Sir: The quality agenda laid out in the NHS plan is embodied by the National Service Framework and evidence-based treatment protocols. However attractive these approaches seem they fail to acknowledge the complex transactional nature of clinician—patient interactions. Inappropriate treatment can occur knowingly, and for many reasons.

For 1 month clinicians from an inner-London community mental health team (CMHT), which uses prescribing guidelines, identified medication they prescribed or dispensed although they believed it was not medically indicated.

Twenty-nine prescriptions for 28 patients were identified. White male patients were over represented. All major diagnostic categories were included and six patients had a drug induced psychosis.

The commonest reported reason for non-indicated prescribing was patient unwillingness to discontinue treatment. The medications prescribed were not trivial and included benzodiazepines, lithium and antipsychotics. In four cases prescribing was thought to facilitate engagement with the CMHT. Two patients received, as required, a supply of antipsychotics to reduce out of hours presentations. Failure to discontinue street drug use was sited in four cases.

Knowing deviation from good practice was readily identified, although uncommon, and made clinicians appropriately uncomfortable. Explanations are complex and include consumer demand, relief of anxiety, attempts to contain workload and lack of skills. We believe this highlights the importance of regular peer review to identify idiosyncratic practice and to consider the reasons why it occurs. We need to guard against our desire for concordance leading us into collusion.