Article Text

Download PDFPDF
Antipsychotics in dementia – mortality risks and strategies to reduce prescribing
  1. Tom G H Smith
  1. Correspondence to Dr Tom G H Smith, Specialist Registrar in Liaison Psychiatry, Paterson Cabin, St Mary's Hospital, London W2 1PF, UK; tomsmith{at}

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

For the UK, the first governmental warning about prescribing antipsychotics in dementia came as a safety briefing in 2004 alerting clinicians to the risks of stroke with two of the atypical antipsychotics.1 Meanwhile, meta-analytic evidence was accumulating across the pond pointing not only to increased risk of stroke but also sedation, urine infections, parkinsonism and cognitive deterioration.2 It was these concerns that led the Department of Health to commission a report from Professor Sube Banerjee on the use of antipsychotics in dementia.3 Published as ‘Time for Action’ in October 2009, Banerjee's detailed review of the RCT literature concluded that, for treatments between 6 and 12 weeks, there was minimal evidence for improvement in global behavioural disturbance (effect size range 0.1–0.2),2 but an increased absolute mortality risk of 1% (NNH 100 (95% CI 50 to 250)).4 Banerjee applied these statistics to the available figures for UK prevalence of antipsychotic prescribing in people with dementia, and calculated the staggering headline figure of 1800 deaths being directly related to these drugs every year.3

However, as acknowledged (if not emphasised) by Banerjee, this alarming figure appears to be a highly conservative underestimate. First, the literature suggests that the prescribing of antipsychotics in dementia continues way beyond the 10–12 weeks used in Banerjee's equation. For example, one small American study (n=58) found a mean prescription length of 16.5 months (median 9 months).5 These findings were replicated in a larger Canadian study in care homes (n=1017) where mean prescription length was 1.02 years (median 34 weeks).6 If the risk of death is cumulative (and linear), and we adopt 34 weeks as an average antipsychotic prescription length then a crude multiplication of Schneider's 1% absolute risk of death gives a new NNH of 33. Applying this in turn to …

View Full Text