We are pleased to see support for our views in the responses received, in particular for the focus on rational debate in this area, rather than an ‘either/or’ debate. However, we would in reply air the following cautions.

Pattanayak claims that antipsychotics may not be as harmful as we have been led to believe. We would urge caution here. The data produced so far do suggest a reliable and quantifiable degree of harm resulting from antipsychotic use. We do not think that it is truly reasonable, now, to suggest that these drugs may be harmless. But we would also argue that Pattanayak's desire to claim a lack of harm is unnecessary. If they are harmful, their use may be justified under the doctrine of double effect by balancing the likely benefits against harm. It is our clinical experience that the discussions of benefit versus harm with relatives and advocates, informed by the principal of double effect, are effective and well understood.

Finally, we note what is, perhaps, an important slip of the pen in Sekhri's letter. He uses the term non-pharmacological as shorthand for alternatives to antipsychotics. He is correct in saying that there are multiple causes of behavioural and psychological symptoms of dementia. But if we describe all alternatives to antipsychotics as non-pharmacological, we may forget the appropriate treatment of physical illness with analgesics or depression with antidepressants. Alternatives to antipsychotic use in distress include both pharmacological and non-pharmacological approaches. One of us remembers replacing haloperidol for behavioural and psychological symptoms of dementia with effective treatment for scabies. Aromatherapy would have done little here! The term ‘non-pharmacological’ may therefore distract clinicians and compound a paucity of response to a complex problem.