Estimating Time Trends in Maternal Mortality for All Countries: A Revised Modeling Approach

Leontine Alkema, National University of Singapore
Doris Chou, World Health Organization (WHO)
Doris Ma Fat, World Health Organization (WHO)
Alison Gemmill, University of California, Berkeley
Daniel Hogan, World Health Organization (WHO)
Colin Douglas Mathers, World Health Organization (WHO)
Ann-Beth Moller, World Health Organization (WHO)
Lale Say, World Health Organization (WHO)
Sanqian Zhang, National University of Singapore

Estimates of the maternal mortality ratio (MMR) are used to track progress in reducing maternal deaths and to evaluate countries' performance related to MDG 5. Given limited data availability and data quality issues for the majority of developing countries, the UN Maternal Mortality Estimation Inter-agency Group obtains estimates for all countries without vital registration systems using a multilevel regression model. While the model works well to assess MMR levels for countries with limited data, it does not provide insights into observed trends in countries where longer time series of observations have become available; estimated trends are covariate-driven and not informed by data-driven trends. In this paper, we propose a revised MMR estimation model, which is a combination of a multilevel regression and a time series model, such that estimates are data-driven for countries with sufficient information. Preliminary results suggest that the revised model provides informative MMR estimates.