A novel methodology for constructing ethnic majority life tables: an analysis of routine data from England and Wales

Abstract

BackgroundLife tables are used to examine mortality by age and sex to calculate life expectancy in a specific population. The use of life tables is of public health interest to produce accurate estimates of mortality for subpopulations. We report work that has produced ethnic life tables for England and Wales. MethodsPopulation data for all 175 434 output areas (OAs) in England and Wales were stratified by age group and sex. OAs are small, socially homogeneous (Office of National Statistics) areas that contain on average 300 people. These population data were matched to the number of deaths by age group and sex occurring in each OA in the years 2000–02. We used 2001 census data to establish the proportion of men and women, grouped into seven age bands, who were black, white, or Asian in each OA, to the nearest 1%, and merged these with mortality and population data. These data were collapsed by age group, sex, and the proportion of each ethnic group in the area (ethnic proportion; 100 groups), which resulted in separate datasets for each ethnic group, containing the number of deaths and the population by age group, sex, and ethnic proportion. We modelled the relation between mortality, age, and ethnic proportion by a multivariable Poisson model with regression splines. We used the predicted age-specific and sex-specific mortality rates for an area that contained 73% black, 98% Asian, or 100% white people to construct ethnic majority life tables for each ethnic group. These corresponded to the maximum proportion of the specific ethnic origin reported in any OA in 2001. FindingsThe model's prediction of overall mortality corresponded closely with existing national life tables. Data at the tails (below 1 year and above 80 years) were unstable and need further investigation, so are excluded from the comparisons reported. The ethnic majority life tables highlighted important differences in mortality between the main ethnic groups: life expectancy from age 1 year was 69·4, 74·7, and 76·1 years for black, Asian, and white men in majority areas, respectively; and 77·7, 79·4, and 80·6 years for corresponding women. Up to age 80 years, the white population had the lowest mortality for both sexes whereas the black population had the highest rates (appendix); for example, at age 60 years, mortality rates for black, Asian, and white men were 23·1, 12·0, and 9·5 per 1000 person-years, respectively; and for women were 10·3, 8·3, and 5·9 per 1000 person-years, respectively. The probability of surviving from age 60 to 80 years also showed a disadvantage for black men at 43·5%, with 53·1% for Asian men and 53·4% for white men. For women, the figures were closer but followed a similar pattern: 61·1%, 65·6%, and 67·2%, respectively (appendix). InterpretationWe adopted a novel ecological method of constructing ethnic majority life tables that quantify differences in mortality between ethnic groups. There will still be diversity within these three broad ethnic groups, and socioeconomic status might also confound these results because it is associated with both ethnic origin and mortality. This factor will be explored further. Future work will also examine the data at the tails of the distributions and assess which life tables perform better in analyses. Inequalities in health outcomes between deprivation groups are well documented, but there has been less quantification of the ethnic differences that can lead to differential access to health care. These ethnic life tables can be used to inform public health planning. FundingFunded by the National Awareness and Early Diagnosis Initiative (NAEDI).