Extraordinary claims need extraordinary evidence – a few thoughts about biological plausibility

Evidence‑based medicine (EBM) was developed to make clinical decisions more reliable by grounding them more solidly in good research. Thus, randomised clinical trials, systematic reviews, and meta-analysis became crucial for healthcare. That development brought undeniable progress, but it also created a problem: if we focus exclusively on such evidence, we might neglect an important question:

IS THE TREATMENT IN QUESTION BIOLOGICALLY PLAUSIBLE?

Put simply, EBM asks “Does it work in this study?” without first asking “Could it reasonably work at all?”

The neglect of biological plausibility can lead to wasted resources, misleading conclusions and, in some cases, the promotion of nonsense. The issue is, of course, particularly relevant in so-called alternative medicine (SCAM) known for its frequent lack of plausibility. A simple example might explain this more clearly: in homeopathy, we see an abundance of poor-quality studies with a positive result. This could easily lead to the overall impression that homeopathy works, while in fact it cannot reasonably work at all.

So, how can we reasonably take account of this complication? It turns out there are several options:

Option 1 Gatekeeping

One way to account for plausibility within EBM is to use it to decide what we test in the first place. Before launching an expensive clinical trial, we can ask for a clear explanation of how the proposed intervention might reasonably work. If no such rationale can be articulated without contradicting science, it is reasonable to conclude that the intervention lacks sufficient plausibility to justify the time, money and ethical burden involved in testing it on patients. In practice, this kind of gatekeeping often happens informally, but making it explicit and mandatory could help keep overtly implausible interventions from consuming scarce resources.

Option 2 Prior probability

Plausibility can also be integrated into how we interpret trial results. Some trialists treat a statistically significant result as an infallible signal that the therapy was effective. When a trial result is “statistically significant”, it means the data we observed would be unlikely if the treatment had no effect.  Prior probability is another way of expressing plausibility. If a hypothesis is highly plausible given existing scientific knowledge, a positive trial fits into a broader, coherent picture. If a hypothesis is highly implausible, a positive trial is more likely to be a false positive, an artefact of bias, chance, methodological flaws, or fraud. In other words, for low‑plausibility claims, we need stronger and more consistent evidence before accepting them as true. The less plausible a claim is, the more extraordinary the evidence must be.

Option 3 Guidelines

Guideline development offers another opportunity to embed plausibility into EBM. When expert panels prepare recommendations, they typically grade the strength of evidence according to study design, risk of bias, and consistency of results. They might also add a distinct step in which they rate the plausibility of the intervention. This rating could be justified explaining how well the intervention fits with established knowledge. Guideline writers could then let this plausibility rating influence the strength of their recommendations.

Health technology assessments have been moving in this direction for some time. It makes guideline documents more transparent: clinicians could see not only what the trials showed, but also how the intervention was judged to fit into or contradict broader scientific understanding.

Option 4 Causation

Finally, causation frameworks are being used to bring plausibility into EBM. When we decide whether an association is causal, we often rely on criteria such as consistency, temporality and strength of association. Biological plausibility is another of these criteria. Using it systematically means asking whether there is a logical pathway from intervention to outcome that passes through known mechanisms and observed effects. If such a pathway can be sketched in a way that accords with science, plausibility is high. If not, plausibility is low, and we should be more cautious about drawing causal conclusions from statistical associations alone.

EBM has revolutionized healthcare, but evaluating evidence in a vacuum can carry the risk of validating the absurd. To minimise this risk, we might consider integrating biological plausibility into EBM, a possibility that has long been discussed by many experts in the field. This approach is not a rejection of EBM, but a vital safeguard for it which ensures that the evidence aligns with and strengthened by fundamental science and existing knowledge. By demanding extraordinary evidence for extraordinary claims, medicine can better protect its resources, maintain intellectual integrity, and ensure that clinical practice rests on a foundation that is both statistically sound and scientifically reasonable.