July 2011
Reviewer: Jeff C. Huffman, MD
Association of cerebrovascular events with antidepressant use: a case-crossover study
Wu CS, Wang SC, Cheng YC, Gau SS
Am J Psychiatry 2011; 168(5):511-521
Background: There has been suggestion of increased rates of stroke in patients receiving antidepressant agents in prior studies, though there have been questions about whether such associations are related to the effects of the antidepressants or the underlying conditions for which the agents were being prescribed.
Methods: The authors performed an assessment of 489,852 persons registered in the large National Health Insurance Research Database in Taiwan (98% of Taiwanese population) using a case-crossover design, comparing rates of antidepressant use during case (cerebrovascular event) and control (no event) time windows of 14 days. Subjects for the analysis must have had a cerebrovascular event, at least one antidepressant prescription in the year prior to the stroke, and no head injury/prior cerebrovascular event or hospitalization within a year. The case-crossover design uses each patient as his/her own control by comparing the rate of antidepressant exposure in the two weeks prior to the stroke (1-14 days before) to the rate of exposure in the two weeks earlier (15-28 days before). The authors also used 7 and 28 day windows. Many additional baseline characteristics of subjects were also recorded to account for them in data analysis. Finally, the ‘type’ of antidepressant — based on extent of serotonin transporter reuptake inhibition — and the dose of the agents were recorded.
Results: A total of 24,214 stroke patients meeting study criteria were identified. Overall, antidepressant use in the 2 weeks prior to stroke was associated with a higher risk of stroke by 48%, adjusting for relevant covariates, and was independent of new-onset depression. Risk of stroke was higher with higher doses of agents and with agents that more strongly affected the serotonin transporter. Antidepressant use was more strongly associated with risk of ischemic stroke (OR 2.52) than risk of hemorrhagic stroke (OR 1.92). These findings also generally held for 7 and 28 windows as well. Patients who were chronically on antidepressants in the long-term had lower risk of stroke associated with ongoing antidepressant use than when antidepressants were discontinued. Finally, during the case period, there were also more outpatient visits and prescriptions of other medications.
Discussion: These potentially important findings are consistent with some of the prior literature suggesting that antidepressants may be associated with increased risk of stroke, potentially through increased risk of bleeding and/or vasoconstriction of large cerebral vessels. The fact that these effects may be mediated by serotonin, and that the risk of both ischemic and hemorrhagic strokes were greater with agents that more strongly affect the serotonin transporter, are potentially consistent with this explanation. Interestingly, the risk appeared only during the first several prescriptions, and, as noted, patients on long-term antidepressants appeared to have a lower risk when they continued, rather than discontinued, the agents.
This study does have some important limitations, as it does not identify indication for antidepressant use (depression/anxiety/migraine), adherence to medications, and does not replace a prospective study of this indication. However, it was a very thoughtful study and may suggest that starting at lower doses, and possibly using agents with lesser serotonergic effects, may be indicated in some patients at risk of stroke. These findings may be the flipside to other observational studies that have found antidepressants (and SSRIs in particular) associated with lower rates of the onset of MI, a more common condition.
In short, this well-designed observational study raises a clinically-important concern about the association between antidepressants and stroke. Still, the jury remains out regarding whether these agents truly cause increased rates of cerebrovascular events, just as the jury remains our whether these agents truly reduce the risk of ischemic cardiac events.