Imagine that you are a medical doctor. You need to tell one of your patients that he has advanced-stage pancreatic cancer, an almost incurable condition. You learn that your patient’s only daughter is getting married five months from now. Without treatment, your patient has about a year left to live. Chemotherapy would increase his chances of being alive in five years by about 20 percent but would also double his chances of dying before his daughter’s wedding. What do you tell him? Of course, the choice is by no means easy or clear-cut.
Doctors are scientists who operate in a world of statistics, odds, and probability. Yet they’ve long been taught that when dealing with patients they should convey a reassuring level of confidence and certainty. As a result, patients expect their doctors to give them a clear diagnosis and a straightforward course of treatment.
But now that information about every medical condition imaginable is just a few clicks away, experts are asking whether doctors’ apparent certainty when communicating with their patients actually does more harm than good. With the information overload brought by the progress of medicine and technology, answers are rarely black or white. Medical schools are only just starting to teach doctors how to deal with this, and patients’ expectations haven’t adjusted, either.
The fact is that medicine has long been steeped in uncertainty and has arguably even thrived on it. To avoid bias when testing a new drug, researchers must have no preconceived notions of which treatment, the new one, the old one, or even a placebo, is the best option. This principle, called equipoise, protects patients, doctors, and researchers alike from making assumptions before scientific proof of efficacy has been gathered.
“Medicine has always fallen short of the sort of certainty that we find in math and geometry,” says Dr. Ross Upshur, a researcher at the Dalla Lana School of Public Health in Toronto. “If you think about knowledge and what it does, it’s about limiting uncertainty, not about creating certainty.”
Doctors in training, like gamblers, need to be comfortable working in a field in which they’re constantly weighing the odds based on a myriad of factors. When Upshur teaches medical students how to diagnose an ailment, he tells them to broaden their inquiry — come up with a list of possibilities — rather than quickly home in on a single solution. “Even when you make a diagnosis that you think is firm, you usually don’t have certainty about what would be the best therapy and what the outcomes will be in the long run.”
Upshur has helped introduce a new course at the University of Toronto that encourages students to take a critical view of evidence and see how it transforms rapidly over time. Thus, as part of the health science research course, students are asked to read a series of articles, old and recent, on stomach ulcers.
“Reading articles on stomach ulcers that were written over the past 100 years made me realize how much things have changed, and that what we know now might not be true any more in 10 years,” says Saba Moghimi, a third-year student who has just completed the course. “Learning about uncertainty taught me to be humble, and to respect those who came before me and didn’t have access to all the knowledge we have now.”
Upshur says that physicians must learn to live with the consequences of uncertainty. “Medicine is fundamentally a moral enterprise,” he told me. “You’re assisting people with their well-being. It is not like putting a dollar into a machine and getting out a candy bar. If it was that easy, we wouldn’t need doctors.”
Technology has helped redefine the quest for certainty. We are reaching a point where we can feed a list of symptoms into a computer and get a more accurate diagnosis than from a doctor. Dr. Richard Schwartzstein, a professor of medicine at Harvard Medical School, sees such developments as both a challenge and an opportunity. On one hand, he says, “technology tries to push you to a maximal level of certainty. Do this test to get a 99 percent level of certainty that you have this disease.”
On the other hand, computers can’t communicate a diagnosis or a treatment to patients in a comforting way. Take a routine screening test for early-stage lung cancer. Based on your age, your smoking status, and your gender, a computer can do a great job of evaluating the chances of finding a cancerous nodule. It can also calculate quite precisely the risk of developing an actual cancer based on the size and shape of a nodule. What it can’t do however, is decide how to break the news that you have a nodule in your lung that has a 1 percent chance of becoming a cancer.
Dr. Steven Hatch, an infectious disease doctor and assistant professor at the University of Massachusetts, is the author of the recent book “Snowball in the Blizzard: A Physician’s Notes on Uncertainty in Medicine.” After working in an Ebola treatment unit in Liberia at the peak of the 2014-’15 epidemic, where uncertainty about his patients’ survival was the bread and butter of his daily work, Hatch used his knowledge of medicine, study design interpretation, and biostatistics to communicate to a lay audience how medicine is often very different from what people think. In his book, Hatch discusses the example of screening mammograms. In 2009, the US Preventative Services Task Force came out with a new set of guidelines on mammography. It recommended that all women between the ages of 50 and 74 undergo a mammogram every two years. This would presumably detect breast cancer early, while it is still easy to treat.
“Within the field, it wasn’t a quantum shift, but the public messaging really hit the public like a thunderbolt,” Hatch says. “The naive understanding is that you can do the mammogram and see the cancer as it is,” he adds. In reality, reading a mammogram is far from a straightforward exercise. Just like screening tests for colon cancer or lung cancer, mammograms often lead to unnecessary procedures and tests, generating more uncertainties than certainties. A recent study from the Canadian Journal of Public Health shows that just over 40 percent of women undergoing recommended mammograph screening will be called in for a falsely positive result.
Some medical schools, including Harvard, the University of Toronto, and the University of California in San Francisco, are starting to incorporate the value of uncertainty into their curriculums. The idea is that, instead of trying to provide clear-cut answers, doctors should help their patients put information into context, while recognizing that patients are the ones who know their bodies best. At the University of California in San Francisco, first-year medical students attend small group sessions each week where they discuss cases where there is no clear consensus on a treatment. One recent point for discussion: Does a study showing improved efficacy of a new asthma medication warrant a medication change for a patient who is perfectly satisfied with her current plan?
While doctors face the biggest challenge in accepting a greater degree of uncertainty, the public — especially patient advocacy groups — must also adjust their thinking. Schwartzstein notes that patient advocacy groups sometimes feed the appetite for certainty by basing their campaigns on highly charged stories of deaths that could have been avoided had a doctor prescribed a specific treatment at an earlier stage of the patient’s illness. Such alarmist messages risk reinforcing the quest for precise answers.
“We have to move away from making decisions based on anecdotes,” says Schwartzstein. “We need to use data, context, and personal preferences.” Patient advocacy groups seldom tell the stories of the large number of patients who suffer only mild consequences as a result of an incorrect diagnosis or an unnecessary surgery or procedure. Schwartzstein argues that we need to hear these stories as well. This would encourage more balanced conversations in situations where the solution to a medical problem is not so obvious.
Sir William Osler, the Canadian physician widely recognized as one of the founding fathers of modern medicine, wisely said: “Medicine is a science of uncertainty and an art of probability.” It is time for doctors to embrace their role as the navigators of uncertainty. They can do this by acquiring better tools to explain such things as odds and probabilities to lay people, a patient population that, in general, has not been trained to understand the complexity of the medical world they are pushed into.