Balancing benefits and harms: the example of non-steroidal anti-inflammatory drugs

To provide safe and effective interventions for people, reliable and valid evidence is needed. This is most easily produced by undertaking trials in samples of people who are as homogeneous as possible and applying the results to similar, well defined groups of patients. To be equitable, however, appropriate care needs to be provided for everyone in the diverse community using health services. Therefore, there is a tension between obtaining scientific evidence that is reliable but which can be applied only to a small subset of the population, and distributive justice that requires that all in need are treated equally appropriately. Drugs have potential harms as well as benefits. Doctors would like to be able to balance any risks against benefits to derive a therapeutic ratio for each patient, but this is difficult. Formal trials can tell a lot about the efficacy of a drug in a specific context, but unless they are huge and pragmatic they tell less about a drug's toxicity. Post-marketing surveillance may uncover more information on toxicity, but the data usually lack sufficient detail to lead to an understanding of the determinants of adverse reactions. Furthermore, extrapolation of the efficacy or toxicity of a drug in one disease or group of patients to those associated with different diseases or groups can be difficult and misleading. We examined aspects of these problems in the context of one commonly prescribed class of drugs—non-selective, non-steroidal anti-inflammatory agents (NSAIDs)—and their use in the management of joint pain. We compared the types of patients in whom trials of non-selective NSAIDs are conducted with those who receive the drugs in practice. Then we examined the prevalence and associations of adverse events in these two groups. ### Methods As NSAIDs are primarily used to treat arthritic pain, we studied trials of these drugs in patients with …