Skip navigation

Homeopathy - dilute information and little knowledge

The effectiveness of complementary medicine (whatever that means exactly) and its place alongside other healthcare services is one that continues to perplex. Arguments are usually unencumbered by evidence. But for homeopathy a new analysis [1] of placebo-controlled studies gives an insight into the effectiveness of this intervention.

Klaus Linde's team have done a great job of finding all the trials, and looking for factors which may have led to over-estimations of treatment effect through publication bias and other forms of bias. The headline result - that on balance homeopathy works - may be presented by others in a more positive light than the conclusions reached by the team - "The results of our meta-analysis are not compatible with the hypothesis that clinical effects of homeopathy are completely due to placebo" and "..we found insufficient evidence from these studies that homeopathy is clearly efficacious for any single clinical condition".

What they did

They sought all the trials on homeopathy, in any condition and with any homeopathic remedy at any dilution, which fulfilled the following inclusion criteria:
  1. Controlled studies
  2. Parallel group design
  3. Randomised, or with descriptions of double-blinding that meant they had to have been randomised
  4. Written reports (including abstracts, theses etc)
  5. Had data on outcomes that could be extracted for data analysis.

They found

Eighty-nine trials could be analysed. They broke down like this:
  • The median number of patients studied in each trial was 60.
  • There were 24 clinical categories.
  • There were four types of homeopathy.
  • There were 50 classes of homeopathic remedy.
So clearly this was not a particularly homogeneous bunch of trials. The number of trials in each clinical class, and the number where homeopathy beat placebo are shown in the Table. Overall, homeopathy beat placebo in only 42% of the trials. So one conclusion was that in 6 out of 10 trials, homeopathy could not be shown to have any benefit over placebo.

Trials in Homeopathy
Number of trials
Clinical class Total Homeopathy beats placebo
Allergy 7 4
Dermatology 9 3
Gastroenterology 9 3
Musculoskeletal 6 2
Neurology 7 4
Obstetrics & Gynaecology 10 5
Chest infection, asthma, ENT 15 4
Rheumatology 7 4
Surgery & Anaesthesia 12 4
Miscellaneous 7 4
This might be seen by some as more instructive than odds ratios, which are the results we are given in the paper. Bandolier has always had trouble with mere statistical outcomes, but in the main that is all we have here. For the record, the odds ratios favoured homeopathy. The odds ratio was 2.5 (2.1 to 2.9) for all 89 trials, though lower at 1.7 (1.3 to 2.1) for high-quality trials.

Special cases

There were two circumstances in which there were similar outcomes and conditions, though predominantly with different homeopathic interventions. For one of those Bandolier could extract information for NNT calculations.

For patients with ocular symptoms from allergic disorders relieved or much better with Galiphimia glauca at four weeks, the relative benefit was 1.4 (1.2 to 1.6) and the number needed to treat (NNT) was 6.6 (4.4 to 13). This means that for every seven patients treated with Galiphimia glauca for four weeks, one will have ocular symptoms improved or relieved who would not have done had they been treated with placebo.

Comment - double standards?

This is an important paper because it applies good meta-analytic principles to a difficult subject in complementary therapy. It examined all comparisons of homeopathy against placebo, and because there was a statistically significant outcome, this will be interpreted by some as "homeopathy works".

Well maybe. But there just is not enough information in any one condition with any one homeopathic treatment to say that homeopathy should be used. If this were a new treatment in conventional healthcare, we would look at it with a very cold and fishy eye. Certainly no conventional therapy would be allowed to have so many different conditions and variations bundled together to try to reach a conclusion.

A sceptical reader might say "If this is the best they can do, why bother?" - but Bandolier couldn't possibly comment.


  1. K Linde, N Clausius, G Ramirez et al. Are the clinical effects of homeopathy placebo effects? A meta-analysis of placebo-controlled trials. Lancet 1997 350: 834-43.

previous or next story in this issue