To compare how cyclooxygenase-2 (COX-2) users with and without cardiovascular (CV) risks responded to the withdrawal of rofecoxib on September 30, 2004.
The data, from a pharmacy claims database, consisted of patients who filled at least one prescription for a COX-2 agent in the 6 months before September 30, 2004 (n = 32,898) or September 30, 2003 for the control cohort (n = 37,930). Baseline drug utilization was used to determine comorbidities, and CV risk status was assessed with surrogate pharmacy markers. The first difference estimator was used to compare changes in utilization after September 30, 2004 with changes in utilization in a control cohort of patients who were treated with COX-2 over a similar time frame in 2003/2004.
The reduction in COX-2 utilization depended on baseline CV risk status. Among celecoxib and valdecoxib users, patients without CV risks reduced their utilization of COX-2, as measured by days of supply, by between 16.2% and 22.7%. The reduction was 32% for patients with one CV risk marker and 55.8% for patients with three or more markers, a proxy for the severity of CV risk. The corresponding figures for rofecoxib users were 47.5% (no CV risk), 55.4% (one marker) and 64.8% (three or more markers).
Our results suggest that patients and physicians used newly available information about COX-2 inhibitor agents and their side effects to reduce treatment. They also provide support for the notion that patients and providers were discriminating in their response to new information as a significant proportion of patients remained on treatment after extensive publicity concerning potential risks.