An emergency physician colleague shared this recent experience:
A 45-year-old healthy male presented to the emergency department with a complaint of palpitations and feeling like he was going to pass out, which began while walking to his car after work. He had had 10 days of worsening shortness of breath, congestion, and generalized malaise, with a negative flu swab at urgent care 4 days ago. His wife had had upper respiratory infection symptoms that began at the same time, but she had been feeling better for several days now. The above presentation in the setting of COVID-19 just beginning to spread through New York state led me to prematurely narrow my differential diagnosis. If it weren't for the good working relationship I have with my colleagues, the final diagnosis of multiple pulmonary emboli would have been missed entirely. I'm grateful that I was at least aware of the possibility of my own cognitive errors, and that I was open to input from others, which helped me avoid a disastrous outcome in this case.
When a single diagnosis is top of mind every waking hour and permeates our brains during sleep, there is no better setup for diagnostic error. Self-awareness of our clinical decision-making process and the possibility of diagnostic error is highly relevant in these times. Availability and premature closure are biases every clinician should be aware of during this crisis.
Availability bias is defined as the propensity to put particular weight on a diagnosis because of the ease with which relevant examples come to mind. COVID-19 is not only available, it is everywhere: in the news, our thoughts, and our fears.
Premature closure occurs when the clinician fails to consider the full differential diagnosis and jumps to the first diagnosis that comes to mind. Premature closure has been likened to looking at a litter of puppies and falling in love with the first one you see without checking all the puppies.
In thinking about the current exam room environment, it is easy to see how a combination of availability bias and premature closure can combine to be deadly for patients.
Another critical factor in diagnostic accuracy is understanding the variability in presentation of a disease. Representative bias occurs when we look for the classic and do not consider the variant presentations. COVID-19 does not always present with fever. Sometimes there is minimal cough. Diarrhea is also said to be a feature. Anosmia and ageusia are also seen. Unfortunately, we do not yet have a clear picture of the variant phenotypical presentations of this illness. This fact means that we must consider COVID-19 to be possible in a variety of presentations, particularly in locations where it has high prevalence.
The enormity of the current crisis and the risk to healthcare personnel is also a large factor in our biases. The decision-maker is at personal risk of becoming a seriously ill patient. Since this is a highly communicable disease, it means that clinicians will assume the patient is SARS-CoV-2 positive to protect the healthcare setting and themselves. With that said, the great challenge of clinical diagnosis in these times means we must understand pre-test and post-test probabilities based on where we are practicing.
The pretest probability of a patient having COVID-19 versus another diagnosis is dependent upon the community base rate of COVID-19. If you are in New York City today, the pre-test probability is very high. Current SARS-CoV-2 testing is said to be 70% sensitive, which means that any patient in New York City with clinical symptomatology of COVID-19 should be treated as positive and quarantined even if the SARS-CoV-2 test is negative. On the other hand, if you have no symptoms suggestive of COVID-19, a negative test is reassuring. In rural areas where there is a low community rate of COVID-19, alternative diagnoses are much more likely.
Hopefully, soon we will have a rapid, highly reliable, and both sensitive and specific SARS-CoV-2 test. When the test universally arrives, we will be able to "rule out" COVID-19 quickly. However, we will still need to leverage our clinical acumen and have a clinically meaningful differential diagnosis. In the meantime, those working in areas with a high prevalence of disease will treat most patients as possibly infected and do their best to consider alternative diagnoses given the reality of their high-pressure circumstances. In areas that have few COVID-19 cases, sound, logical clinical reasoning and judgment remains imperative. For patients presenting with several different symptoms – ageusia, anosmia, sore throat, fever, cough, diarrhea, and/or respiratory failure – COVID-19 is top of mind anywhere, but we must also consider alternative diagnoses. Without widespread testing availability, and with a SARS-CoV-2 test that apparently is not highly sensitive, our clinical logical competence is now more important than ever.
Art Papier, MD, is associate professor of dermatology and medical informatics at the University of Rochester, as well as CEO and co-founder of . Paritosh Prasad, MD, MBA, DTM&H, is associate professor of surgery, medicine, & pediatrics at the University of Rochester Medical Center, where he directs the Highly Infectious Disease Unit, and serves as director of global health at VisualDx.