“But what would a stroke feel like?” my patient asked.
I didn’t quite know how to answer. I again explained that we could give him a blood thinner, which would lower his risk of having a stroke to 3 percent, but increase his risk of bleeding to 5 percent. I again told him that his abnormal heart rhythm, atrial fibrillation, made it more likely a blood clot could shoot off toward his brain, which in turn could lead to difficulty speaking or moving parts of his…
He interrupted me. “You already said that. Would I recover? Be the same person? And where exactly would I bleed from?”
This is partly because while many doctors tend to think in stats, most people think in stories. Salient or visceral experiences, like “my neighbor had a stroke” or vomiting blood into a toilet, are immediately accessible. Abstract risks, like the ones I presented to my patient, can be difficult to grasp.
We also tend to prioritize immediate consequences over long-term possibilities, and are strongly influenced by framing. A 95 percent chance of living, for example, sounds a lot better than a 5 percent chance of dying.
To my patient, then, the choice between risking a stroke or a major bleed might seem like comparing apples to oranges — or rather, three-hundredths of an apple to five-hundredths of an orange. How is he supposed to choose? How am I?
Yet patients and doctors must make these kinds of choices every day. Should you choose six months of life with chemotherapy and intractable nausea, or three months at home chemo-free? Should I prescribe insulin, or something less effective and less dangerous, to treat a patient’s diabetes? Should we operate on an elderly woman with a broken hip — knowing there’s a small chance she won’t live through surgery, but a near-certainty she’ll never walk without it?
Shared decision-making — by which patients and doctors discuss treatments in the context of a patient’s goals and values — is supposed to help us solve this problem, or at least better navigate the options. It emerged in the 1970s alongside greater emphasis on patient autonomy and patient-centeredness, as opposed to the paternalism that had historically dominated medicine. But research suggests shared decision-making isn’t used widely — or effectively — enough, and that our communication with patients suffers.
Fewer than half of patients believe their clinicians understand their goals and concerns, and many people who are seriously ill say their medical care is not alignedwith their preferences. Nearly half of the time when doctors say they discussed prognosis and likely outcomes with patients, their patients say there was no such conversation.
Research also finds that it’s common for doctors to review the benefits of screening and treatment, but discussion of risks — like overtreatment, overdiagnosis and complications — is lacking. One study found that less than 10 percent of patients were told about the potential harms of cancer screening. And more generally, patients tend to vastly overestimate the benefits and underestimate the harms of treatment.
Almost 90 percent of patients have fundamentally mistaken beliefs about how cardiac stents might help them; nearly the same proportion overestimate the benefits of breast cancer and colon cancer screening. For many interventions, the minimum benefit patients say they’re willing to accept is greater than the benefit research tells us they’re likely to receive.
What can be done?
More shared decision-making would be a good start. Shared decision-making initiatives have traditionally used decision aids like educational booklets and DVDs, but new technologies like online interactive tools are increasingly being deployed — with promising results. A recent Cochrane review of over 100 studies and 34,000 patients found that decision aids can improve patients’ knowledge of their options, help them understand what’s important to them, and lead to more accurate expectations about treatments.
The Mayo Clinic and other institutions have developed online decision aids, available to all, to help doctors and patients more easily conceptualize the pros and cons of a given treatment. Mayo’s Statin Choice Decision Aid, for example, allows patients to enter personal health characteristics and, through visually engaging charts and graphics, assess their risk of a heart attack. It then prompts them to select various options to see how likely they are to have that heart attack with treatment or without treatment. It also presents information on how the risks and benefits vary by statin dose, how many people need to be treated for one person to benefit, and what the costs are to both your bank account and your daily routine. Similar decision aids are now available for other common conditions like diabetes, osteoporosis, depression and cancer.
Our medical era is one of complex technologies, seemingly endless information, and imperfect data. It can be difficult for doctors to explain medications and procedures, let alone adjudicate their relative value for each patient. But we must do a better job of guiding patients through the subtle, sometimes counterintuitive, trade-offs inherent in the tests and treatments we propose.
Decision aids may not solve the problem, but they can help. If effectively and creatively deployed, they can translate what we know and feel — our goals, preferences and values — into what we do and experience.
Dhruv Khullar, M.D., M.P.P., is a resident physician at Massachusetts General Hospital and Harvard Medical School. Follow him on Twitter: @DhruvKhullar.