RT Journal Article SR Electronic T1 OP63 Inequalities in general health in England and Wales: an observational study using individual-level census 2021 microdata of 3 million individuals JF Journal of Epidemiology and Community Health JO J Epidemiol Community Health FD BMJ Publishing Group Ltd SP A91 OP A92 DO 10.1136/jech-2024-SSMabstracts.190 VO 78 IS Suppl 1 A1 Cheetham, N A1 Steves, C YR 2024 UL http://jech.bmj.com/content/78/Suppl_1/A91.2.abstract AB Background Associations between social factors and self-rated general health are well-established, and self-rated general health is a strong predictor of morbidity and mortality. However most studies of ‘social determinants’ of health are limited either in breadth of factors available to analyse, or depth in terms of small samples not representative of national populations.Methods We analysed England and Wales Census 2021 microdata, containing a 5% nationally representative sample of census responses (3.0 million individuals). A combination of multivariable poisson regression and matched pair case-control analyses were used to estimate associations between social factors and self-rated general health, reported on a 5-point scale from very bad to very good. Factors tested covered themes of early life, housing, employment and family. Analyses were stratified by age, focusing on adolescent (0-17 years old), transition to adulthood (18-25), working age (26-64), and retirement age (65+) populations. Analyses were designed using a causal inference framework and included extensive adjustment (poisson regression) or exact matching within pairs (case-control analysis) for potential confounders as appropriate for the different age ranges.Results Inequalities in self-rated general health were found in all themes, with large disparities of note found for people living in socially rented housing vs. owning with mortgage (Relative risk ratio, RR, of very bad health = 4.78, 95% CI: 4.59-4.97; case-control analysis mean difference = 0.48 points worse health than controls on 5-point ordinal scale, among pairs matched exactly on confounding factors), and those who cannot speak English vs. English as a main language (RR of very bad health = 4.23, 95% CI: 3.84-4.65, case-control analysis mean difference = 0.63 points worse health than controls). Such disparities were comparable to the difference seen between people aged 80+ years vs. 40-49 years (RR of very bad health = 4.69, 95% CI: 4.50-4.88). Detailed age-stratification revealed how inequalities varied with age, e.g., the average health deficit of social renters from case-control matched analysis increased with age from 0.35 points (on 5-point ordinal scale) among 25-29 year olds to 0.76 points for 60-64 year olds, before declining in older age groups.Discussion Health inequalities based on detailed early life factors, housing, employment and family life were found in a nationally representative sample. Large-scale individual-level administrative census data provides a powerful resource for understanding health inequalities in detail and identifying groups with greater need for health care services.