Determinants of selective cyclooxygenase-2 inhibitor prescribing: are patient or physician characteristics more important?

Background Little is known about which factors influence the widespread use of selective cyclooxygenase (COX)-2 inhibitors. We examined the relative effects of patient risk factors for gastrointestinal toxicity, other patient characteristics, and physician prescribing preferences on the decision to prescribe a selective COX-2 inhibitor. Methods We retrospectively studied a cohort of 28,190 Medicare beneficiaries who were continuously enrolled in a large, state-run pharmacy benefits program that reimbursed for selective COX-2 inhibitors and nonselective nonsteroidal anti-inflammatory drugs (NSAIDs) without restrictions. Half of the study sample filled a prescription for a selective COX-2 inhibitor and the other half for a nonselective NSAID. Multivariable logistic regression models were developed to predict COX-2 inhibitor use. Results Seventeen percent of patients using a COX-2 inhibitor had no identifiable risk factor for NSAID-associated gastrointestinal toxicity, compared with 23% of those using a nonselective NSAID. Established risk factors (age ≥75 years, history of gastrointestinal hemorrhage or peptic ulcer disease, or concomitant warfarin or oral glucocorticoid use) were all significant predictors of COX-2 inhibitor use, but a multivariable model including only these risk factors discriminated poorly between the two patient groups (C STATISTIC = 0.55). Adding other patient clinical and demographic characteristics to the model somewhat improved this association (C STATISTIC = 0.66); however, when physician prescribing preference was included, the model had excellent ability to discriminate between the two treatment groups (C STATISTIC = 0.83). Conclusion Established risk factors for NSAID-associated gastrointestinal toxicity were poor predictors of who was prescribed a selective COX-2 inhibitor; in contrast, physician prescribing preference was an important determinant.