“Non-reproducible single occurrences are of no significance to science”, this quote by Karl Popper often seems to get forgotten in medicine, particularly in alternative medicine. It indicates that findings have to be reproducible to be meaningful – if not, we cannot be sure that the outcome in question was caused by the treatment we applied.
This is thus a question of cause and effect.
The statistician Sir Austin Bradford Hill proposed in 1965 a set of 9 criteria to provide evidence of a relationship between a presumed cause and an observed effect while demonstrating the connection between cigarette smoking and lung cancer. One of his criteria is consistency or reproducibility: Consistent findings observed by different persons in different places with different samples strengthens the likelihood of an effect.
By mentioning ‘different persons’, Hill seems to also establish the concept of INDEPENDENT replication.
Let me try to explain this with an example from the world of SCAM.
- A homeopath feels that childhood diarrhoea could perhaps be treated with individualised homeopathic remedies. She conducts a trial, finds a positive result and concludes that the statistically significant decrease in the duration of diarrhea in the treatment group suggests that homeopathic treatment might be useful in acute childhood diarrhea. Further study of this treatment deserves consideration.
- Unsurprisingly, this study is met with disbelieve by many experts. Some go as far as doubting its validity, and several letters to the editor appear expressing criticism. The homeopath is thus motivated to run another trial to prove her point. Its results are consistent with the finding from the previous study that individualized homeopathic treatment decreases the duration of diarrhea and number of stools in children with acute childhood diarrhea.
- We now have a replication of the original finding. Yet, for a range of reasons, sceptics are far from satisfied. The homeopath thus runs a further trial and publishes a meta-analysis of all there studies. The combined analysis shows a duration of diarrhoea of 3.3 days in the homeopathy group compared with 4.1 in the placebo group (P = 0.008). She thus concludes that the results from these studies confirm that individualized homeopathic treatment decreases the duration of acute childhood diarrhea and suggest that larger sample sizes be used in future homeopathic research to ensure adequate statistical power. Homeopathy should be considered for use as an adjunct to oral rehydration for this illness.
To most homeopaths it seems that this body of evidence from three replication is sound and solid. Consequently, they frequently cite these publications as a cast-iron proof of their assumption that individualised homeopathy is effective. Sceptics, however, are still not convinced.
The studies have been replicated alright, but what is missing is an INDEPENDENT replication.
To me this word implies two things:
- The results have to be reproduced by another research group that is unconnected to the one that conducted the three previous studies.
- That group needs to be independent from any bias that might get in the way of conducting a rigorous trial.
And why do I think this latter point is important?
Simply because I know from many years of experience that a researcher, who strongly believes in homeopathy or any other subject in question, will inadvertently introduce all sorts of biases into a study, even if its design is seemingly rigorous. In the end, these flaws will not necessarily show in the published article which means that the public will be mislead. In other words, the paper will report a false-positive finding.
It is possible, even likely, that this has happened with the three trials mentioned above. The fact is that, as far as I know, there is no independent replication of these studies.
In the light of all this, Popper’s axiom as applied to medicine should perhaps be modified: findings without independent replication are of no significance. Or, to put it even more bluntly: independent replication is an essential self-cleansing process of science by which it rids itself from errors, fraud and misunderstandings.