Individual morbidity and neighbourhood deprivation in a non-metropolitan area

The aim was to replicate, in a non-metropolitan area, a study by Curtis based on data from different parts of London which found a significant relationship between individual morbidity and neighbourhood deprivation.

This study used the same design as the previous study. Information on individual morbidity was obtained, using the Nottingham health profile. Deprivation scores were assigned to respondents according the Jarman (and also Townsend) scores of the enumeration district in which they lived. Logistic regression models were built, using the enumeration district as the unit of analysis, to see if, after allowance for age and sex, the Jarman (or Townsend) score significantly improved the prediction of the enumeration district being above or below the sample median to any of the six dimensions of the profile.

The survey involved households in 10 electoral wards in Colchester and Clacton, Essex, United Kingdom, in 1988.

A systematic sample of 200 persons was drawn from the electoral register in each of 10 wards yielding 2000 names.

Nottingham health profiles were obtained from 1555 respondents out of an initial sample of 2000 names from the electoral register; the response rate among those alive and resident at a valid address was 93%. Jarman and Townsend scores were assigned to 1496 of the respondents, and the 162 enumeration districts were used as a unit of analysis. Contrary to Curtis's finding, Jarman score did not add significantly in a multiple logistic regression model to the prediction of response to any of the six dimensions of the Nottingham health profile, nor did the Townsend score.

Apart from chance variation, there are two possible explanations for this finding. The measures of deprivation may be valid in London but not elsewhere; or there may be a true difference between cities and towns in the effect that deprivation has on subjective health.