Randomized Trials and Scientific Methods

We enjoyed the article by Smolle et al¹ that discussed a randomized clinical trial of homeopathy for warts. We strongly agree with the statement in their introduction that ". . . scientific methods are valuable tools for distinguishing helpful alternative medical methods from superstition and quackery."

Scientific methods do not necessarily require a randomized trial. We can argue that sometimes a randomized trial is harmful. When a trial is performed, a test of statistical significance is done, such as a t test. The standard in clinical medicine for a statistically significant result is a P value less than or equal to .05. Accepting a P value of .05 means that there is a 1 in 20 chance that a therapy without merit could be shown to be significant. One way to decrease the likelihood of getting a spurious result is to limit trials to topics that have a potentially medically explainable result. Is there even 1 scientifically defensible hypothesis for the therapeutic basis of homeopathy? Have any basic science experiments ever shown that the chemically pure water used in homeopathy is different from normal water?

A trial of a therapy with no medical basis is dangerous because by chance alone the trial could suggest that the therapy is effective. A potentially false-positive trial of a novel therapy may be followed by a second trial using a large number of patients. The likelihood that the second, larger trial would also give a false-positive result is low. But what about the interval between the trials? How many patients would spend their money and time chasing an ineffective therapy? Further, some alternative medicine practitioners and customers may choose to look at only "positive" studies. These problems could be avoided by limiting randomized trials to therapies that have a rational physiological explanation.