Another analysis demonstrates: the evidence for homeopathy is not positive

A team of French researchers assessed whether a conflict of interest (COI) might be associated with the direction of the results of meta-analyses of homoeopathy trials. Their analysis (published as a ‘letter to the editor) is complex, therefore, I present here only their main finding.

The team conducted a literature search until July 2022 on PubMed and Embase to identify meta-analyses of randomized clinical trials assessing the efficacy of homoeopathy. They then assessed the existence of potential COI, defined by the presence of at least one of the following criteria:

• affiliation of one or more authors to an academic homoeopathy research or care facility, or to the homoeopathy industry;
• research sponsored or funded by the homoeopathy industry;
• COI declared by the authors.

The researchers also evaluated and classified any spin in meta-analyses conclusions into three categories (misleading reporting, misleading interpretation and inappropriate extrapolation). Two reviewers assessed the quality of meta-analyses and the risk of bias based. Publication bias was evaluated by the funnel plot method. For all the studies included in these meta-analyses, the researchers checked whether they reported a statistically significant result in favour of homoeopathy. Further details about the methods are provided on OSF (https://osf.io/nqw7r/) and in the preregistered protocol (CRD42020206242).

Twenty meta-analyses were included in the analysis (list of references available at https://osf.io/nqw7r/).

• Among the 13 meta-analyses with COI, a significantly positive effect of homoeopathy emerged (OR=0.60 (95% CI 0.50 to 0.70)).
• There was no such effect for meta-analyses without COI (OR=0.96 (95% CI 0.75 to 1.23)).

The authors concluded that in the presence of COI, meta-analyses of homoeopathy trials are more likely
to have favourable results. This is consistent with recent research suggesting that systematic reviews with financial COI are associated with more positive outcomes.

Meta-analyses are systematic reviews (critical assessments of the totality of the available evidence) where the data from the included studies are pooled. For a range of reasons, this may not always be possible. Therefore the number of meta-analyses (20) is substantially lower than that of the existing systematic reviews (>50).

Both systematic reviews and meta-analyses are theoretically the most reliable evidence regarding the value of any intervention. I said ‘theoretically’ because, like any human endeavour, they need to be done in an unbiased fashion to produce reliable results. People with a conflict of interest by definition struggle to be free of bias. As we have seen many times, this would include homoeopaths.

This new analysis confirms what many of us have feared. If proponents of homeopathy with an overt conflict of interest conduct a meta-analysis of studies of homeopathy, the results tend to be more positive than when independent researchers do it. The question that emerges from this is the following:

Are the findings of those researchers who have an interest in producing a positive result closer to the truth than the findings of researchers who have no such conflict?

I let you decide.