Nudging for Better Health
Using techniques from behavioral economics to change individual behavior is very popular these days. The New York Times just published an op-ed in which a physician describes how healthcare can be improved through the use of “nudges” and other behavioral science techniques. Some examples include getting doctors to wash their hands, getting employees to quit smoking, and reminding people to fill their prescriptions and take their medications. Are these behavioral interventions really free of negative "side effects"?
Dr. Richard Thaler, one of the authors of the book “Nudge”, published an essay titled “The Power of Nudges, For Good and Bad.” In this essay he described three guiding principles that result in “good” nudging:
All nudging should be transparent and never misleading.
It should be as easy as possible to opt out of the nudge, preferably with as little as one mouse click.
There should be good reason to believe that the encouraged behavior will improve the welfare of those nudged.
Let us take a look at how the interventions described in the New York Times op-ed do on the Thaler nudge test.
Transparency and honesty
Nudges vary in their degree of transparency. A sign reminding a doctor to wash her hands? Transparent and clear. Notification on a wellness app or a fitness tracker to take one’s pills or walk more steps? Not so much.
Transparency in nudging has two components: choice of direction for the nudge, and the data used to do the nudging.
It is not always clear what facts or standards are used to encourage some behaviors and discourage others. For example, many fitness trackers and wellness programs set a goal of 10,000 steps per day. Although there are many studies showing that an active lifestyle leads to better health than a sedentary one, the 10,000-step standard has never been clinically validated. Some studies show that small levels of physical activity far short of 10,000 steps also improve health. On the other hand, a new study indicates that the goal for health improvement could actually be 15,000 steps per day. To make matters more complicated, many accepted medical standards are changed after rigorous analysis of their risks and benefits. Examples include standards for mammography, cholesterol screening and diabetes management. In order to meet the transparency test, there has to be a way for the person being nudged to find out the scientific basis for the nudge. She should also have confidence that the nudge will change as new knowledge is developed.
Then there’s the question of whether the nudge is general or tailored to the individual. Putting up a sign that tells everyone to wash their hands doesn’t require knowing anything about the people whom it will encourage. However, when nudges are based on individual preferences or behavior, those doing the nudging need data about the individual. Unfortunately, it isn’t always clear what data they use or how they get it. As I have written before, health and wellness apps and fitness trackers seldom offer a clear explanation to users about the data they collect, how they use it, and what happens to the data when someone stops using an app or a tracker. Even those who read privacy policies get, at best, a general idea of what is going on. They learn that their data is “shared” with partners, affiliates or “other third parties.” They may find that their information may be used for marketing or research. There are likely to be few details.
In order to meet the transparency requirement, a “good” nudge must be clear about how those who do the nudging select the target behavior and about how they accomplish the goal. This information should be easily available to those who want to know.
Whether it’s easy to opt out of a nudge depends on the way a nudge is structured. For example, if a doctor ignores a sign and decides not to wash her hands, was the opt-out easy? It is, if no camera records her action and no consequences follow. But what if a recording of the doctor’s actions affects her performance evaluation? What if a doctor’s bonus depends in part on her washing her hands some high percentage of the time before touching patients? In this case, the doctor can still opt out of hand washing, but her decision will become more complicated.
The goal of nudging is to get people to behave in a particular way. As a result, if opting out is too easy, those doing the nudging may make the nudge more forceful. The CVS smoking cessation program described by the op-ed is an example of this.
As described in the op-ed, having people deposit money that they would lose if they didn’t quit smoking was an effective way to get them to quit. What the op-ed did not say is that participants in the study much preferred a program that offered a reward to a program that required them to place a bet on themselves. 90 percent of those offered a reward program signed up, but only 14 percent of those offered a betting program did. A larger percentage of people in the betting group quit smoking, but because the sign-ups for the reward group were higher, more people actually quit smoking and remained tobacco free after 12 months through a program that offered a reward.
Nevertheless, because a higher percentage quit smoking in the betting group, CVS redesigned its smoking cessation program so that everyone who wants to participate must bet on their own success or, as some prefer to call it, “have some skin in the game.” There is no longer any other option. Because it wanted to achieve a specific behavior, CVS made opting out of the nudge more difficult.
Hand-washing and smoking cessation are particularly interesting examples because both encourage behaviors that are so obviously beneficial. The point of nudging, though, is to encourage people in positive behaviors without taking away their autonomy. When an opt-out is difficult or expensive, the ability to opt out is more apparent than real.
Improving the welfare of the person being nudged
Finally, do nudges in healthcare improve the health and welfare of patients? In some cases, they almost certainly do. Unfortunately, many other cases are not at all clear-cut.
As I already mentioned, there is no way to tell whether nudges provided by health and wellness apps or fitness trackers rely on current scientific knowledge. For example, the op-ed praises an app that nudges people to take their medications as prescribed. However, just a week earlier, the New York Times published an essay about the dangers to patients of taking multiple medications. Polypharmacy, as it is known, can be particularly hazardous for older or sicker patients. Shouldn’t the apps that nudge patients to take all their pills also nudge them to ask their doctors whether these pills are necessary?
Medication adherence is only one area where nudging toward particular health behaviors raises questions. Increasingly prevalent workplace wellness programs, with their financial incentives and other nudges, can lead to overtesting and overtreatment, exposing program participants to unnecessary health risks. Overtesting and overtreatment also raise healthcare costs for everyone, including employees who often have to pay some or all of the cost for diagnostic testing when a screening test raises a question. While the goal of screening is early detection of health risks and early intervention, there is often little evidence that benefits justify the risk of harm to everyone who is tested. In fact, this is why standards for mammography and prostate cancer screening have been changed.
Many medical interventions involve risks as well as benefits. Nudging should not be a blunt instrument that channels people into behaviors beneficial to some but potentially harmful to others.
Healthcare and “good” nudges
Doctors take an oath to act in the best interest of their patients or, at least, to do no harm. The willingness of patients to accept medical interventions depends on their trust in doctors and doctors’ helpers. The only nudges that have a place in healthcare are fully transparent, respectful of individual autonomy, and do not harm the people nudged. Given the complexity of medicine and its rapid evolution, these are not easy goals to meet.