Evidence-based medicine is ultimately buttressed on research findings which can be turned into guidelines for everyday patient care. The quality of the research must be such that physicians and patients can rely on the resulting guidelines. Problems with guidelines lead not only to poor quality patient care, but undermine the entire movement to evidence-based medicine. The most trustworthy guidelines would be based on evidence from randomized clinical trials, but these are time-consuming, expensive and often difficult to conduct for certain kinds of treatments. Yet other kinds of studies, based on epidemiology and observation, can conflate cause and effect and lead to guidelines which are inappropriate. An editorial in the Journal of Managed Care Pharmacy explores these issues and explains why non-randomized trials often produce poor guidance. (JMCP Article)
The editorial focuses on several randomized trials of very widely promoted and followed guidelines based on epidemiologic studies; guidelines for common conditions which affected millions of patients. In one instance, promotion of very tight glycemic control for type 2 diabetes patients was shown to have no better ultimate outcomes and to increase adverse events, including death. Similarly, very tight blood pressure control for such patients with cardiovascular risk factors also produced the same outcomes, with worse adverse event rates. The authors identify several reasons why a spurious cause and effect relationship was inferred in the original observational studies leading to promulgation of the guideline, including regression to the mean, confounding, and over-reliance on complex statistical methods.
This editorial is a must read for anyone involved in promoting evidence-based medicine. If we are going to expect physicians to follow, and in some cases be penalized or rewarded for following, care guidelines, they and their patients have to have a very high level of confidence that the guideline will produce better outcomes and less risk of harm. That means that guidelines must be based on well-designed, repeated randomized clinical trials involving patients and circumstances similar to those the physician will see in daily practice. Not cheap, but anything less puts patients at risk.