Interpreting the new illness question in the UK census for health research on small areas

S Martin, T A Sheldon, peter smith

Research output: Contribution to journalArticlepeer-review


Study objective - The study aimed to identify the various factors that seem to influence the average response to the new census question on limiting, long standing illness at the small area level, to assess the extent to which the new question adds to information already available in the census and elsewhere, and to discuss how useful the data are likely to be for those planning health and social services.

Design - This was a cross sectional analysis of the relationship between rates of limiting, long standing illness (standardised for age and sex) and a large number of indicators of health and socioeconomic status at the small area level.

Setting - The study used data relating to 4985 small areas covering the whole of England. The average population was about 10 000.

Participants - The 1991 census of population was addressed to the entire population of England.

Main results - There are wide variations in the levels of self reported long standing illness between small areas, 70% of which are explained by demographic factors. Variation in age/sex standardised responses to the new census question at the small area level can largely be explained by census data on self reported disability among those of working age, standardised mortality ratio, and by indicators of socioeconomic circumstances relating to social class, ethnicity, and the elderly Living alone. There does not seem to be a significant reporting bias due to unemployment.

Conclusions - Unlike the disability question in the census, the standardised, self reported long standing Limiting illness ratio covers the entire population and it is not skewed towards men. Although the variable is a synthesis of the health and social determinants of perceived morbidity, it does not provide much information that was not already available. In addition, it is available every 10 years only and thus may be rather inaccurate as an indicator of relative need towards the end of the decade. Moreover, in future censuses, individuals' answers might be influenced by the knowledge that their responses will affect the volume of resources allocated to the area in which they live.

Original languageEnglish
Pages (from-to)634-641
Number of pages8
JournalJournal of epidemiology and community health
Issue number6
Publication statusPublished - Dec 1995

Cite this