The effect on morbidity of variability in deprivation and population stability in England and Wales: an investigation at small-area level.

We seek to determine whether variability in deprivation at small area level, and population stability, influence standardised morbidity ratios in England and Wales. A regression analysis was conducted with data from the 1991 British Census, in order to explain variation in morbidity. Both an area deprivation score (for electoral wards) and the within-area variability of deprivation scores were examined as possible determinants of morbidity (self-reported, limiting, long-term illness). Particular attention was focused on a spatially-sensitive measure of the variability of deprivation scores within a wider 'locality'. There was a significant, positive relationship between age-standardised limiting, long-term illness and deprivation. The variation in area deprivation scores within the small areas themselves was also significant and positive. However, the variation in deprivation scores calculated for both an electoral ward and its contiguous neighbours (the locality) was slightly more significant. Areas with higher relative levels of in-migration also had significantly lower standardised morbidity ratios. Multivariate models showed that the deprivation score, the variation in deprivation scores for the broader locality, and the measure of migration, were all significant in combination. Residual analysis showed that many areas in London had lower levels of morbidity than expected, while electoral wards in the coal mining valleys of South Wales had higher levels than expected. We conclude that, for small areas (wards) in England and Wales, morbidity is related to deprivation, variation in deprivation within and surrounding each area, and the proportion of the population that are migrants. Variations in deprivation influence standardised morbidity rates, and policies which widen inequalities will influence health outcomes. Resource allocation based simply on measures of deprivation, which ignore population change within the area and variations in deprivation in the locality, may be inefficient.