The reliability of maternal audit instruments to assign cause of death in maternal deaths review process: a systematic review and meta-analysis

Background: Maternal Death Reviews (MDR) can assist in formulating prevention strategies to reduce maternal mortality. To support MDR, an adequate MDR instrument is required to accurately identify the underlying causes of maternal deaths. We conducted a systematic review and meta-analysis to determine the reliability of maternal death instruments for conducting the MDR process. Method: Three databases: PubMed, ProQuest and EBSCO were systematically searched to identify related research articles published between January 2004 and July 2019. The review and meta-analysis involved identification of measurement tools to conduct MDR in all or part of maternal audit. Eligibiliy and quality of studies were evaluated using the Modified Quality Appraisal of Diagnostic Reliability (QAREL) Checklist: Reliability Studies. Results: Overall, 242 articles were identified. Six articles examining the instrument used for MDR in 4 countries (4 articles on verbal autopsy (VA) and 2 articles on facility-based MDR) were included. None of studies identified reliability in evaluation instruments assessing maternal audit cycle as a comprehensive approach. The pooled kappa for the MDR instruments was 0.72 (95CI:0.43–0.99; p < 0.001) with considerable heterogeneity (I2 = 96.19; p < 0.001). Subgroup analysis of MDR instruments showed pooled kappa in VA of 0.89 (95CI:0.52–1.25) and facility-based MDR of 0.48 (95CI:0.15–0.82). Meta-regression analysis tended to show the high heterogeneity was likely associated with sample sizes, regions, and year of publications. Conclusions: The MDR instruments appear feasible. Variation of the instruments suggest the need for judicious selection of MDR instruments by considering the study population and assessment during the target periods. © 2021, The Author(s).