The importance of using placebo controls in nonpharmacological randomised trials

Novel pharmacological compounds must undergo a series of highly regulated steps and have their efficacy demonstrated under strict conditions of placebo-controlled trials before being approved for clinical use. This is often not the case for treatments that do not involve a pharmacological element, such as surgery, physiotherapy, or psychological therapy. Despite the recognised need for evaluation, there are currently no formal requirements to test the efficacy of nonpharmacological medical procedures. Failing to recognise that even large positive or negative effects may be caused by biases, rather than the medical properties of a treatment, may have serious consequences, as ineffective interventionsmay continue to be used, eg, spinal fusion for nonspecific back pain. Alternatively, effective treatments may be abandoned because negative effects are misattributed to the treatment. While some nonpharmacological treatments (Table 1), such as physiotherapy, are generally safe, and if they are not effective, the only harm may be a delay in providing effective therapy, others, such as surgery, are inherently associated with risks. If ineffective surgery continues to be used, not only does it waste time and resources, depriving patients of better treatment, but it also exposes patients to the risks associated with the procedure itself or anaesthesia, without any clinical benefits to justify them. In this topical review, we argue that not testing the efficacy of nonpharmacological procedures is problematic and should be addressed. We also outline possible steps to promote highquality nonpharmacological efficacy trials. 2. Improvement and bias