Sir: MacCleod and Sharp's attention to lithium non-adherence is welcome (Psychiatric Bulletin, May 2001, 25, 183-186). I have concerns, however, that the authors' conclusion may instil complacency, given that they state that “all patients were defined as currently compliant”.

The study uses a definition of non-compliance that requires both subjective and objective criteria to be fulfilled. A patient is then deemed ‘compliant’ by default if they do not meet both sets of criteria. Thus, for instance a patient in the study could have no measurable lithium in his or her serum at all but be deemed ‘compliant’ because he or she and his/her psychiatrist judge him or her to be so. Clinicians' judgements of patients' compliance have been found wanting in almost every study in which they have been tested. In fact the sensitivity of clinical judgement for detecting non-compliance has been quoted as an embarrassing 10% (Reference Stephenson, Rowe and HaynesStephenson et al, 1993).

The authors have not cited any of the work in this field in the past 10 years. In a recent study of compliance in lithium clinics (Reference Schumann, Lenz and BerghöferSchumann et al, 1999) it was noted that 53.9% of patients discontinued lithium prophylaxis at some time. Even more striking is the finding that 76 days is the median duration of continuous lithium adherence before patients elect to discontinue treatment (Reference Johnson and McFarlandJohnson & McFarland, 1996).

Given the potentially catastrophic outcomes of medicine non-adherence in major mental illness there remains a priority to identify strategies that will enhance adherence with the medicines we prescribe.