Mark Tonelli (University of Washington) with Jon Williamson (University of Kent)
Over the last two decades, medical epistemology has been largely focused on epidemiologic information derived from various forms of clinical research and from statistical manipulations of empirical data. With this prioritization, skepticism regarding the value of other kinds of medical knowledge, in particular knowledge of mechanisms, has risen disproportionally. This lopsided skepticism is most evident in the ‘hierarchies of evidence’ promulgated by proponents of evidence-based medicine. Some philosophical attention has been paid to the role of mechanism in medicine, particularly as it relates to determination of causality and, most recently, the potential value for clinical guideline development. The role of mechanism and mechanistic reasoning in clinical practice has been relatively unexamined. We argue that mechanistic understanding and pathophysiologic reasoning remain an integral part of clinical practice, crucial for evaluating reports of clinical research, arriving at diagnostic and treatment decisions, and assessing treatment effects. Some account of biologic plausibility in the form of a putative mechanism of action accompanies virtually every clinical trial describing a new intervention. This mechanistic description may provide support for making a causal inference regarding a positive statistical association and can also be viewed as a component of the background knowledge that clinicians bring to a Bayesian assessment of new information. A lack of a convincing mechanism linking treatment to effect leads to clinicians discounting such an association. In coming to diagnostic and therapeutic decisions regarding patients, clinicians use mechanistic reasoning to extrapolate from population-level research. Pathophysiologic rationale is necessary for both analogical reasoning from “average” patients and for extrapolating from reference classes determined by population-level research. Finally, mechanistic reasoning is crucial for assessing whether a medical intervention is working as anticipated in particular individuals. When not, changes in treatment strategy are usually necessary, changes that will also be guided by pathophysiologic reasoning. The use of mechanistic reasoning in clinical practice is ubiquitous and epistemically justified. The high degree of skepticism toward appeals to mechanism in practice is inappropriate, particularly as understanding of mechanisms in medicine continues to improve. Reasoning about and from mechanism will remain a crucial skill of clinicians aiming to optimize the care of individual patients.
