Theory, Evidence, and Clinical Outcome

To what degree is theory evidence? Scientific theories are supposed to represent the best, most tested and confirmed evidence from observation and hypothesis testing, molded into concepts that aid human understanding. Yet, the modern human mind easily blurs distinctions among theory, hypothesis, mechanism, observation, and opinion, often to someone’s disadvantage or even peril.

This concern has a significant implication for treatment planning. I routinely observe practitioners choosing antidepressants based on presumptions about the efficacy of certain molecular structures or receptor binding characteristics, while ignoring, or at least deemphasizing, clear direction from excellent outcome studies.

Theories are intended as tools we can use to alter undesirable situations. In early human development, our brains were selected to respond to our environment by prioritizing survival – of the species, if not the individual. Our genes still code for these responses, choosing coherence with our self-image and with socialization over our culturally developed rationality, i.e., the scientific method.

Even those of us who strive to utilize the scientific method on a daily basis must struggle to employ the slower Type II reasoning over our innate and rapid Type I programing. As we are uncomfortable with uncertainty, we easily default to the same simple (and often irrational) explanations. This process rewards our self-esteem and reduces our anxiety, yet leads us astray half the time.

The human brain has also evolved to prefer explanations of mechanism (even unproven ones) over clear outcome data. The observationally proven cure for scurvy was delayed implementation for 47 years, only because a mechanistic explanation eluded everyone. This is where a misunderstanding of the nature or inappropriate application of theory leads to harm. This example also illustrates the difference between a scientific theory and a scientific law.

While a theory should represent the best synthesis of current knowledge derived from the scientific method, a law is an observation so reliable that outcomes are predictable. In this case, that diets low in vitamin C will lead to scurvy and that diets containing sufficient amounts of the vitamin will treat and prevent it. Natural laws depend neither on theory nor mechanistic understanding – they simply, and accurately, predict outcomes. In psychiatry, for example, we don’t know why convulsive treatments [e.g., electroconvulsive therapy (ECT)] quickly help desperately depressed people who have ceased to eat or care for themselves, we just know that they do (and how to use these treatments safely).

In medicine, then, well designed outcome studies are superior to theory. Knowing the mechanism of an intervention can help us refine them and identify future novel approaches, but it is the outcomes that drive our clinical impact, not theories about mechanism (including ones that contain some evidence of validity).  

We too often put the cart the before the horse, and ignore what valuable information we already have.   As clinicians, we must guard against our preference for mechanistic explanations over outcome studies, and put the what will happen before the why, lest we misdirect ourselves.