Letters Section Editor: Jody W. Zylke, MD, Senior Editor.
Author Affiliations: Division and Center for Evidence-based Medicine and Health Outcomes Research, University of South Florida, Tampa (Dr Djulbegovic) (email@example.com); and National Health Service, Bedfordshire, Bedford, United Kingdom (Mr Paul).
In Reply: Dr Vohra and colleagues question our statement that inductive reasoning, which relies on past observations, can produce logically guaranteed truthful conclusions. The same applies to n-of-1 trials: there is no logical guarantee that treatment that worked in 1 patient, even if randomized, will generate the same results when applied in the future to the same patient whose circumstances will almost invariably be different from those in the original trial. For example, consider an n-of-1 trial testing the effect of a statin in a patient for primary prevention of coronary heart disease. Such a trial would typically involve 6 random treatment sequences in which the statin would be randomly interspersed with no treatment at 6-week intervals. If the trial had sufficient power (a debatable issue in n-of-1 trials) to detect a beneficial effect on biochemical markers such as total or low-density lipoprotein cholesterol, then both efficacy and effectiveness data should be produced to justify the treatment. However, as the patient ages, starts exercising less, develops diabetes or hypertension, starts smoking, or changes his or her diet, the conditions of the original trial no longer exist. Can the results from the original n-of-1 trial be extrapolated to the “new” patient? Perhaps, but there is no logical way to prove that the effects will remain the same. As a result, physicians will always have to extrapolate beyond available evidence in their attempt to tailor treatments to an individual patient's circumstances.
Djulbegovic B, Paul A. Efficacy Research and Unanswered Clinical Questions—Reply. JAMA. 2011;306(7):709-710. doi:10.1001/jama.2011.1151