Skip navigation

Herbal remedy for depression

Systematic reviews which tell us something we didn't know, not only qualitatively but quantitatively as well, are few and far between. A first class review on the use of the herb St John's wort (Hypericum perforatum) for depression is a superb example [1].

Depression is a major public health issue, affecting many people, many of whom do not wish to use powerful synthetic antidepressants which can have significant adverse effects. Some herbal remedies can be effective, as Bandolier pointed out for gingko biloba in peripheral vascular disorders in Bandolier 18 .

St John's Wort

Extracts of this herb have long been used in folk medicine. In Germany it is licensed for use in anxiety, depression and sleep disorders. The extracts contain many different chemical classes, so the "active agent" is a matter of uncertainty.

Quantitative Systematic Review

The review sought studies which were randomised comparisons of Hypericum extracts against placebo or other antidepressants. They found 23. When judged against a validated quality scoring system, almost half scored 80% or more of the possible points available.

The judgement of success was set by the reviewers as treatment responders. To be a responder, patients had to have a Hamilton depression score of less than 10, a reduction in Hamilton depression score of at least 50%, or be much improved or very much improved on a clinical global impressions index. Treatment or observation periods were usually four to eight weeks.

Seven different preparations had been tested. Daily doses of extract varied considerably in the trials, and dose was not a criterion used to judge efficacy.

Hypericum against Placebo

Thirteen trials had extractable data. The overall response rate with placebo was 22% compared with 55% with Hypericum. Results of these trials are shown here in two plots.

The L'Abbé plot shows that 11 of the trials were in the upper left half of the graph, demonstrating effectiveness. A ladder plot of the risk ratio shows that nine of the 13 trials, and the combined estimate, all showed Hypericum to be significantly better than placebo (combined risk ratio 2.7, 95% CI 1.8 - 4.0).

The pooled NNT calculated from these numbers was 3.0 (95%CI 2.6 - 3.8). This means that for every three patients with depression treated with Hypericum, one more will have been relieved of depression whose depression would not have resolved on placebo.

Hypericum against conventional antidepressants

Three studies compared Hypericum alone against rather low doses of maprotiline, imipramine or amitryptiline in trials involving 300 patients. There was no statistical difference in the trials singly or combined.

Adverse events

In placebo-controlled trials, 4.1% of patients reported adverse events with Hypericum compared with 4.8% on placebo. Major adverse events (withdrawals) were 0.4% and 1.6%.

In active controlled trials, 20% of patients receiving Hypericum reported adverse events compared with 36% with standard antidepressants (but remember the low doses). This yields a number need to harm for standard antidepressant (NNH) of 6 (95% CI 4 - 12); this means that for every six patients with depression treated with standard antidepressant, one fewer will have an adverse event who would not have had one if treated with Hypericum. Adverse event withdrawals were 0.8% and 3% for Hypericum and standard antidepressants respectively.


The authors make several points.

  • That none of the articles would have been found if searches had been restricted to the English language.
  • That multiple publication is a problem. One trial was published five times with two different first authors.
  • That pooling studies of herbal preparations is problematic, because you really don't know what's in them.
  • That despite these promising results, we need trials to evaluate the effectiveness of Hypericum against other antidepressants, and the trials need better defined classification of disease.


  1. K Linde, G Ramirez, CD Mulrow et al. St John's wort for depression - an overview and meta-analysis of randomised clinical trials. British Medical Journal 1996 313:253-8.

previous or next story in this issue