Estimating the completeness of death registration: An empirical method

Area of Science:

• Public Health
• Demography
• Biostatistics

Background:

• Accurate death registration completeness is crucial for national and subnational governments.
• Existing methods (e.g., death distribution, capture-recapture) suffer from inaccuracy and complexity.
• There is a need for a novel, widely applicable method to measure death registration completeness.

Purpose of the Study:

• To present a novel empirical method for estimating death registration completeness.
• To enable routine measurement at national and subnational levels.
• To overcome limitations of existing methodologies.

Main Methods:

• Developed random-effects models using 2,451 country-years (110 countries, 1970-2015) from the Global Burden of Disease 2015 database.
• Utilized predictors: registered crude death rate, under-five mortality rate, population age structure, and under-five death registration completeness.
• Models were stratified by sex (male, female, both sexes).

Main Results:

• Models reliably predict completeness across diverse crude death rates (R-squared 0.85).
• Predicted completeness shows close agreement with observed data for populations outside the training set.
• Method demonstrates flexibility and minimal impact of under-five mortality rate uncertainty on national estimates.

Conclusions:

• The novel method is flexible, applicable with readily available subnational data.
• It can assess completeness from various sources (health facilities, censuses, surveys).
• This approach offers significant potential for widespread government adoption, overcoming limitations of prior methods.