To the Editor: The study by William Coryell, M.D., and colleagues (1) about lithium discontinuation-induced refractoriness, although informative, has two major drawbacks.

First, a study group consisting of 28 patients is inadequate to detect a phenomenon that, although clinically significant and potentially fatal (2), is certainly not frequent. It is useful to mention, is this connection, that although Tondo et al. were unable to find an effect of lithium discontinuation in their study group of 86 patients—to which Dr. Coryell et al. refer in their article—they actually detected this effect in a group of 106 patients (3), in which they found that "the proportion of time ill rose significantly (by 38%)" during the retreatment period.

Second, Dr. Coryell and colleagues provide no information on the treatment received by their patients before entering the study. As far as we know, these patients may already have interrupted their lithium treatment before the index episode, which, of course, would introduce a bias.

We should not forget that the patients described by Post et al. (2) had been receiving successful, continuous lithium prophylaxis for as long as 6 to 15 years before discontinuation and had experienced many relapses during a short period following lithium reinstitution. We should look carefully at the impressive life charts and case reports provided by these authors and by other experienced clinicians such as Goodwin (cited in reference 2) and Koukopoulos et al. (4) before dismissing a clinically meaningful phenomenon on the basis of studies that are carefully conducted but probably do not have adequate statistical power.