[Identification of doctors with high pharmaceutical expenditure]

[corrected] To identify doctors with high pharmaceutical expenditure through multiple regression of variables not exclusively dependent on the doctor's decision. To compare the agreement with the method of the Spanish national health servicem which only considered the percentage of pensioners of the doctor's list.

Cross-sectional, retrospective, descriptive study of the pharmaceutical expenditure of 220 general practitioners in the health district of Toledo (Spain) in 1999. The following variables were collected: pharmaceutical expenditure for policy-holder/month (PEP-H); age; sex; habitat; size of the list; percentage of pensioners; frequency of attendance (consults/policy-holder-year) and workload (consultations per day). A multiple regression model with backwards elimination was constructed, taking like PEP-H as the dependent variable and the remaining as the independent variables. The resultant equation enabled calculation of the expected PEP-H for each doctor and the deviation of their real expenditure ofrom the expected. Doctors were considered to have high pharmaceutical expenditure when the deviation was more than the mean plus 1 standard deviation of this distribution.

The mean PEP-H was 2,584.4 pesetas. Differences were found (F = 11.665; p < 0.005) in PEP-H per habitat (2,723.2 in rural, 2,521.4 in semi urban and 2,168.2 in urban). A significant correlation was found (p < 0.005) between PEP-H and percentage of pensioners (r = 0.728) and frequency of attendance (r = 0.607). Our final model included percentage of pensioners, frequency of attendance, and age (F = 102.33; p < 0.005; r = 0.767; r² = 0.588; β = 206.05; β₁ = 48.27; β₂ = 61.26; β₃ = 9.55). This model were identified 25 DHPC. With the INSALUD model the identification were of 31. The Kappa index showed that agreement between both methods was 0.706 (SE 0.056) and simple disagreement was found in the classification of 24 doctors (10.9%).

Frequency of attendance and the percentage of pensioners on doctors' lists are iclosely related to PEP-H. The indicator currently used by the Spanish national health system to identify doctors with high pharmaceutical expenditure could be improved if other variables, such as frenquency of attendance and the doctors' age, were taken into account. The adoption of this method would make budget allocation more impartial and fair.