Assessing the impact of selective migration and care homes on geographical inequalities in health - a total population cohort study in Sheffield
Ravi Maheswaran (University of Sheffield)
Friday 7th March, 2014 15:00-16:00 Maths 204
Selective migration and moves to care homes may potentially contribute to observed socioeconomic gradients in mortality across cities and regions. Sheffield has striking socioeconomic gradients in area-level mortality across the city. We examined for evidence of selective migration and assessed the contribution of migration to observed mortality gradients. We used a total population cohort (539737 in 2001), linked mortality data (2001-2010) and linked data from a health survey carried out in 2000 (66% response rate yielding 10185 responses). We used lower super-output areas and electoral wards as the spatial units of analysis. We found clear evidence of selective migration. In the 25-44 age band, relative risks of mortality were 1.71 (95% CI 1.37-2.12) in migrants from low to high deprivation areas compared with people remaining in low deprivation areas, and 0.53 (0.42-0.65) in migrants from high to low deprivation areas compared with people remaining in high deprivation areas. Relative risks shrank towards unity with increasing age. Characteristics of migrants and non-migrants (illness prevalence, indicators of socioeconomic status, smoking prevalence) ascertained before migration were largely consistent with the relative risks for mortality. There was also a clear care homes effect, with higher mortality in electoral wards with higher care home bed provision rates. Overall, however, adjustment for selective migration, which included moves to care homes, made little difference to gradients in inequality across the city. Our results suggest that selective migration, including moves to care homes, do not explain existing socioeconomic gradients in area level mortality across Sheffield. However, selective migration can have complex effects including paradoxically decreasing socioeconomic gradients.