Trust but verify.
I heard this phrase frequently in medical training from well-seasoned EM attendings. It means trusting what you are told — and double-checking anyway. In my own practice, I've expanded the concept to include my thought processes, as well. Trust my intuition but verify. No one is out to trick you, but the reality of working in medicine is that it's full of potential missteps that can harm our patients. It's analogous to an operation, where everything is double-checked, from the right patient, to the right location, to the tools and medications that surgeons are handed. In the constantly busy ED, we are susceptible to distractions and biases that can cloud our critical thinking and judgment, which, in this case, are our tools.
This simple adage has served me well in my first year as an attending. We can avoid common pitfalls by maintaining a healthy level of suspicion, not so much of other people, but of ourselves. Some of the most widespread cognitive biases in medicine include confirmation bias and anchoring bias. Confirmation bias is essentially finding the diagnosis you are looking for and ignoring evidence to the contrary. We all want a story that fits. Things that don't fit shouldn't be ignored; they should be met with curiosity. A cousin of confirmation bias is anchoring bias: we find a diagnosis and stick with it. As an attending, I've found myself most susceptible to anchoring bias when working with residents and advanced practice clinicians. They will present a patient with a working diagnosis, and while they are frequently correct, sometimes I must look for clues to the contrary. Undoubtedly, the most difficult diagnoses I have made are the ones where one has already been established.
The first step in overcoming bias is recognizing the kinds of patients and situations where you are most at risk, and it's different for every clinician and specialty. If you've worked in EM, you've probably heard of the term "frequent flyer," a somewhat complicated and disparaging term that refers to a patient who presents frequently to the ED. These patients often have complex psychosocial issues and diseases, which, frankly, the ED is ill-equipped to handle — hence the revolving-door effect. Because these patients present frequently and often are well known to the medical staff, there is a false sense of "I know how this will play out: like every time before." Except for the one time a chronic alcoholic is really a life-threatening head bleed or the person with recalcitrant and thus far idiopathic abdominal pain who has had a CT scan of their abdomen every month for the past year presents just a little bit differently and has a ruptured appendix.
In a busy ED, it can sometimes feel like a nearly Herculean task to approach these patients with a clean slate, but we can only be truly good diagnosticians when the noise is quieted. Once you recognize the instances where you are most at risk for cognitive bias, you can practice what I think of as a forced pause, or cognitive time-out, to review what you could be missing.
There are more subtle ways patient presentations can be biased, too, such as when they are triaged at a lower level or placed in an urgent-care or lower-acuity location in the department. These context clues project that this patient is not sick and lull us into a false sense of security before we even open the exam room door. Triaging is done for good reason — we need to see the sickest patients first — but it is ultimately an informed guess based on a cursory history and initial set of vitals and should be treated as such.
Another less-recognized form of bias that enters the clinical picture comes from patients themselves. This can be thought of as patient-driven bias. Most patients I've cared for present with some idea of what they believe is going on. While early on in my training I found it frustrating and sometimes pushed back, probably more from a defensive position than anything else (wasn't I the one who went to medical school?), I've come to appreciate it as a signifier that the patient is invested in his or her health. It also often spurs more useful and engaging discussions ("I read it could be this … why don't you think it is?") where I can show my thought process and provide reassurance that Dr. Google or ChatGPT can't — and shouldn't.
My second year of residency, I had a patient present for shoulder pain that started when she was moving boxes several days prior. She was sure she had just pulled something, but the pain was becoming excruciating. I felt similarly confident it was the heavy boxes that had caused her injury, until I saw her troponin values and EKG. She was a classic case for atypical chest pain most commonly seen in elderly and female patients. It wasn't that she had intended to mislead me by any means; it was just the most obvious and less scary answer. Now I start many conversations about shoulder pain or abdominal pain with this simple acknowledgment: "Sure, it may well be a pulled muscle, heartburn, gas, or constipation, but here are a few things that I'm also thinking about. It is my job to think of the worst things first." When framed this way, it feels more like a collaborative investigation: a collective effort to avoid missing a diagnosis that could seriously hurt or harm them.
Being cognizant of biases and the potential for human error is not about seeding distrust in other healthcare workers or patients. On the contrary, the information they provide is invaluable, yet still susceptible to error. The more we can be aware of our liability to error and the occasions where we err the most, the more likely we are to succeed — together.
What's the last mistake you caught (or made)? Share in the comments.
