Use of pharmacoeconomic data for formulary decision making in teaching hospitals

A national mail survey of Pharmacy Directors in teaching hospitals was performed to determine the criteria and applicability of the use of Pharmacoeconomics (PE) data for formulary decisions, to identify the determinants of PE-based formulary decisions, and to assess the quality of available PE data. Of the 166 (39.6%) hospitals responding to the survey during April through July, 1999, 83.1% reported using PE-based formulary decisions. The two significant determinants of whether hospitals used or did not use PE data were census and annual pharmacy drug budgets. The two most important criteria for PE data use were the impact of a new medication on institutional costs and the anticipated annual budget for a new medication.;To determine the degree of use of PE-based formulary decisions, principal components analysis was employed to reduce and classify survey-items for a PE-index. PE-index scores were calculated by summing the PE-index item scores. A stepwise multiple regression showed that five determinants jointly explained 49.8% of the variance (Adjusted R2 = 46.7%) of the PE-index scores [F (7,115) = 16.3, p < 0.001]. Hospitals with higher cost-savings from using PE data, higher FTEs, higher number of staff types responsible for PE analyses, and ranking of anti-infective drugs as the most important drug class in requiring PE data, were associated with an increase in PE-index scores. "Hospitals located in the South" was associated with a decrease in PE-index scores. Hospitals reporting a higher degree of comfort if PE data were reviewed and approved by the FDA expected an increase in PE data use.;The study findings suggest that the hospitals use PE data based on cost impact to their institutions. Cost-savings from PE-based formulary decisions would motivate hospitals to devote more resources to PE evaluations and increase the use of PE data. The use of PE data from pharmaceutical companies might increase if the FDA reviewed and approved such data.