“Remember, diseases don’t read books.”
Yousri Youssef Mishriki, MD

“Uncommon manifestations of common diseases are more common than common manifestations of uncommon diseases.”
Yehia Yousri Mishriki, MD

Why can diagnosis in medicine, at times, be so difficult?

Here is a paraphrase of a typical sentence in my internal medicine textbook when I was a student/resident – “In disease x, one finds an elevated white blood cell count. At times, however, the white blood cell count can be normal or low.”

It would not suffice to know the breadth and depth of medicine, which is impossible, to always correctly diagnose one’s patients but one must also be aware of the variability of presentation of all those diseases. Sir William Osler knew as much when he wrote his epic book, ‘The Principles and Practice of Medicine’ in 1892 wherein he described the various ways in which typhoid can present and cautioning that, “This diversified symptomatology has led to many clinical errors…”. The more modern medical literature also describes the varieties of ways diseases can present atypically. For example, Cushing syndrome, due to excess cortisol production, has occasionally been described as “cyclical” rather than continuous, with variable other presentations such as periodic hormonogenesis, unpredictable hypersecretion of cortisol, fluctuating cortisol excretion or intermittent Cushing syndrome. Even in monogenetic illnesses, where one would expect a stereotypical presentation, there can be a variability of presentation due to environmental effects and exposures, post-zygotic mutations, and epigenetic influences.

In his book, ‘Thinking Fast and Slow’, Daniel Kahneman described two methods of thinking when making decisions. In the fast pathway, pattern recognition is fast, intuitive, and often driven by heuristics (cognitive shortcuts for reaching immediate solutions to a problem) and is primarily subconscious. In the slow, analytic pathway, thinking is deliberate, effortful and under conscious control. Physicians toggle back and forth between these two methods depending on the details of the cases they are analyzing, their experience and knowledge and their whim. Tversky and Kahneman posited that most inferential errors produced by heuristics and biases were due to “faulty probabilistic reasoning”. However, Jain did not agree that the diagnostic process was probabilistic, and I believe that he is, at least partially, correct. Nevertheless, once an initial diagnosis comes to mind as a result of a heuristic, biases are bound to arise, particularly anchoring and confirmation biases which interfere with one’s ability to objectively proceed with the evaluation. Furthermore, one must simultaneously avoid the diametrically opposed biases of representativeness restraint, a tendency of looking for typical manifestations of a disease while ignoring atypical variants (“when you hear hoof beats, think of horses and not zebras”) and aggregate bias, a belief that general population data do not apply to one’s patient. A setting for cognitive dissonance, if ever there was one.

What are clinicians to do? Those in the know recommend cognitive bias mitigation which involves “deliberate switching” from intuitive to analytical processing and the use of “debiasing strategies”, what I term ‘metacognition’ or thinking about one’s thinking. After one has generated an initial possible diagnosis or diagnoses, one must step back and carefully question oneself as to whether one has fallen prey to some of the more common biases. Needless to say, this is time consuming, and time is a very rare but critical commodity in accurately diagnosing one’s patients.