Introduction

The benefits for manic-depressive patients of long-term lithium treatment are well established. Lithium is unrivalled for episode prevention in bipolar illness (Schou, 1979a) and is at least as effective as other treatments for prevention in unipolar illness (Schou, 1979b). Also the risks and inconveniences of long-term lithium treatment are well described and understood (Vestergaard, 1983a, b). Some important problems, however, remain unsolved; one is that lithium does not effectively prevent episode recurrence in all patients with manic-depressive illness. Another is that, even when lithium exerts effective prophylaxis, some patients do not adhere to the treatment, this is either because they do not tolerate lithium or because they do not accept the treatment regimen. The first problem may be dealt with in two ways: one is to examine variables which predict good response to lithium treatment; the other is to search for alternative or additional treatments which may benefit patients who do not respond to lithium prophylaxis. The problem of treatment non-adherence may be dealt with through early identification of potential non-compliers and through the examination of the reasons why some patients discontinue lithium treatment against medical advice.

Identification of patients who respond to lithium prophylaxis

Various patient, illness and treatment variables which may predict good response to prophylactic treatment have been described by Carroll (1979) and Pétursson (1979). However, the relative contribution of many different demographic, clinical, genetic, biological and sociomedical variables to treatment outcome has been analysed only by Grof and associates who used multivariate statistics (Grof et al., 1979, 1983).They showed that good response to long-term lithium treatment was related primarily to three clinical variables: (a) recurrent primary affective disease with a clearly episodic course, (b) intervals free of psychopathology and (c) a relatively low number of recent manic-depressive episodes or the absence of rapid cycling. Other variables also examined were either contained in the three clinical variables mentioned above or did not contribute to the variation in outcome. The predictive value of these three variables has been shown in studies by Grof et al. (1983) to be high; the outcome of prophylactic lithium treatment could be predicted correctly in 21 out of 24 patients.