Diffusion of new drug technologies will be studied in a patient- level Medicaid database from a state (New Hampshire) with and without (New Jersey) cost-containment policies. Diffusion will be analyzed using nonparametric, survival methods. These methods will be used 1) to identify physician and patient characteristics predictive of first-use of a new drug technology; 2) to model these descriptive characteristics in order to quantify their relative importance in explaining indicators of diffusion; and 3) to examine the effect of cost-containment policies on new technology diffusion. Continued-use (adoption) of new drug technologies will also be studied and related to diffusion analyses. For these studies, parametric regression models will be employed. To explain use of new technologies over time, regression models will be constructed using descriptive physician and patient characteristics. These models for continued-use of new technologies will be compared with those for diffusion of these new products. Using a times- series regression model, this research will examine the effect of cost-containment on use of new drug technologies over time. The patient-level database to be employed in these studies contains over twenty million prescription drug claims submitted t Medicaid for reimbursement over the years 1980-1984. Demographic and other important descriptive variables are included in the claims files, allowing individual recipients to be tracked over time. These characteristics of the database allow analysis of subgroups of individuals, both patients and physicians, as well as studies in variations of diffusion and continued-use of new technologies across local regions. The very large size of the database allows for a high level of precision in analysis. Results of the analysis will be shared with state Medicaid officials as they examine cost-containment initiatives in reducing rates of spending.
