Damien Bricard, Institut de Recherche et Documentation en Economie de la Santé (IRDES) and Institut National d'Études Démographiques (INED)
Carlo G. Camarda, Institut National d'Études Démographiques (INED)
Emmanuelle Cambois, Institut National d'Études Démographiques (INED)
There is a need for routine life tables by socioeconomic status (SES) to monitor social inequalities in life and health expectancies. However, estimating mortality risks by SES requires large population datasets, with variables of social status, linked to vital statistics. Accurate datasets are scarce and samples are usually relatively small. Routine production of LE by SES therefore requires modeling mortality risks with a great variety in the methods and assumptions that can potentially be used. In this study, we use the census sample mortality follow-up to compare the accuracy of four models for estimating LE by SES. We used the French “Permanent demographic sample”. EDP-Men (aged 30-100) are distributed according to 3 educational levels. We use deaths occurred in a given year between 2008 and 2013 for EDM-men who were surveyed once in the 5 preceding years. Four different Models are used to estimate LE at age 65 for each year between 2008 and 2013. Bayesian Information Criterion (BIC) indicates the “best” estimate for each year, in a statistical point of view. In our sample, LE at age 35 was around 45,5 years. It did not progress much over the 2008-2013 period, but with fluctuations. We confirmed the gap between the men in high-educated group and the men in low-educated group, reaching 6 years in 2013. The four models provide different estimates of LE, differences being smaller than 1 year. Model 1 and 4 provides estimates which are closer to the raw data. Further analysis are needed to determine which of the models is accurate to estimate and monitor LE differentials across educational levels. Replication with data for women and using other criteria for SES should bring new elements to formulate some recommendations.
Presented in Session 93. Socioeconomic differentials in mortality