Addressing Antibiotic Overuse in the Outpatient Setting: Lessons From Behavioral Economics

Introduction

The challenges posed by rising rates of antibiotic resistance cannot be overstated. Antibiotic resistance results in worse outcomes and higher costs. In the outpatient setting, internists too frequently encounter patients with infections that have no effective oral options for treatment.¹ The Centers for Disease Control and Prevention (CDC) estimates more than 2.8 million antibiotic-resistant infections occur in the U.S. each year and result in approximately 35,000 deaths.¹ Healthcare costs attributed to antibiotic-resistant infections surpass two billion dollars annually.

Outpatient clinicians contribute to the rise in antibiotic resistance through antibiotic overuse: the prescribing of antibiotics when not medically necessary, for durations longer than necessary, and the use of broader-spectrum antibiotics when narrower-spectrum agents would be equally, and often more, effective.^1,2 Studies estimate that more than 12% of all ambulatory visits in the U.S. result in an antibiotic prescription, and 23–30% of these prescriptions are “inappropriate” based on their assigned diagnosis.^3,4 Increasing attention has been paid to antibiotic stewardship efforts in the outpatient setting: in 2016, the CDC outlined core elements of outpatient antibiotic stewardship programs (ASPs).⁵ In 2020, the Joint Commission, a non-profit accreditation organization, implemented ASP requirements for outpatient accreditation.⁶

Behavioral motivations for antibiotic overuse are complex and have not been a focus of medical research until recently. Three motivations that affect outpatient clinician practice warrant further attention: (1) risk aversion, (2) action bias, and (3) patient satisfaction (Table, top). Existing strategies to address overuse have taken many forms, including clinician education, prospective audit-and-feedback programs, and preauthorization for restricted antibiotics.⁷ These interventions have been primarily used in the inpatient setting and are more difficult to implement in outpatient practices. Novel strategies focused on restructuring the decision-making environment, a practice known as choice architecture, may play a complementary role in addressing antibiotic overuse without reducing clinician autonomy. These methods show promise as effective strategies in the current campaign against antibiotic resistance.

Table:

Concepts from behavioral economics and their application to antibiotic overuse^a

Concept	Description	Application to Antibiotic Use
Contributors to Antibiotic Overuse
Immediacy effect and the tragedy of the commons	Users are drawn to decisions that produce individualized benefits while delaying risks and distributing them over the entire system.	Clinicians use broad-spectrum antibiotics because the potential harms of non-treatment of a bacterial infection would be directly borne by the patient and clinician whereas the harms of antibiotic overuse are, in the case of drug resistance, distributed over the entire medical system.
Action bias and indication creep	Individuals are motivated to action, even if inappropriate in the specific situation, as it is perceived to be more beneficial than inaction. Clinicians apply interventions to the wrong diagnosis.	Patients with non-specific symptoms and pyuria are misclassified as having urinary tract infection and given antibiotics.27
		Recommendations for antibiotics to start within four hours of presentation for patients with CAP resulted in antibiotic overuse.14,15
Patient satisfaction	Patients may feel heard or validated when they receive a prescription from their clinician.	Clinicians who prescribe antibiotics for upper respiratory tract infections receive higher patient satisfaction scores.20
Solutions to Antibiotic Overuse based on Choice Architecture
Relative social ranking	In times of uncertainty, users may be inclined to choose options that their peers have also chosen.	Comparisons to “best performing” peers reduced antibiotic prescription volume for certain conditions by 5% compared to controls.23
Accountable justification	Users may be less vulnerable to cognitive biases when making a choice if they are asked to justify their decision.	EMR free-text justification resulted in 7% fewer antibiotic prescriptions for respiratory tract infections compared to controls.23
Default practices	In settings with many choices, users often stay with a default option.	EMR antibiotic choices for respiratory tract infections that are grouped or individualized based on spectrum of activity influence prescription behavior in clinical vignettes.25
Pre-commitment	Users who have made a public commitment to certain principles in decision-making are more likely to make choices in line with those principles.	Posters expressing a commitment to reduce antibiotic overuse by clinicians in outpatient practice examination rooms had 20% lower antibiotic use compared to control practices.26

^a.CAP = community-acquired pneumonia; EMR = electronic medical record

Clinician practices regarding antibiotic prescribing

Risk Aversion: Immediacy Effect and the Tragedy of the Commons

Immediacy effect refers to a powerful motivator of human behavior in which individuals value immediate consequences more strongly than delayed consequences. The immediacy effect motivates antibiotic overuse when a clinician is more concerned about the immediate risks of untreated infection than the long-term risks of antibiotic resistance. In circumstances where a patient’s syndrome is caused by bacterial infection, the potential harms of non-treatment (i.e., worsening infection, hospitalization, sepsis) become apparent in the short-term and are directly borne by the patient, family, physician, and care team. Conversely, the harms of antibiotic overuse are often neither immediate nor visible. In the case of drug resistance, the harms are not only delayed but also distributed over the entire medical system. Therefore, even when the likelihood of bacterial infection is low, clinicians who only consider their immediate circumstances have little incentive to restrict antibiotic use: they accrue the benefits of antibiotic overuse by averting the risk of clinical decompensation no matter how small that risk may actually be, while the harms of antibiotic overuse are delayed and also distributed to the remainder of the population and medical system (i.e., antibiotic resistance).

This asymmetric distribution of benefits and risks between individual and system is conceptualized in economics as the “tragedy of the commons.” In the classic description of this concept, a group of self-interested farmers share the availability of a collective resource, such as the land on which their animals graze.⁸ The “tragedy” lies in the subsequent depletion of that resource because each farmer stands to individually benefit from over-grazing while the harmful consequence of resource loss is distributed equally among participants. In the case of antibiotic resistance, it is understandably difficult for individual clinicians to appreciate how each antibiotic they prescribe contributes to the global problem of drug-resistant infections. However, as a collective, their decisions are detrimental to the whole system.

Discomfort with Uncertainty: Action Bias and Indication Creep

A second contributor to antibiotic overuse is action bias, which is the tendency for individuals to take an action, even when it is not appropriate in the specific situation, because action is perceived as more beneficial than inaction.⁹ In antibiotic overuse, this may occur most often in times of diagnostic uncertainty. For example, an elderly patient presenting with non-specific symptoms and pyuria or bacteriuria may be misclassified as having a urinary tract infection even though asymptomatic pyuria and bacteriuria are common in elderly populations. Clinicians are inclined to assign a misdiagnosis with a proven treatment rather than settle in an alternative state of diagnostic uncertainty or with a diagnosis that has no recommended therapy.^10,11

Some authors also refer to this phenomenon as indication creep: when we take interventions with proven benefits in select patient populations and apply them to other patient groups.¹² Indication creep has been evident when antibiotic use is associated with a quality performance measure. In 2003, the Infectious Diseases Society of America (IDSA) recommended that patients with pneumonia receive antibiotics within four hours of presentation.¹³ This practice became a national quality measure and was linked to reimbursement by the Centers for Medicare and Medicaid Services. Subsequently, patients who presented to Emergency Departments (ED) and urgent care centers with respiratory complaints were frequently placed into a clinical treatment pathway for community-acquired pneumonia, which included prompt administration of antibiotics. Two important studies investigated the effect of this recommendation on antibiotic use. One study found that patients with an ED diagnosis of pneumonia were 39% less likely to meet standard criteria for a diagnosis of pneumonia.¹⁴ Another study found that after implementation of these quality measures, patients who received antibiotics as part of an initial admission diagnosis of pneumonia were more frequently discharged with a separate, often non-infectious, diagnosis for their original symptoms: the final diagnosis of pneumonia fell from 75.9% to 58.9%.¹⁵ Together, these studies suggest that clinicians became less accurate in diagnosing pneumonia after implementation of this quality measure with a concomitant increase in antibiotic overuse. These findings motivated a change in the 2007 IDSA guidelines where the recommendation for antibiotics within four hours of presentation was instead generalized to endorse administration at the time of pneumonia diagnosis without a specified target time.¹⁶ They highlight how action bias and indication creep can result in unintended consequences of antibiotic recommendations in clinical practice guidelines.

Patient Satisfaction

Another powerful motivator for antibiotic overuse is clinician desire to earn patient satisfaction. During clinical encounters, patients may perceive receipt of a prescription drug as validation of the illness experience, recognition of symptom severity, and evidence that they were adequately heard by their clinician during the encounter.¹⁷ Studies show that physicians feel compelled to prescribe antibiotics when they believe that it is within the patient expectations.^18,19 In tele-medicine encounters for upper respiratory tract infections, for instance, antibiotic prescriptions are associated with higher patient satisfaction scores.²⁰ This finding is particularly concerning given the substantial shift from in-person to virtual outpatient encounters this year related to the COVID-19 pandemic.

In the outpatient setting, the need to maintain patient satisfaction combines with compelling time pressures to make antibiotic prescription an expedient option during periods of high patient volume. Qualitative studies of physician prescribing for upper respiratory tract infections found that physicians may opt for antibiotic prescriptions when they are unable to provide lengthy counseling on more conservative management options for symptom management.^17,21 Emphasis on patient satisfaction in outpatient quality measures, reimbursement models, and promotion also combine to motivate clinician behavior in favor of antibiotic overuse.

Potential Solutions to Reduce Antibiotic Resistance

Strategies that have been emphasized in the fight against antibiotic resistance include the development of novel diagnostic and therapeutic interventions (such as point-of-care technologies), clinician education, prospective audit-and-feedback, and preauthorization for restricted antibiotics.⁷ However, these strategies are more easily implemented in hospitals than in outpatient practices. They will also have limited success unless the behavioral determinants of antibiotic overuse are also addressed at the clinician level. For example, the use of rapid diagnostic tests to reduce the use of broad-spectrum antibiotics will only prove effective if implemented in concert with informed and rational decision-making by clinicians interpreting the tests.

Choice Architecture: Effective Methods to Changing Prescriber Behavior

A successful method of changing physician behavior may lie in the deliberate presentation of choices within the decision-making environment. This concept, known as choice architecture, has successfully shaped behavior in diverse settings, and does not force practitioners into pre-specified decisions.²²

Clinical care often employs components of choice architecture (Table, bottom). For example, many ASPs compare clinicians’ antibiotic prescription habits with colleagues in the same clinical practice. This makes use of relative social ranking, a phenomenon where individuals, in times of uncertainty, can be influenced to make choices that compare more favorably to their peers. One study randomized primary care practices to email clinicians with their rates of antibiotic prescription for uncomplicated upper respiratory tract infections, comparing them to top-performing practitioners within the same practice.²³ Study investigators found that this intervention lowered the rate of antibiotic prescriptions by 5% compared to control practices. Notably, peer comparisons are most effective when comparison is made with top-performing practitioners rather than the mean performance, since the latter comparison can paradoxically worsen the behavior of top-performers by converging all practitioners towards the mean.²⁴ In explicitly comparing a clinician’s practice to their top-performing peers, it is possible that this strategy counteracts the general aversion to risk that motivates antibiotic overuse.

Many concepts from choice architecture can be incorporated as interventions into the electronic medical record. For example, clinicians are less likely to resort to cognitive biases when they are asked to articulate the rationale for their decisions, a concept known as accountable justification. Physicians who are required to provide free-text justifications for antibiotic indication in the electronic medical record use 7% fewer antibiotics compared to controls.²³ If implemented effectively, accountable justification may overcome action bias as clinicians would be prompted to consider whether their interventions are truly indicated in that specific clinical scenario.

Similarly, the inclination for decision-makers to rely on default practices can be harnessed to address antibiotic overuse. Clinicians may be more likely to choose narrower antibiotics for certain indications when those antibiotics are presented as the default options in electronic ordering systems. One study found that differential framing of broad and narrow-spectrum antibiotics in clinical vignettes reduced selections for broad-spectrum antibiotics for the same clinical indications by up to 11%.²⁵

An effective behavioral strategy that aligns future actions with current preferences is pre-commitment, where subjects publicly state their values or preferences before entering a choice environment. In one randomized study, outpatient primary care practices that included a poster in an examination room displaying a public commitment to judicious antibiotic prescriptions demonstrated a 20% reduction in antibiotic use for acute respiratory infections during influenza season.²⁶ This strategy may be especially effective in outpatient practices where patient satisfaction and patient expectations may be motivating antibiotic overuse.

Implementation of choice architecture into outpatient practice has its challenges. Electronic medical record-based interventions will need to be tailored to specific practice environments and patient populations. Perception of patient expectation and satisfaction may still drive antibiotic overuse, particularly in tele-medicine settings. Finally, outpatient practices will need to monitor patient safety to ensure that appropriate antibiotic use is not affected by changes in clinician behavior. Addressing these and future challenges to effectively achieving antibiotic stewardship goals will require ongoing attention paid by outpatient clinical leadership.

Conclusions

The clinical and economic burden of antibiotic resistance has reached a crisis point in the U.S. Clinician behavior is motivated by a complex intersection of incentives; risk aversion, action bias, and patient satisfaction are three major factors that contribute to antibiotic overuse. An improved understanding of the behavioral contributions to antibiotic overuse may lead us to strategies of behavior change that do not reduce clinician autonomy. These strategies involve restructuring the clinical decision-making environment for more judicious antibiotic use. As medical practices implement CDC-recommended elements of outpatient ASP, clinical leaders should incorporate evidence-based behavioral strategies into their program. In particular, attention should be paid to interventions that make use of relative social ranking, accountable justification, default practices, and pre-commitment.