In recent decades, the burgeoning science of behavioral economics has taught economists what doctors, nurses, and care managers have long known: Faced with a choice, people don’t always do what’s best for them. They go for the easy option instead of the prudent one, or favor immediate pleasure over long-term benefit. But such behavior can be changed, and that fact is fueling a health care innovation that promises to save money while enhancing care. It’s called the “nudge” unit.
Nudge units help redesign “choice architecture”—the way choices are presented—to make the sensible choice also the easy, likely one. Mitesh S. Patel, MD, an assistant professor of medicine and health care management at the University of Pennsylvania, heads the Penn Medicine Nudge Unit, the first such entity embedded in a U.S. health system. Pretty soon, he believes, no self-respecting health system will be without one.
The first nudge unit was created by the British government in 2010 as an experiment to see if behavioral economics and psychology could be applied to induce Britons to make healthier, wiser choices. “It was a huge success,” says Patel, and was quickly emulated by other governments around the world.
The Brits had help. Advising them was University of Chicago economist Richard Thaler, who would go on to win a Nobel Prize in 2017. In 2008, he and Harvard Law School’s Cass Sunstein had published the nudge-unit bible: Nudge: Improving Decisions About Health, Wealth, and Happiness.
When Patel and his colleagues pitched the idea for a health system nudge unit to top management at Penn Medicine, the response was enthusiastic. And no wonder. Penn is a leader in applying behavioral economics to improve health outcomes. Directed by Patel’s colleague, Kevin Volpp, MD, the university’s Center for Health Incentives and Behavioral Economics studies “reducing the disease burden from major U.S. public health problems” by helping people make sound choices.
Even before the system’s unit was introduced in 2016, a small change had saved Penn a lot of money. On electronic health records, the default option—the one easiest to choose—was switched from brand names to their generics.
“That intervention took us about an hour to implement,” says Patel. “And it reduced unnecessary spending by $32 million in two-and-a-half years.” That’s even accounting for pharmacist-guided switches to generics. “We realized we could do things like that more systematically, looking for opportunities to design ‘nudges’ to align things to make the best choice, the most evidence-based choice, also the easiest choice.”
Many health systems have recently been gathering ideas and tinkering with EHRs, but Patel’s unit, once launched, decided to attack the opportunity comprehensively. The team hosted a three-week “innovation tournament,” inviting everyone from hospital orderlies to full professors to submit ideas for possible nudges. They received 225 suggestions and took the 30 best ones to a steering committee that included IT specialists, clinical leaders, and behavioral economists. The committee winnowed that list down to 10 candidates, and proponents of each made their case at a Pitch Day in June 2016 before the health system’s top managers. Three winners got the go-ahead: boosting referrals to cardiac rehabilitation, cutting needless opioid prescribing, and reducing unnecessary imaging during end-of-life palliative radiation therapy for cancer patients. Today the Penn Nudge Unit has some 50 initiatives under way.
Some nudges, like making generics the default option, are aimed at busy clinicians. Others seek to guide the behavior of patients themselves through digital games and incentives—encouraging them, for example, to stay physically active by taking 10,000 steps a day.
Anyone who worries that guiding people’s choices is stealthily Big Brotherish should realize that this barn door has been open for some time. Americans now routinely have their decisions influenced, subtly or not, by everything from Amazon to Aunt Harriet’s Facebook account. And nudges are fundamentally subtle—as Patel points out, they don’t compel, only encourage. That’s why Chicago’s Thaler and other prophets of nudgery challenge our preconceptions by calling their innovation “paternalistic libertarianism.”
Says Patel: “What most people fail to recognize is that we were already being nudged. Now we have a chance to think strategically about it.”
Penn’s Nudge Unit plans to share the fruits of its labor—to spare other health systems its own trial-and-error process of determining what works—through a Nudge Unit Collaborative that is scheduled to launch by year’s end. By paying a membership fee, other organizations will have access to Penn’s “platform”; Patel says he doesn’t know yet what the fee will be.
Here are four examples of what the Penn Medicine Nudge Unit has been up to.
That eligible post-heart attack patients should get into a cardiologist-supervised exercise program seems a no-brainer, but Penn’s 15% rate of successful referrals cried out for a nudge. “We found that it was a manual opt-in process,” says Patel. “It rested on the shoulders of busy cardiologists making rounds to write out referrals on a paper form, because we didn’t have an electronic version yet. Then the patient had to call his or her insurance company to see which cardiac rehab centers were covered, then call a center to see if there was a spot, and set everything up. That’s when they’d just had a heart attack!”
Because EHRs already identified recent heart-attack survivors, the Nudge Unit linked those data to Penn’s secure text-messaging platform. Care managers now get daily messages with the names and locations of patients who need rehab. They also present patients with a list of rehab centers near their homes that Penn has vetted and that their insurance covers. The care manager helps the patient choose and book a rehab facility—and follows up 30 days later to make sure the connection has been made. The 15% successful referral rate was nudged up to 85%.
Heart attack patients themselves are the focus of another nudge. This one takes advantage of a lesson of behavioral economics known as loss aversion: People are more readily impelled to action by the prospect of losing something they have than of gaining something new. In a four-month test, patients were outfitted with a wearable device to measure steps taken each day. Some were given $14 a week in a personal account and told they’d lose $2 for every day they failed to take their daily quota of steps; others were given only the devices. Not only did the incentivized walkers log an average of 1,300 more daily steps than the control group, but walking proved habit-forming: Almost all of that gain—1,250 daily steps—persisted after the incentive was removed.
“Over six months,” says Patel, “those patients walked a hundred miles more than patients who got only the wearable device.”
Last September, when Penn hosted its first annual Nudge Units in Health Care Symposium, many representatives of the 22 health systems from around the country that attended said a key concern was opioid abuse. Penn’s unit reported that in its own emergency department, prescriptions had defaulted to a 30-tablet quantity, which is perfectly appropriate for many medications. “But you don’t need 30 pills for opioids,” says Patel. The default was changed to 10 tablets, a three-to-five-day supply, and unnecessary prescribing was cut in half.
If there’s one thing the Nudge Unit folks like almost as much as success, it’s failure—because failure is rich with lessons for how to do better. Half of the high-cardiovascular-risk patients in the U.S. who should be taking statins aren’t doing so, and that lack seemed nudge-worthy. Penn designed an intervention to encourage clinicians to prescribe statins and to show them how they stacked up against their peers as statin prescribers. “We actually tripled statin prescribing,” says Patel, but he regards that increase—from 2.5% to 8%—as a too-small drop in a very empty bucket.
“We learned from that early failure always to include clinicians in the design of an intervention,” says Patel, explaining that the statin prompt had proved impractical for many clinicians. “They told us, ‘Hey, I don’t have time to look at a list of 200 patients. And I usually prescribe statins during a visit; you’re asking me to do this outside a visit.’ And there wasn’t a good mechanism for patients to call in with questions.” Now the Nudge Unit is working with Penn’s clinicians to identify a new version of the intervention that this time benefits from clinician input. Starting with a $250,000 NIH grant, the unit is studying three approaches, with nudges for clinicians, patients, or both, to see which works best.
Patel believes that too many health systems decide on an improvement and roll it out to everyone right away, thus they can’t study its effectiveness. Instead, when Penn designs a nudge, it tries it out in a testable way—say, in just one hospital—to study its efficacy (and note any unintended consequences) before rolling it out to the whole health system.
Asked about clinician resistance, Patel concedes that the occasional doctor has had a huffy moment. “One of them might say, ‘We had a decade of training; we’re experts in our field, we don’t want someone else influencing our decisions,’” he says. “But then we show them examples of how the previous EHR design for our patient portals often influenced their decisions toward inefficient ones that obliged them to do extra work.”