Don Klein begins his comment by recalling the background of the 1984 Quitkin et al. study that concluded that antidepressants (ADs) lagged several weeks behind the initial “clinical” effects on the CNS.
I am not persuaded that his description of the fine points of that study changes any of the facts cited in my essay. The Quitkin study was innovative in its approach to distinguishing specific drug responders from placebo responders. I also, with other interpreters of their results, noted that the Quitkin study was not designed to determine the time of onset AD clinical actions. The results could not be generalized to inform about onset because of the modest dosage of imipramine used to treat study patients, the graduated dosage schedule applied so that even this modest dosage level was not achieved until the end of the first two weeks and the insensitivity of the measures used to detect clinical effects, if they existed, during this period. I am not critical of the study otherwise, only that it has little to tell us about onset of AD clinical actions. On the issue of when “full response” to antidepressants is achieved, all would agree with Klein that that that does not usually occur before several weeks of treatment. Nevertheless, the question here is not about full response, but whether the drug is inducing significant clinical actions early, within the first two weeks, of the course of treatment. Here the jury is not out; several independent multisite, large sample studies have affirmed that such actions do occur, sometimes as early as the first week. All also agree about the importance of such findings toward understanding the neurobehavioral mechanisms underlying the efficacy of antidepressant drugs and the practical implications of such findings for the future design of clinical trials and for treatment practice.
The other studies referred to by Klein, which also included placebo controls (Stassen et al; Szegredi et al; Katz et al), sought to determine onset by various techniques, one of which involved testing significance of difference in changes at one and two weeks of treatment between drug and placebo, another by comparing the number of patients who showed a >20% decrease in the Ham-D total score (a decrease validated by Stassen et al to be clinically significant) at the early time points between drug and placebo.
It was further demonstrated in these studies that this amount of “early improvement”, i.e., >20% decrease, was clinically significant in that it could predict that 70% of patients showing this early improvement would go on to respond at 6 or 8 weeks to the experimental treatment; even more telling, that less than 10% of patients who did not show this early improvement, did not change course and respond at outcome.
Klein’s concern is that these findings could not be applied to individual patients (he apparently believes despite the finding of drug-placebo differences, that is “not the usual situation”). The facts evidenced from the studies referenced above indicate that on the contrary, on average, patients will respond in the manner described in these carefully designed studies.
We do make a case as he notes, that these results support shortening the standard clinical trial, and that our conclusions are strongly supported by the accumulated evidence on prediction in this area. Klein turns to our 2011 paper and analysis in the second part of his comments, to dispute the interpretation of the results presented. I am, however, unclear about how he has reanalysed the data from our table in that paper, what procedures he actually carried out, so am unable to respond intelligibly to this section of his comments.
Martin M. Katz
February 26, 2015