Geographic variation in the utilization of new prescription drugs: A diffusion perspective

The objectives of this study were to describe the extent of variation in the utilization of innovative prescription drugs across geographic areas over time, and to explain geographic variation in the utilization of the drugs across geographic areas using a diffusion framework. Prescription drugs claims data for the Iowa Medicaid program were used. The new drugs of interest selected were selective serotonin reuptake inhibitors, HMG-CoA reductase inhibitors, and second-generation sulfonylureas. The number of claims for new drugs in each therapeutic class per eligible Medicaid enrollee and the proportion of total claims in each therapeutic class accounting for new drugs were calculated by year for each of the 99 counties. Coefficients of variation were calculated to measure the extent of geographic variation in each utilization measure. Multiple regression models for cross-sectional data and for combined cross-sectional and time series data were used to explain geographic variation in the utilization of prescription drugs per capita and the proportion of claims for new drugs. Explanatory variables in the model included measures reflecting characteristics of physicians, Medicaid beneficiaries, and residents in each county.;The descriptive analysis demonstrated that there was considerable variation in the utilization of new prescription drugs across counties. The extent of geographic variation was greater in the use of more recently introduced drugs and diminished markedly over time. These results suggest that the extent of geographic variation in the utilization of new drugs is related to the diffusion of new and innovative drugs.;The regression analysis showed that geographic variation in the utilization of new prescription drugs was explained in part by some county characteristics. The success of the regression model in explaining geographic variation, however, was limited. A substantial portion of geographic variation remained unexplained. None of independent variables were significant across drug classes for each time period. The interaction effects between some county characteristics and time were significant, suggesting that the effects of county characteristics on the utilization of new prescription drugs varied over time.