Variance in practice emergency medical admission rates: can it be explained?

BACKGROUND Emergency admission rates have been rising steadily in recent years, with the majority of the increase owing to emergency medical admissions. Possible causative factors include changing demography, incidence of disease, admission thresholds, multiple admissions, and appropriateness of admission. AIM To investigate the impact of patient and practice factors on variance in general practices' emergency medical admissions rates. DESIGN OF STUDY Multiple regression analysis relating emergency medical admission rates of general practices to patient and practice characteristics. SETTING The study was conducted between 1996 and 1997 in the acute hospital trust serving the study area, Dundee, Scotland. METHOD Scottish Morbidity Record 1 (SMR1) data, which contains details of all hospital consultant episodes, was used to calculate individual practices emergency medical admission rates. These rates were then standardised to an expected value of 100. Forward selection was used to find a suitable multiple regression model to predict each practice's emergency medical admission rate from practice and patient variables. RESULTS Crude emergency medical admission rates for general practices showed a 1.8-fold variation between the top and bottom deciles. The deprivation status and age of patients explained 42% of the variance in admission rates (64% with the exclusion of one practice that had a poor fit to the model). After correcting for age and deprivation there was a 1.2-fold variation in general practices' emergency medical admission rates. CONCLUSION The most important factors in explaining variance in general practices' emergency medical admission rates are socio-demographic, with age and deprivation explaining a large proportion of the variation. We found no evidence to support the contention that general practice factors were linked with admission rates.