What’s the Right Way to Present Safety Data?
We have an enormous responsibility when we treat patients, for all the treatments we give carry risks. We must give patients adequate, accurate information so that they can make good judgments about their health care.
Presenting risk information is inherently tricky, because the human mind is fuzzy in how it interprets information. For example, let’s say there is a drug with a 1 in 100 risk of some adverse event. We could tell patients that, and it might seem scary. When people hear “1 in 100,” they often imagine being the one. On the other hand, if we wanted to reassure patients who heard of this side effect, we could tell them there is a 99 of 100 chance for the event not to happen. When patients hear 99 of 100 they do not have a problem, they envision being one of the 99.
I have a patient who was afraid of getting liver disease from biologic treatment. (Biologics do not cause liver disease, but that does not matter.) I asked him would he take the drug if there was only a 1 in 1000 chance of cirrhosis. “No way, doc!” was his response. I asked him would he take another biologic if 99 of 100 did not get cirrhosis with it; he said, “Sure. I would take that.”
Can we give patients an objectively accurate perception of risks? No, we cannot. We humans interpret information subjectively.
This applies not just to patients but even to our own perceptions. Should I be using drugs that have potential risks for patients? If I were to look at the risks of a drug that doubled the chance of infection, I might get scared. But if I focus on the absolute risk (2 in 1000 instead of 1 in 1000) rather than the relative risk, I would be more assured. I could even reassure myself further if I only looked at the attributable risk (which would be only 1 in 1000). I make myself feel even more comfortable if I focus on the 999 of 1000 who do not get an infection due to drug.
Even in totally objective numbers, when we have them, there is huge potential for subjective interpretation. There is no purely objective way to evaluate the data. And that is in the best of circumstances, when we have unequivocal numerical data, something we usually do not have.
I do not think we have to worry about the practice of medicine becoming less art and all science anytime soon.
Steven R. Feldman, MD, PhD
Chief Medical Editor
Dr Feldman is with the Center for Dermatology Research and the Departments of Dermatology, Pathology and Public Health Sciences at Wake Forest University School of Medicine in Winston-Salem, NC.