When “nudge” doesn’t work: Medication Reminders to Outcomes After Myocardial Infarction

Gur Huberman points to this news article by Aaron Carroll, “Don’t Nudge Me: The Limits of Behavioral Economics in Medicine,” which reports on a recent study by Kevin Volpp et al. that set out “to determine whether a system of medication reminders using financial incentives and social support delays subsequent vascular events in patients following AMI compared with usual care”—and found no effect:

A compound intervention integrating wireless pill bottles, lottery-based incentives, and social support did not significantly improve medication adherence or vascular readmission outcomes for AMI survivors.

That said, there were some observed differences between the two groups, most notably:

Mean (SD) medication adherence did not differ between control (0.42 [0.39]) and intervention (0.46 [0.39]) (difference, 0.04; 95% CI, −0.01 to 0.09; P = .10).

An increase in adherence from 42% to 46% ain’t nothing, but, yes, a null effect is also within the margin of error. And, in any case, 46% adherence is not so impressive.

Here’s Carroll:

A thorough review published in The New England Journal of Medicine about a decade ago estimated that up to two-thirds of medication-related hospital admissions in the United States were because of noncompliance . . . To address the issue, researchers have been trying various strategies . . . So far, there hasn’t been much progress. . . . A more recent Cochrane review concluded that “current methods of improving medication adherence for chronic health problems are mostly complex and not very effective.” . . .

He then describes the Volpp et al. study quoted above:

Researchers randomly assigned more than 1,500 people to one of two groups. All had recently had heart attacks. One group received the usual care. The other received special electronic pill bottles that monitored patients’ use of medication. . . .


Those patients who took their drugs were entered into a lottery in which they had a 20 percent chance to receive $5 and a 1 percent chance to win $50 every day for a year.

That’s not all. The lottery group members could also sign up to have a friend or family member automatically be notified if they didn’t take their pills so that they could receive social support. They were given access to special social work resources. There was even a staff engagement adviser whose specific duty was providing close monitoring and feedback, and who would remind patients about the importance of adherence.

But, Carroll writes:

The time to first hospitalization for a cardiovascular problem or death was the same between the two groups. The time to any hospitalization and the total number of hospitalizations were the same. So were the medical costs. Even medication adherence — the process measure that might influence these outcomes — was no different between the two groups.

This is not correct. There were, in fact, differences. But, yes, the differences were not statistically significant and it looks like differences of that size could’ve occurred by chance alone. So we can say that the treatment had no clear or large apparent effects.

Carroll also writes:

Maybe financial incentives, and behavioral economics in general, work better in public health than in more direct health care.

I have no idea why he is saying this. Also it’s not clear to me how he distinguishes “public health” from “direct health care.” He mentions weight loss and smoking cessation but these seem to blur the boundary, as they’re public health issues that are often addressed by health care providers.

Anyway, my point here is not to criticize Carroll. It’s an interesting topic. My quick thought on why nudges seem so ineffective here is that people must have good reasons for not complying—or they must think they have good reasons. After all, complying would seem to be a good idea, and it’s close to effortless, no? So if the baseline rate of compliance is really only 40%, maybe it would take a lot to convince those other 60% to change their behaviors.

It’s similar to the difficulty of losing weight or quitting smoking. It’s not that it’s so inherently hard to lose weight or to quit smoking; it’s that people who can easily lose weight or quit smoking have already done so, and it’s the tough cases that remain. Similarly, the people for whom it’s easy to convince to comply . . . they’re already complying with the treatment. The noncompliers are a tougher nut to crack.

The post When “nudge” doesn’t work: Medication Reminders to Outcomes After Myocardial Infarction appeared first on Statistical Modeling, Causal Inference, and Social Science.