One intervention, many ways to explain risks

Which of these drugs would you be most likely to take—one that prevents a heart attack in one out of 13 patients taking the drug, one that postpones a heart attack for two months in all patients, or one that postpones a heart attack for eight months in one out of four patients?

When researcher Peder A. Halvorsen, MD, asked a similar question of 1,754 Norwegian patients, he wondered whether most patients would opt for the scenario that gave them the certain benefit of postponing disease. In fact, his study, published in the June 19, 2007 Annals of Internal Medicine, found that patients were most likely to agree to the drug when told that it prevented heart attacks in one in 13 patients, known as the number-needed-to-treat (NNT) format.

“In my opinion, the most interesting thing about this study is that these different ways of explaining things may matter to lay people,” said Dr. Halvorsen. In fact, the differences mattered quite a lot. The rate of patients consenting to treatment varied from 93% in the NNT group to 69% in those who were told about the two-month postponement for all patients.

The findings of Dr. Halvorsen's study point out both the importance and the challenges of communicating with patients about risk. While it may be difficult for physicians to decide on the way to explain statistical concepts to patients, it's clear that the explanation they choose greatly affects the way patients respond.

The dilemma is a fairly new one in medicine, but in recent years a smattering of researchers have begun working to uncover the best possible way to explain risk and the statistics that quantify it to patients.

Make it absolute

By studying how patients make both real and hypothetical decisions about medications, screenings and other intervention choices, researchers have come to some general conclusions about effective risk communication.

“The literature would bear out that presenting risk in certain formats is easier to understand and less likely to bias patients. The formats that seem to do better are the absolute risk reduction and relative risk reduction formats,” said Stacey L. Sheridan, ACP Member and assistant professor of medicine at the University of North Carolina.

Both formats use percentages to show patients how an intervention would affect their risk of developing a certain outcome. The difference is that absolute risk quantifies the change in the patient's risk from the baseline, while relative risk reveals only the change in risk accomplished by the intervention. The net effect is that changes in relative risk tend to be much more dramatic—for example, a reduction in risk from eight out of 100 to four out of 100 is a 50% relative risk reduction, but only a 4% absolute risk reduction.

Both calculations may be easy for patients to understand, but they can have very different effects on their decisions, according to Dr. Sheridan. “Something notable about relative risk reduction is that it's persuasive. We have showed time and time again that people who are presented with a relative risk reduction format are more likely to choose to take a medication. Doctors who get information on the benefits of a medication in terms of relative risk reduction are more likely to prescribe the medicine,” she said.

For example, consider a case of a woman deciding whether to take adjuvant therapy for breast cancer, said Albert Mulley, MD, chief of general medicine at Massachusetts General Hospital. He described a scenario in which adjuvant therapy reduces the risk of recurrence by 30%. As a relative risk reduction, that sounds significant, but the absolute risk reduction would vary dramatically depending on whether the women's overall risk of recurrence is 40% or 10%.

“With more and more women with smaller and smaller node-negative tumors being found with mammography, we see lots of women with recurrence risks of well less than 10%, so the 30% reduction is just tiny in the baseline scheme of things,” said Dr. Mulley.

Comprehension, not persuasion

The possibility that some of those low-risk women might choose therapy because they have an exaggerated sense of the benefits concerns Dr. Sheridan. For that reason, she has focused her research on the question of how well patients understand statistics, in addition to how they react to them.

“Most of the studies about risk perception really have looked at persuasiveness. We've spent very little time figuring out what people understand,” she said. Her research has found that understanding to be fairly low. In one study, at best only 21% of patients could calculate the effect of a drug on baseline risk of disease.

The patients had more success when they were asked to compare the effectiveness of two different interventions. “Even people who have high levels of education and literacy have trouble with numbers,” said Dr. Sheridan. “People do much better at comparative tasks than calculating tasks, so the extent to which we can help them make comparisons, we can help them do better.”

Given the difficulty of explaining statistics to patients, some experts have begun searching for alternative methods to compare individual risks. Steven A. Grover, MD, professor of medicine at McGill University, recently tried out one technique, cardiovascular age, on more than 2,000 Canadian patients.

Patients in the study, which was published in the Archives of Internal Medicine, were told their cardiovascular age (current age minus calculated change in life expectancy based on cardiovascular risk factors) and offered advice on lifestyle changes and pharmacotherapy to treat dyslipidemia. Controls were not told their cardiovascular age.

Although overall the study recorded fairly modest impact on the patients' achievement of lipid targets, the patients with the greatest difference between their real age and cardiovascular age had significantly larger drops in cholesterol levels than those with the lowest difference. Conversely, those patients whose cardiovascular age was lower than their chronologic age were less likely to get to targets than those receiving usual care, which indicated that risk assessment also served to reassure low-risk individuals, noted Dr. Grover.

“All we know for certain is that low-risk people were less likely to respond to the risk profile,” he said. “People clearly understand when their risk factors are high enough that their cardiovascular age is higher than their chronological age, and when they've done something to bring that cardiovascular age down, they know they are going in the right direction,” he added.

Decision aids

After his study was complete, Dr. Grover put the cardiovascular age tool on the Internet so that interested patients could access it themselves. Such resources (commonly called decision aids) that enable patients to learn more about their risk profile without a physician present may be key to helping patients make good decisions, experts said.

Evidence-based decision aids, which can take the form of picture boards, videos or booklets, can prepare patients for a conversation about their risks and options, a big benefit for time-pressed physicians. “The doctors don't have to spend as much time. They can spend time answering questions,” said Dr. Sheridan.

Time and office space could be used even more efficiently if decision aids turn out to work as take-home assignments for patients. Dominick L. Frosch, PhD, assistant professor in the UCLA department of medicine, has studied video decision aids used in the office, and someday plans to examine what would happen if physicians prescribed the aids.

“There are a lot of open questions, such as ‘After the physician prescribes it, will the patient actually watch it at home? After the patient watches the program, will the patient return to the doctor?’” he said.

The entire field of research on patient understanding of risk will have many big, open questions for some time to come, noted Dr. Mulley. Even if a method is found to make all patients perfectly understand the risks of an intervention, they would still have dramatically different perspectives on the statistics.

“If there is a 3% risk of death from a procedure, that 3% might be acceptable to me and wholly unacceptable to you,” Dr. Mulley said. “Often medical decisions depend more on those subjective attitudes that a patient brings to a decision than they do on what they learn from the clinical research.”

For a side-by-side comparison of the risks, see Table.