Learning to Savor Uncertainty
Author: Dan Dworkis, MD-PhD, FACEP (@TheEmergMind, Emergency Medicine, USC Keck School of Medicine) // Reviewed by: Alex Koyfman, MD (@EMHighAK) and Brit Long, MD (@long_brit)
“You must not merely tolerate uncertainty, you must savor it. Or you won’t last long.”
Uncertainty is one of the core realities of emergencies, so learning to operate under uncertain conditions is one of the fundamental jobs of the emergency medical provider. When a patient arrives needing emergency care, there is a vast landscape of uncertainty we need to chart a course through as we work to diagnose and treat their illness.
Why are they having trouble breathing? Is it a severe pneumonia, are they choking on a sandwich, do they have advanced lung cancer? What is the underlying driving process of this crisis, and what are our options or resources? What might happen if we choose to perform one treatment over another?
Uncertainty is everywhere during an emergency, and it can easily feel overwhelming, especially when we are early in our careers. What we do know though is that we cannot always wait for certainty to act – in other words we need to become comfortable acting with limited and imperfect information.
For example, if our patient arrives gasping for air and covered in an itchy rash, we often need to assume they are having an anaphylactic reaction and consider delivering an injection of epinephrine to halt their progression to life threatening shock. If we first required confirmation from the local restaurant that the dish they just ate does, in fact, have peanuts in it, we might be too late, and the patient might suffer or die.
A first step toward increasing our comfort with uncertainty is to dive into the multiple ways in which uncertainty can present during an emergency. Sometimes, the uncertainty we face is obvious: we know that we lack knowledge in a particular area, but the shape of that hole is itself informative and it lets us prioritize our actions and the care we deliver.
Why is this patient short of breath? We are not sure, but we can provide oxygen and obtain studies like a chest x-ray or lung ultrasound to help us better understand the situation. How many patients will be arriving from the multi-car accident? We do not know, but we think there will be at least three critical so we can set up four trauma bays to be prepared.
Other times, the uncertainty is less obvious, and we have to purposefully look for it in order to see it. A crucial example of potentially hidden uncertainty is the idea of test error. When we perform a test, we tend to take the answer that the test comes up with as actual fact for that patient; however, since no test is perfect, the result a test returns is best interpreted not as the ground truth, but as a guess of what the patient is actual experiencing. Depending on the situation, that guess might be a good guess or a bad guess, but either way it is a guess, not the absolute truth. Underestimating the uncertainty with which a test delivers an answer can lead to significant errors in our understanding of the disease process and optimal treatment for a particular patient.
As an example, consider a test to measure the sodium level in a patient’s blood—which we routinely run as part of every basic metabolic panel. Sodium measurements are generally fairly accurate, and the “true” value of the patient’s sodium is typically thought to be within 1-2 mEq/L of the reported value in most cases. While this level of uncertainty might seem small, when treating a critically hyponatremic patient where you are attempting to slowly and safely raise their sodium levels to a more appropriate range, being aware of the range of potential values around the reported value can make a large difference in your treatment decisions.
Additionally, tests can sometimes be “tricked” by other factors outside of what they are directly measuring (making the test result a “bad guess”). In the case of measuring sodium, elevated levels of blood glucose distort test results markedly, leading to falsely low sodium readings. Without awareness of the “hidden” uncertainty in these test results, treatment plans might be ineffective or even potentially harmful. 
Since uncertainty is such a large part of operating during a crisis, learning to “savor the uncertainty” is one of the key steps in learning to practice emergency care. How can we accomplish this? As a first step, early career providers can recognize that generating absolute certainty is not the goal: efficient and skillful action in the face of uncertainty is.
Often, we do not need to completely understand what is happening with a particular patient to know what we need to do to help her. Simply noting to ourselves when we are striving (quixotically) for certainty instead of action can help us reorient our priorities to providing emergency care now, then going back to try to resolve uncertainty.
Second, we can work to learn the physiological signs of uncertainty in our own body and mind. When faced with decisions in uncertain conditions, do we physically tense up? Experience abdominal pain? Are we short with our friends or coworkers? Sometimes what really is uncomfortable is not the uncertainty itself, but how we have conditioned our body to react to it.
Addressing the signs of uncertainty—for example by using deep breathing techniques to counter uncertainty-induced shortness of breath—can often help mitigate the fear of uncertainty and help us generate a more rational response to uncertain conditions. After all, the same mystery of not knowing what is around the next curve is mentally mapped to fear by some and adventure by other and to some extent how we interpret these physiological signals is a choice, not a given.
Finally, where possible, we can identify potential outcomes and train for contingencies. If we are uncertain about the result of a particular event but have developed well-tested backup plans with which we feel confident and prepared, then we will become more tolerant of the potential downside of the situation and therefore the uncertainty it involves. Becoming excellent at airway control and ventilator management for example, allows us to better tolerate the uncertainty of how to help an individual who is short of breath because we know if the situation deteriorates further, we will be in a better position to provide care.
Ultimately, becoming comfortable with uncertainty is not something that can be taught to us externally the way we can teach doses of medications or steps of procedures. Instead, learning to savor operating under uncertain conditions is something we must take responsibility for ourselves and consciously study. Uncertainty is not going away, but we can take the advice of mindfulness expert Jon Kabat-Zinn, who said, “You can’t stop the waves, but you can learn to surf.”
- The accuracy of a particular test (sodium or otherwise) for a particular patient is impacted by the characteristics of the test itself and how well the test applies to the patient at hand. The estimate of 1-2 mEq/L provided here is based on a historical conversation in 2013 with laboratory technicians at Massachusetts General Hospital in Boston, MA, whom I asked to estimate the error of the sodium measurements based on the methods they were using at the time, and might or might not be representative of the test error at a different hospital or at a different point in time. The reference laboratory at a hospital should be able to tell a provider the characteristics of a particular test at that hospital, which depends on the particular assay and equipment used. That said, the art of mapping the test to the patient, is on the provider, not the lab. Sometimes, as in the case of correcting sodium levels for hyperglycemia, there is a simple formula which allows us to mitigate the uncertainty of how a particular test result applies to a particular patient. See for example https://www.mdcalc.com/sodium-correction-hyperglycemia, which describes the classic equation correcting sodium measurements in this case. Other times, the relationship between the reported value and the ground truth is more complicated or might not be known.