OBJECTIVES: To show the advantages and disadvantages of a multi-dimensional small area classification in the analysis of child health data in order to measure social inequalities in health and to identify the types of area that have greater health needs. DESIGN: Health data on children from the district child health information system and a survey of primary school children's height were classified by the census enumeration district of residence using the Super profiles neighbourhood classification. SETTING: County of Northumberland, United Kingdom. SUBJECTS: One cohort comprised 21,702 preschool children age 0-5 years resident in Northumberland, and another cohort 9930 school children aged 5-8.5 years. MAIN OUTCOME MEASURES: Variations between types of area in the proportions of babies with birthweight less than 2.8 kg; births to mothers aged less than 20 years; pertussis immunisation uptake; child health screening uptake; and mean height of school children. RESULTS: Areas with the poorest child health measures were those which were most socially disadvantaged. The most affluent areas tended to have the best measures of health, although rural areas also had good measures. Problems in analysis included examples of the "ecological fallacy", misleading area descriptions, and the identification of the specific factors associated with poor health measures. Advantages included a wider view of social circumstances than simply "deprivation" and the ability to identify characteristic types of areas with increased child health needs. CONCLUSIONS: There is a limited place for multidimensional small area classifications in the analysis of health data for both research and health needs assessment provided the inherent drawbacks of these data are understood in interpreting the results.
OBJECTIVES: To show the advantages and disadvantages of a multi-dimensional small area classification in the analysis of child health data in order to measure social inequalities in health and to identify the types of area that have greater health needs. DESIGN: Health data on children from the district child health information system and a survey of primary school children's height were classified by the census enumeration district of residence using the Super profiles neighbourhood classification. SETTING: County of Northumberland, United Kingdom. SUBJECTS: One cohort comprised 21,702 preschool children age 0-5 years resident in Northumberland, and another cohort 9930 school children aged 5-8.5 years. MAIN OUTCOME MEASURES: Variations between types of area in the proportions of babies with birthweight less than 2.8 kg; births to mothers aged less than 20 years; pertussis immunisation uptake; child health screening uptake; and mean height of school children. RESULTS: Areas with the poorest child health measures were those which were most socially disadvantaged. The most affluent areas tended to have the best measures of health, although rural areas also had good measures. Problems in analysis included examples of the "ecological fallacy", misleading area descriptions, and the identification of the specific factors associated with poor health measures. Advantages included a wider view of social circumstances than simply "deprivation" and the ability to identify characteristic types of areas with increased child health needs. CONCLUSIONS: There is a limited place for multidimensional small area classifications in the analysis of health data for both research and health needs assessment provided the inherent drawbacks of these data are understood in interpreting the results.