COX-2 inhibitors associated with greater risk for death after ischemic stroke

Cyclooxygenase (COX)-2 inhibitor use before hospital admission for ischemic stroke was associated with a higher 30-day mortality rate, according to a new study.

Researchers performed a nationwide population-based cohort study in Denmark to investigate whether nonselective NSAID use or COX-2 inhibitor use before hospital admission for stroke was associated with 30-day mortality. Medical databases were used to identify all hospitalizations for first-time stroke between 2004 and 2012, as well as subsequent death. NSAID use was defined as current use (i.e., redeeming a prescription within 60 days pre-hospital admission), former use, and nonuse. Current use was then classified as new or long-term, and hazard ratios for death within 30 days were calculated. Study results were published online Nov. 5 by Neurology.

A total of 100,043 patients with first-time stroke were identified. Of these, 51,224 (51%) had ischemic stroke, 11,779 (12%) had intracranial hemorrhage, 4,528 (5%) had subarachnoid hemorrhage, and 32,512 (32%) had unspecified stroke. Median patient age in these groups was 74 years, 72 years, 58 years, and 76 years, respectively. Overall, 10,835 patients (10.8%) reported currently using NSAIDs (54.2% ibuprofen, 3.2% naproxen, 27.0% diclofenac, 10.7% etodolac, 1.0% celecoxib, and 0.5% rofecoxib). In addition, 8,402 (8.4%) reported former NSAID use and 80,806 (80.8%) reported no use. A total of 15.5% of patients used NSAIDs after discharge for ischemic stroke, 6.1% used them after intracranial hemorrhage, and 4.9% used them after subarachnoid hemorrhage.

Thirty-day mortality rates among NSAID users were 8.7% for ischemic stroke, 35.1% for intracranial hemorrhage, 24.5% for subarachnoid hemorrhage, and 14.3% for unspecified stroke. At 30 days, the adjusted hazard ratios for death were 1.19 (95% CI, 1.02 to 1.38) in patients who had ischemic stroke and were current users of selective COX-2 inhibitors compared with nonusers. The mortality hazard ratio for new users in this group was 1.42 (95% CI, 1.14 to 1.77). When different COX-2 inhibitors were compared, the hazard ratio was 1.42 (95% CI, 1.14 to 1.78) for new users of older, traditional COX-2 inhibitors; separately, etodolac use had a hazard ratio of 1.53 (95% CI, 1.02 to 2.28) and diclofenac use had a hazard ratio of 1.28 (95% CI, 0.98 to 1.68). No association was seen between use of nonselective NSAIDs or COX-2 inhibitors and death from intracerebral hemorrhage.

The researchers noted that patients were not randomly assigned and that detailed data were not available on such characteristics as smoking, body weight, or indications for NSAIDs, among other limitations. However, they concluded that use of COX-2 inhibitors before hospital admission for ischemic stroke was associated with an increased risk for 30-day mortality. This association was not seen for hemorrhagic stroke or with nonselective NSAIDs. They also pointed out that the increased mortality rate for ischemic stroke was noted only in current users of COX-2 inhibitors, increasing the likelihood that the finding was a true drug effect.

“Our study adds to the increasing body of evidence concerning the vascular risk and prognostic impact associated with use of COX-2 inhibitors,” the researchers wrote. “The prognostic impact also needs to be considered when prescribing older or newer COX-2 inhibitors to patients at increased risk of thromboembolic events.” A true causal association would strongly support efforts to ensure that COX-2 inhibitors aren't prescribed to patients at high risk for stroke if other treatments are available, they noted.