To the Editor The recent article by Kravitz and colleagues1 reporting the results of a randomized clinical trial comparing n-of-1 trials with standard care for treatment of chronic musculoskeletal pain raises a number of interesting issues that merit further examination.
Weighed on a continuum of probability, it is certainly possible that there is a positive effect of these trials on pain interference score, which is unlikely to be because of chance alone based on the observed results; however, this effect is not demonstrated to the arbitrary level that is customary for statistical significance. This speaks directly to 2 points: (1) the P value standards we set are subjective and should be viewed not as dichotomous yes/no evidence, but as a continuum of probability as put forth in the American Statistical Association statement on P values,2 and (2) the clinical significance of the findings of this trial are ignored in favor of identifying statistical superiority for pain-interference score. The stated goal of the authors was to establish the “benefits of participating in an n-of-1 trial, not to assess the superiority or inferiority of any particular treatment.”1(1369) The improvement in shared decision making regarding medications is perhaps even more meaningful than a narrow statistical focus on pain interference score as an outcome owing to the mistrust of health care professionals that has become pervasive in the medical system and litigation that is a prominent feature of US medicine.