Table 1 Overview of differences in modelling approach used by IHME, IBGE and DDM-R for estimating completeness of death registration. Albuquerque and Senna [10], IHME [11] and IBGE [12]
Method | Overview |
|---|---|
GGB | • Requirements: 2 population age structure, death counts by age • Assumptions: no age misreporting, completeness constant by age, closed population or very small migration flows • Basic idea: balancing equation, death rates = birth rates − growth rates • Model as a regression model. Intercept gives the relative coverage from the two-population age structure, slope gives an estimate of relative completeness of death registration |
SEG | • Requirements: 2 population age structure, death counts by age • Assumptions: no age misreporting, completeness constant by age, closed population or very small migration flows, coverage constant across population counts • If we follow a cohort, number of people aged 0 is equals to the number of deaths aged 0 in year t, plus deaths aged 1 in year t + 1 and so on. Use age-specific growth rates and current deaths to estimate future cohort deaths • Completeness is obtained by comparing observed population to population obtained from death counts |
SEG-adj | • Requirements: 2 population age structure, death counts by age • Basic idea: uses GGB to obtain relative coverage from the two-population age structure. Adjust population so they have the same level of completeness. Apply SEG after this adjustment. |
DDM-R | • Provides estimates of the 3 basic death distribution methods • Age segments used to obtain intercept and slope in GGB are given by testing different adjustment and selecting the best fit using RMSE • Same age group is used to estimate SEG and SEG-adj |
IBGE | • In 1980–1991 and 1991–2000 uses the original version of the Growth Balance that assumes stable population • 2000–2010 used GGB, but does a top-down adjustment to the country level to correct subnational level |
IHME | • Documentation suggests that uses all three versions of DDM • SEG 55–80, GGB 40–70, and GGBSEG 50–70 are the best methods that can be currently used to estimate relative completeness of death registration • Documentation is not clear on which age range is used to obtain estimates for Brazil and subnational level. |