of social, economic and political determinants, health system and policy determinants, and health programmes and interventions to this success.
Methods: For each of the years between 1990 and 2006, we obtained an estimate of U5MR for 30 Chinese provinces from the annual China Health Statistics Yearbook. For each year, we also obtained data describing the status of 8 social, 10 economic, 2 political, 9 health system and policy, and six health programmes and intervention indicators for each province. These government data are not of the same quality as some other health information sources in modern China, such as articles with primary research data available in Chinese National Knowledge Infrastructure (CNKI) and Wan Fang databases, or Chinese Maternal and Child Mortality Surveillance system. Still, the comparison of relative changes in underlying indicators with the undisputed strong general trend of childhood mortality reduction over 17 years should still capture the main effects at the macro-level. We used multivariate random effect regression models to determine the effect of 35 indicators individually and 5 constructs defined by factor analysis (reflecting effects of social, economic, political, health systems and policy, and health programmes) on the reduction of U5MR in China.
Results: In the univariate regression applied with a one-year time lag, social determinants of health construct showed the strongest crude association with U5MR reduction (R(2) = 0.74), followed by the constructs for health programmes and interventions (R(2) = 0.65), economic (R(2) = 0.47), political (R(2) = 0.28) and health system and policy determinants (R(2) = 0.26), respectively. Similarly, when multivariate regression was applied with a one-year time lag, the social determinants construct showed the strongest effect (beta = 11.79, P < 0.0001), followed by the construct for political factors (beta = 4.24, P < 0.0001) and health programmes and interventions (beta = -3.45, P < 0.0001). The 5 studied constructs accounted for about 80% of variability in U5MR reduction across provinces over the 17-year period.
Conclusion: Vertical intervention programs, health systems strengthening or economic growth alone may all fail to achieve the desired reduction in child mortality when improvement of the key social determinants of health is lagging behind. To accelerate progress toward MDG4, low- and middle-income countries should undertake appropriate efforts to promote maternal education, reduce fertility rates, integrate minority populations and improve access to clean water and safe sanitation. A cross-sectoral approach seems most likely to have the greatest impact on U5MR.