Presentation at the Global Maternal Newborn Health Conference, October 21, 2015
Background: Accurately measuring maternal mortality is critical for designing mortality reduction interventions and reaching maternal health goals. Differential reporting of causes of facility-level maternal deaths often mask differences in maternal mortality magnitude. Enhanced detection methods can strengthen facility-based reporting and identify data sources in greatest need of improvement.
Methods: Facility deaths were retrospectively identified and reviewed using a Rapid Ascertainment Process for Institutional Deaths (RAPID) methodology. RAPID detects all institutional deaths among women of reproductive age and reviews those that are pregnancy related. RAPID-identified maternal deaths were compared with those identified in the Heath Management Information System (HMIS) reproductive health (RH) reports. Confidence intervals were calculated for the total number of maternal deaths by source, and for the difference in deaths reported between the two sources. Further, the proportion of maternal deaths due to different causes from RAPID was compared with global WHO estimates.
Results: RAPID identified 53% more maternal deaths in 2008-2012 than the HMIS reports (rapid: 427, 95% CI 386–468; RH report 280, 95% CI 247–313; difference 147, 95% CI 95–199). Obstetric hemorrhage (21%), obstructed labor/uterine rupture (OL/UR) (18%), pregnancy-induced hypertension (PIH) (10%), sepsis (11%), and abortion (15%) were the main direct obstetric causes identified by RAPID. Hemorrhage (32%), sepsis (17%), and PIH (15%) were the main causes in the RH reports, where fewer abortion (5%), and OL/UR deaths (9%) were reported. Indirect obstetric causes were similarly reported (26% vs 22%). The RAPID cause of death distribution resembled the WHO global cause of death estimates (hemorrhage 25%, indirect causes 20%, and infections 15%).
Conclusions: HMIS reporting in Kigoma underestimates maternal mortality. Enhanced detection of facility-based maternal mortality of all causes is urgently needed to design effective interventions to reduce maternal deaths.