Atypical antipsychotic drugs and risk of ischaemic stroke: population based retrospective cohort study

BMJ 2005; 330 doi: (Published 24 February 2005)
Cite this as: BMJ 2005;330:445

Recent rapid responses

Rapid responses are electronic letters to the editor. They enable our users to debate issues raised in articles published on Although a selection of rapid responses will be included as edited readers' letters in the weekly print issue of the BMJ, their first appearance online means that they are published articles. If you need the url (web address) of an individual response, perhaps for citation purposes, simply click on the response headline and copy the url from the browser window.

Displaying 1-6 out of 6 published

The study by Gill et al (1) does indeed leave some important questions unanswered but nevertheless data such as these make a helpful contribution to those real-world decisions where one has to chose between two (or more) drugs rather than one drug and no treatment. In answer to Martin Duerdin (2), observational data trumps RCT data where no RCT data exist. It most improbable that we will ever have a quantity of RCT data in relation to the safety of older antipsychotic drugs in patients with dementia that approaches that available for risperidone and olanzapine etc.

1. Gill SS, et al. Atypical antipsychotic drugs and risk of ischaemic stroke: population based retrospective cohort study. BMJ 2005;330:445-451. 2. Duerdin MG. Since when has observation trumped the RCT? BMJ rapid response, 27 Feb 2005

Competing interests: I have from time to time received lecturing fees and honoraria from pharmaceutical companies that manufacture atypical antipsychotic drugs

Competing interests: None declared

Richard Prettyman, Consultant Old Age Psychiatrist

Bennion Centre, Glenfield Hospital, Leicester LE3 9DZ

Click to like:

Editor – the large observational study by Gill et al [1] has demonstrated that the risk of cerebrovascular events associated with the use of atypical antipsychotic agents is far from clear, and further studies are required to better define this apparent risk. Previous randomised trials have not been adequately powered to study the risk of stroke, and the total number of strokes reported in these trials was relatively small (stroke rate was 1.3% for olanzapine and 3.4% for risperidone), leading to substantial uncertainty about the true size of this risk. All the patients in the trials who experienced cerebrovascular adverse events also had other risk factors for stroke such as atrial fibrillation, hypertension and diabetes. Moreover, many of them had vascular dementia as the original diagnosis. This meant that confounding might have contributed towards reducing the internal validity of this outcome.

There are many examples where confounded results had been generated by randomised trials. The most famous of which was ISIS-2, a study of thrombolytic drugs for probable myocardial infarction, which found that the drug was ineffective for those patients born under the astrological star sign Gemini [2]. Another example was the ALLHAT trial, which showed that during follow-up, the number of patients with new-onset heart failure was 19% higher in the lisinopril group compared with the chlorthalidone group (statistically significant) [3]. In the latter example, should we then stop prescribing lisinopril in fear of increasing the risk of heart failure? Naturally, the answer is no, since there is already robust evidence that ACE inhibitors are beneficial for patients with cardiovascular risk factors. When a clinical trial throws up unexpected (usually rare) but harmful secondary associations, it is tempting to over- interpret it, which often lead to labelling restrictions or even outright ban, especially in the current climate of medico-legal hype.

Every drug is associated with a list of side effects, and if we continue to make hasty prescribing ‘rules’ based on small number of adverse events, then in time there will be very limited number of drugs availble to be used. Take aspirin for instance, clinical trials have shown that gastrointestinal haemorrhage occurs in 2.5% of all patients taking aspirin [4]. So, should aspirin be banned, or should we (physicians and patients) make informed choices by carefully weighing up the benefits and risks of taking aspirin? Surely the latter is more appropriate and acceptable. Let us continue to exercise the art of evidence-based medicine.


1. Gill SS, et al. Atypical antipsychotic drugs and risk of ischaemic stroke: population based retrospective cohort study. BMJ 2005;330:445-451.

2. ISIS-2 (Second International Study of Infarct Survival) Collaborative Group. Randomised trial of intravenous streptokinase, oral aspirin, both, or neither among 17,187 cases of suspected acute myocardial infarction: ISIS-2. Lancet 1988;ii:349-360.

3. The ALLHAT Officers and Coordinators for the ALLHAT Collaborative Research Group. Major outcomes in high-risk hypertensive patients randomized to angiotensin-converting enzyme inhibitor or calcium channel blocker vs diuretic: the antihypertensive and lipid lowering treatment to prevent heart attack trial (ALLHAT). JAMA 2002;288:2981–97.

4. Derry S and Loke Y-K. Risk of gastrointestinal haemorrhage with long term use of aspirin: meta-analysis. BMJ 2000;321:1183-1187.

Competing interests: None declared

Competing interests: None declared

Joseph Kwan, Registrar in geriatric medicine

Elderly Care Research Unit, University of Southampton

Click to like:

Dear Editor;

Gill and colleagues provide useful information for clinicians regarding the safety of treatment with antipsychotics in patients with dementia.(1) Their data implied that there is not likely to be any more than a small difference in risk for stroke between modern and traditional antipsychotics. This assertion is based on the extremes of the hazard ratio confidence interval (HR=1.01; 95% CI: 0.81, 1.26) and the baseline event rate of 22.3 stroke admission/1000 patient years. That is, for every 1000 patient years there may be as many as 4.2 stroke admissions avoided when modern agents are used preferentially or as many as 5.8 avoided when traditional agents are used. A difference larger than this is unlikely to exist (assuming relative equivalence in risk was achieved after adjustments in this study).

However, it is very important that the question this study helped to answer is distinguished from the question addressed by the warnings released by Health Canada, the US FDA, the Committee on Safety of Medicines, and other regulatory bodies related to modern antipsychotic use and stroke risk. The study by Gill and colleagues retrospectively compared two active treatments, modern vs. traditional antipsychotics, and helps clinicians determine if one choice is safer than the other in terms of stroke. Their study did not compare patients using and not using antipsychotics. As such, their study does not address the question of whether antipsychotics (modern, traditional or both groups of agents) independently increase the risk for stroke above baseline. This issue was partly addressed by the warnings, which are based on the cumulative findings of RCTs in which modern antipsychotics (specifically risperidone and olanzapine) were compared to placebo and were found to increase the risk for stroke approximately 2-3-fold.(2,3)

As such, the relevant questions left without satisfactory answers are "Do all antipsychotics increase the risk of stroke in patients with dementia?" and “Is their a clinically important difference in the magnitude of the increase in risk among the antipsychotics?” It is hoped that Gill and colleagues, and others, will follow up with similar retrospective studies addressing these questions, as it is unlikely that they will be answered by RCTs any time soon.

1. Gill SS, Rochon PA, Herrmann N, et al. Atypical antipsychotic drugs and risk of ischaemic stroke: population based retrospective cohort study. BMJ 2005;330:445-448.

2. Risperdal (risperidone) and cerebrovascular adverse events in placebo-controlled dementia trials: letter to healthcare professionals. Toronto, Janssen-Ortho. October 11, 2002. dgpsa/tpd-dpt/risperdal1_e.html. Accessed on March 1, 2005.

3. Zyrexa (olanzapine) and cerebrovascular adverse events in placebo- controlled dementia trials: letter to healthcare professionals. Toronto, Eli Lilly Canada Inc. March 10, 2004. dgpsa/tpd-dpt/zyprexa_hpc_e.html. Accessed on March 1, 2005.

Competing interests: None declared

Competing interests: None declared

David M Gardner, Associate Professor of Psychiatry and Pharmacy

Dalhousie University, Halifax, NS, Canada, B3H 2E2

Click to like:

This study by Gill and colleagues is definitely noteworthy and brings to fore the debate on the superiority of the study methods.

However, what is going unnoticed among the various concerns raised by the concerned health authorities and the deescalation of these concerns by such studies is the fact that whether the changes and decisions that are made really in the best interest of the patients with dementia.

The atypical antipsychotics increase the probability of cerebrovascular events in a minority of patients!! However, it does bring in significant improvement in the quality of life in the rest; if not all. The real issue here as in all other decisions is weighing the pros and cons of the risks. At one hand, we have the atypicals( with the lone exception of Quetiapine, now that even Aripiprazole has joined the list) which increase the risk of CVA and on the other hand we have the typicals which at the first instance were taken over by the atypicals for their sheer advantages.

We need to exercise judgement at least in a few situations where the well being of the patients is at stake.

Competing interests: None declared

Competing interests: None declared

DINESH KUMAR KAMALA KANNAN, SHO in Genral adult Psychiatry

St Luke's Hospital, South West Yporkshire Mental Health NHS Trust, Huddersfield, HD4 5RQ

Click to like:

Can anyone cite a scientific, as opposed to a marketing definition of an "atypical antipsychotic"? The more I see of quetiapine, for example, the more it reminds me of chlorpromazine: a "dirty drug" acting on multiple receptors, whose main distinction is that it is more expensive than its predecessors.

Thomas L. Perry, M.D., FRCPC

Competing interests: None declared

Competing interests: None declared

Thomas L. Perry, Dept. of Pharmacology & Therapeutics, University of British Columbia

University Hospital, Vancouver, B.C. V6T 2B5

Click to like:

The study by Gill and colleagues[1] and the accompanying headline in This Week in the BMJ “Choice of antipsychotics for dementia doesn't affect risk of stroke” sets a dangerous precedent.

There is a hierarchy of evidence and traditionally observational data has been used to provide suggestions of an association which can generate hypotheses; randomised- controlled-trials (RCTs) help to prove or disprove these hypotheses. Inadvisably, such associations are often taken to heart before more robust evidence is established, as was seen when hormone replacement therapy was proposed to reduce coronary heart disease. This was widely believed and influenced prescribing until the theory was disproved by the Women’s Health Initiative RCT[2].

The study by Gill et al. looks at observational data and implies that this disproves evidence from RCTs, which in the UK caused the Committee on Safety of Medicines to caution against using atypical antipsychotics where there is a stroke risk[3]. Observational data are fraught with bias and confounding factors; one suspects that data on admission with stroke garnered from administrative databases is too crude as a proxy for overall stroke and transient ischaemic event rates. A fundamental concern here is that the licenses for atypical antipsychotics do not include an indication for behavioural problems in dementia. There is a reason for this: studies performed to develop this licence indication have shown the stroke hazard – disappointing but true.

Have we not learnt the lesson that RCT evidence, that is experimentation, trumps observation?[4]

[1]. Gill SS, Rochon PA, Herrmann N, et al. Atypical antipsychotic drugs and risk of ischaemic stroke: population based retrospective cohort study. BMJ 2005;330:445-448.

[2] Writing Group for the Women’s Health Initiative Investigators. Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principle results from the Women's Health Initiative randomized controlled trial. JAMA 2002;288:321-333.

[3] Atypical antipsychotic drugs and stroke: message from Professor Gordon Duff, Chairman, Committee on Safety of Medicines (CEM/CMO/2004/1).

[4] Petitti DB. Hormone replacement therapy and heart disease prevention: Experimentation trumps observation. JAMA 1998;280:650-52.

Competing interests: None declared

Competing interests: None declared

Martin G Duerden, General Practitioner

Meddygfa Gyffin, Conwy, LL32 8LT

Click to like: