Pharmacoeconomic modeling of prior-authorization intervention for COX-2 specific inhibitors in a 3-tier copay plan.

OBJECTIVE To determine from a health plan perspective the cost-effectiveness of cyclooxygenase-2 (COX-2) specific inhibitors, with and without a prior-authorization (PA) process. METHODS A modeling exercise was employed, based on prescription drug claims for a managed care organization with 3.8 million health maintenance organization (HMO) and preferred provider organization (PPO) members. Drug claims revealed 96154 members (2.9% of the 3.3 million members with a pharmacy benefit) who received either one or more prescriptions for a COX-2 drug or a nonspecific nonsteroidal anti-inflammatory drug (NSAID). These patients were stratified into 2 groups for further analysis, those having a concurrent proton pump inhibitor (PPI) and those without a concurrent PPI. Decision analysis was used to estimate the cost-effectiveness of COX-2 therapy. Actual health plan drug claims data were used to determine utilization and prescribing patterns of nonspecific NSAIDs, COX-2 specific inhibitors, and PPIs. Results from the literature from 8 clinical trials were employed to determine the probability of a serious gastrointestinal (GI) event. Cost-effectiveness analysis (CEA) was used to determine the cost of each therapy, including the predicted cost to treat a serious GI event in a drug benefit design with PA versus a benefit design without PA. RESULTS Cost-effectiveness analysis (CEA) showed that the cost per success no serious GI event) for Cox-2 specific inhibitors with PA was US dollars 278 versus US dollars 422 without PA. CONCLUSIONS The one-year model predicted that costs associated with an increase in COX-2 utilization after removal of PA would exceed the costs to administer PA and treat NSAID-related serious GI events in the managed care population. Based upon this CEA, PA appears to be an effective tool to manage COX-2 pharmacy costs. Further examination of the medical claims would be useful to validate the assumed GI event rates with or without PA and to further demonstrate more definitively the value of a PA program for COX-2 drugs.