Navigating the Uncertainties of Medicine

In the early 2000s, Gordon Schiff set out to answer a nagging question: Why were so many patients being misdiagnosed? It was a timely effort, as just a few years earlier the Institute of Medicine had reported that medical errors led to as many as 98,000 deaths each year in the U.S., a finding that stunned the health care industry.

Schiff, now a Harvard Medical School associate professor of medicine at Brigham and Women’s Hospital, was interested in how diagnostic errors could harm patients. He and several colleagues at Chicago’s John H. Stroger Jr. Hospital surveyed physicians from around the country to find examples of misdiagnosis, with the goal of understanding how and why such errors occur. But to Schiff’s surprise, the small group of experts reviewing the cases often couldn’t agree on the correct diagnosis, even with the benefit of hindsight.

“There was uncertainty about what the diagnosis was and whether the doctor should or could have made the diagnosis,” Schiff recalls.

About a decade later, in 2016, Galina Gheihman, MD ’19, was a second-year medical student at HMS. She was working with a patient with Guillain-Barré syndrome, a neurological disorder, when, over the course of about ten to twelve hours, the patient lost the ability to move her body. Gheihman felt well prepared to talk with the patient’s family about the causes of the disorder and potential treatments, but she had trouble answering the questions the family members actually asked: Would the patient be able to walk again? If so, when? What kind of care would she need over the long term? And why was her condition more severe than it was for many other patients with Guillain-Barré?

“No textbook had those answers,” Gheihman says. “That was a very difficult position to be in, and one that felt very uncomfortable.”

Complicating matters was that, as a medical student, she wasn’t even sure why she didn’t have answers to the family’s questions. Was it a gap in her own knowledge, or was there a gap in the field as a whole? “There was uncertainty about the causes of the uncertainty,” she says.

These experiences have stuck with Schiff and Gheihman. Now an HMS instructor in neurology at Brigham and Women’s and Massachusetts General Hospital, Gheihman says that this was one of her first encounters with uncertainty in medicine, but far from her last. “I don’t think I can think of a patient case where there wasn’t some kind of uncertainty,” she says.

Schiff’s work convinced him that diagnostic error was closely tied to uncertainty. “That really crystalized for me this idea that we need a new science of uncertainty,” he says.

Acknowledging the unknown

“The whole of medicine is to some degree uncertain,” wrote the medical ethicist Kathryn Montgomery in her 2005 book How Doctors Think. And Atul Gawande, MD ’95, in his 2002 book Complications, called uncertainty “the core predicament of medicine.”

Examples of uncertainty in medicine abound. In a 2008 study, for example, researchers at HMS and Mass General examined the records of about 600 patients who had gone to an emergency department because they were experiencing shortness of breath. The researchers found that for almost a third of the patients, there was uncertainty about the diagnosis, and that those patients were likely to have longer hospital stays and had a higher risk of dying within the next year.

Sonja Solomon lives with uncertainty every day in her practice as a primary care physician. “There is so much uncertainty just inherent in the practice of medicine,” says Solomon, an HMS instructor in medicine at Brigham and Women’s. “I think patients often come to us expecting things to be clear, and so often they’re complicated.”

She says that uncertainty can result from limitations in medical knowledge or from the complexity of applying what is known to the care of specific patients. A doctor might know that, in general, 70 percent of patients respond to a treatment, but for any single patient, they are unlikely to know whether that patient will be part of the 70 percent who benefit or the 30 percent who do not. “We don’t have a way of saying that this is exactly what’s going to happen with this treatment for you, with few exceptions,” Solomon says.

“We often talk in ranges,” Gheihman says. "For example, I might say that we expect to see improvement on the order of weeks to months, as opposed to days to weeks. It is rare that we are able to give specific answers that are directly individualized.”

Schiff cites the prostate-specific antigen (PSA) test as another example. The test is used to detect signs of prostate cancer, but there are conflicting guidelines about who should be offered it and what the results mean for treatment. “There’s uncertainty about whether the test should be done in the first place,” he says. “Even under the best circumstances, some of these tests are just riddled with uncertainty.”

Despite the abundance of examples and the fact that the topic has long been studied in medical journals, uncertainty often still goes unacknowledged by physicians and unrecognized by patients.

“Although physicians are rationally aware when uncertainty exists, the culture of medicine evinces a deep-rooted unwillingness to acknowledge and embrace it,” wrote Arabella Simpkin, MMSc ’16, and Richard Schwartzstein, MD ’79, the Ellen and Melvin Gordon Distinguished Professor of Medical Education at HMS, in a 2016 article in the New England Journal of Medicine. “Doctors often fear that by expressing uncertainty, they will project ignorance to patients and colleagues, so they internalize and mask it.”

That’s something that many clinicians and researchers — including Gheihman, Schiff, and Solomon — are trying to change.

Communicating clearly

“I think it’s really critical that we are as explicit as we can be with patients about what we know,” Solomon says, “that when we really do think that something is true, we’re clear about it, and when we don’t know, we’re also really clear about that.”

Indeed, the downsides of uncertainty — and of a lack of communication about it — are well documented. According to one recent estimate, about 800,000 patients are harmed each year in the U.S. due to diagnostic error, and uncertainty plays a significant role. A 2018 review found that diagnostic uncertainty can lead to overtreatment, including unnecessary tests, hospitalizations, and surgeries, all of which incur higher costs as well.

In addition to the effects on patients and the health care system, uncertainty can take a toll on clinicians, especially when they don’t feel free to discuss it openly. The stress caused by uncertainty has been associated with burnout, depression, and lower job satisfaction among physicians, and it can affect the decisions they make about patient care.

“The ability to tolerate uncertainty is somewhat correlated with well-being and sense of self-efficacy in practice,” Solomon says. “And so, I think to the extent that it's teachable and that we can actually build that as a skill where people can learn to tolerate it more, then I think that it will actually increase people's satisfaction in practice.”

Solomon sees other benefits of acknowledging uncertainty. “It humanizes you with the patient, it is an opportunity for deep engagement with patients, and it is an opportunity for learning,” she says.

Gheihman agrees. “I increasingly recognize the benefits of uncertainty,” she says. “There’s more opportunity for relationship-building.”

But embracing uncertainty isn’t easy, which is why Gheihman, Schiff, and Solomon are developing concrete guidance for clinicians to help them manage it. As Schiff puts it, saying that clinicians should be transparent with patients sounds good, “but what we need is to figure out ways of communicating this.”

Creating structure

Over the past several years, Schiff has collaborated with several HMS colleagues to create a practical approach to ensuring that clinicians and patients have a shared understanding when there is uncertainty in a diagnosis. They published the details of their work in a paper last year. “Diagnostic uncertainty needs to be practically, acceptably, and efficiently integrated into the clinical workflow,” Schiff and his coauthors write.

To start, they interviewed primary care physicians, asking them how they would manage uncertainty in four specific sample patient visits. In one, a 56-year-old patient has been experiencing headaches. The symptoms indicate that it’s a muscle tension headache, but the patient is concerned that it’s something more serious. In another, a 25-year-old patient has symptoms consistent with the seasonal flu but asks about getting antibiotics. For each encounter, the physicians were asked, “How would you discuss what you think this could be, and the uncertainties you have with the patient about her diagnosis?”

Even where there is a high degree of uncertainty, there’s always a way forward.

The research team used the results of these interviews to draft a handout that could be given to patients at the end of a visit and a guide to talking about uncertainty intended to be used by clinicians. They tested the materials with patients and clinicians and revised them based on feedback.

Schiff and a fellow coauthor and primary care physician, Narath Carlile, an HMS assistant professor of medicine at Brigham and Women’s, used the final versions of the materials in talking with their own patients as a test. During the visits, they discussed any specific areas of uncertainty with their patients and used voice recording software to capture the details of the discussion and automatically fill out a form with that information. At the end of the visit, they gave the patients copies of the form to take with them. The process was designed to be integrated into a physician’s existing workflow and to ensure that both patient and physician come away from the visit with the same understanding of any uncertainties.

“Patients really liked it even more than we thought,” Schiff says. “We were very pleasantly surprised.”

While the communication tool developed by Schiff and colleagues is intended for primary care settings when there is uncertainty about a diagnosis, Gheihman, Solomon, and other HMS researchers recently published a framework that can be applied to discussions of uncertainty in any clinical situation. They describe it as “a practical approach to managing uncertainty that can be easily integrated into clinical practice and that goes beyond the act of communication.” They break the management of uncertainty into four aspects, using the mnemonic “RAPS” to refer to them (for Recognize, Acknowledge, Partner, and Seek support).

The first step is simply for the clinician to recognize that uncertainty exists, and the second is to acknowledge it with the patient. Solomon says that could mean telling the patient that more information is needed or that it will take time to know with more certainty what the best course of action is. One goal of this step is to ensure that the patient and physician share an understanding of where the uncertainty lies. Gheihman adds that another benefit of talking about uncertainty with a patient early in the process is that clinicians can get a sense of how comfortable that patient is with these conversations.

But acknowledging uncertainty isn’t enough, Solomon says. Clinicians also have to develop plans to manage both the uncertainty and what the patient is experiencing in terms of symptoms, which Solomon and Gheihman refer to as partnering with the patient. The clinician should be empathetic and transparent and work with the patient to determine the possible next steps in the patient’s care.

“The conversation can’t just stop with ‘I don’t know,’ Solomon says. “Even where there is a high degree of uncertainty, there’s always a way forward.”

Finally, Solomon says, it’s essential that clinicians can discuss uncertainty with their colleagues, which is where seeking support comes in. “It’s really important for the clinician not to feel alone, and to harness that uncertainty for their own learning and growth,” she says.

Embracing uncertainty

Gheihman compares past studies of uncertainty to basic science research and says that their framework is an attempt to translate what’s known about clinical uncertainty into practice. What’s needed next, she says, is to study the use of the framework to validate its effectiveness. Schiff and his coauthors note that their work was a small pilot study and also has not yet been tested in a larger study. But they all hope their efforts will help make discussions of uncertainty a routine component of clinical practice.

“This is an important part of what we do,” Solomon says. “I hope that spelling it out in a framework gives people some additional guidance beyond what we all learn through our personal experience.”

Gheihman often follows the steps laid out in the framework in her own clinical practice. For example, she says, she recently met with a patient with an uncertain diagnosis. They discussed two options to narrow down the diagnosis, an imaging test and a spinal tap, and Gheihman explained what would be learned from each option to narrow down the possible diagnoses. Based on the patient’s preferences, they decided on starting with the less invasive imaging test. Finally, Gheihman talked with a colleague about what they would have done in that situation to affirm the decision.

Gheihman notes that she still sometimes gets a sense of unease when she realizes there is something uncertain that could affect a patient, but she also has started to see the benefits of acknowledging it. “The more you can embrace uncertainty is, if anything, a mark of your dedication to your patient,” she says. “You’re saying, ‘I don’t know, and I won’t give up until I figure it out.’ ”

Amos Esty is the editor of Harvard Medicine magazine.