OBJECTIVE: To obtain a profile of adult patients at risk for nonsteroidal anti-inflammatory drug (NSAID) induced gastrointestinal (GI) complications in a large managed care health plan in the Southwest United States. METHODS: Patients with NSAID prescription claims between July 1996 and June 1997 were identified from a health plan claims database. Patients with claims associated with ICD-9 and CPT codes indicating serious GI complications were then identified. The ICD-9 codes used were those associated with GI ulcers (531.x, 532.x, 533.x, and 534). A total of 19 CPT codes for GI procedures indicative of a GI complication were used. A forward stepwise logistic regression analysis, using the likelihood-ratio (LR) test, was performed to identify predictors of GI complications. Predictors included in the model were individual NSAIDs and the following potential risk factors: age, gender, previous GI drug usage, previous steroid usage, and total days supply of NSAIDs during the study period. RESULTS: A sample of 15,772 patients with prescription claims for NSAIDs was identified. Of these patients, 213 (1.4%) had an ICD-9 or CPT code suggestive of serious GI complications secondary to a NSAID. The logistic regression results indicated that women (OR=0.65, 95% CI=0.48 - 0.87) were less likely to develop GI complications. However, patients with previous GI drug usage (OR=5.97, 95% CI=4.51 - 7.90), those who used ketorolac (OR=2.01, 95% CI=1.10 - 3.67) and those who used oxaprozin (OR=1.82, 95% CI=1.10 - 3.00) were more likely to develop GI complications. CONCLUSION: Users of ketorolac and oxaprozin, as well as those with previous GI drug usage were at a higher risk, while women were at a lower risk of GI complications in this managed care population.