Everyone has cognitive biases that can lead to errors in diagnosis and have a profound impact on patient management decisions. In this thought-provoking session, Dr. Daniel Restrepo, a hospitalist in the department of medicine and the associate program director for point-of-care ultrasound for the internal medicine residency program at Massachusetts General Hospital, and assistant professor of medicine at Harvard Medical School, both in Boston, went through heuristics, common cognitive biases, and mitigation strategies you can implement in your practice. He brought new life into these topics with relatable real-life examples.
The session started by going through two heuristics, or mental shortcuts, hospitalists take to avoid rounding in the hospital all day. The first is representativeness. If a patient presents with classic symptoms of a very rare disease, some clinicians may be tempted to go down the diagnostic rabbit hole and order a slew of tests. However, this embodies the trap of representativeness, where you identified a constellation of symptoms that seemed to fit a pattern that you recognized but forgot to consider the low probability of the disease itself. Consider this—the prevalence of cholelithiasis in adults living in the U.S. is between 175 and 685 cases per 100,000 people, whereas the prevalence of autoimmune hepatitis is estimated between 10 and 15 cases per 100,000 people in European countries. That’s orders of magnitude more prevalent. In this memorable case, Dr. Restrepo ignored the base rate of disease or the pretest probability and began to look for autoimmune hepatitis because the symptoms fit so perfectly. In the end, probability won, and the patient was diagnosed with an atypical presentation of cholelithiasis. Remember that common things are common. Dr. Restrepo said it best. “Craft the differential based on pattern. Reshuffle it based on probability.”
The second heuristic covered in this session was availability. When it comes to diagnosis, many clinicians have heard about availability bias. This is the idea that diagnoses that come to mind more easily are overrepresented in a clinician’s thought process. For example, after just seeing a case of new anti-neutrophil cytoplasmic autoantibody (ANCA) vasculitis, subsequent patients might also start to look like ANCA vasculitis. This talk highlighted how availability bias can have a profound impact on management decisions. If a physician starts a patient with atrial fibrillation and a CHA2DS2-VASc of 6 on anticoagulation, and the patient then comes back with a massive head bleed, that physician is less likely to prescribe anticoagulation to the next patient with a CHA2DS2-VASc of 6 because their assessment of the bleeding risk is influenced by this recent experience. To combat this natural tendency, Dr. Restrepo recommends stepping back and thinking about alternative outcomes. Remember the patient who was NOT started on anticoagulation and then presented with a stroke. Play the odds. Remember the probability of outcomes.
After heuristics, the session moves through six biases frequently seen in hospital medicine: anchoring, premature closure, confirmation, framing, diagnostic momentum, and search satisfying. In the context of a hospitalist’s day-to-day work, for instance, once a urinary tract infection (UTI) or pneumonia is diagnosed, other diagnoses may not be considered. The search has been satisfied. However, misdiagnosis rates of pneumonia, cellulitis, and UTI are shockingly high: between 10 and 30%. That elderly patient coming in with a UTI might actually have a perforated bladder or bladder cancer. I particularly liked how Dr. Restrepo highlighted the collective tendency for “hyposkepticemia” in hospital medicine or the lack of skepticism when you take over a patient’s care from a colleague, which can lead to diagnostic momentum and add to chart lore. If a colleague hands off a patient with a chronic obstructive pulmonary disease exacerbation, how often does the next hospitalist go back and look for pulmonary function tests to confirm the original diagnosis? This also ties into the framing effect and can have profound implications on a clinician’s perception of the patient. If that same patient keeps coming in with a chronic obstructive pulmonary disease exacerbation, the next clinician might not recognize ischemia as the cause of the patient’s progressive dyspnea on exertion.
Lastly, Dr. Restrepo covered some mitigation strategies. He went through the effortful pause, playing the odds, diagnostic frameworks, working in teams, the note, follow up, and asking why. When coming on service, use the opportunity to pause and really examine each case. Trust but verify. Consider the probability of the disease in question and play the odds. As Dr. Restrepo put it, are all these cases really zebras, or are some of them just weird-looking horses? Diagnostic frameworks are also a great tool that not only helps teach learners about diseases but also acts as an automatic checklist. Many of us have the privilege to work in teams with great colleagues who can help see the case from multiple different lenses. After rounding, when finally sitting down to write the note, take the time to craft a helpful medical note that not only documents what happened to the patient but also crystallizes your thought process and diagnostic reasoning. Follow up on patients even when you go off service and even after they leave the hospital. Sometimes, the diagnosis is not clear until much later.
Lastly, stay curious and ask why.
Key Takeaways
- When you first take over a patient’s case, use it as an opportunity to think it through from a fresh perspective.
- Learn common diseases that are frequently misdiagnosed and examine these more closely.
- Play the odds. Don’t forget about disease prevalence when crafting your differential.
- Pause and re-examine your diagnosis, especially if the clinical course isn’t adding up or you’re jumping through hoops to make data fit.
- When something goes wrong, try also to remember all the times it went right.
- Don’t forget to ask why.
Dr. Shi
Dr. Shi is an adult hospitalist and clinical assistant professor of medicine at UC Davis Medical Center in Sacramento, Calif. She is currently involved in medical education and curriculum development.
I just retired, it was not “early” but it was partially triggered by tiring picking up patients with insufficient initial evaluation and the cognitive errors cited resulting in “wrong road” diagnosis and treatment. More frustrating was checking up on a patient handed off and finding that the secondary diagnosis found during the admission and noted in sign out was not addressed upon discharge, including at times the treatment initiated prior to the hand off not being ordered at the time of discharge. In my 16 years as a hospitalist I have noted many (but thankfully not all) hospitalists showing a lack of commitment to clinical excellence and a greater concern for being able to come late and leave early than optimal patient management.