New use of nonsteroidal anti-inflammatory drugs of the first generation that selectively inhibits the cyclooxygenase-2 (COX-2) enzyme, such as diclofenac and etodolac, was associated with increased 30-day mortality following ischemic stroke in a Danish database analysis.
Compared with patients who were not taking NSAIDs other than aspirin, those who had started on old-line COX-2 inhibitors within 60 days of admission for ischemic stroke showed a 42% increase in multivariable-adjusted risk for 30-day mortality (hazard ratio 1.42, 95% CI 1.14-1.78), according to Morten Schmidt, MD, of Aarhus University Hospital in Aarhus, Denmark, and colleagues.
No significant increase in risk was seen for other NSAIDs, including members of the newer coxib class of COX-2 inhibitors as well as nonselective agents such as naproxen, the researchers reported online inNeurology. None of the drugs examined in the study was associated with increased or decreased risk of death following hemorrhagic stroke.
"The increased mortality rate associated with COX-2 inhibition in ischemic stroke was observed only among current users, which may indicate an actual drug effect," Schmidt and colleagues noted.
Although the study offered no definitive explanation for the association, the researchers suggested several possibilities for causal mechanisms:
- Drug-induced thromboembolism leading to larger and more severe occlusions
- Increased risk of post-stroke cardiovascular events including but not limited to recurrent stroke
- Impairment in beneficial inflammatory responses to ischemic stroke
The current study, looking at the most severe adverse event that can occur after stroke onset, adds to the reasons to avoid such agents, the researchers contended.
"If the association is truly causal, it constitutes a strong argument for increasing the efforts to ensure that patients with a high predicted risk of arterial thromboembolism (e.g., atrial fibrillation patients with high CHA2DS2-VASc score) are not prescribed COX-2 inhibitors when alternative treatment options are available," they wrote.
On the other hand, they acknowledged that COX-2's role in the body is complicated, and inhibitors of it may help as well as harm stroke patients -- possibly both at once.
While the harm of COX-2 inhibition could come from suppression of neuroprotective factors, it is also the case that inflammatory reactions involving release of nitric oxide, interleukin-1-beta, and tumor necrosis factor -- which are believed to be neurotoxic to some degree -- are reduced by COX-2 inhibitors.
Adam Kelly, MD, of the University of Rochester Medical Center in New York state, said the study would be valuable in providing more information for prognosis in stroke patients. "We're always looking for possible predictors of mortality after stroke," said Kelly, who was not involved with the analysis.
He also noted that the authors' discussion of inflammatory processes "adds some biological plausibility" to the reported associations.
Study Details
Because the possibility that preadmission use of COX-2 inhibitors prior to stroke onset may worsen outcomes has not been studied in detail, Schmidt and colleagues decided on a relatively simple way to take a first look at the issue -- by examining registry data from Denmark, where individuals' medical histories can be accessed on a population basis.
From July 1, 2004, to Dec. 31, 2012, the researchers found just over 100,000 individuals who had experienced a first-time stroke. Of these, about half had ischemic strokes, 12% had intracerebral hemorrhages, 5% had subarachnoid hemorrhages, and for 32% the type was unspecified.
Some 11,000 of the patients were current users of prescription non-aspirin NSAIDs; 8% had records of previous use but none within 60 days of admission; and 81% showed no record of use. (Over-the-counter use may not have been captured in the data.)
Ibuprofen was the most commonly prescribed agent (51% of current users), followed by diclofenac (27%), etodolac (11%), naproxen (3%), and celecoxib (1%). Records indicated that 15.5% of ischemic stroke patients were using NSAIDs after discharge, with smaller proportions of hemorrhagic stroke patients showing postdischarge use.
Death within 30 days of admission was recorded for 8.7% of the ischemic stroke patients, 35% of those with intracerebral hemorrhage, 24.5% of the subarachnoid hemorrhage group, and 14.3% of those with unspecified types.
Any NSAID use started 60 days or less before admission for ischemic stroke was associated with increased risk of 30-day death with a nonsignificant hazard ratio of 1.18 (95% CI 0.95-1.47) versus nonusers, after adjustment for propensity scores and a host of other potential confounders. The risk associated with new use of any COX-2 inhibitor (both old-line and coxib) was nearly the same at 1.21.
Schmidt and colleagues found a hint that the coxib class may also increase the risk of 30-day mortality following ischemic stroke -- a propensity- and multivariable-adjusted hazard ratio of 1.41 versus nonusers -- but there were so few patients taking these agents that the result did not achieve statistical significance (95% CI 0.37-5.31).
In contract, there was not even the faintest suggestion that NSAIDs of any type worsened outcomes after hemorrhagic strokes. New use of nonselective NSAIDs in people later diagnosed with subarachnoid hemorrhage, in fact, was associated with reduced mortality after adjustments (HR 0.58, 95% CI 0.37-0.91).
Kelly told MedPage Today that the findings, in and of themselves, would not justify withholding older COX-2 inhibitors from all patients. He suggested that these medications may be the best choices "from a quality-of-life standpoint" for some patients with musculoskeletal conditions. "It's a really important patient-centered question that the patient and the provider need to address," he said.
