OBJECTIVES: To analyze the accuracy of an AMCP budget impact model designed to forecast the first-year use of a new osteoporosis therapy in a managed care/pharmacy benefit setting and describe characteristics of patients who initiated therapy. METHODS: Following AMCP guidelines, a model was developed to predict first-year (2003) use and treatment costs for a new osteoporosis therapy (teriparatide) using plan characteristics (total number of insured lives, age and sex distribution) as model inputs. Using administrative claims data from an employer database (n=3.5 million), predicted versus actual use and treatment costs were compared for eight health care plans of varying types (indemnity, PPO, POS, HMO). Demographic characteristics and medical histories of patients initiating the new therapy were compared to other osteoporosis patients in the database (ICD-9 CM = 733.0x). RESULTS: The model predicted a total of 131 patients would initiate therapy; the actual number was 133. The number was overestimated in four plans with the margin of error ranging from +113% (34 predicted vs. 16 actual) to -30% (40 predicted vs. 57 actual). The difference in predicted versus actual treatment costs ranged from -$19,527 to +$103,315. Patients initiating the new therapy were on average older compared to other osteoporosis patients (P<0.001). In the year prior to initiating the new therapy, these patients also had more outpatient, inpatient, office, and prescription drug claims per month and were more likely to have seen a specialist or have had an osteoporotic fracture compared to other osteoporosis patients (P<0.001). CONCLUSION: AMCP-style models for newly approved therapies can provide important insight to health plan administrators when making formulary decisions during the initial year of therapy availability. Updating such models by incorporating inputs based on patient characteristics and actual first-year use of therapy may yield more accurate estimates of future use and treatment costs.