Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

The effectiveness of behavioral economics-informed interventions on physician behavioral change: A systematic literature review

  • Sophie Y. Wang ,

    Roles Conceptualization, Data curation, Formal analysis, Methodology, Validation, Writing – original draft

    s.wang@optimedis.de

    Affiliations OptiMedis AG, Hamburg, Germany, Hamburg Center for Health Economics, Hamburg, Germany

  • Oliver Groene

    Roles Funding acquisition, Methodology, Resources, Supervision, Writing – review & editing

    Affiliations OptiMedis AG, Hamburg, Germany, London School of Hygiene & Tropical Medicine, London, England, United Kingdom

The effectiveness of behavioral economics-informed interventions on physician behavioral change: A systematic literature review

  • Sophie Y. Wang, 
  • Oliver Groene
PLOS
x

Abstract

Objective

Interventions informed by behavioral economics have the potential to change behaviors governed by underlying cognitive biases. This has been explored extensively for various use in healthcare including changing patient behavior and, more recently, physician behavior. We aimed to systematically review the literature on the use and effectiveness of behavioral economics-informed interventions in changing physician behavior.

Method

We searched Medline, Cochrane Library, EBM Reviews, PsychINFO, EconLit, Business Source Complete and Web of Science for peer-reviewed studies published in English that examined the effectiveness of behavioral economics-informed interventions on physician behavioral change. We included studies of physicians in all care settings and specialties and all types of objectively measured behavioral outcomes. The reporting quality of included studies was appraised using the Effective Public Health Practice Project tool.

Results

We screened 6,439 studies and included 17 studies that met our criteria, involving at least 9,834 physicians. The majority of studies were conducted in the United States, published between 2014 and 2018, and were in the patient safety and quality domain. Reporting quality of included studies included strong (n = 7), moderate (n = 6) and weak (n = 4). Changing default settings and providing social reference points were the most widely studied interventions, with these studies consistently demonstrating their effectiveness in changing physician behavior despite differences in implementation methods among studies. Prescribing behavior was most frequently targeted in included studies, with consistent effectiveness of studied interventions.

Conclusion

Changing default settings and providing social reference points were the most frequently studied and consistently effective interventions in changing physician behavior towards guideline-concordant practices. Additional theory-informed research is needed to better understand the mechanisms underlying the effectiveness of these interventions to guide implementation.

Introduction

The integration of research evidence into routine clinical practice to ensure safe and effective care for patients and reduce unnecessary expenditures has been a long-standing challenge [1,2]Studies in the United States, Netherlands, and Canada have shown that 30% to 40% of patients do not receive guideline-concordant care, and that more than 20% of care provided is unnecessary or potentially harmful [3,4]. Accordingly, given physicians’ role as key decision makers in healthcare, an increased focus on physician behavioral change has emerged [46].

Behavioral change is complex as it is influenced by a wide range of intrinsic and extrinsic factors such as an individual’s motivation, skills and knowledge as well as the social and physical environment [7]. Traditional behavioral change approaches to better align clinical practice with research evidence have mainly focused on improving access to information, such as guideline dissemination and education seminars. These methods are based on conventional economic theory and presume that physicians are perfectly rational decision-makers. However, research in behavioral economics, an evolving field rooted in economics and psychology, finds that humans have a “predictable” bounded rationality and rarely behave as the utility maximizers conventional economics theory would predict [8].

Physicians make numerous complex decisions on diagnostic and treatment plans daily, often with limited information and under time pressure [9]. When making high-volume of decisions under conditions of uncertainty, the decision-making process may be guided by environmental cues and heuristics [9,10]. While these strategies can be adaptive in clinical environments where high volume rapid decision-making is required, they can also be vulnerable to cognitive biases [11]. Systematic cognitive biases may affect the decisions of physicians thereby distorting probability estimation and impairing information synthesis [12]. Such biases may underlie why physicians sometimes continue to deliver care that robust evidence has shown to be of low value [13].

Interventions that aim to counteract the adverse effects of these cognitive biases are receiving growing attention, and have been applied in a variety of fields including energy, finance, taxation, and environmental studies [14,15]. In healthcare, behavioral economics-informed interventions have primarily been applied to changing patient behaviors, such as improving dietary choices [1618], increasing preventative health screening participation [19,20] and increasing vaccination rates [21]. There has also been an increasing interest in using such interventions for provider behavioral change, but a systematic synthesis of the empirical evidence is lacking. Therefore, we systematically evaluated the literature to date on the use and effectiveness of behavioral economics-informed interventions in changing physician behavior. We focused specifically on physicians given their professional autonomy and their key decision-making role in team-based care models.

Methods

We conducted and report the systematic review in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Protocols (PRISMA-P) guidelines [22]. We searched the International Prospective Register of Systematic Reviews (PROSPERO) prior to conducting this systematic literature review to ensure that no similar reviews have been conducted and registered the review protocol on PROSPERO in June 2019 (ID: CRD42019134956).

Definition of behavioral economics-informed intervention

A “nudge” is defined as an intervention that predictably changes human behavior without limiting free choice or changing financial incentives significantly [23]. In the context of changing physician behavior, an example might be setting generic medications as the default when trying to decrease branded prescribing by physicians with the assumption that physicians prescribe more branded drugs due to an underlying salience bias. In this systematic review, we chose the term “behavioral economics-informed intervention” to broaden the concept, and we define this as an intervention designed to change behavior within a decision context by counteracting an underlying cognitive bias [2426].

Search strategy

We first conducted a broad and exploratory search to identify key terms under the umbrella term of behavioral economics-informed intervention; this took place in November 2018. We then conducted a second search in Medline using a combination of MeSH terms and free text terms within each of the population and intervention categories in the title and abstract search field (Table 1). We adapted and optimized the syntax individually for other databases. We searched databases from both medical, economics, and business administration fields including Medline, Cochrane Library, EBM Reviews, PsychINFO, EconLit, Business Source Complete and Web of Science. Our search included articles from inception until September 2019 to ensure we captured the evolution of this growing field. We then augmented the search using references from included articles and relevant reviews.

We included the general terms such as “behavio?ral economic*”, “nudg*”, asymmetric paternalism” and “choice architect*” under intervention. To identify specific key terms to use for specifying the intervention, we consulted current literature on existing classification systems of behavioral economics-informed interventions. Of the various classification systems that have been proposed [2729], we found the taxonomy proposed by Münscher et al [26] to be appropriate for our review as it classifies interventions into mutually exclusive and exhaustive categories, thereby facilitating inter-study generalizability and knowledge accumulation. Thus, our search terms included for intervention were informed by the 9 subcategories proposed by Münscher et al [26] along with terms identified from our initial exploratory search (Table 1).

Eligibility criteria

We included primary research articles published in the English language that met the following inclusion criteria: (i) the study design was randomized and controlled or quasi-experimental; (ii) if study was conducted on a mixture of different healthcare providers, the majority were physicians; (iii) physicians of all specialties in all care settings; (iii) a behavioral outcome was objectively measured, as opposed to attitudes or preferences. Studies were excluded if they met the following criteria: (i) the full text could not be obtained; (ii) set in low- or middle-income countries; (iii) targetted patient behavioral change. Healthcare system challenges in low- and middle-income countries such as unstable governance structures, limited resources in service delivery, and limited access to healthcare differ substantially from high-income countries [30]. Thus, to ensure external validity we decided to restrict our review to focus on high-income countries as interventions may affect providers differentially in these different settings.

Data extraction and analysis

We imported all retrieved studies into Zotero reference management software and removed duplicates. One reviewer (SW) conducted a title screen on all retrieved articles based on the previously described criteria. Two reviewers (SW, NL) then independently screened the abstract and full text of the remaining studies and documented reasons for exclusion. Any disagreements at this stage were resolved by an independent third reviewer (OG).

We developed a standardized data extraction form in Microsoft Excel based on a Cochrane collaboration form and on reporting guidelines for randomized controlled trials in behavioral medicine [31,32]. Data we collected includes study context, research design, intervention and outcome. We piloted the form on 10% of included articles to ensure feasibility, completeness, and consistency of extraction by reviewers, and iteratively refined the form as needed.

We categorized the interventions of included studies using a taxonomy proposed by Münscher, Vetter, & Scheuerle (2016), with interventions classified into three broad categories and nine subcategories (Table 2). Interventions categorized as “Decision Information” focus on the presentation of information relevant to the decision and includes translating information, making information visible, and providing social reference points [26]. While “translating information” aims to change the format or presentation of information while retaining original content, “making information visible” aims to bring previously hidden information to the forefront [26]. Interventions categorized as “Decision Structure” alter the arrangement of the options or the decision-making format. This includes changing choice defaults, changing option-related efforts, changing range or composition of options, and changing option consequences [26]. An example of “changing option-related efforts” in a healthcare setting can be installing hand sanitizers at eye level or at entrances to decrease the effort required. An example of “changing the range or composition of options” in the healthcare setting is having different medication options presented to prescribers either spread horizontally or stacked vertically. Interventions categorized as “Decision Assistance” aims to help decision-makers follow through with their intentions, and includes providing reminders and facilitating commitment [26]. We categorized physician behaviors inductively by reviewing all measured outcomes in included studies.

thumbnail
Table 2. Categorization of interventions in included studies based on taxonomy developed by Menscher et al (2016).

https://doi.org/10.1371/journal.pone.0234149.t002

Two reviewers extracted and assessed the data independently, and discrepancies were resolved through discussion with a third reviewer or clarification from study authors. We developed a template to query authors via email in the case of missing data. If the study author did not respond to the query, the requested data were treated as missing data.

Included studies were highly heterogenous in the intervention type and implementation, the target population and setting, the outcome measures reported, and the measures used to determine intervention effect. Thus, we conducted a narrative synthesis by comparing results of studies within each intervention category instead of a meta-analysis.

Quality assessment

Included studies were assessed for methodological quality and risk of bias using the Effective Public Health Practice Project (EPHPP) tool [33]. This tool evaluates the overall quality of a study based on internal validity (study design, confounders, data collection methods) and external validity (sampling, actual participation). Each included study received a score of weak, moderate, or strong.

Results

Of the 6,439 citations our search returned, we screened 237 abstracts and identified 17 studies involving at least 9,834 healthcare providers that met our prespecified inclusion criteria (Fig 1).

Description of studies

Most of the studies (n = 14) were conducted in the United States, and in the United Kingdom. Studies were mainly experimental (n = 10), were set in multiple clinics or practices (n = 8), and most were published between 2014 and 2018 (Table 3.), illustrating the development and interest in this field in recent years. Sample size ranged from 14 (clinic-level study) [34] to 5,055 (physician-level study) [35]. Most studies were of general practitioners (n = 13), and in the patient safety and quality domain (n = 7). Most of the included studies were rated as moderate (n = 8) and strong (n = 7).

thumbnail
Table 3. Descriptive statistics of included papers investigating the effectiveness of behavioral economics informed interventions on physician behavioral change (n = 17).

https://doi.org/10.1371/journal.pone.0234149.t003

The 17 included studies investigated 10 distinct interventions belonging to six categories, including changing choice defaults (n = 7), providing social reference points (n = 7), and facilitating commitment (n = 3) (Table 4). Most interventions (n = 13) investigated the effectiveness of one intervention compared to usual care, while three studies simultaneously compared the effects of different interventions separately [3638] or in combination [39] on one behavioral outcome. Most studies found a significant positive effect on targetted behavior. Only three studies [37,38,40] identified an underlying theory that specifies the mechanism of action for each studied intervention and target behavior. The most common behavioral outcomes studied were prescribing (n = 12) and diagnostic test ordering (n = 4).

thumbnail
Table 4. Overview of interventions examined in included studies by intervention category and primary outcome category.

https://doi.org/10.1371/journal.pone.0234149.t004

Intervention outcomes

Change choice defaults.

Interventions in this category can be further subdivided into prompted choice which includes active choice and no-action defaults which includes default and order set (Table 4). Active choice is an intervention where decision makers are prompted to choose without a pre-selected option within the choice environment. Default is presented as a pre-selected option that freely allow decision makers to choose another option. Order sets are a prepackaged group of orders that apply to a specific diagnosis and is used in healthcare to standardize patient care and reduce errors. Studies in this category were of strong [39,40,48], moderate [46,49,50] and weak [47] quality and targeted behaviors from all categories (Table 5).

One US-based research group examined the effectiveness of active choice in three separate studies, and found positive impacts on increasing the rate of cancer screening, influenza vaccination orders, and on increasing guideline concordant statin prescribing [39,49,50]. Active choice was implemented as a prompt delivered prior to the patient’s clinical visit, alerting the decision-maker of the patient’s eligibility for the desired action—statin prescription, cancer screening, and influenza vaccination. Active choice intervention increased the rate of cancer screening tests [49] and flu vaccination [50] ordered. For statin prescribing, the positive impact was only noted when implemented in conjunction with peer-comparison intervention but not as a stand-alone intervention [39]. Researchers noted the importance of integrating the intervention into the clinical workflow as a contributing factor to effectiveness.

No-action default was examined in four studies—two in defaults and two in order set redesign, and all studies found these interventions to be effective in changing prescribing practices (Table 5). The studies varied in their implementation. One study examined the effectiveness of a lowered default opioid pill count set for postoperative analgesia and provided an education session for physicians as part of its implementation [46]. Another study examined the use of defaults to decrease medication costs by changing the electronic prescribing system to only initially display generics, instead of displaying them alongside the brand name drugs for which prescribers searched. Changes in defaults were associated with an increase in generics prescribing, though not for all medications studied (statins, proton pump inhibitors and beta blockers) and not among resident physicians [40]. Authors attributed the insignificant increase in generic proton pump inhibitor prescribing to its availability as an over the counter drug, so patients may already have a preference [40]. Additionally, authors postulated that the smaller effect of default setting on resident physician prescribing was due their higher baseline prescribing of generics [40]. The two studies investigating the use of order sets took place in busy clinical settings and found positive impacts. One study aimed to decrease unnecessary urine test orders and had an education session with physicians regarding the removal of the urine test option [48]. The other study aimed to improve guideline concordant care of mechanically ventilated patients in intensive care units by implementing evidence-based order sets with prepopulated medication orders [47].

Provide social reference point.

Providing a social reference point for decision-makers was the second most studied intervention (n = 7) (Table 4). Interventions in this group provided physicians with feedback on their performance relative to their peers on select quality indicators. We included in this group studies using interventions termed peer comparison, social comparison, and social norm feedback. All studies used a randomized controlled design and were of strong [38,39,43], moderate [34,44,52] and weak [35] quality (Table 5). Studies examined interventions’ impact on antibiotic prescribing behavior [37,43,44], statin prescribing [39], antipsychotic prescribing [35], controlled substances prescribing [38] and laboratory test ordering [52].

Four of the seven studies found peer comparison feedback to positively impact prescribing behavior [35,43,44], one of which was implemented in conjunction with active choice [39]. While Hallsworth et al found social comparison to have a significant positive impact on decreasing inappropriate antibiotic prescribing, it was not possible to disentangle the independent effects of social comparison, high profile messenger, and recommended actions in the single letter that was sent as the intervention [44].

Of the three studies that did not find this intervention to be effective, two were implemented as pilots [35,37] with larger scale randomized trials implemented later that did show intervention effectiveness [35,43]. Persell et al found peer comparison to be effective at reducing antibiotics prescribing for all acute respiratory illnesses but not specifically antibiotic-inappropriate diagnoses in the pilot study [37]. However, in a later and larger randomized trial, the same research group found peer comparison to be effective at reducing inappropriate antibiotic prescribing [43]. Authors attributed the null effects of peer comparison in the pilot to the small sample size and to potential contamination which were both addressed in the later study by increased sample and block randomization. In the other set of studies, Sacarny et al found peer comparison letters to increase the inappropriate prescription of controlled substances among high-prescribers, though the effect was not statistically significant [38]. In a subsequent study, peer comparison was found to be effective in reducing antipsychotic prescribing among high-prescribers [35]. Authors attributed the positive findings in the latter study to the increased intensity of the interventions (peer comparison letter sent with two subsequent reminders), refined target population using more recent data and stronger wording that triggered action [35]. The last study to find null effect of peer comparison was designed to decrease unnecessary laboratory testing [52]. As potential explanations for the null findings, authors noted a lack of engagement among study participants with the peer comparison dashboard, cross-contamination, and an undifferentiated target group [52]. Additionally, this study targetted interns and residents during the two week service block and utilized the entire medical team’s prescribing data rather than individual prescribing data [52].

Make information visible.

We identified two studies [41,42,53] that aimed to bring often invisible behavioral consequences to the attention of decision makers through feedback provision (Table 4). Study quality was rated as moderate [42], and weak [41] (Table 5). Providing feedback to change physician behavior is an extensively researched field [54]. Included studies examined the effectiveness of feedback on reducing opioid prescribing [42], and spending on antibiotics prescriptions [41].

Two studies found the intervention positively changed the target behavior [41,42]. Implementation and the feedback content differed drastically across studies. Doctor et al sent one letter to prescribers notifying them of their patients’ deaths due to opioid overdose and observed a significant decrease (9.7%, 95%CI 6.2 to 13.2) in milligram morphine equivalents filled three months after the intervention [42]. Lastly, Langley et al found that having the drug cost continuously displayed on the electronic prescribing system reduced the mean weekly expenditure of antibiotic prescriptions by £3.75 (95%CI –6.52 to –0.98) per patient at the start, however cost increased steadily by £0.10 per week (95%CI 0.02 to 0.18) afterwards [41] during the 12 week study period.

Change option-related effort and change range or composition of options.

We identified two studies (pilot and follow-up) conducted by one US-based research group that concurrently examined effectiveness of interventions grouped into these two categories aimed at decreasing antibiotic prescribing (Table 4). Accountable justification is an intervention whereby clinicians receive alerts via the electronic prescribing platform containing prescribing guidelines and prompting a justification for the antibiotic prescription. Suggested alternative is an intervention whereby alternative options such as non-antibiotic prescriptions, non-prescription medications, and patient-oriented educational materials are presented in the electronic prescribing system when an antibiotic is entered.

Both pilot [37] and follow up studies [43] concurrently examined the efficacy of accountable justification, suggested alternatives, and peer comparison in reducing antibiotic prescribing (Table 5). Participants in both studies received an education module on guideline-concordant antibiotic prescription prior to intervention. While the pilot study [37] did not find either of the interventions to significantly reduce antibiotic prescribing (OR = 0.98, 95%CI: 0.42–2.29), the follow-up, a higher powered study using same methods [43] did significantly reduce inappropriate antibiotic prescribing (difference in difference -7.0%, 95%CI: -9.1 to -2.9).

Facilitate commitment.

Facilitating commitment bridges the intention-behavior gap by counteracting self-control problems making individuals more likely to follow through with actions [26]. Effectiveness of commitment was examined in reducing antibiotic prescriptions and unnecessary diagnostic imaging in two studies, both of moderate-quality [34,51] (Table 5). One study explored the impact of a public commitment poster—written in accessible language, signed by the physicians, and posted in the examination room for three months. This was effective in reducing the rate of inappropriate antibiotic prescribing up to 12 months at follow-up. Another study found precommitment to have a small but statistically significant effect on reducing low-value imaging for lower back pain. In this study, precommitment was reinforced via a paper-based reminder provided to physicians at the point of clinical encounter. However, the effect was not sustained after 3 months [51].

Discussion

Our systematic review including seventeen articles investigating the effectiveness of behavioral-economics informed interventions on changing physician behavior found the two most studied interventions—changing default settings and providing social reference points, were consistently effective in behavioral change. A large volume of research exists examining the effectiveness of audit and feedback and reminder provision on physician behavior, and is reviewed extensively elsewhere [2,55]. However, few studies in these two intervention categories were included in our review as most were not used to address a behavioral change with an underlying cognitive bias. The most common behavioral outcome targeted was prescribing behavior, with most of these studies examining decreasing inappropriate antibiotic prescribing. To our knowledge, this is the first systematic review of the use and effectiveness of behavioral economics-informed interventions on physician behavior.

Changing default settings is a common approach, likely due to its relative ease of implementation and effectiveness in different contexts [56]. In our review, we included both prompted choice and no action default in this intervention category. We noted differences in implementation of active choice between included studies which may be associated with the intervention effectiveness. Specifically, we found that studies targeting both clinicians and medical assistants found active choice to be effective as a standalone intervention, but not when targeting only clinicians. This was further explored in a later study by the same US-based research group where they found active choice to have an even larger effect on increasing clinician ordering of cancer screening tests when targeting only medical assistant [57]. Researchers noted the importance of considering relieving physician burden and particularly alert fatigue among physicians when implementing active choice interventions.

All included studies investigating no-action default found a significant positive effect on behavioral change. This finding is consistent with that from a recent meta-analysis of 58 default intervention studies that found sizeable and robust effects [56]. However, there was substantial variation in the effect size across studies which authors noted may be due to an imperfect understanding of the mechanisms of action [56]. Indeed, only one [40] of four studies that examined the effectiveness of no-action defaults referenced the theorized mechanism of effect.

A theoretical framework proposed by Dinner et al outlines three possible mechanisms of how default interventions influence behavior—endowment, endorsement and effort [58]. It postulates that individuals are more likely to choose a preselected option because: (1) alternatives are evaluated in reference to the preselected option which is already endowed (2) preselected option is viewed as endorsement from the choice architect (3) less effort is exerted when a preselected option is chosen [58]. Literature suggests that effectiveness may increase when more mechanisms of action are activated [56]. Additionally, increasing evidence suggests that disclosure of a default may further increase the intervention effectiveness [59,60]. It is theorized that transparent disclosures can foster positive inferences such as trust and credibility of the default intervention implementor thereby increasing compliance, which aligns with the postulated endorsement mechanism. This may have contributed to the success of two included studies that briefed physicians on changes made to an existing order set [48] and implementation of a default change in opioid prescribing prior to implementation [46].

Social comparison was the second most frequently studied intervention in our review and generally resulted in significant positive effects on changing prescribing behavior [35,39,43,44]. We found notable differences in how this intervention was implemented across studies. When comparing studies that showed a positive significant effect [35,43,44] to those that did not find a significant effect [40,52], we did not find consistent explanatory characteristics such as mode of communication (email vs paper), frequency (one time vs recurring) and duration of intervention.

Providing a social reference point is theorized to change behavior by engaging individuals in “upward comparison”, with individuals evaluating their own performance against superior performers [61]. While all studies compared participants with better performance, the choice of comparator differed. Persell et al argued that comparing individuals with high performers sustains high performance; their study compares antibiotic prescribing rates of all participating physicians with rates of the top 10 percentile [37]. In contrast, Patel et al segmented participating physicians into three groups and provided different comparators [39]. While both interventions showed a significant positive effect on changing prescribing rates, research suggests that individuals engage in upward social comparison with the goal of self-improvement when they perceive the comparator to be similar to oneself [61]. This supports the comparison to similar performing others in the latter study [39].

Based on our findings, default interventions seem most effective when targeting behaviors contained within an electronic system such as prescription or test orders due to its simplicity in implementation and workflow integration. It may also be helpful to be transparent when implementing default interventions research has shown disclosure to engender trust [59,60]. For organizations engaged in a culture of feedback, we found incorporating social reference points for individuals to be effective in guiding individuals to evaluate their own performance and engage in upward comparison. Among included studies, we found studies that employed a moving target, utilized up-to-date data, and chose a comparator that target individual can identify with saw positive impact.

Limitations

Our study has important limitations. First, no consensus exists on which interventions are informed by behavioral economics and no common terminology is utilized in reference to the same concepts. While we were comprehensive in our search by incorporating not only terms from our chosen taxonomy but also known related concepts, it is conceivable that relevant studies may inadvertently be overlooked. In addition, this further limits our ability to compare and contrast our findings of intervention effectiveness with other fields. Second, as most studies (n = 13) did not include a follow-up period of more than 6 months, it was not possible for us to determine the intervention’s sustainability and longer-term effects. Thus, conclusions from this review should be interpreted as effectiveness for behavioral change and should not be confused with behavioral maintenance. Third, this review is heavily weighted by evidence from general practitioners, thus indicating a need for studies examining the applicability of these types of interventions among specialists due to differences in practice environments. Finally, although we have selected a taxonomy with mutually exclusive categories, categorizing interventions is difficult when intervention theory or mechanism of action is not described, as was the case for most of the included studies.

Recommendation for future research

Researchers should aim to design theory-informed intervention while considering the underlying barrier to behavioral change. This can not only increase the potential for intervention effectiveness, but also improve our understanding of the intervention’s mechanism of action. Further, outcomes should be examined over a longer follow-up period. This not only sheds light on intervention sustainability but also potential negative impacts. Finally, a common terminology for the interventions informed by behavioral economics should be adopted to improve evidence accumulation. Although not without its limitations, we recommend the use of Münscher et al’s taxonomy [26].

Conclusion

We find that changing default settings and providing social reference points were the most commonly studied behavioral economics-informed interventions employed to change physician behavior. Both classes of interventions were generally effective in positively changing physician behavior, particularly in prescribing.

Supporting information

S1 Table. Overview of intervention type and study results.

https://doi.org/10.1371/journal.pone.0234149.s001

(DOCX)

Acknowledgments

We thank William Fleischman and Tom Stargardt for their constructive criticism of the manuscript. We also thank Henricke Schwickert and Nicolas Larrain for their support in data extraction and categorization of interventions. And lastly, members of the HealthPros research network for their feedback on the project throughout.

References

  1. 1. Grimshaw JM, Eccles MP, Walker AE, Thomas RE. Changing physicians’ behavior: what works and thoughts on getting more things to work. J Contin Educ Health Prof. 2002;22: 237–243.
  2. 2. Johnson MJ, May CR. Promoting professional behaviour change in healthcare: what interventions work, and why? A theory-led overview of systematic reviews. BMJ Open. 2015;5: e008592. pmid:26423853
  3. 3. Bouck Z, Pendrith C, Chen X-K, Frood J, Reason B, Khan T, et al. Measuring the frequency and variation of unnecessary care across Canada. BMC Health Serv Res. 2019;19: 446. pmid:31269933
  4. 4. Grol R, Grimshaw J. From best evidence to best practice: effective implementation of change in patients’ care. The Lancet. 2003;362: 1225–1230.
  5. 5. Grimshaw J, Shirran L, Thomas R, Mowatt G, Fraser C, Bero L, et al. Changing Provider Behavior: An Overview of Systematic Reviews of Interventions. Medical care. 2001;30: II2–II45.
  6. 6. Grimshaw JM, Eccles MP, Walker AE, Thomas RE. Changing physiciansʼ behavior: What works and thoughts on getting more things to work: Journal of Continuing Education in the Health Professions. 2002;22: 237–243.
  7. 7. Michie S, Johnston M, Francis J, Hardeman W, Eccles M. From Theory to Intervention: Mapping Theoretically Derived Behavioural Determinants to Behaviour Change Techniques. Applied Psychology. 2008;57: 660–680.
  8. 8. Kahneman D. Maps of Bounded Rationality: Psychology for Behavioral Economics. American Economic Review. 2003;93: 1449–1475.
  9. 9. Blumenthal-Barby JS, Krieger H. Cognitive Biases and Heuristics in Medical Decision Making: A Critical Review Using a Systematic Search Strategy. Med Decis Making. 2015;35: 539–557. pmid:25145577
  10. 10. Scott IA, Soon J, Elshaug AG, Lindner R. Countering cognitive biases in minimising low value care. MEDICAL JOURNAL OF AUSTRALIA. 2017;206. pmid:28490292
  11. 11. Saposnik G, Redelmeier D, Ruff CC, Tobler PN. Cognitive biases associated with medical decisions: a systematic review. BMC Medical Informatics and Decision Making. 2016;16: 138. pmid:27809908
  12. 12. Dawson NV, Arkes HR. Systematic errors in medical decision making:: Judgment limitations. J Gen Intern Med. 1987;2: 183–187.
  13. 13. Scott IA, Soon J, Elshaug AG, Lindner R. Countering cognitive biases in minimising low value care. Medical Journal of Australia. 2017;206: 407–411. pmid:28490292
  14. 14. Halpern D, Sanders M. Nudging by government: Progress, impact, & lessons learned. Behavioral Science & Policy. 2016;2: 52–65.
  15. 15. OECD. Behavioural Insights and Public Policy: Lessons from Around the World. Paris: OECD; 2017. https://www.oecd-ilibrary.org/governance/behavioural-insights-and-public-policy_9789264270480-en
  16. 16. Bucher T, Collins C, Rollo ME, McCaffrey TA, De Vlieger N, Van der Bend D, et al. Nudging consumers towards healthier choices: a systematic review of positional influences on food choice. Br J Nutr. 2016;115: 2252–2263. pmid:27185414
  17. 17. Skov LR, Lourenço S, Hansen GL, Mikkelsen BE, Schofield C. Choice architecture as a means to change eating behaviour in self-service settings: a systematic review: Choice architecture in self-service eating settings. Obes Rev. 2013;14: 187–196.
  18. 18. Wilson AL, Buckley E, Buckley JD, Bogomolova S. Nudging healthier food and beverage choices through salience and priming. Evidence from a systematic review. Food Quality and Preference. 2016;51: 47–64.
  19. 19. Barnes AJ, Groskaufmanis L, Thomson NB. Promising Approaches From Behavioral Economics to Improve Patient Lung Cancer Screening Decisions. Journal of the American College of Radiology. 2016;13: 1566–1570. pmid:27888943
  20. 20. Purnell JQ, Thompson T, Kreuter MW, McBride TD. Behavioral Economics: “Nudging” Underserved Populations to Be Screened for Cancer. Prev Chronic Dis. 2015;12: 140346.
  21. 21. Korn L, Betsch C, Böhm R, Meier NW. Social nudging: The effect of social feedback interventions on vaccine uptake. Health Psychology. 2018;37: 1045–1054. pmid:30221969
  22. 22. Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gøtzsche PC, Ioannidis JPA, et al. The PRISMA Statement for Reporting Systematic Reviews and Meta-Analyses of Studies That Evaluate Health Care Interventions: Explanation and Elaboration. PLoS Med. 2009;6: e1000100. pmid:19621070
  23. 23. Thaler R, Sunstein C. Nudge: Improving decisions about health, wealth, and happiness. New Haven, CT, US: Yale University Press; 2008.
  24. 24. Hansen PG, Jespersen AM. Nudge and the Manipulation of Choice: A Framework for the Responsible Use of the Nudge Approach to Behaviour Change in Public Policy. Eur j risk regul. 2013;4: 3–28.
  25. 25. Hausman DM, Welch B. Debate: To Nudge or Not to Nudge*. Journal of Political Philosophy. 2010;18: 123–136.
  26. 26. Münscher R, Vetter M, Scheuerle T. A Review and Taxonomy of Choice Architecture Techniques: Choice Architecture Techniques. J Behav Dec Making. 2016;29: 511–524.
  27. 27. Dolan P, Hallsworth M, Halpern D, King D, Metcalfe R, Vlaev I. Influencing behaviour: The mindspace way. Journal of Economic Psychology. 2012;33: 264–277.
  28. 28. Johnson E, Shu S, Dellaert B, Fox C, Craig R, Goldstein D, et al. Beyond Nudges: Tools of a Choice Architecture. Marketing Letters. 2012;23: 487–504.
  29. 29. Sunstein CR. Nudging: A Very Short Guide. J Consum Policy. 2014;37: 583–588.
  30. 30. Mills A. Health Care Systems in Low- and Middle-Income Countries. N Engl J Med. 2014;370: 552–557. pmid:24499213
  31. 31. Davidson KW, Goldstein M, Kaplan RM, Kaufmann PG, Knatterud GL, Orleans CT, et al. Evidence-based behavioral medicine: What is it and how do we achieve it? ann behav med. 2003;26: 161–171. pmid:14644692
  32. 32. Effective Practice and Organisation of Care (EPOC). Data collection form. EPOC Resources for review authors. Norwegian Knowledge Centre for the Health Services;
  33. 33. Thomas BH, Ciliska D, Dobbins M, Micucci S. A Process for Systematically Reviewing the Literature: Providing the Research Evidence for Public Health Nursing Interventions. Worldviews on Evidence-Based Nursing. 2004;1: 176–184. pmid:17163895
  34. 34. Meeker D, Knight TK, Friedberg MW, Linder JA, Goldstein NJ, Fox CR, et al. Nudging Guideline-Concordant Antibiotic Prescribing: A Randomized Clinical Trial. JAMA Intern Med. 2014;174: 425. pmid:24474434
  35. 35. Sacarny A, Barnett ML, Le J, Tetkoski F, Yokum D, Agrawal S. Effect of Peer Comparison Letters for High-Volume Primary Care Prescribers of Quetiapine in Older and Disabled Adults: A Randomized Clinical Trial. JAMA Psychiatry. 2018;75: 1003. pmid:30073273
  36. 36. Linder JA, Meeker D, Fox CR, Friedberg MW, Persell SD, Goldstein NJ, et al. Effects of Behavioral Interventions on Inappropriate Antibiotic Prescribing in Primary Care 12 Months After Stopping Interventions. JAMA. 2017;318: 1391. pmid:29049577
  37. 37. Persell SD, Doctor JN, Friedberg MW, Meeker D, Friesema E, Cooper A, et al. Behavioral interventions to reduce inappropriate antibiotic prescribing: a randomized pilot trial. BMC Infect Dis. 2016;16: 373. pmid:27495917
  38. 38. Sacarny A, Yokum D, Finkelstein A, Agrawal S. Medicare Letters To Curb Overprescribing Of Controlled Substances Had No Detectable Effect On Providers. Health Affairs. 2016;35: 471–479. pmid:26953302
  39. 39. Patel MS, Kurtzman GW, Kannan S, Small DS, Morris A, Honeywell S, et al. Effect of an Automated Patient Dashboard Using Active Choice and Peer Comparison Performance Feedback to Physicians on Statin Prescribing: The PRESCRIBE Cluster Randomized Clinical Trial. JAMA Netw Open. 2018;1: e180818. pmid:30646039
  40. 40. Patel MS, Day S, Small DS, Howell JT, Lautenbach GL, Nierman EH, et al. Using Default Options Within the Electronic Health Record to Increase the Prescribing of Generic-Equivalent Medications: A Quasi-experimental Study. Ann Intern Med. 2014;161: S44. pmid:25402402
  41. 41. Langley T, Lacey J, Johnson A, Newman C, Subramanian D, Khare M, et al. An evaluation of a price transparency intervention for two commonly prescribed medications on total institutional expenditure: a prospective study. Future Healthc J. 2018;5: 198–202. pmid:31098566
  42. 42. Doctor JN, Nguyen A, Lev R, Lucas J, Knight T, Zhao H, et al. Opioid prescribing decreases after learning of a patient’s fatal overdose. Science. 2018;361: 588–590.
  43. 43. Meeker D, Linder JA, Fox CR, Friedberg MW, Persell SD, Goldstein NJ, et al. Effect of Behavioral Interventions on Inappropriate Antibiotic Prescribing Among Primary Care Practices: A Randomized Clinical Trial. JAMA. 2016;315: 562. pmid:26864410
  44. 44. Hallsworth M, Chadborn T, Sallis A, Sanders M, Berry D, Greaves F, et al. Provision of social norm feedback to high prescribers of antibiotics in general practice: a pragmatic national randomised controlled trial. The Lancet. 2016;387: 1743–1752. pmid:26898856
  45. 45. Ryskina K, Jessica Dine C, Gitelman Y, Leri D, Patel M, Kurtzman G, et al. Effect of Social Comparison Feedback on Laboratory Test Ordering for Hospitalized Patients: A Randomized Controlled Trial. J Gen Intern Med. 2018;33: 1639–1645. pmid:29790072
  46. 46. Chiu AS, Jean RA, Hoag JR, Freedman-Weiss M, Healy JM, Pei KY. Association of Lowering Default Pill Counts in Electronic Medical Record Systems With Postoperative Opioid Prescribing. JAMA Surg. 2018;153: 1012. pmid:30027289
  47. 47. Bourdeaux CP, Davies KJ, Thomas MJC, Bewley JS, Gould TH. Using ‘nudge’ principles for order set design: a before and after evaluation of an electronic prescribing template in critical care. BMJ Qual Saf. 2014;23: 382–388.
  48. 48. Munigala S, Jackups RR, Poirier RF, Liang SY, Wood H, Jafarzadeh SR, et al. Impact of order set design on urine culturing practices at an academic medical centre emergency department. BMJ Qual Saf. 2018;27: 587–592. pmid:29353243
  49. 49. Patel MS, Volpp KG, Small DS, Wynn C, Zhu J, Yang L, et al. Using active choice within the electronic health record to increase physician ordering and patient completion of high-value cancer screening tests. Healthcare. 2016;4: 340–345. pmid:28007228
  50. 50. Patel MS, Volpp KG, Small DS, Wynne C, Zhu J, Yang L, et al. Using Active Choice Within the Electronic Health Record to Increase Influenza Vaccination Rates. J GEN INTERN MED. 2017;32: 790–795. pmid:28337690
  51. 51. Kullgren JT, Krupka E, Schachter A, Linden A, Miller J, Acharya Y, et al. Precommitting to choose wisely about low-value services: a stepped wedge cluster randomised trial. BMJ Qual Saf. 2018;27: 355–364. pmid:29066616
  52. 52. Ryskina K, Jessica Dine C, Gitelman Y, Leri D, Patel M, Kurtzman G, et al. Effect of Social Comparison Feedback on Laboratory Test Ordering for Hospitalized Patients: A Randomized Controlled Trial. J GEN INTERN MED. 2018;33: 1639–1645. pmid:29790072
  53. 53. Eccles M, Steen N, Grimshaw J, Thomas L, McNamee P, Soutter J, et al. Effect of audit and feedback, and reminder messages on primary-care radiology referrals: a randomised trial. The Lancet. 2001;357: 1406–1409.
  54. 54. Ivers N, Jamtvedt G, Flottorp S, Young JM, Odgaard-Jensen J, French SD, et al. Audit and feedback: effects on professional practice and healthcare outcomes. Cochrane Effective Practice and Organisation of Care Group, editor. Cochrane Database of Systematic Reviews. 2012 [cited 30 Apr 2020]. pmid:22696318
  55. 55. Cunningham BP, Bakker CJ, Parikh HR, Johal H, Swiontkowski MF. Physician Behavior Change: A Systematic Review. Journal of Orthopaedic Trauma. 2019;33: S62–S72. pmid:31596788
  56. 56. Jachimowicz JM, Duncan S, Weber EU, Johnson EJ. When and why defaults influence decisions: a meta-analysis of default effects. Behav Public Policy. 2019;3: 159–186.
  57. 57. Hsiang EY, Mehta SJ, Small DS, Rareshide CAL, Snider CK, Day SC, et al. Association of an Active Choice Intervention in the Electronic Health Record Directed to Medical Assistants With Clinician Ordering and Patient Completion of Breast and Colorectal Cancer Screening Tests. JAMA Netw Open. 2019;2: e1915619. pmid:31730186
  58. 58. Dinner I, Johnson EJ, Goldstein DG, Liu K. Partitioning default effects: Why people choose not to choose. Journal of Experimental Psychology: Applied. 2011;17: 332–341. pmid:21707203
  59. 59. Bruns H, Kantorowicz-Reznichenko E, Klement K, Luistro Jonsson M, Rahali B. Can Nudges Be Transparent and Yet Effective? SSRN Journal. 2016 [cited 30 Apr 2020].
  60. 60. Paunov Y, Wänke M, Vogel T. Ethical defaults: which transparency components can increase the effectiveness of default nudges? Social Influence. 2019;14: 104–116.
  61. 61. Guyer JJ, Vaughan-Johnston TI. Social Comparisons (Upward and Downward). In: Zeigler-Hill V, Shackelford TK, editors. Encyclopedia of Personality and Individual Differences. Cham: Springer International Publishing; 2018. pp. 1–5.