Death audits and reviews for reducing maternal, perinatal and child mortality

Background

The United Nations' Sustainable Development Goals (SDGs) include reducing the global maternal mortality rate to less than 70 per 100,000 live births and ending preventable deaths of newborns and children under five years of age, in every country, by 2030. Maternal and perinatal death audit and review is widely recommended as an intervention to reduce maternal and perinatal mortality, and to improve quality of care, and could be key to attaining the SDGs. However, there is uncertainty over the most cost‐effective way of auditing and reviewing deaths: community‐based audit (verbal and social autopsy), facility‐based audits (significant event analysis (SEA)) or a combination of both (confidential enquiry).

Objectives

To assess the impact and cost‐effectiveness of different types of death audits and reviews in reducing maternal, perinatal and child mortality.

Search methods

We searched the following from inception to 16 January 2019: CENTRAL, Ovid MEDLINE, Embase OvidSP, and five other databases. We identified ongoing studies using ClinicalTrials.gov and the World Health Organization (WHO) International Clinical Trials Registry Platform, and searched reference lists of included articles.

Selection criteria

Cluster‐randomised trials, cluster non‐randomised trials, controlled before‐and‐after studies and interrupted time series studies of any form of death audit or review that involved reviewing individual cases of maternal, perinatal or child deaths, identifying avoidable factors, and making recommendations. To be included in the review, a study needed to report at least one of the following outcomes: perinatal mortality rate; stillbirth rate; neonatal mortality rate; mortality rate in children under five years of age or maternal mortality rate.

Data collection and analysis

We used standard Cochrane Effective Practice and Organisation of Care (EPOC) group methodological procedures. Two review authors independently extracted data, assessed risk of bias and assessed the certainty of the evidence using GRADE. We planned to perform a meta‐analysis using a random‐effects model but included studies were not homogeneous enough to make pooling their results meaningful.