How to read a homeopathy trial without fooling yourself

Why this matters for practitioners

You will be asked about evidence — by patients, by colleagues in other systems, sometimes by regulators. Answering well requires being able to read a trial properly, including the ones whose conclusions you would prefer.

The discipline is the same one you apply to a case: judge the method before the outcome.

The checklist

1. What was actually randomised and blinded?

Double-blind, placebo-controlled is the claim to look for, and it should be stated explicitly. An open-label or observational study can be interesting, but it cannot separate the remedy from expectation, natural course, and the effect of the consultation itself.

2. How many participants?

Small trials produce dramatic results in both directions by chance alone. A positive result in 30 patients is a reason to run a larger trial, not a conclusion. Be as sceptical of a small positive as you would be of a small negative.

3. Individualised or fixed prescription?

Specific to our field, and often skated over. A trial giving every participant the same remedy for the same diagnosis is not testing homeopathy as it is practised. A trial permitting individualised prescription models practice better — but is harder to blind and to replicate. Neither design is free; know which one you are reading.

4. Was the outcome decided in advance?

Look for pre-registration. If the primary outcome was chosen after the data came in, the result means much less than it appears to. This is not a homeopathy-specific problem — it affects trials across all of medicine — but it applies here too.

5. Who summarised it, and did you read past the abstract?

Abstracts overstate. Advocacy sites and debunking sites both quote selectively. Read the methods section, or be honest that you are repeating someone else''s reading.

Meta-analyses need the same scrutiny

A review is only as good as its inclusion criteria. Two reviews of the same literature can reach opposite conclusions purely by choosing differently which trials qualify. When someone cites a meta-analysis at you, the useful question is "which trials did it include, and on what grounds?"

The honest position

The literature contains a substantial proportion of inconclusive results, some positive trials, and a real replication problem — a summary of the current numbers is in where the evidence base stands. Overstating this does the profession no favours; nor does refusing to look at it.

Being able to say "here is what the good trials show, here is what remains unresolved" is a stronger professional position than confidence in either direction.