OBJECTIVES: To derive a predictive model based on the morbidity, demographic and socio-economic characteristics of district populations to explain variations in prescribing costs in England. METHOD: Inter-relations between morbidity, demographic, socio-economic, general practice supply characteristics and net ingredient cost per age, sex and temporary resident originated prescribing unit (ASTRO-PU) were explored statistically for 90 districts in England using 1994 cost data. The possibility of mutual inter-relationship between 'supply' and 'demand' was examined; then the associations between a range of factors and prescribing costs were estimated using ordinary least squares regression and the predictive power of the possible models was systematically examined. RESULTS: Whilst there was a relatively weak relationship between the supply factors that were measured, there did not appear to be any reciprocal relationship. Three parsimonious models estimated using ordinary least squares multiple regression techniques based on combinations of permanent sickness, low birth weight and the proportion of general practitioners registered for postgraduate certificate of education were identified. The models explained up to 61% of variation between districts in prescribing costs. CONCLUSIONS: 'Need' and 'supply' characteristics are independently associated with variations in prescribing costs at district level. The negative association between the proportion of general practitioners eligible for postgraduate education allowance and prescribing costs may reflect 'better' prescribing but could not be introduced into a resource allocation formula without introducing perverse incentives. The combination of permanent sickness and low birth weight complement each other by providing a proxy measure of morbidity mostly applicable to adult males (permanent sickness) and mothers (low birth weight being a measure of maternal health). These variables should be considered further for use in the process of allocating resources for prescribing to districts.
OBJECTIVES: To derive a predictive model based on the morbidity, demographic and socio-economic characteristics of district populations to explain variations in prescribing costs in England. METHOD: Inter-relations between morbidity, demographic, socio-economic, general practice supply characteristics and net ingredient cost per age, sex and temporary resident originated prescribing unit (ASTRO-PU) were explored statistically for 90 districts in England using 1994 cost data. The possibility of mutual inter-relationship between 'supply' and 'demand' was examined; then the associations between a range of factors and prescribing costs were estimated using ordinary least squares regression and the predictive power of the possible models was systematically examined. RESULTS: Whilst there was a relatively weak relationship between the supply factors that were measured, there did not appear to be any reciprocal relationship. Three parsimonious models estimated using ordinary least squares multiple regression techniques based on combinations of permanent sickness, low birth weight and the proportion of general practitioners registered for postgraduate certificate of education were identified. The models explained up to 61% of variation between districts in prescribing costs. CONCLUSIONS: 'Need' and 'supply' characteristics are independently associated with variations in prescribing costs at district level. The negative association between the proportion of general practitioners eligible for postgraduate education allowance and prescribing costs may reflect 'better' prescribing but could not be introduced into a resource allocation formula without introducing perverse incentives. The combination of permanent sickness and low birth weight complement each other by providing a proxy measure of morbidity mostly applicable to adult males (permanent sickness) and mothers (low birth weight being a measure of maternal health). These variables should be considered further for use in the process of allocating resources for prescribing to districts.