Abdulnour and associates have announced the availability of a new tool called NEJM Healer, which provides deliberate practice of diagnostic reasoning at a virtual bedside. Their rationale for developing this new tool was explained thusly.
The acquisition of expertise requires highly motivated, focused repetition combined with specific and timely feedback to hone and perfect performance. Studies have shown that time spent on this particular type of practice, called deliberate practice, is a stronger predictor of clinical performance than either academic aptitude or experience.
As I read their report, I couldn’t help but recall several specific instances from the past. I remember proctoring a medicine resident as he did a history and physical examination on a real patient, selected by me, from the general medicine service. The patient had been diagnosed with milk-alkali syndrome and had a good understanding of how that affected his health. He was also alert, cooperative, and anxious to help. The resident, on the other hand, tried to take a short-cut, and knowing I was a nephrologist, kept cutting off the patient when he talked about his dyspepsia, a classic instance of premature closure. As I pointed out to the resident at the time, if he had just kept quiet and listened, he would have gotten the whole story quickly and accurately.
On another occasion, I had just sat down for lunch when the beeper went off with a page to the ER. A resident got on the line and presented the case of a man found down and brought to the ER unresponsive. Lab work had shown he had a severe anion-gap acidosis that had not responded to their attempts to correct it. I immediately suspected a toxic alcohol poisoning, and gave instructions for emergency measures while I arranged for acute hemodialysis. By the time I got to the ER, the resident had read up on toxic alcohol syndromes, and commented to me: “It sure is hard to think of something you’ve never seen before.” The truth of that comment has stuck with me ever since.
While both of these anecdotes illustrate timeless truths, I think there is another issue at work that may be different now than in the past. When I was learning the craft, the elders emphasized the primacy of the history, and the need to construct a coherent narrative based on what the patient said and what was found, or not found, on physical examination. Laboratory testing was used to confirm or reject the original diagnostic hypothesis. I don’t remember any particular shame if the original diagnostic hypothesis was wrong, provided the thinking was logical. Of course, everyone knew the problem of breakdowns and miscommunication between patient and physician, which is why it was thought of as a hypothesis.
Now, however, lab is often obtained before the patient is even seen. If lab tests were highly sensitive and highly specific, this might be okay, but the problem stated in Bayes’ theorem, wherein most tests are useful if and only if there is at least a 10% prior probability of disease, is routinely ignored. Even though the EMR requires a diagnosis before ordering tests, many “order sets” have batteries of such tests, some of which are required for payment purposes as much as for diagnosis. And working from the lab tests to the patient is both error prone and more expensive.
While this is an issue, I think a bigger issue is the virtual disappearance of the clinical narrative, being replaced by a computer checkbox with a series of pre-programmed positives or negatives, again designed to make sure the coders are satisfied. Over the years the “best doctors” I worked with were those who could present a coherent narrative that turned out to be an accurate representation of the data. The true stars could do so with few unnecessary words. Developing this skill obviously takes time and practice, but in our current environment, such behavior is not rewarded. The pay is the same, regardless of the quality of the work, and excellence is almost irrelevant.
I have discussed previously the notion of tacit knowledge, which is acquired through practice. It clearly leads to more efficient, cost-effective practice, but where is that captured in all the discussions about value-based care? Clinical reasoning, along with seeing more rare conditions, are the one skill that can improve with age and practice. I suspect driving the computer program rapidly is a skill that does not improve. And the computer, which does not “forget” anything, brings with it another challenge.
Over the past three years I have had the opportunity to provide virtual consultations to primary care providers whose patients have limited access to specialty consultation. Of course, the nature of the questions asked varies widely, sometimes reflecting the nature of the patient’s problem, but many times the question arises because the provider has been overwhelmed by data. They have many data points, but no story, and therefore no way to link the data to the patient in a way that allows for action. I find myself trying to suggest a story, based on the limited data provided and without talking to the patient myself, then encouraging them to explore this hypothesis to see if it makes sense after getting back to the patient. The feedback suggests they find this helpful and gives them renewed confidence to address the patient’s issues.
For many years it has been clear that lab (and X-ray) data are seen by many physicians as both more “true,” and more valuable, than information from the history and physical exam. Now, I suspect we find the computer more “reliable” and truer than the patient. While I applaud the notion of using the computer to train people in diagnostic reasoning, I think the challenge of improving clinical reasoning is greater than the authors recognize.
22 May 2022
 Abdulnour R-EE, Parsons AS, Muller D, Drazen J., Rubin EJ, Rencic J. Deliberate Practice at the Virtual Bedside to Improve Clinical Reasoning. NEJM 2022;386(20):1946-1947. doi:10.1056/NEJMe2204540.
Are We Too Task Oriented?
The number of tasks doctors must complete grows exponentially. Have we become too task oriented at the expense of our patients?
Fluid Intelligence Versus Crystallized Intelligence
More on Biases
A recent series of articles in the New England Journal of Medicine provide more insight into the issue of bias in medical decision making.
Physician Decision Making
Physician decision-making is both complex and deals with uncertainty dooming current simple approaches to changing physician behavior.
The Anchoring Heuristic
Businessmen and health policy experts fail to recognize the limits imposed by the experiential nature of medical practice, both of which impact achieving the "triple aim."
What Makes a Successful Physician?
What skills are required to be a successful practicing physician?