Background: COX-2 and NSAID differ on their gastrointestinal (GI) and cardiovascular (CV) toxicity.
Objectives: Describe the patterns of use of NSAID and COX-2 in the health improvement network (THIN) database and determine whether prescribing practices are influenced by the presence of CV, hepatic or GI risk factors.
Methods: We looked at the experience of ten distinct cohorts of new users who were 18 years old or greater: diclofenac, naproxen, ibuprofen, piroxicam, other NSAID, meloxicam, celecoxib, etoricoxib, rofecoxib, and valdecoxib. The study period was 1 January 1995–31 March 2004. We collected information on the presence of history of upper GI disease, CV disease, renal failure and hepatic disease, dose, concomitant medication, and visits to a rheumatologist. We used proportional hazard models to compute the ‘switch ratio’ and 95%CI of dose increase or decrease.
Results: There were 486,076 unique patient-drug pairs in the study. Of these, 78,201 (16%) were COX-2 users. Diclofenac and ibuprofen were the most frequently prescribed (80.7% combined) NSAIDs, and celecoxib and rofecoxib were the most frequently prescribed (73.7% combined) COX-2 agents. More COX-2 users than NSAID users were 75 years (19–25% vs. 7–11%), received concomitant gastroprotective gastroprotective agents (GPA) (20.2 vs. 6.3%), corticosteroids (6.2 vs. 3.2%), and anti-platelet therapy (10.6 vs. 4.1%), had a history of thromboembolic (TE) events (13.6 vs. 6.4%), hypertension (30% vs. 15%), and were chronic users. NSAID users had lower chronic disease score (34% had 0 score) compared with COX-2 users (>10% had 7). Approximately 1% of rofecoxib use was at 50 mg/day. There was a higher ‘dose change ratio’ for increases in valdecoxib and rofecoxib doses and for decreases in the doses of etoricoxib. We observed an inverse relationship between dose increase and older age.
Conclusions: Celecoxib and rofecoxib were prescribed more frequently for chronic conditions. Naproxen was not subject to CV channeling whereas COX-2 users showed evidence of both GI and CV channeling. Most prescriptions followed the dosing regimen specified in the labels. Etoricoxib use suggested some off-label use. Piroxicam use suggested lower GI tolerance. Restricting the use of COX-2 to patients at higher risk of GI events may have had the paradoxical effect of exposing patients at higher risk of CV events preferentially to COX-2 agents, when some of these agents may be associated with higher CV risk.
