OBJECTIVES: This study examines the patient level factors that play role in initiating the process of prior authorization, and evaluates whether prior authorization (PA) process has an effect on future gastrointestinal (GI) outcomes (e.g. bleeding, peptic ulcer disease (PUD)) in managed care population. METHODS: Patients who encountered a PA system edit for Cox-2 inhibitors during January 1, 2001 to July 31, 2002 were followed up for 1 year post rejection date. Patients were stratified to 2 groups; 1) initiated the process of PA, 2) never initiated the process of PA. Using multivariate logistic regression, significant covariates in predicting future GI bleed or PUD, and applying for PA were identified. Charlson Comorbidity Index (CCI) was calculated for each patient and used in the regression analysis. RESULTS: Those who initiated the PA process were significantly older, had higher CCI, higher total medical and pharmacy costs, and more GI events than those who never applied for a PA. Significant covariates in predicting future GI bleed were aged = 65, previous GI bleed or PUD, and higher CCI. In addition to gender similar covariates also predicted future PUD where females were at higher risk of having a PUD. Whether prior authorization process was initiated or not was not a significant factor in predicting future GI event. Patients who are females or age = 65, with osteoarthritis, rheumatoid arthritis, previous ulcer or GI bleed event, or high CCI score were more likely to apply for PA. CONCLUSION: PA process appears to select high risk patients to receive Cox-2 inhibitors. Patients who did not apply for PA appear to have similar adjusted future GI event rate than those who applied.