The holy grail of clinical reasoning is, in a word, assessment. Ought we to measure clinical reasoning as a function of experience, knowledge base, or as a process measure? In medical parlance, there is no “gold standard.”

In 2015, to tell whether your doctor is a great diagnostician is based more on reputation than hard evidence. This gap in evidence is all the more interesting given recent press about the inattention given to diagnostic error in decision making.

In a recent grand rounds, an expert in clinical reasoning outlined practical strategies to improve thinking, which would be familiar to an elementary school educator: Prime students before setting them loose on tasks, focus on knowledge as “context-specific” and promote reflective practice. Despite these excellent teaching strategies, progress in medical education appears stagnant compared to the exponential rise of specialty knowledge. The reasons, I believe, have to do with the contexts in which medical education takes place. Here are a few principles that may guide a more data-driven approach to assessment.

Leave the academic medical center

Students and residents currently spend the majority of their training in large academic medical centers (AMC). This occurred nearly accidentally, as the post-war period yielded increased demand of hospitals, and consequently, trainees. The educational rationale for training students in AMCs is simple: That’s where the “real” disease is. In order to diagnose it, you must see it!

While initially intuitive, this argument fails the context-specificity test. Students and residents are asked to rotate through highly subspecialized services, with the concept that this “will be relevant” to future practice. It is easy to cherry-pick examples on both sides of this argument. But the point is that inpatient education is at odds with the reality that after training most care takes place in the outpatient setting. If we agree that students ought to focus more on nuances of symptoms versus laboratory findings, providing more robust training in the community may be a good place to start. Transition to community-based third-year curriculum is taking hold at medical schools across the country, including within Harvard Medical School’s new curriculum.

Develop new tools for assessing diagnostic accuracy

Currently, the best methods we have for assessing clinical reasoning include faculty evaluations, peer reviews, and knowledge assessments. It is important to mention clinical reasoning is not an end in itself, but a means to patient outcomes. This feels akin to an investment banking firm evaluating interns on their process for choosing stocks, while ignoring if their picks bankrupt the firm. In the zeal to define clinical reasoning, we have overlooked the potential value of including patient data as a component of resident feedback. I have written elsewhere about how to develop competencies in population health management.

On the inpatient side, there may be low-hanging fruit in evaluating resident diagnostic accuracy from overnight rotations, and using diagnostic errors as platforms for refining clinical knowledge. Other metrics for patient outcomes could be related to appropriateness of testing — for instance, what percentage of tests ordered changed management decisions on real patients. To create valid metrics for diagnostic accuracy will be difficult, but likely easier than continually seeking the distant holy grail of assessment.

Focus on collective competence over individual competence

Dr. Lorelei Lingard, a leading researcher in health communication, has written extensively about the over-reliance in medical training on individual over collective competence. Despite the push towards team-based collaboration in medical care, the dominant educational methods are individual, with the exception of team-based training for responding to medical codes. How do you define the effectiveness of inpatient medical ward teams? What are the characteristics of teams that reduce patient length of stay, reduce readmission, or result in fewer diagnostic and systems errors? To approach clinical reasoning through the lens of team-based care fits with the realities of inpatient medicine — nearly 90 percent of thinking does not take place at the bedside, where more than one provider has unlimited access to medical information.

It may be that the best way to the holy grail of assessment is to ignore it, and focus more on the realities of contemporary medical practice: the move toward community health, the promise of big-data, and expansion of the medical team.

Tom Peteet is an internal medicine resident.