This blog post summarizes Roland Mady’s presentation during the 2025 annual meeting, which presented new tools and methods for assessing institutional maternal mortality, stillbirth, and neonatal mortality before discharge.  Mr. Mady is a research associate at the Johns Hopkins Bloomberg School of Public Health.

In many countries, increasing levels of facility births mean that routine health facility data can be a valuable data source for mortality estimation.  These data are also increasingly relied upon in the context of declining survey frequency and constrained funding for population-level data collection.

Why Institutional Mortality Data Matters

Institutional mortality data drawn from routine health management information systems like DHIS2 are a core source of evidence for national health monitoring. However, underreporting remains a widespread issue, especially for maternal deaths. Factors such as fear of punitive measures and inconsistent documentation practices contribute to data gaps, undermining the utility of mortality indicators for policy and planning.

Therefore, transparency and practical interpretation are important factors. It is important to interrogate the data’s plausibility, consistency, and representativeness.

Understanding Institutional Mortality Metrics

The following three approaches can be used to evaluate data quality:

1. Plausibility of Institutional Mortality Rates

After calculating institutional maternal mortality ratios (MMRs) and stillbirth rates (SBRs), it is important to be aware that extremely high or low values—particularly those deviating by 50% or more from a five-year median, or falling below thresholds observed in high-income countries—are flagged as potential outliers requiring further review or imputation. For instance, SBRs below 6 per 1,000 or MMRs below 25 per 100,000 suggest potential underreporting or data quality issues

2. Ratio of Stillbirths to Maternal Deaths

The expected institutional ratio of stillbirths to maternal deaths typically falls between 5 and 15. Ratios outside this range can signal reporting imbalances—such as underreporting of maternal deaths relative to stillbirths or vice versa.

• A ratio greater than or equal to 15 may indicate that maternal deaths are being underreported more severely than stillbirths.
• A ratio between 5 and 15 can suggest one of two things, depending on the plausibility of overall mortality levels (indicator 1): either both maternal deaths and stillbirths are underreported, or both are reported with reasonable quality.
• A ratio below 5 implies that maternal deaths may be better reported than stillbirths.

By examining this ratio in tandem with overall mortality plausibility, country teams can better interpret discrepancies and identify specific gaps in death reporting systems.

3. Completeness of Death Reporting

The third indicator compares the observed MMR from facility data with an expected MMR, calculated based on two key assumptions:

• An estimate of the population-level MMR, such as UN estimates or alternative sources like high-quality surveys;
• An assumed ratio of community to institutional maternal mortality, typically ranging from 0.5 to 2.5, reflecting the expected relationship between deaths occurring inside and outside health facilities.

Scenarios can be created to estimate likely completeness ranges. In one example, with 80% institutional birth coverage and a reported MMR of 100, the estimated completeness of maternal death reporting ranged from 59% to 64%, depending on the assumptions applied.

Moving Forward: Data for Accountability and Action

Strengthening the quality and use of routine data is essential for informing responsive, evidence-based health programs. By diagnosing where and why mortality data fall short, country teams are better equipped to target interventions and improve the accuracy of reporting systems over time.

The institutional mortality module promotes country-led analysis and capacity building. As countries face growing pressure to “do more with less,” such approaches will be increasingly vital to achieve progress on maternal and child health outcomes.

• Download the full presentation here
• Access the Shiny app and Github repository