Laboratory-reproducible effects deserve more weight than human trial results.

Evidence

But there is also a good reason for assigning a different value to laboratory-tested mechanisms of action than to a “proof of effect” in humans. As already explained in “He who heals is right”, there are a large number of interfering factors in experiments on humans which can only be controlled with great effort – and these experiments can only ever be repeated to a limited extent. In the case of laboratory results, the transferability to humans must always be checked, but the reliability of the data itself can be raised to an incomparably higher level. There is therefore a considerable difference between whether a study is simply intended to show that a well-understood effect that can be clearly demonstrated in the laboratory still works in the complex overall human system, or whether an effect is claimed solely on the basis of error-prone testing on an inevitably limited number of people. — Holm Gero Hümmler, Ulrike Schiesser, Fact and Prejudice, 771