Ken Hill1, Cynthia Stanton2, Yoonjoung Choi2, and Kevin Thomas3. (1) Harvard University Initiative for Global Health, 104 Mount Auburn Street, Cambridge, MA 01238, 617-495-8231, email@example.com, (2) Johns Hopkins Bloomberg School of Public Health, Bloomberg School of Public Health, Baltimore, MD 21505, (3) Harvard Initiative for Global Health, 104 Mount Aurbun Street, Cambridge, MA 02138
The Millennium Development Goal of improving maternal health sets targets for reducing national maternal mortality ratios (MMR) from 1990 to 2015 in the developing world. However, data constraints have made it impossible to measure progress: the MMR is inherently difficult to measure.
Three exercises to make global and national estimates of MMR have been undertaken by WHO, UNICEF and UNFPA. However, these exercises were strictly cross-sectional and did not draw any conclusions about trends. The bases for these exercises were validated national estimates of MMR for dates ranging from 1980 to 2000. To explore possible trends further, we pool these data bases and apply random and fixed effects models that include independent variables representing time to 113 estimates of MMR for 73 countries for reference dates between 1987 and 2000. We also introduce control variables for proportions of births with a professional attendant, for national per capita income, for the level of fertility, and whether the country is in sub-Saharan Africa or not.
Across these countries, we find a small but significant decline in MMR of between 2.5% and 3.3% per annum. However, further exploration reveals that this decline is entirely limited to countries with moderate (<=200) initial MMR; countries with high initial MMR show no trend. Introduction of the control variables into the models indicated that the trends could be entirely accounted for by the controls.
The countries included in this exercise represent 75% of the births in the developing world. However, the trend estimates from the random effects models could be distorted by differential selectivity over time, while the fixed effects models could be distorted by selectivity on good performers. Since deriving valid estimates of MMR for all developing countries is still well in the future, analyzing trends in more easily measured process indicators represents a more viable short term strategy for monitoring maternal health.
Keywords: Maternal Health, Maternal and Child Health
Presenting author's disclosure statement:
Any relevant financial relationships? No
The 134th Annual Meeting & Exposition (November 4-8, 2006) of APHA