Focus on Alternative and Complementary Therapies
www.pharmpress.com/fact
Focus Alternat Complement Ther©2005 Pharmaceutical Press
Focus Altern Complement Ther 2006; 11: 28–9
The recently published meta-analysis on homoeopathy by Shang et al. (Lancet 2005; 366: 726–32) has been criticised to be prone to selection bias: Its main conclusion is based on a subset of only eight trials out of 21 high-quality trials. The aim was to assess the sensitivity of the results according to which subset of trials is analysed.
We re-extracted all relevant data from the 21 original publications and defined various subsets of trials according to sample size, type of homoeopathy, type of publication and treated disease/condition. For each subset we estimated the overall treatment effects (odds ratio, OR) from random effect meta-analyses.
All 21 trials were highly heterogeneous (I2 = 62.2%) but showed a significant effect beyond placebo (OR = 0.76; 95% CI 0.59 to 0.99; P = 0.0392). This vanished when only the eight largest trials where analysed (OR = 0.88, CI 0.66 to 1.18; P = 0.4090; identical to Shang’s results). Sensitivity analyses showed that Shang’s negative results were mainly affected by one single trial on the effectiveness of Arnica montana in preventing muscle soreness.
Although the overall ORs were similar irrespective of which subset was analysed, the P values were not. Moreover, the high heterogeneity made the meta-analyses less reliable. Thus, Shang’s results – and even more the respective conclusions – are not as definite as they had been presented to be.
