3772 PART 19 Consultative Medicine

noncardiac surgery should be avoided in this vulnerable period. The

duration of DAPT thereafter is dictated by the circumstances in

which PCI was performed and whether the indication was stable

ischemic heart disease or acute coronary syndrome. For the former

among patients treated with a DES, DAPT should be given for at least

6 months. For the latter, DAPT should be given for at least 12 months.

However, DAPT may be interrupted to allow for noncardiac surgery

30 days after BMS and 6 months after DES, respectively. Elective, noncardiac surgery should be delayed for 5 days since the last dose of clopidogrel; 7 days since the last dose of prasugrel; and 3–5 days since the

last dose of ticagrelor. The use of cangrelor, an intravenous reversible

P2Y12 receptor antagonist, may be an appealing bridging strategy (with

an infusion given at a bridging dose of 0.75 μg/kg per min, although

studies of its use in those undergoing cardiac and noncardiac surgery

are limited. If P2Y12 inhibitor therapy (clopidogrel, prasugrel, or ticagrelor) is interrupted or discontinued in patients who have received

intracoronary stents, aspirin should be continued perioperatively (save

select circumstances where the risk of bleeding may be catastrophic as

in neurosurgical or spinal procedures) and the P2Y12 receptor inhibitor

should be restarted as soon as possible postoperatively. Decisions surrounding antiplatelet management in the perioperative setting among

patients who have received intracoronary stents are complex and

should involve multidisciplinary decision-making.

Α2

 AGONISTS Based on the results of POISE-2 (a large multicenter, international, blinded randomized clinical trial of aspirin and

clonidine), α2

 agonists for prevention of cardiac events are not recommended in patients who are undergoing noncardiac surgery. In

this trial, clonidine increased the rate of nonfatal cardiac arrest and

clinically important hypotension, while reducing the rate of death or

nonfatal MI.

CALCIUM CHANNEL BLOCKERS Evidence is lacking to support the

use of calcium channel blockers as a prophylactic strategy to decrease

perioperative risk in major noncardiac surgery.

ANESTHETICS Mortality risk is low with safe delivery of modern anesthesia, especially among low-risk patients undergoing low-risk surgery

(Table 480-4). Inhaled anesthetics have predictable circulatory and

respiratory effects: all decrease arterial pressure in a dose-dependent

manner by reducing sympathetic tone and causing systemic vasodilation,

myocardial depression, and decreased cardiac output. Inhaled anesthetics also cause respiratory depression, with diminished responses to both

hypercapnia and hypoxemia, in a dose-dependent manner; in addition,

these agents have a variable effect on heart rate. Prolonged residual neuromuscular blockade also increases the risk of postoperative pulmonary

complications due to reduction in functional residual lung capacity, loss

of diaphragmatic and intercostal muscle function, atelectasis, and arterial

hypoxemia from ventilation–perfusion mismatch.

Several meta-analyses have shown that rates of pneumonia and

respiratory failure are lower among patients receiving neuroaxial

anesthesia (epidural or spinal) rather than general anesthesia. However, there were no significant differences in cardiac events between

the two approaches. Evidence from a meta-analysis of randomized

controlled trials supports postoperative epidural analgesia for >24 h for

the purpose of pain relief. However, the risk of epidural hematoma in

the setting of systemic anticoagulation for venous thromboembolism

prophylaxis (see below) and postoperative epidural catheterization

must be considered.

PREOPERATIVE PULMONARY

RISK ASSESSMENT

Perioperative pulmonary complications occur frequently and lead to

significant morbidity and mortality. Clinical practice guidelines recommend the following:

1. All patients undergoing noncardiac surgery should be assessed for

risk of pulmonary complications (Table 480-5).

2. While select studies have suggested that quitting smoking shortly

before surgery increases the risk for postoperative complications

through increased sputum production and/or decreased cough,

meta-analysis of the available data has challenged this, and all

patients should be advised of the imperative to stop smoking

presurgically.

3. Patients undergoing emergency or prolonged (3–4 h) surgery; aortic aneurysm repair; vascular surgery; major abdominal, thoracic,

neurologic, head, or neck surgery; and general anesthesia should

be considered to be at elevated risk for postoperative pulmonary

complications.

4. Patients at higher risk of pulmonary complications should undergo

incentive spirometry, deep breathing exercises, cough encouragement, postural drainage, percussion and vibration, suctioning and

ambulation, intermittent positive-pressure breathing, continuous

positive airway pressure, and selective use of a nasogastric tube for

postoperative nausea, vomiting, or symptomatic abdominal distention to reduce postoperative risk. Multiple pulmonary risk indices

are available to estimate the postoperative risk of respiratory failure,

pneumonia, and other pulmonary complications; among these is

the ARISCAT risk index, which accounts for the following seven

risk factors: age, low preoperative oxygen saturation, respiratory

infection within the preceding month, upper abdominal or thoracic

surgery, surgery lasting >2 h, hemoglobin <10 g/dL, and emergency

surgery (Table 480-6).

TABLE 480-4 Gradation of Mortality Risk of Common Noncardiac

Surgical Procedures

Higher • Emergent major operations, especially in the elderly

• Aortic and other noncarotid major vascular surgery

(endovascular and nonendovascular)

• Prolonged surgery associated with large fluid shift and/

or blood loss

Intermediate • Major thoracic surgery

• Major abdominal surgery

• Carotid endarterectomy surgery

• Head/neck surgery

• Orthopedic surgery

• Prostate surgery

Lower • Eye, skin, and superficial surgery

• Endoscopic procedures

Source: Reproduced with permission from LA Fleisher et al: ACC/AHA

2007 Guidelines on perioperative cardiovascular evaluation and care for

noncardiacsurgery. Circulation 116:1971, 2007.

TABLE 480-5 Predisposing Risk Factors for Pulmonary Complications

1. Upper respiratory tract infection: cough, dyspnea

2. Age >60 years

3. Chronic obstructive pulmonary disease

4. Cigarette use

5. American Society of Anesthesiologists Class ≥2

6. Functional dependence

7. Congestive heart failure

8. Serum albumin <3.5 g/dL

9. Obstructive sleep apnea

10. Impaired sensorium (confusion, delirium, or mental status changes)

11. Abnormal findings on chest examination

12. Alcohol use

13. Weight loss

14. Spirometry threshold before lung resection

a. FEV1 <2 L

b. MVV <50% of predicted

c. PEF <100 L or 50% predicted value

d. PCO2 ≥45 mmHg

e. PO2 ≤50 mmHg

Abbreviations: FEV1

, forced expiratory volume in 1 s; MVV, maximal voluntary

ventilation; PCO2

, partial pressure of carbon dioxide; PEF, peak expiratory flow rate;

PO2

, partial pressure of oxygen.

Source: A Qaseem et al: Ann Intern Med 144:575, 2006. Modified from GW Smetana et al:

Ann Intern Med 144:581, 2006, and from DN Mohr et al: Postgrad Med 100:247, 1996.

3773 Medical Evaluation of the Surgical Patient CHAPTER 480

TABLE 480-6 Risk Modification to Reduce Perioperative

Pulmonary Complications

Preoperatively

• Smoking cessation

• Training in proper lung expansion techniques

• Inhalation bronchodilator and/or steroid therapy, when indicated

• Control of infection and secretion, when indicated

• Weight reduction, when appropriate

Intraoperatively

• Limited duration of anesthesia

• Avoidance of long-acting neuromuscular blocking drugs, when indicated

• Prevention of aspiration and maintenance of optimal bronchodilation

Postoperatively

• Optimization of inspiratory capacity maneuvers, with attention to:

• Mobilization of secretions

• Early ambulation

• Encouragement of coughing

• Selective use of a nasogastric tube

• Adequate pain control without excessive narcotics

Source: From VA Lawrence et al: Ann Intern Med 144:596, 2006, and WF Dunn, PD

Scanlon: Mayo Clin Proc 68:371, 1993.

5. Preoperative spirometry and chest radiography should not be used

routinely for predicting risk of postoperative pulmonary complications but may be appropriate for patients with chronic obstructive

pulmonary disease or asthma.

6. Spirometry is of value before lung resection in determining candidacy for coronary artery bypass; however, it does not provide a

spirometric threshold for extrathoracic surgery below which the

risks of surgery are unacceptable.

7. Pulmonary artery catheterization, administration of total parenteral

nutrition (as opposed to no supplementation), or total enteral nutrition has no consistent benefit in reducing postoperative pulmonary

complications.

PERIOPERATIVE MANAGEMENT

AND PROPHYLAXIS

■ DIABETES MELLITUS (SEE ALSO CHAPS. 403–405)

Many patients with diabetes mellitus have significant symptomatic or

asymptomatic CAD and may have silent myocardial ischemia due to

autonomic dysfunction. Intensive (versus lenient) glycemic control in

the perioperative period is generally not associated with improved outcomes and may increase the risk of hypoglycemia. Practice guidelines

advocate a target glucose range of 100–180 mg/dL in the perioperative

period. Oral hypoglycemic agonists should not be given on the morning of surgery. Perioperative hyperglycemia should be treated with IV

infusion of short-acting insulin or subcutaneous sliding-scale insulin.

Patients whose diabetes is diet controlled may proceed to surgery with

close postoperative monitoring.

■ INFECTIVE ENDOCARDITIS (SEE ALSO CHAP. 128)

Prophylactic antibiotics should be administered to the following

patients before dental procedures that involve manipulation of gingival

tissue, manipulation of the periapical region of teeth, or perforation

of the oral mucosa: those with prosthetic cardiac valves (including

transcatheter prosthetic valves); prosthetic material used in valve

repair (annuloplasty ring or artificial chord); previous infective endocarditis; cardiac transplant recipients with valvular regurgitation from

a structurally abnormal valve; and unrepaired cyanotic congenital

heart disease or repaired congenital heart disease, with residual shunts

or valvular regurgitation at the site adjacent to the site of a prosthetic

patch or prosthetic device.

■ AORTIC STENOSIS

Previous 2007 ACC/AHA guidelines have cautioned against surgery

in patients with severe aortic stenosis, citing a 10% mortality risk.

More recent guidelines, rooted in contemporary data, offer greater

latitude for noncardiac surgery in appropriately selected patients with

severe aortic stenosis. Among those who are asymptomatic with severe

aortic stenosis, the ACC/AHA assigns a Class IIa, Level of Evidence C

recommendation for moderate-risk surgery. In an analysis of patients

undergoing moderate or high-risk surgery at the Mayo Clinic from

2000 to 2010, there was no significant difference in 30-day mortality

between those with severe aortic stenosis and matched controls (5.9%

vs 3.1%, p=0.13); however, those with severe aortic stenosis had more

major adverse cardiac events (18.8% vs 10.5%, p=0.01), mainly due to

heart failure. In sum, severe aortic stenosis is associated with adverse

outcomes in patients undergoing noncardiac surgery; however, in

contemporary cohorts this risk is less than has previously been stated.

Balloon valvotomy is usually not recommended but may serve a role

in the minority of patients who need “bridging” to a necessary surgery

or procedure.

■ VENOUS THROMBOEMBOLISM

(See also Chap. 279)

Perioperative prophylaxis of venous thromboembolism should follow established guidelines of the American College of Chest

Physicians. Aspirin is not supported as a single agent for thromboprophylaxis. Low-dose unfractionated heparin (≤5000 units SC bid), lowmolecular-weight heparin (e.g., enoxaparin, 30 mg bid or 40 mg qd),

or a pentasaccharide (fondaparinux, 2.5 mg qd) is appropriate for

patients at moderate risk; unfractionated heparin (5000 units SC tid) is

appropriate for patients at high risk. The use of direct oral anticoagulants may be an alternative to the use of prophylactic doses of low-dose

unfractionated heparin and low-molecular-weight heparin; among

patients immobilized after nonmajor orthopedic surgery, rivaroxaban

10 mg once daily when compared to enoxaparin was associated with

a decrease in venous thromboembolic events, without a significant

change in bleeding. Graduated compression stockings and pneumatic

compression devices are useful supplements to anticoagulant therapy

or as alternatives to anticoagulant therapy in patients at excessive

bleeding risk.

■ FURTHER READING

Angiolillo DJ et al: Bridging antiplatelet therapy with cangrelor in

patients undergoing cardiac surgery: A randomized controlled trial.

JAMA 307:265, 2012.

Fleisher LA et al: 2014 ACC/AHA Guideline on perioperative cardiovascular evaluation and management of patients undergoing noncardiac surgery. Circulation 130:e278, 2014.

Levine GN et al: 2016 ACC/AHA guideline focused update on duration of dual antiplatelet therapy in patients with coronary artery

disease. A Report of the American College of Cardiology/American

Heart Association Task Force on Clinical Practice Guidelines. J Am

Coll Cardiol 68:1082, 2016.

Myers K et al: Stopping smoking shortly before surgery and postoperative complications: a systematic review and meta-analysis.

JAMA Intern Med 171:983, 2011.

Nishimura RA et al: 2017 AHA/ACC focused update of the 2014

AHA/ACC guideline for the management of patients with valvular

heart disease. A report of the American College of Cardiology/

American Heart Association Task Force on clinical practice guidelines. Circulation 135:1, 2017.

Samama CM et al: PRONOMOS Investigators. Rivaroxaban or enoxaparin in nonmajor orthopedic surgery. N Engl J Med 382:1916, 2020.

Smetana GW et al: American College of Physicians. Preoperative

pulmonary risk stratification for noncardiothoracic surgery: Systematic review for the American College of Physicians. Ann Intern Med

144:581, 2006.

Tashiro T et al: Perioperative risk of major non-cardiac surgery in

patients with severe aortic stenosis: a reappraisal in contemporary

practice. Eur Heart J 35:2372, 2014.

This page intentionally left blank

Frontiers PART 20

481

Tobacco use, physical inactivity, unhealthy diet, excessive alcohol use,

and other individual behaviors are estimated to underlie 40% of premature mortality in the United States. Approximately 75% of the $3.5

trillion currently spent on health care in the United States is attributable to cancer, heart disease, type 2 diabetes, and obesity, and each of

these conditions is strongly influenced by behavior. Nearly one-half

of patients prescribed medications to lower their cholesterol within

1 year following myocardial infarction stop taking these drugs—even

when they are provided free of charge. Despite great advances in the

science and technology of health care, a large gap separates theoretically achievable goals in health and health care from what individuals

and populations actually achieve. Human behavior is both a major

contributor to health problems and a barrier to the successful implementation of solutions to address them.

Recognizing that there are many reasons people do not take action

to improve their health, behavior change experts have focused efforts

on strategies targeted at individual behavior, introducing incentives for

weight loss, medication adherence, and smoking cessation, environmental strategies such as mandated food labeling, and combination

approaches. For example, Section 2705 of the Affordable Care Act

(ACA) allows employers to provide incentives of up to 50% of total premiums based on outcomes such as reduced body mass index, lowered

blood pressure or cholesterol, and smoking cessation; this legislation

puts as much as $300 billion worth of employee health incentives in

play annually.

Many of these approaches have built-in limitations, largely because

they have been designed around the pervasive view that people reliably

act to improve their self-interest. Existing policy solutions presuppose

that health care decisions are rationally based economic transactions

and that rational people will dispassionately assess the net present value

of the costs and benefits of alternative courses of behavior and pursue

the best path forward. These approaches are normatively appealing

but seem better suited to support the health of people who behave as

economists assume they do, and perhaps less effective when exposed to

such realities of human behavior as limited attention, overconfidence,

and problems of self-control. It is not just the magnitude of incentives

that matters, but also other critical features such as the specific nature

of rewards, feedback frequency, saliency, and framing. Public health

programs, including those involving financial incentives, are more

likely to achieve their goals if designed based not on how perfectly

rational people ought to make health decisions but rather on how

humans actually make them. The field of behavioral economics uses

insights from psychology to identify ways that human decision-making

often falls short of the ideal. The field aims to offer a more realistic

account of human behavior, providing a natural framework for efforts

to encourage changes in behavior.

■ CONCEPTS OF CLASSICAL ECONOMICS AND HOW

THEY DIFFER FROM BEHAVIORAL ECONOMICS

Classical economics posits that individuals are rational utility maximizers, meaning that they are able to dispassionately identify alternative decisions, calculate the probabilities of and utility/disutility

for each potential outcome, and then, through a process of backward

induction, implement the decision that has the highest net present

value. When it comes to health behaviors, classical economic theory

would assume that if people are obese they must have decided that the

costs of obesity are outweighed by the benefits of the behaviors that

lead to it and that if people smoke they must have decided that the

pleasures of doing so outweigh the costs. This assumption of rational

utility maximization has two major consequences for public policy,

both of which are applicable in the health domain.

The first is that conventional economic thinking radically limits

the range of situations in which it makes sense to intervene at a policy

level. Under the assumptions of conventional economics, regulatory

interventions such as targeted taxes and subsidies are considered

appropriate only in situations characterized by externalities (costs that

an individual’s actions impose on others such as second-hand cigarette

smoke), the presence of “market failures” (e.g., monopolies), or the

presence of certain types of information asymmetries. The second is

that standard economics offers a relatively restricted array of policy

tools, including taxes, subsidies, and mandates regarding patient education and information provision, and makes unrealistic assumptions

about how each of these approaches will influence the behavior of

individuals.

Behavioral economics builds on conventional economics by enriching its conception of individual behavior, offering enhancements on

each of these dimensions. First, it broadens the range of situations in

which policy interventions make sense by introducing the notion of

“internalities.” While externalities are costs (or benefits) individual

behaviors impose on others, internalities are the costs individuals

impose on themselves—typically their future selves. While interventions to address smoking and obesity can be justified on externality

grounds (e.g., the health care costs are borne by individuals other than

the immediately affected individual), they can also be justified on

grounds of internalities—e.g., people often irrationally discount the

delayed consequences of their behavior. In that sense, people might

want to be protected not just from others, but from the decisions of

their prior selves. Policy intervention can be further justified by the

ubiquitous exploitation of individual weaknesses by businesses. Commercial enterprises may take advantage of such vulnerabilities for individual profit rather than customer health, such as with the formulation

of tempting but unhealthy processed foods, the selling or prescribing

of addictive substances such as cigarettes or opioids, or the pricing of

meal “deals” that do not account for the health consequences of economically predatory offers. This behavior is found across industries:

Credit card companies and automobile manufacturers lure new customers with “$0 down” and fleeting but appealing teaser rates of “0%

interest,” playing on the common propensity to focus on the present

rather than on the future, as well as on people’s overoptimism when it

comes to their own future finances. Banks earn revenue by charging

high fees for minor mistakes such as account overdrafts or breaches

of minimum balance rules, often hiding the description of such fees in

small print and complicated jargon. States market lottery tickets that

return $0.45 on the dollar and promote these games in ways that ignore

more realistic expectations using one-sided messages such as “you can’t

win if you don’t play” rather than, for example, the equally accurate

message, “you can’t lose if you don’t play.”

Second, behavioral economics substantially broadens the range

of potential policy interventions beyond those offered by traditional

economics. Behavioral economics has become best known for the

concepts of “libertarian paternalism” and “asymmetric paternalism.”

In contrast to “heavy-handed” paternalism, asymmetric paternalism

attempts to protect people without limiting freedom of choice. That is,

it is asymmetric because it seeks to help individuals who are prone to

making irrational decisions without restricting the freedom of choice

of those making informed, deliberate decisions. For example, arranging

the presentation of food in a cafeteria line so that the healthy foods

appear first is likely to increase the amount of healthy food chosen,

without depriving those who want the unhealthy foods of the opportunity to purchase them. People who believe that individuals behave

Behavioral Economics

and Health

Kevin G. Volpp, George Loewenstein,

David A. Asch

3776 PART 20 Frontiers

optimally should not object to asymmetric paternalism because it does

not limit freedom, but those who acknowledge the limits of human

rationality should endorse such measures, assuming that they are

designed with individuals’ interests at heart. Frequently, these measures

are called “nudges.”

A nudge has been defined as “any aspect of the choice architecture

that alters people’s behavior in a predictable way without forbidding

any options or significantly changing their economic incentives. To

count as a mere nudge, the intervention must be easy and cheap to

avoid.” The most prominent, and to date successful, application of

nudges has been the use of defaults to increase enrollment in defined

contribution retirement savings plans and, secondarily, the use of

automatic escalation to encourage higher savings rates. These ideas

and research findings have had a major impact on retirement savings

policies worldwide, including the Pension Protection Act of 2006 in the

United States. Building on this success story in savings, and bolstered

by the establishment of so-called “nudge units” worldwide, the nudge

agenda has positioned behavioral economics at the center of public

policy.

The applicability of behavioral economics to policy, including

health-related policies, however, goes well beyond nudges. Many

health-relevant insights from behavioral economics do not fit the definition of a nudge. For example, there have been incentive programs aimed

at changing health behaviors for maximum cost-effectiveness, improvements in the delivery of health-related information, such as nutrition

labels, and new designs for physician incentives or health insurance.

A common theme of this work is its harnessing of predictable decision errors that usually harm people to, instead, help them to achieve

longer-term health goals or other socially valuable purposes, much the

way some martial arts redirect an adversary’s strength against him.

Many of the same messages, incentives, and choice structures used

so effectively to lure people into self-destructive health behaviors can

be redirected to attract them to healthier choices that improve their

long-term health and well-being. In addition, some features of human

decision-making that are not recognized by traditional economics

but are not really examples of decision errors, such as our propensity

to experience regret when we make a bad choice and our aversion to

putting ourselves into such situations (a phenomenon known as “regret

aversion”) can also be exploited in the service of improving health

behaviors and outcomes.

There is a common misconception that if you deploy financial

incentives in order to promote behavior change, then you are engaged

in behavioral economics. But that kind of activity is not behavioral

economics; it is simply economics. Indeed, a large number of everyday

transactions, such as being paid to go to work or getting a fine for

parking in the wrong place, reflect traditional economic incentives

to encourage or discourage certain behaviors. However, introducing

incentives in situations, such as those characterized by internalities, in

which traditional economics would suggest the incentives are unnecessary can be viewed as applications of behavioral economics.

Perhaps more importantly, a central lesson from the field of behavioral economics is that how incentives are delivered can matter more

than their objective magnitude. There are ways of delivering large

incentives that make them ineffective in changing behavior, and there

are ways that can greatly magnify the effectiveness of comparatively

small incentives. This observation is a source of optimism because

it suggests that, with careful design, we can leverage relatively small

investments to improve public health.

■ USING BEHAVIORAL ECONOMICS TO PROMOTE

SELF-BENEFICIAL HEALTH BEHAVIORS

Behavioral economics builds on neoclassical economics, which has at

its core expected utility theory. Expected utility theory both presumes

to describe how people make decisions and offers a prescription for

how such decisions should be made. While utility maximization is a

powerful normative model of how we ought to behave, it turns out to

be a poor descriptive model of how real people actually behave. Efforts

to alter human behavior that rely on this inaccurate and incomplete

model often fall short.

Over the past several decades, behavioral economics has described

various ways in which people’s decisions differ from standard economic models (Table 481-1). While economists mapped out concepts

of “bounded rationality” in the 1950s and identified limitations in the

dominant expected utility model of decision making under risk, the

publication of the article “Prospect Theory” by Kahneman and Tversky

in 1979 is widely credited with being seminal in the development of

behavioral economics. Prospect theory provides an overarching conceptual framework for describing observations about human behaviors

that could not be explained by expected utility theory.

Although these deviations from expected utility theory can be seen

as psychological foibles and errors, the real value of this work has come

less from identifying these errors and more from recognizing that they

occur in predictable ways. It is the predictability of these errors that

allows the design of strategies to overcome them (Table 481-2). Indeed,

it is the predictability and reliability of a number of key features of

human decision-making identified by behavioral economics, including but not limited to those connected to prospect theory, that allows

them to be used in order to promote self-advantageous health-related

behaviors.

Loss Aversion and Framing Effects Key tenets of prospect

theory include: (1) how people feel about a set of possible outcomes

depends on their starting point; this is the notion of referencedependence in which decision-makers evaluate outcomes as gains or

losses depending on their starting point; (2) people dislike losses much

more than they like equivalent-sized gains, a phenomenon known as

TABLE 481-1 Traditional vs Behavioral Economics

TRADITIONAL ECONOMICS BEHAVIORAL ECONOMICS

Core theory: Expected utility

maximization

Core theory: Prospect theory

Assumes perfect rationality Recognizes that people make decision

errors

Starting point independent Assessment depends on your starting

point

Framing doesn’t matter Framing affects assessment even

when utilities are the same

Stable preferences Time-inconsistent preferences

People discount the future at constant

rates

People discount the near future to

a greater degree and have timeinconsistent discounting

Intervene only when my actions

adversely affect others (negative

externalities)

Consider interventions when

people will harm their future selves

(internalities)

Regulations and policies generally

geared to protecting people from the

actions of others

Regulations and policies often geared

to protecting people from themselves

TABLE 481-2 Key Decision Errors and Suggestions for Addressing

Them

Present-biased preferences Provide feedback and rewards quickly

Nonlinear probability

weighting

Motivate people with probabilistic rewards

(lotteries)

Overoptimism and loss

aversion

Get people to precommit and put money at risk

as an effective motivational tool

Peanuts effect Deliver rewards in bundles, avoiding many small

rewards

Narrow bracketing Frame rewards in terms of effort per day rather

than per month or year

Regret aversion Help people anticipate the regret of poor choices

Defaults/status quo bias Change the architecture or environment of

choice to shift the path of least resistance to

favor healthy decisions

Rational world bias Move beyond the assumption that simply

providing information will lead to desired

behaviors

3777Behavioral Economics and Health CHAPTER 481

“loss aversion”; and (3) there is diminished sensitivity to both gains and

losses (Fig. 481-1).

One example of how we can apply the insights of loss aversion can

be seen in designing provider payment systems. Financial incentives

to physicians to achieve quality goals might be presented—“framed”

in the language of behavioral economics—as either rewards (in which

physicians receive a bonus if, for example, they increase their colorectal

cancer screening rates or increase the percentage of their patients who

reach a glycosylated hemoglobin target) or penalties (in which physicians would fail to receive an expected payment if they do not meet

these targets). Classical economists would consider equivalently sized

rewards and penalties as identically motivating because each reflects a

structure in which one receives $X for a certain outcome (Fig. 481-1).

However, loss aversion reminds us that the disutility of losing money is

much greater than the utility of gaining the same amount of money. A

number of studies have shown that people have a “loss aversion ratio”

in a range of 1.5 to 2.5. That means that a potential penalty of $1000 for

failing to meet a quality target ought to be about as potent a motivator

as a potential reward of $1500 to $2500 for meeting the same target.

That multiplier is nonsensical from the standpoint of classical economics, but as an empirically verifiable descriptor of human behavior,

it can be exploited in the designs of programs for clinicians or patients

to improve health.

All of us—including physicians—are highly susceptible to how

information is framed. In a set of experiments, patients, students, and

physicians were presented cases of lung cancer that could be treated

either by surgery or radiation therapy. Among all three groups, the

choice of surgery was more popular when its outcomes were framed

in terms of the probability of survival (e.g., a 68% chance of living for

>1 year) rather than in terms of the probability of death (e.g., a 32%

chance of dying by the end of 1 year). We can say that such sensitivity

to framing is irrational since a 68% chance of survival is logically equivalent to a 32% chance of dying—but these irrational decisions fall into

predictable patterns of behavior, and that predictability can be used to

influence those decisions. This means that clinicians have enormous

opportunities to lead patients toward particular decisions by framing

the outcomes of those decisions in specific ways, even as they remain

truthful. Such an understanding could lead to the view that clinicians

should be careful to balance their framing (for example, by following

up the statement of 68% chance of survival with a statement like “That

means there is a 32% chance of death”) in order to provide information

in a way that is likely to lead to a particular choice. Alternatively, it

could lead to the view that clinicians should deliberately frame outcomes in certain ways in order to lead patients to particular choices—a

much more paternalistic stance. Patients often rely on trusted clinicians to help them make the best decisions, and in some settings,

that reliance may justify using the principles of behavioral economics

strategically even if the same actions in other settings might be seen as

paternalistic, anti-libertarian, or coercive.

Loss Aversion and Overoptimism The power of loss aversion can

be most effectively leveraged when combined with a well-documented

decision error: overoptimism, or unrealistically high expectations

about future outcomes. Overoptimism is especially pronounced in the

context of people’s predictions about their own likelihood of exerting

self-control, sometimes referred to as the “false hope syndrome.” The

optimism that “This time, my efforts at weight loss will succeed!” or

“This year, I will follow my New Year’s resolutions!” often reveals itself

to be false. Although in some contexts overoptimism seems to be

beneficial, it can also result in suboptimal patterns of behavior. For

example, people prefer paying a flat rate for gym memberships even

though they would spend less if they were to pay on a per-visit basis,

in part because they overestimate their future gym attendance. Indeed,

many people enter into commercial weight loss programs and pay

for a full year up front because they are overly optimistic about their

chances of success.

Loss aversion can produce a variety of undesired behaviors, from

excessive risk aversion to the tendency for people to hold on to losing

investments, such as houses or stocks, for too long. However, loss aversion can advance individual or social goals. Framing reward outcomes

in terms of losses by “fronting” a sum of money that gets lost if goals

are not met, is often more motivating than providing equivalent gains

for meeting goals (the economic but not psychological equivalent). Yet,

despite the greater behavioral potency of losses, program administrators are often reluctant to use loss framing, perhaps because such programs can seem more punitive than organizations may wish to appear.

However, it is possible to take advantage of loss aversion by designing

programs in which people voluntarily put their own money at risk in

the service of achieving health-behavior goals that they themselves

desire.

The combination of overoptimism and loss aversion has been used

to help people lose weight by giving them the opportunity to participate in deposit contracts, in which they could deposit $0.01–$3.00 per

day of their own money, with a 100% match. Participants reported their

weight daily and received the sum of the deposit and the matching

funds each day they were on track to meeting their monthly weight loss

targets, but they forfeited both if they were not on track. The deposit

contract leveraged participants’ overly optimistic self-predictions of

how much weight they would lose as well as loss aversion once deposits

were made. Once deposited, however, such optimism can become a

self-fulfilling prophecy as loss aversion provides extra motivation to

meet goals. In this 16-week study, average weight loss was 14.0 pounds

in the deposit group compared with 3.9 pounds in the control group.

This work was extended in a 32-week study in which weight loss was

sustained for the duration of the intervention (8.7 vs 2.2 pounds in the

control group).

Although these results are promising, to be effective as a health

strategy, this approach needs to achieve high ongoing participation

rates to sustain its population effect. Deposit contract approaches are

powerful motivators of behavior change, but they are not always popular, even among those who initially opt to try them, posing a challenge

to long-term success.

Peanuts Effects The prospect theory value function assumes

diminishing marginal utility, which means that small gains and losses

are disproportionately motivating relative to larger ones (e.g., two $500

rewards would be more potent than one $1000 reward). However, this

may be too simple for small rewards (e.g., two $5 rewards may be less

motivating than one $10 reward). This “peanuts effect” may be part

of the reason why charities and retailers often describe costs in terms

of “pennies a day.” This observation challenges the expected efficacy of

programs that emphasize efforts to repeatedly achieve small changes

such as in weight loss programs. It is easy for a patient to rationalize

that no single cigarette causes lung cancer, that no single snack leads

to obesity, or that no single trip to the gym prevents heart disease.

Value (−)

Value (+)

Gains

(+ dollar

amount)

Losses

(− dollar

amount)

FIGURE 481-1 Prospect theory. (Adapted from D Kahneman: Thinking, Fast and

Slow. New York, NY: Farrar, Straus and Giroux, 2011.)

3778 PART 20 Frontiers

If self-destructive patterns of behavior, such as cigarette smoking,

weight gain, or cell phone use while driving, are seen as individual

instances rather than parts of a composite whole, it is easier to understand how they can be seen as acceptable. The pleasure of smoking a

cigarette or eating a dessert and the convenience and engagement of

conducting business or socializing in otherwise “dead” travel time are

immediate and tangible, but the marginal costs—increased risks of

developing lung cancer, being overweight, or having a car accident—

seem inappreciably small. Across a lifetime or a population, however,

the cumulative costs and/or probabilities are not small at all.

The tendency to underweight small outcomes can also be used to

people’s advantage, such as by inducing people to put away small sums

for retirement savings in short cycle lengths or to make small periodic

investments in their health via medication adherence. In each of these

efforts, we need to think asymmetrically. In our incentive programs we

should provide frequent (often daily) feedback on rewards because of

present-biased preferences, but if we are delivering financial rewards,

we want to aggregate them so that the rewards appear substantial

enough to warrant attention. To use the peanuts effect to advantage,

one might alert people to their rewards daily but then deliver them

monthly to create larger aggregate payments.

■ PRESENT-BIASED PREFERENCES

Another central observation from behavioral economics is the concept of hyperbolic discounting, or “present bias.” It is standard in

conventional economics to assume that people discount the future; for

example, $1000 today is worth more than $1000 a year from now, since

money can be invested and earn interest. However, people tend to discount outcomes that are close in time more steeply than outcomes that

are farther off in time; the degree of time discounting is disproportionately greater for short time delays than for long ones, in contrast to the

assumption of standard economic models.

The medical implications of present bias are profound. For example, most people would desperately like to avoid a stroke, and many

patients with hypertension have an understanding that taking their

antihypertensive medications is one of the best ways to avoid a stroke

in the future. While the classical economist would see daily adherence

to antihypertensive treatment as a trivial cost to avoid the major cost

of a future stroke, the stroke that is avoided is far in the future and

uncertain; moreover, the stroke that is prevented is never noticed. In

contrast, even the relatively small effort required to stay on antihypertensive medication is incurred in the present and comes without any

immediate compensatory benefit. To the extent that patients overly discount the future harms of a later stroke, they will be less motivated to

invest today in their own medication adherence. To the classical economist, these patients are behaving irrationally because they are failing to

make small investments that, if they did the calculations, would clearly

make them more likely to be substantially better off; this reasoning is

parallel to how classical economists consider people who undersave

for retirement (the vast majority of Americans) to be irrational. To the

behavioral economist, these errors are targets for therapy, in the way

that we can see genetic mutations or defects in chemical pathways as

therapeutic targets in the management of illness.

Present bias is somewhat subtler than simply steep discounting of

the future. In fact, it reflects two behavioral tendencies: (1) the tendency to overweight immediate costs and benefits relative to those

occurring in the future, as just discussed, and (2) the tendency to take

a more evenhanded approach to future costs and benefits. People are

much more willing to begin dieting tomorrow, because the overweighting of immediate costs deters us from the immediate deprivation of

dieting, and the more balanced perspective on delayed deprivation

makes us willing to impose these costs on ourselves in the future.

Although present-biased preferences typically promote unhealthy

behaviors, policymakers can use them for beneficial effects. The

motivational impact of benefits and costs, such as rewards for good

behavior and punishments for bad behavior, can be increased substantially if they are made immediate. These consequences should coincide

as closely as possible with the timing of the behaviors they are meant

to encourage or deter. Funds for this could be provided by employers

or insurers, for whom this might be a cost-effective way to improve

worker health and productivity.

Such programs have been shown to have dramatic effects in the

area of drug addiction. This success is particularly striking, because

many individuals with drug addiction already face major adverse

consequences, such as loss of livelihood and disenfranchisement from

their families; yet these costs are often insufficient to motivate abstinence. Similarly, small incentives offered on proof of abstinence have

succeeded in tripling smoking-cessation rates where the far larger (but

delayed) incentives in terms of improved health have failed. Small,

daily, lottery-based incentives have significantly increased medication adherence and weight loss in part because they bring immediate

rewards (money, excitement) to a situation in which the benefits of

avoiding ill health are typically distant and uncertain.

Thus, rather than requiring individuals to make decisions based

on consideration of their long-term best interests, it might be useful

to change short-term incentives so that beneficial actions are easier

and more attractive to choose or small costs—“hassle factors”—are put

in the way of behaviors adverse to health. Some school districts have

begun to use this approach by removing various products such as soda

and candy from vending machines so that the cost of obtaining them

now includes a walk off campus, while healthier food and beverage

options are immediately and readily available. In addition, people’s

willingness to commit to future changes can be leveraged by giving

them choices between health-benefiting and health-harming behaviors

well before they actually have to act on them. An example of this is

scheduling gym visits and laboratory tests to monitor cholesterol ahead

of time or having employees preorder lunch shortly after breakfast

when they are more likely to select healthy meals both due to present

bias and to their current lack of acute hunger.

Nonlinear Probability Weighting Prospect theory also encompasses a second important dimension: the way that people weight

probabilities. In contrast to the standard expected utility model,

which assumes that people weight outcomes according to their raw

probabilities, prospect theory assumes that people overweight, but are

insensitive to differences in, small probabilities—for example, between

a 0.001 and a 0.00001 chance of winning a prize, even though the

probabilities differ by several orders of magnitude—except where they

provide a transition to certainty. Such overweighting of small probabilities is partly responsible for the enormous attraction of lottery tickets,

which are widely seen by experts as a poor use of money; yet, like present-biased preferences, this overweighting can be used to advantage in

public health interventions.

Following these cognitive pathways, lottery-based reward systems

have been introduced in programs aimed at motivating diverse health

behaviors (described more fully below). These interventions exploit

overweighting of small probabilities and also play on other psychological insights. Because people tend to be motivated by both the

experience of past rewards and the prospect of future rewards, these

lottery-based systems provide frequent small payoffs and infrequent

large payoffs. This approach has been demonstrated to be effective in a

variety of areas, including helping people lose weight (52.6% of people

achieve 16-week weight loss goals compared with ~10.5% in a control

group) and reducing medication nonadherence (from ~23% to ~3%).

However, results across studies and contexts have been inconsistent,

suggesting that the active ingredients of these lottery-based incentives

have not been fully elucidated.

Regret Aversion People dislike regretting decisions they have

made, often voicing laments such as “If only I had . . . .” Contemporary

rhetoric recasts one form of regret aversion as “FOMO,” the fear of

missing out. Moreover, people are sufficiently far-sighted to anticipate

possible future regret and seek to make decisions today that reduce

that risk in the future. The avoidance of anticipated regret is a useful

exception to present-biased preferences.

Regret aversion helps to explain the success of the Dutch postal code

lottery, in which winning postal codes are selected and those living

within the selected areas who purchased tickets receive prizes. Those

3779Behavioral Economics and Health CHAPTER 481

who did not purchase tickets learn that they would have won had they

done so. Individuals see their neighbors winning large prizes that they

do not share, or fear that this will occur, and their desire to avoid future

regret drives subsequent lottery participation.

Anticipated regret has been shown to affect a variety of preventive behaviors, such as the significant increase in vaccination use

among people who experienced illness after failing to get vaccinated.

Lottery-based incentive programs where eligibility is conditioned on

adherence (for example, you aren’t eligible to play the lottery unless

you had taken your medication or checked your blood pressure the

previous day) can incorporate regret aversion into their design by

notifying both winners and losers. Those whose number comes up but

are ineligible for a reward because of nonadherence are told they would

have won had they only taken their medication, checked their blood

pressure, or done whatever it was the lottery program was designed to

promote. Because people hate the feeling of regret, they are more likely

to engage in behaviors to avoid that feeling. Indeed, the advertising

campaigns of some traditional lottery systems take advantage of regret

aversion. Many people play the same favorite number when the buy lottery tickets. It easy to imagine the disappointment you would feel if you

missed buying a ticket one week and that was the week your favorite

number came through. Advertising slogans like “don’t let your number

win without you” keep people in the game in order to escape that feeling of regret. The same techniques used to promote the marketing of

lottery tickets can be deployed toward health promotion.

Defaults Although many interventions playing on behavioral economics do not qualify as “nudges,” nudges remain a powerful way to

influence choice. Traditional economic thinking ignores the power

of the default option—the path that is “selected” when no selection

is made. However, the default bias, or status quo bias, reflects our

tendency to take “the path of least resistance”—to continue doing

what we have been doing, or to do what comes automatically, even

when superior alternatives exist. Defaults have been blamed for a

wide range of suboptimal outcomes, from the failure of employees to

put aside retirement funds in companies whose default contribution

rate is zero, to suboptimal allocation between investment alternatives,

to excessive consumption of French fries and large sodas as part of

“supersized” meals at fast-food franchises. Likewise, countries and

states that have “opt-in” policies for organ donation—that is, the

default is nonparticipation—tend to have lower donation rates than

those with opt-out defaults or (as is increasingly viewed as the best

compromise) those in which citizens are required to make an active

choice about whether they want to be listed as a donor. Defaults that

prioritize comfort over so-called “heroic measures” have been shown to

increase the rate at which patients with terminal lung diseases choose

comfort-oriented plans of care, and they could be used more widely to

encourage the choice of beneficial health options that would vary based

on the clinical context.

If used tactically, a choice architect (the person who makes decisions

on how choices are presented to the end user) can utilize defaults such

as changing scheduled automatic prescription refills from 30 to 90 days

(or longer) for patients requiring lifelong chronic-disease therapy or

changing the default option in fast-food restaurants in combination

meals from large sodas to small sodas or water or from French fries

to carrots unless you ask for French fries to help propel people toward

self-beneficial behaviors. A default toward prescription of generic

medications embedded in a health system electronic order entry system moved generic prescribing to nearly 100% in a context in which

previous attempts at physician education had shown no effect at all.

Changing the default framing of an invitation to patients with poorly

controlled diabetes to join a free remote glycemic monitoring program

tripled enrollment from ~13% to 37%.

Such approaches cost nothing, since a default has to be set one way

or another, preserve freedom of choice, and could change behavior

substantially. We miss out on many opportunities to nudge people toward healthier lifestyles when we do not consider the default

option as a tactical choice to be actively incorporated into an overall

health-promoting strategy.

The Rational-World Bias Perhaps the most consequential decision error affecting health-related behaviors is the tendency of public

health officials and private-sector benefit designers to make policy

decisions based on the assumption that people’s choices are deliberative

and rational. This in turn leads to assumptions that information provision is all that is needed for optimal decision-making and that, when

financial incentives are offered, the amount is all that really matters.

One significant manifestation of the rational-world bias is the complexity of health insurance plans. Health insurance is complicated for

many reasons, but one reason is that health care itself is complicated

and health insurance plans are full of deductibles and copayments

and other financial incentives that are designed by economists and

actuaries to encourage insured patients to direct their care in specific

cost-minimizing and health-maximizing ways. For these built-in

incentives to exert their desired influence, however, patients need to

understand the incentives they are facing, and there is considerable

evidence that they do not. This complexity of insurance benefit design,

rather than creating perfectly tailored incentives, is part of insurance

design’s undoing—because an incentive that cannot be understood

cannot be effective. Most consumers lack an understanding of the most

basic insurance concepts, such as deductibles, copays, and coinsurance,

and, when given a simplified version of a traditional insurance plan, are

unable to compute the costs they would incur for basic services.

Giving a nod to the benefits of simplicity, the ACA in the United States

requires insurance plans to be presented in a standardized document

that describes plan features such as premiums, deductibles, and coinsurance. However, describing something that is inherently complex in

simple terms raises the risk of glossing over important subtleties. A

more productive approach would be to provide consumers with a truly

simplified insurance product. For example, insurers might create a

plan designed with copayments only, since copayments are most easily

understood by consumers: copayments are analogous to paying a price

for a good or service when shopping, whereas coinsurance and deductibles make it difficult for people to estimate how much their care would

cost. It is difficult, if not impossible, for a consumer to estimate 10%

coinsurance on a hospital stay or an emergency room visit since they

have no idea of the full amount on which the 10% will be calculated.

For medical markets and patient cost-sharing elements embedded in

plan design to affect decision-making ex ante, consumers need some

way of accurately gauging not only benefits and risks of treatments but

also the costs. This should occur before receiving a diagnostic test or

treatment as opposed to what typically happens now: a medical bill is

received after the fact.

Another manifestation of the rational-world bias is the more insidious and pervasive belief that if only people knew more about how

to advance their health they would do so. This error leads to making

education the centerpiece of many health interventions even though

public health experts have long recognized that knowledge alone rarely

translates to health-enhancing behavior. For example, nearly everyone

knows the health hazards of smoking. While raising and maintaining awareness of the dangers of smoking is an important goal, the

financial budgets of the Centers for Disease Control and Prevention,

or the time budgets of clinicians, might be more efficiently allocated

toward efforts to change behavior that do not presume that the deficit

is a lack of knowledge. Indeed, it has been argued that smokers tend

to overestimate the hazards of tobacco, just as many women tend to

overestimate their risks of developing breast cancer. In these cases,

better or more accurate education might lead people to lower their

estimates of risk and might, if they were perfectly rational, lead them to

reduce health-promoting behaviors. In cases where the deficit is not in

knowledge but in behavior, reliance on education as a primary avenue

for reducing health risks may divert efforts and resources from other

activities that might be much more effective.

■ APPLICATIONS

Weight Loss Efforts to combat obesity using incentives started

in the 1970s. This early work was motivated by the observation that

participants who deposited money and other valuables with a therapist

3780 PART 20 Frontiers

and signed contracts in which return of their valuables was contingent

on progress toward prespecified goals lost tremendous amounts of

weight. Even though participants received no training in weight loss or

maintenance strategies, participants lost an average of 32 pounds. This

small initial study lacked control groups and long-term follow-up but

provided an important proof of concept.

In the first systematic study of deposit or precommitment contracts, participants responding to an advertisement for a weight loss

program were informed that study participation required a deposit

of $200 (1974 dollars), which would be fully refunded contingent on

satisfactory weight loss. After an 11-session, 10-week program, those

who received incentives for losing weight or for limiting calories

lost significantly more weight than those who received incentives

for attending sessions. Of the participants in the former two groups,

70% lost >15 pounds. The major limitation to this approach was that

only 15% of the prospective participants who responded to the initial

newspaper advertisement ended up enrolling, suggesting that deposit

contracts that require participants to commit substantial funds up

front are very effective for people who agree to participate, but that this

requirement likely deters a substantial portion of high-risk participants

from entering and its effectiveness may just as much reflect selection of

those highly motivated to lose weight.

A subsequent study tested the effects on weight loss of deposit

contracts of $30, $150, or $300, with the deposits returned based on

either individual or group weight loss over 15 weeks. Participants in

the intervention group could win $1, $5, or $10 per pound lost up

to a maximum cumulative weight loss of 2 pounds per week (either

individual or group average). Mean weight loss was large in all three

groups but did not differ significantly based on contract size. However,

the proportion who reached the goal of a 30-pound weight loss was

significantly higher in the larger dollar groups.

Because it is increasingly difficult to shed pounds as weight is lost,

the investigators also tested whether a deposit contract with increasing

payments ($5, $10, $20, $40, or $75) for each 5-pound increment of

weight loss would be more effective than offering $30 for each 5-pound

increment of weight loss. Participants in both conditions were also

offered a maintenance program requiring a $100 deposit, returned in

$25 increments for attendance at follow-up visits every 3 months. The

increasing contract resulted in qualitatively larger weight loss during

the weight loss phase, but the maintenance program did not prevent

weight regain, likely because the magnitude of the deposit contract

for maintenance was small and feedback was infrequent (only every

3 months).

Paying participants for weight loss using direct payments was less

effective than deposit contracts. In a randomized trial, cash payments

up to $25 per week for making 100% of proportional progress toward

goal, $12.50 for 50% of goal, and $2.50 for not gaining weight did not

result in greater weight loss in the payment group than among control

subjects.

Studies that have shown no effects on either initial weight loss or

maintenance typically have used incentives of small magnitude or

were targeted at behaviors like attendance at weight loss programs

that, by themselves, do not ensure weight loss. In recent years, weight

loss incentives have become a common feature in programs used by

employers and health plans, and a variety of start-ups like stickk.com

and dietbet.com use deposit contracts as a way of trying to help people

lose weight.

Newer approaches have provided proof of concept that daily lotterytype incentives and precommitment contract incentives promote

initial weight loss over 16 weeks (lottery = 13.1 pounds; p = .014 for

lottery vs control; deposit contract = 14.0 pounds; p = .003 vs control;

Fig. 481-2). However, participants regained most of the weight they

had lost over the following 3 months. Longer trials (8 months) found

no difference in effectiveness of deposit contracts for continuous

8-month weight loss versus 6-month weight loss with 2 months of

maintenance, but both were successful in achieving a mean weight loss

of ~10 pounds. Subsequent tests included an employer-based study

showing greater effectiveness of team competitive versus individual

incentives, although there was a confound in that participants in the

teams had the potential to win more money than those in the individual arms.

While both of these studies suggest that financial incentives promote weight loss, many employers use premium adjustments (increases

or decreases to health insurance payments) as their standard approach

to using financial incentives for health promotion. The effectiveness of

premium-based financial incentives to promote weight loss has been

assessed in a workplace wellness program and with a goal of losing

5% of initial weight over the next year. Participants were randomly

assigned to a control group with no other intervention or one of three

financial incentive groups. Two intervention groups were offered a

premium reduction of $550 if they achieved their weight loss goal by

1 year. The “delayed” group would receive the premium adjustment in

the following year, spread across each pay period. The “immediate”

group would have their premiums adjusted as soon as the weight loss

goal was met. A third intervention group was offered a daily lottery

with about a 1 in 5 chance of winning $10 and a 1 in 100 chance of

winning $100. To be eligible to win each day, participants had to weigh

in and be on track to lose 5% of their initial weight by 6 months, with

maintenance for the subsequent 6 months. After 1 year in the program,

none of the intervention approaches showed a significant degree of

weight loss. The control group gained <0.1 kg.

The relative ineffectiveness of premium-based financial incentives is

not surprising given their design. Premium adjustments are logistically

appealing because the infrastructure to do so is already in place, there

isn’t strong evidence for their effectiveness and theory argues against

it. Such incentives are typically hidden in paychecks that are directly

deposited in bank accounts and may go unnoticed by the individual.

While a $550 incentive seems like a large amount, it is only $20 in each

biweekly paycheck. Typically, these incentives are administered on an

all-or-nothing basis contingent on meeting a specific threshold such as

a body mass index of ≤25 kg/m2

, meaning that those who are close to

that target may be motivated by a goal within reach, but those who are

farther away (and have the most weight to lose and the most health to

gain) may be less motivated (perhaps even demotivated) by a goal that

does not seem attainable. The best evidence and theory now suggest

that the standard approach of using premium-based incentives is not

very effective and that employers should consider alternative delivery

channels for financial incentives to promote health.

A common problem employers and health plans struggle with is

getting high levels of engagement among employees/health plan members. Weight Watchers, the largest commercial weight loss program

in the United States, worked with a team of academic investigators to

conduct a randomized controlled trial involving >23,000 participants

testing the impact of offering employer subsidization of Weight Watchers membership fees of 50%, 80%, 100%, or 50% that could turn into

100% conditional on attending at least three Weight Watchers classes a

month. Higher subsidies led to higher program enrollment (p <.0001).

Enrollment differed significantly by subsidy level (p <.0001). The 100%

subsidy produced the highest enrollment (7.7%), significantly higher

than each of the lower subsidies (vs 80% subsidy: 6.2%, p = .002; vs 50%

subsidy: 3.9%, p <.0001; vs hybrid: 3.7%, p <.0001). Enrollment in the

80% subsidy group was significantly higher than both lower subsidy

groups (vs 50% subsidy: 3.9%, p <.0001; vs hybrid: 3.7%, p <.0001).

Among enrollees, however, there were no differences among the four

groups in attendance or weight loss. In all groups, overall weight loss

0

5

10

15

Control Lottery Deposit

Weight loss (lbs)

Control

Lottery

Deposit

FIGURE 481-2 Weight loss in incentives versus control. (Figure created using

data from KG Volpp et al: Financial incentive-based approaches for weight loss: A

randomized trial. JAMA 300:2631, 2008.)

3781Behavioral Economics and Health CHAPTER 481

was modest, with a mean weight loss of 2.6 pounds (95% confidence

interval [CI], 5.7-pound loss to 0.3-pound gain) in the 100% subsidy

arm; 1.5 pounds (95% CI, 5.6-pound loss to 1.3-pound gain) in the 80%

subsidy arm; 3.8 pounds (95% CI, 7.9-pound loss to 0.4-pound gain)

in the 50% subsidy arm, and 4.0 pounds (95% CI, 8.1-pound loss

to 0.1-pound gain) in the hybrid subsidy arm. In all arms, attendance

rates dropped steadily over time, suggesting that while the hypothesized trade-off between higher subsidies and lower ongoing engagement in the program did not exist, ongoing participation is a challenge

across the board, and subsidies that lower the cost of participating are

insufficient to achieve high levels of ongoing engagement (Fig. 481-3).

In another study, 281 overweight and obese adults were randomly

assigned to a control group or one of three incentive groups for a

13-week physical activity program. All participants were given a goal

of achieving 7000 steps a day, tracked automatically using their smart

phones. Participants in each of the three incentive arms were offered

the same magnitude incentive, $1.40 per day, and were told that

accumulated earnings would be sent via a check at the end of each

month. However, the incentive in each group was framed differently.

In the standard gain incentive group, participants were told that they

could earn $1.40 each day they achieved the goal. In the regret lottery

incentive group, participants were in a daily regret lottery in which

they had an 18 in 100 chance of winning $5 and a 1 in 100 chance

of winning $50, which averages to $1.40 per day. In the loss framing

incentive group, participants were told at the beginning of each month

that $42 had been placed in a virtual account and that they would

lose $1.40 each day the goal was not achieved. During the 13-week

intervention, participants in the control, standard gain, and regret gain

arms achieved their daily step goal ~30%, 35%, and 36% of the time,

respectively, but the gain-framed incentive arms were not significantly

different from the control arm. However, in the loss framing group,

participants achieved the goal 45% of the time, a 50% relative increase,

which was significantly greater than the control arm (p = .001). This

study demonstrated how loss framing can be used to motivate behavior

change. It is also one of the first studies to create a loss frame without

requiring participants to put their own money at risk using a deposit

contract. This is important because fewer people are willing to engage

in deposit contract-based incentives than reward-based incentives.

Experience with behavioral economics and weight loss provides

several general lessons that are transferable to other health applications. While early studies largely emphasized research in the size of

financial incentives, later studies have revealed that the design of the

incentive strategy is at least as important. Moreover, designs can vary

considerably based on the timing of incentives that can be immediate

or delayed, or frequent or one time; the setting of targets that can be

achievable, aspirational, or demotivating; the certainty of incentives

that can be fixed or probabilistic; the channel for incentives that can

either be delivered separately or bundled through payrolls; or the

framing of incentives as gains or losses. Classical economists would

see only the size of the incentive as an available lever for motivation,

but behavioral economists face a much larger set of considerations in

designing and testing effective therapies.

Medication Adherence Numerous studies have shown that at

least a third of patients fail to adhere to medication regimens. One

approach to improving medication adherence is to change some of

the underlying defaults by, for example, using 90-day prescriptions for

chronic illness medications as opposed to 30-day prescriptions. While

we are unaware of empirical evidence supporting longer prescription

cycles, it seems logical that adherence rates would be higher over a

year if people had to get refills three times as opposed to 11 times—the

latter provides more opportunities to forget, experience delays, or fall

off the wagon. Automatic refills through prescription mail order might

similarly prevent some people from inadvertently falling off the wagon

of medication adherence, giving them one less thing to think about or

fail to think about.

Ninety-day prescriptions or automatic refills could be set up as

the default, and patients or their providers could opt out if desired.

Of course, opt-out defaults are not always possible. A large pharmacy

benefits manager wanted to encourage automatic refills for patients

on long-term medication but could not have members opt out of such

a system because of the potential that those who missed the implications of the opt-out would be surprised or angry about finding credit

card charges for automatic refills. Using active choice—i.e., requiring

customers to make an explicit decision rather than only providing the

service to those who proactively opt in—the company also presented

options in a way that highlighted the convenience of automatic refills:

“We can send your refills to you automatically or you can get your

refills manually if you prefer.” This resulted in more than twice as many

patients choosing to be in the automatic-refill program as compared

with opt-in.

A feature of many health insurance plans is that they require

patients to pay some costs out of pocket and hence discourage the use

of a number of high-value elements of care, such as the treatment of

Month

Average # of meetings attended per month

1

0.0

0.5

1.0

1.5

2.0

2.5

3.0

23456789 10 11 12

Arm 50% subsidy 50–100% subsidy 80% subsidy 100% subsidy

FIGURE 481-3 Attendance rates at Weight Watchers over time in different subsidy groups. (Reproduced with permission from LK John et al: The effect of cost sharing on

an employee weight loss program: A randomized trial. Am J Health Promot 32:170, 2018.)

3782 PART 20 Frontiers

hypertension or the use of statins by patients with diabetes—care that

is widely seen as worth its cost. Support for the use of health insurance deductibles that require patients to have “skin in the game” for

their health expenditures derives from insurance theory as well as the

seminal RAND health insurance experiment, which demonstrated that

these deductibles help overcome moral hazard and reduce the consumption of health care services. Copayments, deductibles, and other

out-of-pocket costs make consumers more cost conscious and so aim

to make them better shoppers for health care services. Indeed, the rise

of high-deductible health plans largely aims to increase patients’ skin in

the game, to make them more value-conscious shoppers.

However, while deductibles and copayments make sense as a way

to reduce overutilization of some lower-value health care services,

deductibles and copayments make considerably less sense for medications and services that are high value and low cost—e.g., when

patients receive medications to manage their hypertension, diabetes, or

hyperlipidemia. Given that deductibles and copayments are designed

to reduce utilization, why would we ever want to apply them to antihypertensives or statins or insulin given the high health value of these

drugs? Why put any barriers between patients and these drugs? Indeed,

as the RAND health insurance study showed, high-deductible health

plans are as likely to discourage the use of high-value services as the

use of low-value services. Because patients lack knowledge about what

tests or services are of high or low value and do not have information

about the relationship between price and quality, such plans discourage

spending on all tests and services, including those of high value.

Value-based insurance design—which involves discounting, or making free, services that are deemed to be high in value—is an attempt to

sharpen the blunt incentives inherent in deductibles and copayments.

Value-based insurance design was inspired by research that showed

the use of higher copayments significantly reduced the use of services

such as prescriptions but ultimately raised costs because lower rates of

medication nonadherence led to higher rates of emergency department

visits and adverse outcomes. Extrapolating from these results, it was

natural to conclude that lowering cost sharing for high-value activities,

such as taking medications for chronic conditions, would increase

adherence and potentially thereby reduce costs. The ACA incorporates

a kind of value-based insurance design in its requirement that preventive services be offered to patients at no charge.

Unfortunately, value-based insurance design has not delivered on the

hope that it would both save money and significantly improve health.

From the perspective of the purchaser (for example, the employer

or insurer), the economic impact of value-based insurance design

depends on whether it can make enough people adherent who were

previously nonadherent—and on the health and cost consequences of

that improved adherence—to offset the loss of the copayments from

those who were already adherent. Although some experimental tests of

value-based insurance design have found that copayment reductions

increase adherence, those effects have typically been small, in the range

of 3–6 percentage points. Even among patients who had recent heart

attacks and were given their cardiovascular medications for free, average adherence was only ~45%, just a few percentage points higher than

that seen with regular copayments. One reason for these disappointing

results is the “dog that didn’t bark” problem. People who are nonadherent do not notice that their copays have been reduced because they are

not using (and thus are not paying for) the service.

Indeed, one of the valuable lessons learned from efforts to introduce

value-based insurance design has been a reminder of the asymmetry

of the forces that surround patient engagement. Based on conventional

economic thought, it might seem reasonable to assume that decreasing

copayments would create effects equal and opposite to those of increasing copayments. However, behavioral economic research reveals that

framing matters and that losses (in this case, higher copayments

intended to reduce use) loom larger in patients’ minds than gains

(lowered copayments). Furthermore, people who would be deterred

by higher copayments are different from people who might become

adherent with lower copayments, because the first group consists of

those who take their medications while the second group consists of

those who do not. Behavioral economic thinking, therefore, helps to

explain what has been observed: increases in copayments have larger

effects in reducing adherence than decreases in copayments have in

raising adherence. In general, copayment increases lead to far smaller

decreases in medication adherence than copayment decreases lead to

increases in medication adherence.

Value-based insurance design is an appealing idea. But its benefits

could be increased through the application of ideas from behavioral

economics, such as simple changes in reward delivery to increase

salience (e.g., retaining the copay, but sending a rebate) and communications from insurers to patients so that even those who are nonadherent are aware of the benefit. Better designs might also reflect that most

medication adherence happens at least daily, and so reinforcements to

that behavior probably need to occur more frequently than the 30- or

90-day cycles coinciding with prescription refills.

A series of studies have used daily lottery-based financial incentives

to improve medication adherence. Early work tested the impact of a

lottery on medication adherence among patients on warfarin. Participants were eligible for the lottery daily if they correctly took their

warfarin the day before. In two pilot studies, we found that the rate of

incorrect pill taking was 1.6–2.3%, compared with a historic mean of

22% incorrect pill taking. In a two-arm randomized trial of lotteries

for warfarin adherence, the a priori subgroup with baseline international normalized ratios (INRs) below the therapeutic range showed

no change relative to control, but in the a priori subgroup with out-ofrange INRs, there was a significant reduction in out-of-range INRs in

the lottery arm versus the control arm (adjusted odds ratio, 0.39; 95%

CI, 0.25–0.62).

A four-arm National Heart, Lung, and Blood Institute–funded randomized controlled trial tested the impact of daily lotteries and daily

reminders in a 2 × 2 factorial design on warfarin adherence to address

the question of the degree to which a daily lottery is effective due to

the fact that it also constitutes a daily reminder. While participants

in the reminder group had the lowest percentage of time out of target

INR range, with an adjusted odds of an out-of-range INR 36% lower

than among those in the control group (95% CI, 7–55%), the only

group with significant improvement in incorrect adherence was the

lottery group (incorrect adherence: 12.1% compared with 23.7% in the

control group; difference of 7.4%; 95% CI, –14 to –0.3%). There was

no relationship between changes in adherence and anticoagulation

control in the lottery group, highlighting that participants may appear

to change their behavior without perhaps taking the medication and

the importance of serologic or biometric confirmation when possible.

A more recent study of 1509 patients hospitalized with heart attacks

from 45 states did not find an effect of lottery incentives on time to

first vascular rehospitalization (hazard ratio, 0.89; 95% CI, 0.73–10.09;

p = .27) or death (hazard ratio, 1.04; 95% CI, 0.71–1.52; p = .84). Mean

medical costs did not differ between control ($29,811) and intervention

($24,038) (difference, –$5773; 95% CI, –$13,682 to $2137; p = .15).

Although it is not clear why the intervention did not succeed, it seems

likely that the intervention was mistimed: patients were enrolled a

mean of 40.8 days following index discharge due to the time required

for adjudication of insurance claims. Post hoc analyses suggested that

patients who had already been readmitted prior to joining the intervention did achieve lower subsequent readmission rates as a result of

the intervention. To be successful, interventions to change behavior not

only must reflect good principles of behavioral science but must also

fit logistically into the information systems and patient and clinician

journeys that already exist.

The 5000 Hours Problem Indeed, a major challenge is determining how to reach patients and reinforce their behavior each day if

we want to significantly improve their health behaviors. Even patients

with chronic illnesses may spend only a few hours a year with a doctor

or nurse, but they spend ~5000 waking hours a year doing just about

everything else. Those 5000 h are when they live their lives and make

choices about what to eat and whether to exercise, smoke, take their

medications, or visit the doctor.

Although what people do in those hours almost certainly affects

their health outcomes, the hours are typically ignored by the U.S. health

3783Behavioral Economics and Health CHAPTER 481

care system, in part because current approaches to U.S. health care

financing support health care during visits to the doctor, not between

them, and because “hovering over” people during the hours between

visits is personnel-intensive, often requiring nurses or other clinicians

to call or visit patients or to staff telemedicine programs. Hovering also

requires a fair amount of the very kind of engagement in their own

health and health care that is so often missing in the patients these

interventions aim to reach. As a result, many of the most promising

efforts in telemedicine and home health care have been disappointing.

If some form of hovering is required to engage people who are otherwise hard to engage during the 5000 h, it almost certainly has to become

substantially more automated—both because providers must reduce

the need for expensive personnel and because patients have in many

cases already revealed limits to their willingness to exert themselves to

improve their health. Nevertheless, there is reason for optimism based

on the increasing use of cell phones and other wireless devices that

make it technologically easier to embed reminders and other forms of

touch into patients’ lives. Indeed, one key lesson from behavioral economics is that rather than trying to change people’s behavior patterns

to promote health, it is better to restructure their environment and

circumstances so that their existing behavior patterns are more likely

to lead to better outcomes. Those efforts require a substantial amount

of hovering over patients. Cell phones and other wireless devices do not

necessarily change behavior on their own, but because they are already

part of many patients’ everyday lives, they allow previously private

behaviors to be witnessed and at times acted upon.

Indeed, an error in early approaches to technology and health

behavior overgeneralized from the technologies that support the

“quantified self ” movement. Apps and wearables that track your diet,

physical activity, and biometrics were largely designed for people passionate about measuring themselves. Such individuals do not need a lot

of encouragement to wear devices or enter data. The same approaches

are far less likely to be useful for patients with difficult-to-manage

chronic illness. Many of the internal and external challenges that make

their chronic illness hard to manage also make such monitoring hard

to manage. A patient who is nonadherent to medication is likely also

to be nonadherent to using a new electronic device, but devices like

cell phones that are already in use or other devices that require much

less active involvement (like wireless pill bottles) offer more conceptual

appeal.

■ REFLECTIONS ON SARS-CoV-2

Since SARS-CoV-2 began spreading early in 2020, it has become

increasingly clear that behavioral decisions among both policymakers

and individuals play a major role in spread. Within countries that

exercised strict control over individual behavior through measures

like mandated testing, isolation, quarantine, use of contact tracing

apps, and fines for noncompliance, containment was achieved and the

number of new cases dropped nearly to zero. Within countries like the

United States in which there was no comparable national policy and

policy decisions were left to the states, the rate of spread and degree

of containment varied enormously. Leaders of many states relied on

the rational-world bias, assuming individual decisions would be deliberative and rational. Individual decisions about whether to maintain

social distance, wear masks, and otherwise modify routines to reduce

risk reflect many of the behavioral challenges described earlier in

this chapter—present bias, overoptimism, anticipated regret, narrow

bracketing, status quo bias, and the peanuts effect all contributed

to people either taking more or less precautions depending on their

personal decision-making proclivities. Vaccination acceptance poses

similar challenges; if governments leave it up to individuals to decide

and do not use strong incentives or mandates, significant variability

in behavior will likely ensue with corresponding consequences for the

populations they govern.

■ FUTURE PERSPECTIVES

Human health derives from the interaction of basic biologic processes,

environmental exposures, social structures, and behavior. The field of

behavioral economics has greatly contributed to our understanding

of behavior and has made significant contributions to the science of

public policy. Given that understanding, we have the opportunity to

replace older policies based on unrealistic normative models of rational

decision-making with newer policies reflecting our most up-to-date

understanding of how humans actually make decisions. Individual and

population health outcomes would be very different if people were

able to weigh the present and future costs of their actions carefully and

dispassionately and had the necessary information and self-control to

implement behavioral plans and overcome decision errors that contribute to unhealthy behaviors. Because few people can meet any of those

challenges, we should not structure our behavior change interventions

and public health policies around such models of behavior.

There is broad potential for improving private and public approaches

to health behavior by drawing upon a more realistic understanding

of human motivation rooted in behavioral economics. One major

question is whether developed economies will continue to invest the

majority of their health care dollars in treatment (typically ~97% of

health care dollars) rather than shifting it toward innovations that seek

to keep people healthy. This will be particularly important in settings

where treatment options are limited and highly effective methods of

prevention exist. For example, it has been estimated that a combination

of low-cost cardiovascular drugs could reduce cardiovascular events by

62–88% with perfect adherence, revealing that reducing atherosclerotic

cardiovascular disease risk is largely a behavioral challenge, given that

adherence to medications remains low despite effective pharmacologic

solutions. Shifts in health care financing away from fee-for-service

toward various forms of payment that require health delivery systems

to take on financial risk for populations of patients may drive greater

interest in addressing these behavioral and social determinants of

health. Research expenditures, which in many developed countries are

also roughly allocated ~97% to new treatments and ~3% to prevention, similarly could be shifted to focus more on testing of innovative

approaches to improve population health.

The same errors that misdirect patients and providers also misdirect policymakers. In part because of present bias, preventive services

often are covered by insurance only if they show a positive return on

investment, yet treatments of existing disease are not held to the same

standard. An employer wondering whether to introduce a smoking

cessation program for employees wonders about the return on that

investment in terms of reduced illness and its cost. The same employer

might never question the return on investment of treating lung cancer

in the same employee pool, despite what would almost certainly be a

negative return on investment. These asymmetries are so embedded

in policy making as to be nearly invisible or at least unchallengeable.

In fact, the cost of treatments is not even allowed to be considered in

Medicare coverage decisions. This prohibition naturally leads to overinvestment in treatments of low value and underinvestment in prevention. The same standard for assessing the impact of health programs,

with the goal of achieving the most health possible with the available

resources, should be used for both preventive and therapeutic services.

Despite the promise of behavioral economics in structuring policy

solutions to social goals, plenty of existing policies that have nothing to

do with behavioral economics are effective. For example, raising taxes

on cigarettes or sugar-sweetened beverages or other unhealthy goods

where it is in the public interest to consume less is a powerful policy

tool derived from classical economics. Indeed, tobacco taxes represent

one of the most effective ways to curb the use of tobacco and its initiation among youth. Behavioral economics can help make such policies

more effective but should not be seen as a substitute for them.

For private-sector entities, the implications of choosing defaults

wisely are recognized by many organizations that aim to shift the “path

of least resistance” toward healthier choices. Setting up defaults in

benefit program design to favor health plans that provide better coverage of preventive services, changing the environment in workplaces

to make it easier to take the stairs, and serving more healthful food in

cafeterias represent approaches to gently lead people toward individual

and population goals.

While medical research continues to generate new tests, interventions, and drugs, all of which successfully target conditions recently