Patient Engagement

Applying Behavioral Insights to Improve Health and Health Care

Article · May 14, 2017

Policymakers often try to achieve their goals by influencing citizen behavior. This is particularly true when it comes to health: in developed countries, most of the burden of disease arises from unhealthy eating, smoking, or alcohol consumption. However, the last 40 years have seen a growing body of evidence that challenges traditional policy approaches to influencing behavior. Decisions are often less deliberate and considered than has been assumed and more habitual, automatic, and influenced by the environment in which they are made. Policy approaches that neglect this evidence are less likely to be successful.

This idea was the impetus behind the creation of the Behavioural Insights Team (BIT) within the UK Government in 2010. The team was set up to apply findings from the behavioral sciences in order to improve policy and service design. After working in policy areas such as health, employment, and tax compliance, the team was spun out of government in 2014 to become a social purpose consulting company (still partly owned by government).

BIT uses a simple framework to apply behavioral science to policy: in order to encourage a behavior, make it Easy, Attractive, Social, and Timely (EAST). The sections below give some examples of how this framework can be applied to improve health and health care. Many of these initiatives were tested through low-cost randomized controlled trials; we believe that such trials could be used much more by health care providers and policymakers to improve their everyday activities.


The more effort that a behavior requires, the less likely someone will be to do it. This may seem obvious, but behavioral science research has shown that even small amounts of effort can have a disproportionately large effect on outcomes. For example, a UK law that made it slightly more difficult to buy painkillers in large quantities (and to release them from their packaging quickly) was found to have prevented around 800 suicides over 11 years. The slight increase in effort required was enough to save lives.

A study from Imperial College London, supported by BIT, applied the same principle to patient care. As part of the study, the investigators redesigned a hospital prescription chart to make clear communication easier. For example, rather than requiring doctors to write out milligrams (mg) and micrograms (mcg) by hand, the new chart allowed the user to circle one or the other. The idea was to “design out” errors. A trial showed that the chart improved the accuracy and completeness of prescribing. These examples represent the kinds of changes that can be easily scaled throughout health care systems.


Individuals are exposed to a vast amount of information every day. In response, they develop strategies for filtering out most of this information and focusing only on a few pieces. Therefore, information is only likely to influence behavior if it is delivered in a way that attracts attention effectively. We believe that there are many opportunities to improve the ways in which health systems try to attract the attention of their users.

For example, in the UK, 1 in 10 hospital appointments is missed. Sending cell phone text messages to remind patients about scheduled appointments does increase attendance, but there is little or no evidence about how the wording of these messages might affect no-shows. BIT ran a trial to test this idea in a real-life environment. We randomly allocated patients to receive several different reminder messages. We found that the best-performing message, which stated the cost of a missed appointment to the National Health Service, reduced no-shows by almost 25%. Moreover, it was the specific cost and wording that attracted attention: the same message expressed in general terms was less effective. Again, this is a simple change that is both cheap and scalable.


Individuals are strongly influenced by what they see others do or by what they are told others do in the same situation. Behaviors often spread through social networks rapidly and unpredictably, and people may not realize just how powerful the actions of others can be.

In a recent randomized trial, led by BIT and Public Health England, this idea was applied in an effort to reduce the use of antibiotics in primary care, a core goal of the UK’s Antimicrobial Resistance Strategy. One obvious way of achieving this goal would be to pay doctors to prescribe fewer antibiotics. But social norms may offer another option. As part of this trial, we used online prescribing data to identify the primary care practices that were prescribing antibiotics at a higher rate than other practices in the local area. We then sent half of those prescribers a letter, signed by England’s Chief Medical Officer, pointing out that “The great majority (80%) of practices in [local area] prescribe fewer antibiotics per head than yours.”

The study showed that the 800 practices that received the letter reduced their antibiotic prescribing rate by 3.3% (Figure 1), which amounted to an estimated 73,406 fewer doses of antibiotics being dispensed over 6 months and savings of £92,356 in direct prescription costs. We calculated that, if letters had been sent to all participants, England’s antibiotic prescribing rate would have been reduced by just under 1%. As a point of comparison, in the same year, the National Health Service spent £23m to incentivize general practitioners to achieve the same goal (a 1% reduction in their prescribing rate).

Impact of a Simple Social Norm Feedback Intervention on General Practitioner Prescribing Rates in the UK

This figure compares the antibiotic prescribing rates of doctors who received feedback that their rate was higher than that of their peers (“Treatment”) with the rates of doctors who did not receive such feedback (“Control”). In this randomized trial, prescribing rates were statistically indistinguishable before the feedback was received (September); however, the rates subsequently diverged, with the rate for the Treatment group being 3.3% lower than that for the Control group during the period presented. Click To Enlarge.


Timing is an often-neglected part of policy and service design. The likelihood that an individual will accept an offer can vary greatly depending on when it is offered. Certain moments can disrupt existing patterns of behavior and provide the opportunity for change. In one study, for example, the experience of having even minor surgery was shown to increase the likelihood of quitting smoking by nearly 30%, and receiving a short smoking cessation intervention at that time increased this effect even further.

Another effective way of influencing behavior is to provide timely feedback on an individual’s actions. BIT recently used this idea in an effort to reduce unnecessary visits to hospital emergency departments. We first identified people who had recently sought treatment in an emergency department when they could have gone elsewhere, as indicated by hospital statistics. We randomly allocated half of these people to receive a supportive letter that suggested alternative options for care (the other half received no letter). The letter was timely: it arrived within a week of the emergency department visit.

We then measured whether the people who had received the letter were more or less likely to return to the emergency department unnecessarily. In this instance, we saw no difference in the rates of unnecessary returns. It may be that there are systemic issues that make it too difficult to change this behavior at the individual level. Nevertheless, this result shows the value of testing, which allows us to identify options that are not effective and to focus our attention on areas that are most likely to bring improvement.

Looking Forward

These approaches offer the possibility of improving the quality and efficiency of care — at low cost — by focusing on drivers of behavior that health systems often neglect. Excitingly, their use appears to be growing, with many more interventions being tested in health care settings around the world. This process of testing will help us to understand what works best, where, and for whom. We argue that these EAST principles should be part of the basic approach for anyone working in health care policy and service design. They will be integral to achieving the changes in behavior that are necessary to make health systems sustainable in the future.


This article originally appeared in NEJM Catalyst on October 12, 2016.

Call for submissions:

Now inviting expert articles, longform articles, and case studies for peer review


A weekly email newsletter featuring the latest actionable ideas and practical innovations from NEJM Catalyst.

Learn More »

More From Patient Engagement

Data Graphic: CMOs and Staff Physicians Lead Patient Engagement Efforts

Survey respondents say CMOs and staff physicians head patient engagement efforts within health care organizations, but it’s important to note who does the ranking.

How Chronic Pain Treatment Falls Short of Patient-Centered Care

Healing the opioid epidemic requires nothing less than a comprehensive theory of patient health.

patient engagement vs patient experience

Patient Engagement versus Patient Experience

The difference between the two terms is muddled but important, says Cleveland Clinic’s Chief Experience Officer.

Evaluating Complex Care Programs: Is It a Zero-Sum Game?

Policymakers see programs for complex patient populations as a way to bend the health care cost curve, but are reduced health care utilization and costs the right measures of success?

Toyin Ajayi Talk Still: The Difference Between Noncompliance and Defiance

The Difference Between Noncompliance and Defiance

How do we align our goals for patient engagement with even the most complex, difficult patients?

Hardwiring Patient Engagement to Deliver Better Health

Imagine a world in which health care was not “one size fits all,” but customized for each patient individually.

Health Care — A Final Frontier for Design

Design must move beyond narrow projects and encompass complex systems.

Engineering Social Incentives for Health

It’s important to engineer social engagements that promote health, but we must also test their acceptability and effectiveness.

When — and How — Can Incentives Actually Work?

Incentives can work if they’re designed a bit better.

My Favorite Slide: When Patients Are Prevented from Engaging

If patient engagement is about action, why don’t we let them speak?


A weekly email newsletter featuring the latest actionable ideas and practical innovations from NEJM Catalyst.

Learn More »


Patient-Centered Care

277 Articles

Information Asymmetry: The Untapped Value of…

The knowledge and preferences that patients could — and should — share with clinicians would…

Information Asymmetry: The Untapped Value of…

The knowledge and preferences that patients could — and should — share with clinicians would…

Patient Incentives

75 Articles

Taxonomy of the Patient Voice

While health care pursues the important trend of putting patients at the center of care,…

Insights Council

Have a voice. Join other health care leaders effecting change, shaping tomorrow.

Apply Now