OBJECTIVE: To explore the possibility of using routine Hospital Episode Statistics, census data, and vital statistics to derive weights for an equitable capitation formula for setting general practitioner fundholding budgets for buying acute hospital services. DESIGN: Analysis of a routine dataset of 9 million hospital episodes in 1991-2, extracting elective general practitioner fundholding procedures, combined with 1991 census variables, vital statistics, and data on supply of health care at ward level. Costs were attached to each procedure according to the average cost of the relevant "Mersey" band category. MAIN OUTCOME MEASURES: Variation in age and sex adjusted expenditure per head on fundholding procedures across wards modelled for the impact of health and social needs variables after adjusting for variations in supply. RESULTS: No sensible simple model including determinants of use other than age and sex could be derived. The most parsimonious but statistically acceptable model showed that though standardised mortality ratio and self reported illness and several social class variables were associated with utilisation, the signs and the size of the coefficients were contradictory. The most important explanation of variation was provided by age and sex differences between wards. CONCLUSIONS: An equitable system of setting general practitioner fundholders' budgets is needed. In the short term age and sex weighted capitation should form the principal basis of fundholder budgets. Utilisation data at ward level are inadequate for developing a formula which adequately adjusts for the differences in the health care needs of populations. A capitation formula based on information derived from individual cohort data may be the only means of promoting equity and efficiency and of avoiding discriminating against patients with known high cost health problems.
OBJECTIVE: To explore the possibility of using routine Hospital Episode Statistics, census data, and vital statistics to derive weights for an equitable capitation formula for setting general practitioner fundholding budgets for buying acute hospital services. DESIGN: Analysis of a routine dataset of 9 million hospital episodes in 1991-2, extracting elective general practitioner fundholding procedures, combined with 1991 census variables, vital statistics, and data on supply of health care at ward level. Costs were attached to each procedure according to the average cost of the relevant "Mersey" band category. MAIN OUTCOME MEASURES: Variation in age and sex adjusted expenditure per head on fundholding procedures across wards modelled for the impact of health and social needs variables after adjusting for variations in supply. RESULTS: No sensible simple model including determinants of use other than age and sex could be derived. The most parsimonious but statistically acceptable model showed that though standardised mortality ratio and self reported illness and several social class variables were associated with utilisation, the signs and the size of the coefficients were contradictory. The most important explanation of variation was provided by age and sex differences between wards. CONCLUSIONS: An equitable system of setting general practitioner fundholders' budgets is needed. In the short term age and sex weighted capitation should form the principal basis of fundholder budgets. Utilisation data at ward level are inadequate for developing a formula which adequately adjusts for the differences in the health care needs of populations. A capitation formula based on information derived from individual cohort data may be the only means of promoting equity and efficiency and of avoiding discriminating against patients with known high cost health problems.