The two new studies reconstructed the Lancet review and the main conclusions of that reconstruction are:

* That the results of the Lancet review were very sensitive to the definition of ‘large’ trials.
* Because of heterogeneity between the trials included in the review, its results are less definite than claimed. The conclusion that homeopathy is, and that conventional treatment is not, a placebo effect, was not based on a comparative analysis and is unjustified because of the heterogeneity of trials and lack of sensitivity analysis.
* The review did however, demonstrate that the quality of homeopathy trials was on average better than the conventional trials analysed.

‘The review gave no indication of which trials were analysed nor of the various vital assumptions made about the data. This is not usual scientific practice. If we presume that homeopathy works for some conditions but not others, or change the definition of a ‘larger trial’, the conclusions change. This indicates a fundamental weakness in the conclusions: they are NOT reliable”, said George Lewith, Professor of Health Research at Southampton University

The background to the ongoing debate is as follows:

In August 2005, The Lancet published an editorial entitled ‘The End of Homeopathy’, prompted by a review comparing clinical trials of homeopathy with trials of conventional medicine. The claim that homeopathic medicines are just placebo was based on 6 clinical trials of conventional medicine and 8 studies of homeopathy but did not reveal the identity of these trials. The review was criticised for its opacity as it gave no indication of which trials were analysed or the various assumptions made about the data.

Sufficient detail to enable a reconstruction was eventually provided and these two new studies are based on such a reconstruction and challenge the Lancet review. Specifically these two studies show:

* Analysis of all high quality trials of homeopathy yields a positive conclusion.
* The 8 larger higher quality trials of homeopathy were all for different conditions. Homeopathy works for some of these but not others, implying that homeopathy is not placebo.
* The comparison with conventional medicine was meaningless.
* Doubts remain about the opaque, unpublished criteria used in the review, including the definition of ‘higher quality’.

The Lancet review, led by Prof Matthias Egger of the Department of Social and Preventive Medicine at the University of Berne, started with 110 matched clinical trials of homeopathy and conventional medicine, reduced these to ‘higher quality trials’ and then to 8 and 6 respectively ‘larger higher quality trials’. Based on these 14 studies the review concluded that there is ‘weak evidence for a specific effect of homoeopathic remedies, but strong evidence for specific effects of conventional interventions’.

Due to a lack of funding, there are a limited number of homeopathic studies. As a result, it is quite possible to interpret homeopathic data selectively and unfavourably, which is what appears to have been done in the Lancet paper. If we assume that homeopathy does not work for just one condition (Arnica for post-exercise muscle stiffness), or alter the definition of ‘larger trial’, the results are positive. The comparison with conventional medicine was meaningless: the original 110 trials were matched, but matching was lost after the trials were reduced to 8 in one group and 6 in the other. Interestingly, the quality of homeopathic trials was better than conventional trials.

This reconstruction casts serious doubts on the Lancet review, showing that it was based on a series of hidden judgments unfavorable to homeopathy. An open assessment of the current evidence suggests that homeopathy is probably effective for a number of conditions including allergies, upper respiratory tract infections and ‘flu, but more research is desperately needed.

Prof Egger has declined to comment on these findings.