Lithium augmentation in antidepressant-resistant patients.

SIR: Austin et a! (Journal, October 1991, 159, 510— 514) recently presentedtheir meta-analysisof five placebo-controlled trials oflithium augmentation in depressed patients “¿ resistant to a standard trial of an antidepressant―. They concluded that lithium aug mentation had significant efficacy in these patients. However, as the authors acknowledged, while the statistical procedure used to arrive at their con clusions may be †̃¿ elegant',its utility is limited by the quality ofdata being analysed. Two issues deserve further discussion. The first relates to the duration of antidepressant medication before a patient can be considered treatment refractory. The studies reviewed used antidepressants for a minimum ofthree weeks before adding lithium. However, there is evidence (Quitkin et al, 1984; Georgotas et a!, 1986) that up to 25% of patients treated with an adequate dose of antidepressant medication will not respond until weeks four to six of treatment. Given that many of the patients used in the meta-analysis were prescribed lithium after only three weeks, one has to question whether clinical improvement was attributable to lithium or was, in fact, a delayed response to the primary antidepressant. Dr Austin et al state that “¿ it is doubtful that many clinicians would in practice wait more than four weeks before changing to another treatment if no response is seen―.However, if the evidence suggests that six weeks is required for an adequate trial of an antidepressant, then it is incumbent on the clinician to persist with that treatment rather than pre maturely adding another medication with its own risks and side-effects. The second issue pertains to the criteria used to define response to lithium augmentation. Three of the five studies used a @40—50%reduction in Hamilton Rating Scale for Depression (HRSD; Hamilton, 1960) score and a fourth study used a @2 point decrease on a 15-point nurses rating scale. Given that the mean HRSD entry scores for those studies ranged from 20—34,many patients fulfilling criteria for response could still have had scores con sistent with mild to moderate levels of depression. Therefore, in these studies, response did not necessarily equate with remission of the depressive illness. Given these considerations, we must await further