New Analysis Questions Antipsychotics' Stroke Risk

Using atypical antipsychotic medications in elderly demented patients to control agitation and aggression is a standard—if off-label—practice, but one tempered by warnings in recent years that they raise the risk of stroke.

Now several recent studies complicate that issue. One reanalysis of earlier clinical trials found that increased rates of stroke in one trial may have been due to inclusion of patients at higher cerebrovascular risk or the mislabeling of the events they experienced. A large, population-based cohort study reported no difference in risk of ischemic stroke between those taking typical and atypical antipsychotics, while a systematic review of four drug classes concluded that atypicals showed modest efficacy but with an increased risk of stroke.

No randomized controlled studies testing atypical antipsychotics in this population have been carried out, according to Nathan Herrmann, M.D., an associate professor of psychiatry and medicine, and Krista L. Lanctôt, Ph.D., an associate professor of psychiatry and pharmacology at the University of Toronto, and Sunnybrook and Women's College Health Sciences Centre. That has not kept the drugs from playing an important role, especially in institutional settings. Guidelines issued by the American Academy of Neurology, American Geriatrics Society, and others recommend their use once behavioral methods have failed.

Herrmann and Lanctôt reanalyzed results from 11 randomized controlled trials of risperidone and olanzapine. Beginning in 2002, drug regulators in Canada and Switzerland, and then in the United States, issued warnings about elevated risks of stroke on the basis of those trials. Although these warnings were first published for risperidone and olanzapine, the FDA recommended a similar warning about stroke for aripiprazole in December 2004.

The initial conclusions were based on a review of six studies of risperidone indicating that 3.3 percent of risperidone-treated subjects experienced strokes, compared with 1.1 percent of those on placebo. Five persons (0.5 percent) on risperidone died, as did one (0.1 percent) subject on placebo. In olanzapine trials, 1.3 percent of patients recorded a cerebrovascular event (with four deaths), compared with 0.4 percent on placebo (and one death).

“Taken at face value, without critical review, these are disturbing figures,” wrote Herrmann and Lanctôt in the February issue of CNS Drugs. However, their detailed analysis found less clarity below the surface of the numbers. For one, the definition of cerebrovascular adverse events included unclear or overlapping terms: “stroke, transient ischemic attacks, cerebrovascular accidents, cerebral ischemia, cerebral infarct, cerebrovascular disturbance, and cerebrovascular disorder.”

Comparing the risperidone and olanzapine trials, they found that the risperidone trials had included disproportionately large numbers of patients who had vascular dementia and mixed dementia and thus would have begun the trials at higher risk for cerebrovascular events. Trial subjects were not stratified by stroke risk factors. Eleven patients who experienced cerebrovascular events in these trials had hypertension, eight had atrial fibrillation, and 10 had prior strokes, all important risk factors for stroke.

Then, by dividing events into the serious (death, needing hospitalization, leading to serious disability) and nonserious, Herrmann and Lanctôt found that the rate of serious events (15 of 1009, or 1.5 percent) was not significantly different statistically from the rate of those on placebo (4 of 712, or 0.6 percent, p=0.27). The rate of nonserious events of subjects on risperidone (18 of 1009, or 1.8 percent) was significant compared with the rate of those on placebo (4 of 172, or 0.6 percent, p=0.026).

“Given that the frequency of cerebrovascular adverse events differs only for the nonserious events, it is possible that miscoding of events might explain the differences,” the authors stated.

Statistically different event and mortality rates were also found in olanzapine trials, but differences between serious and nonserious outcomes were not recorded. Several small, unpublished trials found olanzapine and risperidone had similar event and mortality rates, as did trials comparing olanzapine and typical antipsychotics.

“This is a group of patients prone to having strokes anyway,” said William Regenold, M.D., an assistant professor and director of the Division of Geriatric Psychiatry at the University of Maryland in Baltimore. The majority of strokes occur in patients over 75 years of age, and cognitive impairment is an independent risk factor. Low absolute rates of stroke in these drug trials make it hard to tease out the causes.

A population-based retrospective cohort study examined 32,710 older adults with dementia living in Ontario, Canada, from 1997 to 2002. Of those, 17,845 received atypical antipsychotics, and 14,865 were prescribed typical antipsychotic medications. The adjusted hazard ratio for incidence of admission to hospital with stroke of those using atypical compared to typical antipsychotics was 1.01. Although the risk of stroke was the same, clinicians should weight other risks and benefits of typical or atypical antipsychotics, wrote Gill, Rochon, Herrmann, and colleagues in the February 26 BMJ (British Medical Journal).

Authors of a systematic review published in the February 2 JAMA examined efficacy and adverse effects among typical antipsychotics, atypical antipsychotics, mood stabilizers, and cholinesterase inhibitors in older patients with dementia.

“Pharmacological therapies are not particularly effective for management of neuropsychiatric symptoms of dementia,” wrote Kaycee M. Sink, M.D., Karen F. Holden, M.D., and Kristine Yaffe, M.D. “Of the agents reviewed, the atypical antipsychotics risperidone and olanzapine currently have the best evidence for efficacy. However, the effects are modest and further complicated by an increased risk of stroke.”

“Clinicians may be using medications in place of more supportive environments,” said University of Maryland's Regenold. “A lot of patients are on antipsychotic drugs when other causes should be investigated. Patients may be in pain, fearful, or confused by their surroundings. Modifying the conditions that produce symptoms should be the first step.”

Sink, Holden, and Yaffe came to the same conclusion in their report and recommended that physicians and caregivers should try nonpharmacological treatment first. If that does not alleviate behavioral problems, they suggested beginning drug treatment with a cholinesterase inhibitor unless the patient is already receiving one, while avoiding typical antipsychotics and benzodiazepines. They presented a more complete algorithm for managing neuropsychiatric symptoms of dementia in their article.

“Do Atypical Antipsychotics Cause Stroke?” is posted online at<www.ingentaconnect.com/content/adis/cns/2005/00000019/00000002/art00001>.“ Atypical Antipsychotic Drugs and Risk of Ischaemic Stroke: Population Based Retrospective Cohort Study” is posted at<http://bmj.bmjjournals.com/cgi/content/full/330/7489/445>. An abstract of “Pharmacological Treatment of Neuropsychiatric Symptoms of Dementia” is posted at<http://jama.ama-assn.org/cgi/content/abstract/293/5/596>.▪

BMJ 2005 330 445

JAMA 2005 293 596