We read with interest the paper by Paton & Gill-Banham (Psychiatric Bulletin, June 2003, 27, 208-210). They say there are no systematic studies of prescribing errors in psychiatry. We have recently published an analysis of prescribing errors detected by pharmacists working in a 400-bed specialist psychiatric hospital (Reference Haw and StubbsHaw & Stubbs, 2003). A panel of three assessors (two consultant psychiatrists and the head pharmacist) screened and classified errors according to the definition and classification of Dean et al (Reference Dean, Barber and Schachter2000). Error severity was rated on a four-point scale.

During the one-month study period, 311 errors were detected that met the study definition. Fifty-six per cent were rated as clinically insignificant, with only 9% rated as having the potential to cause the patient harm, and none as potentially life-threatening. A greater proportion of the more serious errors had been made by non-consultant psychiatrists. Prescription writing errors (e.g. transcription errors) (88%) were more common than decision-making ones (e.g. prescribing a drug to which the patient has a known allergy) (12%). The error rate for non-psychotropics was twice that for psychotropics, perhaps reflecting psychiatrists' greater familiarity with the latter.

Our findings are broadly similar to those of Paton & Gill-Banham (Reference Paton and Gill-Banham2003), but we found most errors to be of the prescription writing (clerical) type. We agree that clinical pharmacists have a role to play in detecting errors. However, we found that in 42% of cases, the drug involved had already been administered. Most errors would probably have been detected at the source by electronic prescribing with computerised physician decision support.