What works for whom? Determining the efficacy and harm of treatments for pain.

There has been a tension between the needs of regulators and industry to demonstrate that interventions are effective and safe, and the needs of professionals to understand how well interventions will work for their patients, and patients to understand what might work for them as individuals. The custom has been to focus on statistical outcomes based on average results, but in-depth analysis based on outcomes obtained by individual patients demonstrates that few are average. Rather, a minority of patients achieve very large reductions in pain (responders), while the majority achieve little (nonresponders). Those who benefit in terms of pain also benefit in other areas, with improved sleep, fatigue, mood, function, quality of life, and ability to work. This changes how benefit and risk are seen; nonresponders should stop treatments that don't work and not, therefore, be exposed to risks, while responders have very large benefits to offset against rare but potentially serious harm. This alternative view, patient-centred and practice-orientated, has major implications for clinical practice, how and why we do clinical trials and how they are designed, how health economic evaluations are done, for decisions made by regulatory and other bodies, and for the theory and practice of evidence-based medicine.