Response by Martin M.Katz

Martin M. Katz: Onset of antidepressant action

Martin M. Katz’s response to Donald F. Klein’s response to his reply to Carlos Morra’s comment

Dr. Klein has gone to great pains to reanalyze the data from our paper (Katz et al 2011) that reported on a secondary analysis from the 2004 Texas study.  I and colleagues consider that we did well by these results, as previously explained, but understand that others can have different perspectives on how to go about it. Nevertheless, as explained in my last note, we stand by the reported results, decided that the study has had sufficient scrutiny, and we would avoid any further  “over-analysis” of this relatively small size sample study.

More important, however, this dispute about how to analyze this small study, serves to distract from the main thrust of the controversy over “the onset of antidepressant clinical actions”. In my initial essay on the topic, I summarized the evidence which stems not from this small study, but from many studies, including at least two meta-analyses that involved thousands of patients, the entire range of antidepressant drugs and the necessary placebo controls, i.e., Stassen et al 1993, Szegedi et al 2009. Among these studies was the Texas study (Katz et al 2004) that had as its major aim the determination of onset. That study also contained a placebo control and met all requirements of a controlled, direct study of the issue. The results of these many studies are summarized in my essay with the appropriate references. Included also in that review is the chronology of events and the studies that resulted in the controversy.
In brief, that initial summary began with early clinical observations by such astute clinical investigators as Kuhn, in his original paper (1958) and Paul Kielhoz (1968), who all observed positive clinical effects by the end of the first week of treatment; the Quitkin et al 1984 study, not initially designed to determine “onset”, found it would take several weeks for drug-induced clinical effects to be detected; through the Katz et al 1987 study, which although also not designed to identify onset, reported significant clinical actions to occur within 1 to 2 weeks. The latter study did not have a placebo group and was not designed to determine onset, so the results were not apparently acceptable to Dr. Klein, one of the coauthors of the Quitkin et al study. In the follow-up studies, starting with the Stassen et al study in 1993, and then later, Szegedi et al tackle the problem of onset directly. All of these studies came up with roughly the same results, i.e., onset occurs within the first two weeks of treatment. The work of Szegedi and Stassen among others cited also found that “failure of the patient to respond with early improvement (within 2 weeks) almost certainly (>90%) results in failure of clinical response at treatment outcome”.

Responding to a published critique of our work by the Quitkin, Klein et al group in the early  90’s , we did  an analysis of their 1984 study that had reported a “lag” in onset of clinical actions, delayed for several weeks, the work that appeared to ignite the controversy.   We pointed out (Katz et al 1997) that the study, allowing for the fact that it was not targeted on the onset issue, was flawed in achieving that goal in several respects, notably (1) study dosage of drugs was relatively low, administered gradually during the first two weeks, so that the effective dosage was not reached until 2 weeks; (2) the study outpatients were only mild to moderately depressed; (3) measures of clinical actions during the first two weeks were inadequate. Quitkin, Klein and colleagues apparently disturbed by this analysis published a not very convincing rejoinder. (See exchange in Neuropsychopharmacology referenced in the initial essay).

I have returned the discussion to the central theme of this controversy because the over-scrutinizing of the secondary analysis from our 2004 study has clouded the basic issues, and diverted attention from the main, indisputable facts on the determination of onset, supported by the several meta-analytic and directly targeted studies on the issue. I am now satisfied that I have presented the case as clearly as possible.

It is most important to get the facts straight on this issue since the “time of onset” is critical, not only for clinicians for effective clinical practice, but for basic investigators attempting to uncover the mechanisms underlying therapeutic drug actions and seeking to develop new more effective drugs.


Katz MM, Berman N, Bowden CL, Frazer A. The componential approach enhances the effectiveness of 2-week trials of new antidepressants. J Clin Psychopharmacology 2011; 37: 193-218.

Katz MM, Koslow SH, Maas JW, Frazer A, Bowden CL, Casper R, Croughan J, Kocsis J, Redmond E. The timing, specificity, and clinical prediction of tricyclic drug effects in depression. Psychol Med 1987;17:297-309.

Katz MM, Tekell J, Bowden CL Brannan S, Houston JP, Berman N, Frazer A. Onset and early behavioral effects of pharmacologically different antidepressants and placebo in depression. Neuropsychopharmacology 2004;29:566-79.

Katz MM, Bowden CL, Stokes P, Casper R, Frazer A, Koslow S, Kocsis J, Secunda S, Swann A, Berman N. Can the effects of antidepressants be observed in the first two weeks of treatment? Neuropsychopharmacology 1997;17:110-1.

Kuhn R. The treatment of depressive states with G22355 (imipramine hydrochloride). American Journal of Psychiatry 1958;115:459-64.

Kielholz P. Die Behandlung endogener depresionen mit psychopharmaia. Deutch Med Wsche 1968;93:701.

Quitkin F M, Rabkin JG, Ross D, Stewart JW. Identification of true drug response to antidepressant. Arch Gen Psychiatry 1984;41:782-6.

Redmond E. The timing, specificity, and clinical prediction of tricyclic drug effects in depression. Psychol Med 1987;17:297-309.

Stassen HH, Angst J, Delini-Stula A. Time course of improvement under antidepressant treatment: a survival–analytic approach. Eur Neuropsychopharmacol 1993;3:127-35.

Szegedi A, Jansen WT, van Wugenburg AP. Early improvement in the first two weeks as predictors of treatment outcome in patients with major depressive disorder: a meta-analysis including 6,562 patients. J Clin Psychiatry 2009;70:344-53.

Martin M. Katz
March 17, 2016