candy-pieces

Perspectives

Using behavioral nudges to treat diabetes

A new report in 'Harvard Business Review' by Thomas H. Davenport, James Guszcza, and Greg Szwartz

Thomas H. Davenport is a professor at Babson College, a research fellow at the MIT Initiative on the digital economy, and a senior adviser at Deloitte Analytics. James Guszcza is the US chief data scientist for Deloitte Consulting LLP. Greg Szwartz is a managing director of advanced analytics for Deloitte Consulting LLP.

Health care practitioners and payer organizations increasingly use big data to overcome what might be called a “flaw of averages” in traditional medicine: a treatment that has been tested at a population level might, in fact, work better for some individuals than others. The goal of precision medicine is, therefore, to identify treatments appropriate to an individual, rather than a population, based on granular genotype and phenotype data from his or her medical records. The individualized, data-driven nature of such treatment protocols improves the odds that a specific treatment will work for a specific patient.

But both traditional and precision medicine confront a “last mile” problem involving patient behavior: even the most appropriate medical treatment will be effective only if the patient follows through with it. The cost of medication non-adherence is conservatively estimated at more than $250 billion a year in the United States, and the majority of hospital readmissions after surgery are due to non-adherence to discharge protocols.

Issues of adherence to treatment have been acknowledged and studied for years, but only recently have techniques derived from behavioral science been used to actually improve patient follow-through. Dr. Mitesh Patel, director of Penn Medicine’s “Nudge Unit” notes, “Human behavior is the final common pathway for the application of nearly every advance in medicine.” Patel and others—including a group at Deloitte—are now exploring whether data-driven precision medicine can also incorporate precision behavioral nudges that improve patient willingness to follow clinicians’ recommendations by offering choices within an incentive scheme designed to address the patient’s specific motivational profile.

Behavioral scientists are cultivating an ever-expanding repertoire of nudge techniques: subtle design tweaks to the environment that prompt individuals to automatically adopt behaviors that are in their best interests. For example, the Wharton behavioral economist Katherine Milkman and her collaborators tested the efficacy of “pre-commitment strategies” in prompting people to get vaccinated. It turned out that specifying—in writing—the exact time and place of vaccinations resulted in measurably better follow-through compared to a vague or general commitment. Thus a small, inexpensive bit of choice architecture provides a real shot in the arm for population health. Gamification, “buddy” programs, and text message reminders are other familiar nudge-for-health tactics.

But not everyone will be interested in a game to earn health points, or motivated by a pre-commitment agreement, or a buddy keeping tabs on their health behaviors. And just as a particular treatment won’t work equally well for each patient, not every behavioral intervention will nudge everyone equally effectively.

This is where “precision engagement” comes in. Only recently applied to health care, it combines data and behavioral sciences to customize the behavioral intervention most likely to motivate a patient to change a behavior and stay engaged in a treatment protocol. In a precision engagement program, instead of only personalizing medicines or treatment protocols, interventions are also personalized. Patients are encouraged to adopt recommended health behaviors with interventions that are tailored to their own personalities, motivations, care journeys, and engagement challenges. With the help of artificial intelligence technologies and machine learning algorithms, it is becoming possible to recommend not only medications, but also targeted behavioral interventions. In short, big data can be used to deliver the right nudge to the right patient (or physician), at the right time, and in the right way, to improve adherence to evidence-based therapies.

For precision engagement to work, the provider needs to know a lot about the patient. This includes demographic and socioeconomic information, how the patient best receives recommendations in person or through an app, the progress of the care journey, and how he or she responds to social cues and incentives. These all serve as predictive or control variables in analytical models that assess which factors are associated statistically with the desired health behaviors.

To aid in this analysis, the Penn Medicine Nudge Unit is conducting a clinical trial sponsored by Deloitte. The trial, led by Patel, seeks to understand how different people respond to different types of motivators to influence exercise behaviors among overweight and obese adults. Patients are assessed and placed randomly into one of three different socially driven interventions—competitive, collaborative, and supportive. Typical behavioral interventions like gamification and social incentives are explored with patients.

Another attempt at precision engagement is being put into action in Mexico. Clinics called Clinicas de Azucar (“sugar clinics”), or CdA, which make up the largest private network of diabetes and hypertension clinics in the country, are focused on diabetes treatment for middle- and low-income populations. When patients first arrive, their HbA1c (blood glucose) levels are usually around 11% (a “crisis” level), but within a couple of visits focusing on treatment and behavioral change, their levels stabilize at approximately 7%. This drop results in a 200% to 400% reduction in negative outcomes. This short-term success is worth celebrating. But the fear is that unless patients come back year after year, they may revert to old behaviors and gains will be lost.

Miguel Garza, chief operating officer of CdA, says data can be part of the answer. He notes, “We have been providing diabetes care for more than seven years and we have seen more than 30,000 patients in a one-stop-shop model. So we have a lot of data. We need to find the nudges that eliminate barriers and motivate our patients to exercise more, to eat better, and to take the drugs that are prescribed. And we need to leverage the data to find out which nudges work for which patients.”

CdA and Deloitte are now establishing a Behavior Change Learning Lab to gather and analyze precision engagement data and build predictive algorithms applicable to the larger population. Javier Lozano, the founder and chief executive officer of CdA, says that the goal is to make sure that precision engagement insights learned from the current clinics can be scaled to support millions of people with diabetes in Mexico and beyond. The typical changes that patients with diabetes have to make in their lifestyle are usually difficult. If we can use precision-engagement data to recommend only the interventions that are most likely influence each patient, we’ll lower patients’ costs and see better outcomes.

At the CdA Learning Lab, the impacts of various types of precision engagement on different care goals will be measured.

The early success in engagement that CdA clinics are seeing with diabetic patients is encouraging, but should become even more impressive as precision engagement plans are rolled out. If these go as planned, the goal will be to test and scale what we learn for a variety of diseases that can be managed with behavior change, including other aspects of metabolic syndrome (such as hypertension and obesity) and substance use disorders.

The field of disease management has had a somewhat troubled history. We believe this is because effective behavioral nudges have not been widely used. Just as individuals respond differently to drugs, it seems increasingly clear that they will respond differently to well-designed, personalized engagement interventions.

Read the article on Harvard Business Review. 

© 2018 Harvard Business School Publishing corp. Distributed by the New York Times Licensing Group.

Fullwidth SCC. Do not delete! This box/component contains JavaScript that is needed on this page. This message will not be visible when page is activated.

Did you find this useful?