This Month’s Expert: Michael Thase, M.D., on Deciphering Antidepressant Research

This Month’s Expert: Michael Thase, M.D., on Deciphering Antidepressant ResearchTCPR: I know that you published a famous meta-analysis several years ago in which you compared the remission rates of patients on venlafaxine with SSRIs and placebo. Can you remind us of what that study showed?

Dr. Thase: This report, which was done in collaboration with two researchers from Wyeth, was based on the first eight randomized, double-blind, head-to-head comparisons between venlafaxine and SSRIs (Thase et al, Br J Psychiatry 2001 Mar;178:234-241). Five of the eight studies used fluoxetine, two used paroxetine and one used fluvoxamine. It didn’t include any sertraline or citalopram studies, and escitalopram wasn’t an approved medication at the time, so it wasn’t included either. That paper was heavily dependent on the studies that used fluoxetine.

TCPR: And I remember clearly that there was kind of a rule of thumb that everybody was discussing after your study came out which was that we can expect remission rates of 45% on Effexor, 35% on SSRIs and 25% on placebo. Is that an accurate description of your findings?

Dr. Thase: Yes, it is an accurate summary. There is so much cynicism today about the results of industry-sponsored research. The findings became controversial, both because there’s honest room for disagreement and because the findings were marketed very vigorously by Wyeth and “counter-detailed” by the manufacturers of the SSRIs – Lilly, GlaxoSmithKline, Pfizer, and then later, Forest.

TCPR: Looking back on the results with the perspective of time, do you think the main result is still valid – that Effexor is better at producing remission than the SSRIs?

Dr. Thase: I think that it is a true finding, although the degree of venlafaxine’s advantage may have been overestimated in the original analysis, because the main SSRI comparator was fluoxetine. Recall that fluoxetine and its metabolite have very long elimination half lives, so that it takes about six weeks before it gets to steady state. Because of this, you may not see fluoxetine’s full effectiveness until you get out to at least eight or nine weeks. But even this point is still pretty controversial. It also may be true that a dual reuptake inhibitor has an advantage in some patient groups – such as older patients with more severe depression – but little if any advantage in others, such as younger women with less severe depressive episodes.

TCPR: Has the venlafaxine advantage held up in more recent studies?

This article originally appeared in The Carlat Psychiatry Report -- an unbiased monthly covering all things psychiatry.
Want more, plus easy CME credit?
Subscribe today!

Dr. Thase: A somewhat smaller average difference has been observed in the more recent studies, perhaps because most of those trials use lower doses of the extended release formulation of venlafaxine. So in the more recent wave of studies, the remission rates on venlafaxine have been more in the lower 40s on average, whereas the SSRI remission rate has remained steady at 35%. I have always felt that when you use the whole dose range of venlafaxine – up to 375mg/day – you do get a few more percent of people better. One caveat that I should mention is that there are still only two direct comparisons of venlafaxine and escitalopram and, pooling these two studies, there is no evidence of an advantage for venlafaxine.

TCPR: Has the reported remission rate from SSRIs started to creep up at all from that 35% figure?

Dr. Thase: Actually, that 35% remission rate on the SSRIs has held rock solid compared to about a 25% remission rate on placebo. The only studies that report higher remission rates for SSRIs have also reported higher placebo response rates. You’ll see a few studies showing a 45% remission rate on SSRIs, but they don’t emphasize that in those studies there is a 35% remission rate on placebo.

TCPR: I’m not sure I understand. Why do the placebo rates increase? Is it possible to manipulate placebo rates to make your drug look better?

Dr. Thase: It’s not really a matter of “manipulation.” Over the years, as the benefits of antidepressants have become increasingly well known, the proportion of people with significant depression who are willing to come into a placebo-controlled study has been shrinking and shrinking.

TCPR: In other words, the people who come into antidepressant trials now are not as sick as they used to be?

Dr. Thase: Yes. The placebo response rate is often an indication of how easy to treat the patient population is. There is a paper by Walsh in which they looked at the temporal trend in placebo response rates to antidepressants from 1970 to 2000, and it has been increasing slowly and steadily (Walsh et al, JAMA 2002;287(14):1840-1847). So it doesn’t surprise me to see placebo response rates in new studies of 45%, 48%, even 50%.

TCPR: And how should a higher placebo response rate affect our interpretation of these studies?

Dr. Thase: The real benefit is measured by how much the active drug surpasses the placebo response. Typically, a good, strong antidepressant will cause 10-20% of people to respond more than a placebo, but it will also lose about 5-10% more of the participants because of dropouts due to side effects, because placebo doesn’t have strong side effects, by and large. For example, if you take an initial 15% difference, subtract the 5% due to dropouts, then that is your standard 10% difference in an intent-to-treat analysis. Thus, if you see a 55% remission rate vs. a 45% placebo rate, this is really no better than a 35% vs. 25% placebo rate.

TCPR: Meaning that the seemingly higher remission rates that we have been seeing in recent SSRI studies can’t fairly be compared with the venlafaxine remission rates from older studies, because the placebo rates were lower back then.

Dr. Thase: Right.

TCPR: More recently, we have seen results from the STAR-D study. I know you were one of the investigators on those trials. The results have been viewed by many as discouraging; what’s your take?

Dr. Thase: STAR-D really illustrates how pernicious a condition depression is, how difficult to treat it is, and how in our optimism to help people we sometimes overestimate the likely benefits of particular treatments. In STAR-D’s case we also were somewhat surprised by the risk of relapse after successful treatment. Everybody in STAR-D got free medication and excellent followup care at a very low cost and for an entire year after they were successfully treated. If you think about most health service delivery systems, it doesn’t get any better than that. And in that context, about 50% of the patients had sufficient increase in symptoms over a year to have met the definition of relapse. So when you do finally bring about remission in somebody with a difficult-to-treat depression, you can’t simply relax. At the very least you need to provide psychoeducation to enhance treatment adherence.

TCPR:How do STAR-D’s results fit in the context of your belief that venlafaxine has an advantage over SSRIs?

Dr. Thase: When patients were switched from citalopram to venlafaxine, they had a 25% remission rate vs. an 18% remission rate on sertraline. Interestingly, the ratio of 25 to 18 is very similar to the ratio of 45 to 35, the response rates in my original meta-analysis.

TCPR:But the difference wasn’t statistically significant in STAR-D.

Dr. Thase: The difference wasn’t statistically significant because the treatment groups contained only about 220 patients and in order to find a difference of that size, you needed to have about 500 patients per cell.

TCPR:Can you elaborate a bit on that?

Dr. Thase: Most studies comparing active treatments aren’t large enough to truly resolve whether one active treatment is better than another. Studies that are calibrated to show the difference between a good drug and a placebo don’t need huge numbers, but in order to show a difference between a good drug and a better drug, you need a very large number of subjects, and there are very, very few studies that are that large. Regrettably, STAR-D is not one of those studies.

Our policy on financial conflict of interest

Readers have wondered why we sometimes interview experts who have financial ties with pharmaceutical companies. TCPR accepts no money from the pharmaceutical industry, nor do we allow any of our writing staff or editorial board members to accept such money. HOWEVER, it is extremely difficult to locate national experts who do not have such ties. Furthermore, involvement with industry often leads to experiences and perspectives that are helpful in understanding crucial issues in research and treatment. Dr. Carlat conducts all expert interviews personally, and edits any content that appears unbalanced. Hopefully, this allows us to showcase the major figures in psychiatry without the injection of commercial bias. But ultimately, you, the reader, must be the judge.

This Month’s Expert: Michael Thase, M.D., on Deciphering Antidepressant Research

This article originally appeared in:

The Carlat Psychiatry Report
Click on the image to learn more or subscribe today!

This article was published in print 1/2007 in Volume:Issue 5:1.


APA Reference
Thase,, M. (2013). This Month’s Expert: Michael Thase, M.D., on Deciphering Antidepressant Research. Psych Central. Retrieved on December 4, 2020, from


Scientifically Reviewed
Last updated: 20 Aug 2013
Last reviewed: By John M. Grohol, Psy.D. on 20 Aug 2013
Published on All rights reserved.