June, 2017
By Charles A. Pilcher MD FACEP
In a March 2017 article in Emergency Medicine News, Dr. Justin Morgenstern astutely observes that “Doctors don’t make definitive diagnoses, [but] our patients are certainly under the illusion that we do.”
Dr. Morgenstern hits the nail on the head here. Uncertainty is a significant part of the practice of medicine, and how both doctors and patients deal with that uncertainty and communicate with each other about that uncertainty can play a significant role in improving patient safety. He proposes a novel concept: Determine the probability of a disease and quantify the level of uncertainty present in each “diagnosis.”
What Dr. Morgenstern is talking about here is one of the 4 most common “cognitive biases”: Diagnosis Momentum. Once a “diagnosis” is made, it tends to perpetuate itself, even if the patient’s course is not going as expected. Every provider must be astute enough to stop and ask “What else might this be?”
In the past 4 years, I have posted 3 “Perspective” articles on the role of cognitive biases in diagnostic error. [See links below. / CP] Besides diagnosis momentum, the other cognitive biases in the top 4 are anchoring bias, confirmation bias and over-confidence bias. Every med mal lawsuit contains an element of at least one of these – if not all 4. As a patient safety advocate, my goal is to do whatever we can to eliminate such biases and improve the quality, efficiency and safety of patient care.
Whenever patients are discharged from the ED, the last thing on their discharge instructions is something like this: “If you are not improving or have new or worsening symptoms, contact your doctor or return to the emergency department.” The problem is that when we mistakenly assume that we know exactly what is wrong, nothing unexpected should happen. Both we and our patients are failing to acknowledge the elephant in the room: medicine is not an exact science. Telling a patient that their chest pain is “pleurisy” does not guarantee that they don’t have a pulmonary embolus. Telling a patient that their abdominal pain is “gastroenteritis” does not guarantee that they don’t have a mesenteric thrombosis. Medical decisions are made on the basis of training, experience and statistical probability – not certainty.
The same problem with certainty extends to the rest of the care team when a patient is admitted. Med mal attorneys can all attest to patients for whom an admitting diagnosis from the ED was carried over to the inpatient setting – as if it were certain, e.g., CHF vs. pneumonia or meningitis vs. spinal epidural abscess. Choosing one over the other without leaving room for uncertainty. It creates “diagnosis momentum” and delays the critical thinking that is necessary should the diagnosis be wrong and the patient deteriorate. Based on one patient’s classic presentation, the diagnosis of CHF may be nearly 100% certain. However, should the presentation be less than classic, the diagnosis of CHF may be little more than a highly educated guess. Care for each patient may need to be markedly different. However, when a “diagnosis” like “CHF” is ascribed, it implies certainty.
Dr. Morgenstern’s point is that when handing off a patient, there needs to be some method of establishing the level of certainty regarding the diagnosis into the patient’s care plan. He proposes using “probabilistic notations to indicate our level of certainty about a diagnosis.” He suggests appending the percentage of certainty to every diagnosis, e.g., “CHF 99%” or “CHF 55%.” The care team is thus alerted to apply more critical thinking to the latter situation if the patient is not responding to treatment as expected.
The level of certainty also helps establish the time frame for outpatient followup. An obvious tibia fracture will need followup sooner than an equivocal knee sprain. Dr. Morgenstern uses the example that “The time frame for follow-up with rheumatology might be different for ‘temporal arteritis 99%’ versus ‘temporal arteritis 5%.'” When communicating among physicians, these percentages matter. Another example is that of a migraine patient being seen in the ED. “Wouldn’t it be nice to know if this was a definitive diagnosis (migraine 100%) or a provisional diagnosis (migraine 60%)? A provisional diagnosis is fine in theory; [doctors] understand that it is probabilistic. But provisional diagnoses quickly become permanent diagnoses in practice.”
“Calling a change in a provisional diagnosis an error is wrong,” asserts Dr. Morgenstern. “Unfortunately, the act of transcription into a chart has a way of transforming a provisional diagnosis into the final diagnosis.”
He acknowledges that his proposed solution is “inelegant” and needs further work, but I agree with his closing comment that “Whatever the solution, we need to embrace the role of probability in all medical diagnoses.”
[The July “Perspective” will cover this topic from a slightly different angle.
Perspective: Stop! Think! Listen! “Cognitive pause” will reduce medical errors