Problematic placebos in physical therapy trials

The function of a placebo control in a randomised trial is to permit blinding and reduce risk of bias. Adopting Grűnbaum’s definitional scheme of a placebo, all treatments must be viewed as packages consisting of characteristic and incidental features. An adequate placebo for an experimental treatment contains none of the characteristic features, all of the incidental features, and nothing more. For drug treatments, characteristic features can be readily identified, isolated, and separated. By contrast, physical therapy treatments often involve features such as patient-therapist contact and sensory feedback that make this separation difficult both conceptually and practically. It is therefore unsurprising that attempts to construct placebos for physical therapy treatments have in the past led to biased estimates of treatment effects. In this perspective piece, we describe the problem with constructing placebos for physical therapy trials drawing upon Grűnbaum’s definition and using paradigmatic examples from existing literature. We conclude by submitting that in the many cases where an adequate placebo cannot be achieved, alternative trial designs, e.g. dose-response or comparative-effectiveness trials, carry a lower risk of bias and should be favoured.