Country estimates of maternal mortality: an alternative model

Area of Science:

• Global Health
• Demography
• Epidemiology

Background:

• Country-level maternal mortality estimates by WHO and UNICEF have faced controversy.
• Previous revisions using modified multivariate models did not resolve these controversies.
• Disagreements persist regarding the higher figures generated by the WHO/UNICEF multivariate modeling approach.

Purpose of the Study:

• To address the controversy surrounding WHO/UNICEF maternal mortality estimates.
• To propose and evaluate an alternative multivariate model using maternal mortality ratios (MMRs) as the dependent variable.
• To provide revised country-level maternal mortality estimates for 1995.

Main Methods:

• Developed an alternative multivariate model using MMR as the dependent variable.
• Incorporated age standardization for demographic and health surveys (DHS) direct sisterhood data.
• Compared the results of the new model with existing WHO/UNICEF estimates.

Main Results:

• The proposed model using MMR as the dependent variable demonstrated smaller root mean square relative errors.
• While global maternal death estimates were similar to WHO/UNICEF figures, significant country-level discrepancies were observed.
• The study highlights conceptual and practical disadvantages of using PMDF, including its dependence on non-maternal deaths.

Conclusions:

• Using MMR as the dependent variable in multivariate models is preferable to PMDF for estimating maternal mortality.
• The alternative model offers more reliable country-specific maternal mortality estimates.
• Further discussion is needed on optimally incorporating DHS direct sisterhood data into these models.