Methodological rigour and results of clinical trials of homoeopathic remedies

We studied security of randomisation vis-g-vis treatment effect in 26 clini cal trials of homoeopathic remedies, designated >>high quality<< in a publi shed meta-analysis, An invited international panel of acknowledged experts scored security of randomisation at two stages: firstly using the published trial reports, and secondly using these plus further clarificatory informa tion provided on request by trialists. The quality of blinding of outcome a ssessment was also scored, by two of the present authors. Security of rando misation at the two stages, and quality of blinding, were each plotted agai nst treatment effect size as calculated in the meta-analysis. A regression analysis featuring stage 1 judgements showed no positive trend; that featur ing stage 2 judgements showed a visible but not statistically significant t rend. No conclusions could be drawn from the data concerning quality of bli nding in relation to treatment effect. Of the 26 trials, only seven showed no effect or an effect favouring non homoeopathic treatment, and all three studies with objectively verifiable outcomes fell into this category. Retro spective verification of study quality is not an exact science, and the ava ilable data are thin; our findings are thus unlikely to sway anyone's antec edently held opinion about the scientific plausibility of homoeopathy, Larg e, multi centre trials of homoeopathic remedies, with minimum involvement o f practitioners in random allocation, and careful blinding, represent the b est way of advancing the debate.