Regional variations in pharmaceutical consumption in Sweden

Secondary, non-sample data, mainly from official statistics from 259 municipalities in Sweden, are used to investigate systematic variations in the consumptian of pharmaceuticals on the county and municipality level. The theoretical framework is taken from health economics. The utilization of health care, and consequently the consumptian of pharmaceuticals, is determined simultaneausly by the demand for health care, derived from a demand for health, and the supply of health care. The model is exploratory, aiming at the identification of important factors for further study. In particular the importance of the economic budget restriction is examined. An analysis of variance shows that there is no reason to assume systematic differences in the consumptian of pharmaceuticals caused by actions on the county level. The existing variation more likely stems from differences in characteristics of the municipalities, thus reflecting intovariations on the county level. The impact of a number of explanatory variables on drug consumptian is measured by means of multiple regression analyses. The variables included are sick-leave, disability pension, unemployment, age, alcohol consumption, religiousness, occupational structure, commuting, physician density, rural population, presence of a hospital or a regional hospital and income. The regression results indicate that age structure, the number of doctars and commuting explain approximately one third of the variation. Detailed regression studies of the consumption of 19 pharmaceutical subgroups show a more diversified result, but the main pattern remains. Using predictians from the 19 subregressions, a case study is performed for Gotland, known as a divergent municipality with respect to drug consumption. Two pharmaceutical groups, cardiovasculars and analgesics, were identified as particularly divergent campared to the general consumptian behaviour of the municipalities of Sweden. An analysis using the LISREL technique is performed in order to study the relative importance of the latent, unmeasured variables 'morbidity', 'life style', 'availability' and 'economy', grouped from the measured variables. The analysis indicates that 'availability' is the most important factor underlying pharmaceutical consumptian and that 'econorny' is more important than 'morbidity' and 'life style'. The analysis also shows, that low income is associated with high consumption of drugs. Whether this relationship on the municipality leve! also holds on the individual levelis not known.