by Vadim Shteyler
A.F. was an elderly patient admitted to our service for a diagnostic work-up and management of a large pocket of pus surrounding her lungs. Until recently, she was very independent and in good health; this was her third hospitalization for the same reason in one month. Radiographic imaging was consistent with pneumonia but other causes could not be ruled out. She had not responded to antibiotics, she had no other signs of infection, and numerous cultures from her blood, pus, and sputum failed to grow microbes. Extensive testing for other possible causes was also negative. At that point, we all had the same suspicion—cancer. Some tumors in the chest can cause inflammation that may look like a pneumonia and result in a collection of pus. That inflammation can also hide the tumor on imaging. In fact, it would be a few weeks, after we drained all of the pus and the inflammation subsided, until we would have a clearer image of the lungs. Though cancer was a plausible explanation, we had no evidence at that time. Should we have discussed our concerns with A.F.? The diagnosis was not certain, so we didn’t…
In daily clinical practice, uncertainties come in many forms. Outcomes for most medical interventions are probabilistic (they are not 100% predictable). And those probabilities are often ambiguous (they are more often ranges than specific percentages) or simply unknown. At a broader level, science is underdetermined, medicine is inductive, and innumerable non-medical forces influence the medical landscape (biases, conflicts of interest, values, etc.).
How effectively providers communicate uncertainty is…well, uncertain. One study showed that in at least some situations, such as diagnosing the cause of abdominal or back pain, primary care providers did a good job. In more stressful settings, they did not fare as well.
Why providers don’t address uncertainties has been widely discussed in medical literature. Some providers may be uncomfortable with uncertainties themselves. Some blame medical education for fostering unwavering confidence. Some blame time constraints. Providers may fear that disclosing uncertainties would sacrifice trust, confidence, and satisfaction. They may worry that probabilities and ambiguities of complex medical information are too difficult to understand. And they may feel that recognizing uncertainty can be overwhelming, especially for serious conditions and impactful interventions. Trying to provide comfort is human. Discussing uncertainty is hard. And sometimes, platitudes are the best people can come up with.
Recent attention to awareness about cancer overdetection with routine screening (finding cancers that would have never caused symptoms) may have stumbled upon another reason for omitting uncertainties—maximizing utilization of interventions categorically considered “good.” Preventive medicine is often seen as uniformly beneficial. Early diagnosis and treatment decrease morbidity and mortality and save future healthcare dollars. In fact, 8 of the only 33 ACO quality metrics focus on preventive medicine, 2 of which are colorectal and breast cancer screening. Well-intentioned providers, biased in favor of cancer screening, may unwittingly omit discussing the uncertainties of screening tools. Surveyed women, for instance, largely reported being unaware of overdetection on mammography but found the information important. Some clinics, afraid of dissuading patients from screening, omit this information purposefully.
If society agreed that universal screening were worth the risk of overdetection, then omitting this information might be less consequential. A recent survey, however, found very large variation in peoples’ tolerance for overdetection in breast, colorectal, and prostate cancer screening. When preferences differ, allowing patients to make individual decisions becomes so much more important. Further, a randomized control trial providing women with decision aids for mammography, but only one group with information about overdetection, demonstrated that having this information measurably impacted women’s intentions to screen.
As this trial showed, people were able to understand uncertainty and make more informed decisions. When uncertainty is not disclosed, patients assume medical certainty and are more likely to defer to medical expertise. When they are more aware of uncertainties, they realize when a decision is less medical and more value-based.