International Clinical Pharmacology, (1996), 11, 137-145

Efficacy of venlafaxine and placebo during long-term treatment of depression: a pooled analysis of relapse rates

A.R. Entsuah1, R.L. Rudolph1, D. Hackett2 and S. Miska1

1Clinical Research and Development, Wyeth-Ayerst Research, PO. Box 8299, Philadelphia, Pennsylvania 19101-1245, USA, and 2Wyeth-Ayerst Research, Paris, France

Correspondence to: A. R. Entsuah at above address

The objective of this analysis was to determine the efficacy of venlafaxine in comparison with that of placebo during long-term treatment. A pooled analysis of relapse rates in outpatients with major depression continuing long-term treatment (up to 12 months) after responding to short-term treatment (6 weeks) was performed combining the data from four randomized, double-blind, placebo-controlled clinical trials. Relapses were defined as two consecutive Clinical Global Impression (CGI) severity scores greater than 3 (mildly ill), as a CGI severity score greater than 3 at withdrawal regardless of the reason for withdrawal, or as withdrawal due to lack of efficacy. Data from 304 patients (185 venlafaxine, 119 placebo) well balanced for baseline characteristics were included in the pooled analysis. Percentages of patients completing the long-term phase were 38% venlafaxine and 26% placebo (p = 0.034). Cumulative relapse rates by 6 months of long-term treatment were 11 % venlafaxine and 23% placebo (p = 0.019). Cumulative relapse curves for the venlafaxine and placebo groups over the 1-vear long-term treatment differed significantly (p = 0.022). The results from this analysis indicate that long-term treatment with venlafaxine in patients with major depressive disorder is effective in maintaining the initial response compared with placebo and suggest that venlafaxine will be effective in the prevention of relapse. (Authors' abstract)


Comment by CM: The design of this long-term study seems significantly biased towards a positive finding for the active drug, and would over-estimate the long-term effectiveness of the drug in general practice. This is because the design excluded: [a] several classes of more vulnerable patients; [b] people whose depression markedly improved after a short course on placebo; and [c] anyone who had previously demonstrated intolerance to the study drug, by failing to complete a six week trial period under treatment.

The results of this study suggest that, if 100 patients came to a general practice for treatment for depression, perhaps 40 would either not need the active drug (strong placebo responders) or find it intolerable (drop outs), or be at greater risk of having unwanted effects. Of the 60 patients remaining, 27 on the active drug would not have a relapse or recurrence of depression for one year - though 18 patients would have done as well on placebo. In other words, the advantage claimed might be in practice be available to about one in patient in every ten.

As a statistically significant result in favour of venlafaxine, this legimises the claim that "long-term treatment with venlafaxine in patients with major depressive disorder is effective ...". Nor, if the advantage were real, should one begrudge it to the small minority of patients who might benefit.However, there is room for doubt that the benefit is as great as claimed, notably because:

[a] Twice as many placebo-treated patients as those on venlafaxine discontinued treatment at the patient's request - for reasons which included "feeling better";

[b] "About three quarters of the patients in each treatment group who met the criteria for relapse discontinued because of lack of efficacy"; and

[c] the proportion of placebo-treated patients who discontinued because an adverse reaction (7.6%) seems very high in relation to those on venlafaxine (10.8%). These figures bear no close resemblance to the data reported in all phase II and phase III studies (Physicians' Desk Reference, 1996):


"Nineteen percent 537/2897 of venlafaxine patients in phase 2-3 depression studies discontinued treatment due to an adverse event.The more common events (>1%) assocated with discontinuation and considered to be drug-related (i.e., those events associated with dropout at a rate approximately twice or greater for venlafaxine compared to placebo) included:

CNS Venlafaxine Placebo
Somnolence 3% 1%
Insomnia 3% 1%
Dizziness 3% -
Nervousness 2% -
Dry mouth 2% -
Anxiety 2% 1%
Nausea 6% 1%
Abnormal ejaculation* 3% -
Headache 3% 1%
Asthenia 2% -
Sweating 2% -
* Percentage based on No. of males

Wyeth was asked to comment on these points, but declined.

Back to 4.2