Prescribing in healthcare: behavioral factors and management levers

The behaviors and decisions of public and private stakeholders play a crucial role in determining how effectively healthcare systems can fulfill their mission of improving health and well-being. These choices directly influence the appropriateness of care and indirectly affect the efficiency of resource use and the sustainability of healthcare operations. Recent studies on the management of medical decision making focusing on heuristics, cognitive biases, and nudging interventions converge on two conclusions. On the one hand, “medical errors occur in 1.7–6.5% of all hospital admissions, causing up to 100,000 unnecessary deaths each year, and […] in 2008, medical errors cost the USA $19.5 billion” (Saposnik et al., 2016, p. 2). On the other hand, “biases and heuristics have been under investigation in medical personnel” (Blumenthal-Barby and Krieger, 2015, p. 539), irrespective of the fact that choice architecture interventions successfully promote behavior change (Kahneman et al., 2021; Mertens et al., 2022). For instance, adequate training may be useful because variability in skill among physicians “can explain 44% of the variation in diagnostic decisions” (Kahneman et al., 2021, p. 256). Kahneman's research highlights systematic biases and judgment noise in decision-making (Kahneman, 2011; Kahneman et al., 2021), while Thaler and Sunstein show how managerial choice architecture and nudges can shape behavior without relying on strong incentives or prohibitions (Thaler and Sunstein, 2008, 2021). In parallel, Gigerenzer et al. emphasize the adaptive value of heuristics, demonstrating how simple decision rules often enable effective action under uncertainty rather than leading to error (Gigerenzer et al., 2022). Very recent empirical research that has been published in the last couple of years builds on these theoretical pillars to advance understanding of decision-making across diverse management domains (e.g. Belle et al., 2024a, b). Practical implications are so meaningful that, just as one notable example, the Penn Medicine Nudge Unit was launched in 2016 as the world's first behavioral design team embedded within a health system, to design, implement, evaluate, and disseminate scalable nudges to steer decision-making toward high-value care, improve patient outcomes, enhance public health, and increase health equity (Patel, 2021; Patel et al., 2018).

In this context, we systematically review the academic literature to understand the behavioral factors behind prescribing decisions, and seek to respond to the following research questions: how do behavioral factors – either external interventions or internal mechanisms – shape healthcare prescribing decisions? How have existing theoretical frameworks been applied to explain prescribing, and how can they inform the management of the decision-making environment? By analyzing existing research, this review aspires to provide actionable insights, bridge research streams in management and healthcare, and define a forward-looking agenda for practitioners and scholars alike. Our work aims to make an original contribution by addressing a significant gap in the extant literature that is both theoretical and methodological in nature. Although foundational behavioral theories have been developed or selectively contextualized within management research, second-level theoretical frameworks are frequently proposed without an explicit and systematic connection to these foundational theories. In addition, a comprehensive mapping of the state of the art in empirical research has not yet been undertaken. Prior reviews have focused on specific contexts or factors. For instance, Kasse et al. (2024) review antibiotic prescribing in primary health care, identifying physician-, patient-, and system-level drivers of variability. Courish et al. (2024) examine educational interventions to improve exercise counseling and prescription by physicians. Redvers et al. (2023) explore barriers and facilitators of patient–planetary health co-benefit prescribing, mapping them onto behavioral frameworks. While valuable, these studies adopt narrower lenses, being restricted either to specific types of prescriptions or to specific intervention modalities.

By contrast, our review adopts a broader scope, with no limitation on either the typology of prescribing or the behavioral factors considered, and with a focus on the micro-level of individual behavior. In doing so, we compile evidence from a substantially larger body of primary studies than prior reviews. In line with recent contributions in management studies (e.g. Cantarelli, 2023), we argue that a managerial perspective offers a crucial complement to medical and psychological sciences. This applies to both scholars and practitioners. On the scholarly side, public management research has expanded rapidly in applying behavioral lenses to decisions that are crucial for the economy and society (e.g. Battaglio et al., 2019; Belle et al., 2024b). More broadly, management research has increasingly tested hypotheses derived from decision sciences across managerial domains (e.g. Belle et al., 2024a; Guercini et al., 2024; Hess and Bacigalupo, 2011; Kenworthy and Balakrishnan, 2016; Thürmer et al., 2020), with high-stakes questions – such as the hidden costs of unwanted variability in decision-making – adding to the prominence initiated almost a decade ago in management work (Kahneman et al., 2016). On the practitioner side, nudge units worldwide embody this interdisciplinary logic, integrating clinicians, psychologists, managers, and informatics experts to improve prescribing practices. Managers play a pivotal role, for example, in designing training programs, conducting sludge audits, or selecting digital tools to support professionals (Patel, 2021; Patel et al., 2018).

The theoretical foundation of this research draws on the contributions of leading scholars in the behavioral sciences. Daniel Kahneman and his collaborators have extensively explored cognitive biases and heuristics, revealing systematic departures from rational judgment and shedding light on consistent patterns of error in decision-making (Kahneman, 2011), as well as various forms of judgment noise (Kahneman et al., 2021). The conceptualization of decision noise and its various forms represents the most recent development in management theory concerning professionals' judgments. Consequently, empirical research in this domain – both quantitative and qualitative – remains at a very early and largely nascent stage. Building on this work, Thaler and Sunstein (2008, 2021) helped bridge behavioral science with economics and management by showing how nudging – seemingly irrelevant modifications to the decision-making context – can steer individuals toward improved choices without limiting their autonomy or significantly altering economic incentives. Importantly, in situations where strong financial rewards or penalties are not feasible, nudges crafted by managers acting as choice architects offer a third, practical strategy for influencing behavior. Heuristics play an unparalleled role in ensuring progress for individuals and organizations. In those circumstances in which they help the decision-making process rather than leading to predictable deviations (Gigerenzer et al., 2022), heuristics guarantee smooth procedures and good decisions (Guercini, 2019; Guercini and Milanesi, 2020).

Under this foundational umbrella, the growing interest in the behavioral roots of decisions and practices has attracted contributions from diverse disciplines – such as management (e.g. Belle et al., 2024a, b; Guercini and Milanesi, 2020), marketing (e.g. Guercini, 2019), healthcare (e.g. Patel, 2021), public administration (Battaglio et al., 2019) and applied psychology (Gigerenzer, 2014) – which each offer complementary perspectives. For example, management studies may examine how organizational incentives influence prescribing habits, while applied psychology explores the cognitive biases underpinning clinical decision-making. These multidisciplinary models thus enable a more comprehensive understanding of prescribing behavior and how it can be steered toward more consistent and evidence-based practices. The concepts of behavioral interventions and behavioral mechanisms are related but distinct and can collectively be considered behavioral factors. Behavioral interventions are external tools or actions designed to influence behavior. For example, the adoption of clinical guidelines in healthcare can be seen as a behavioral intervention (e.g. Gandolfi et al., 2025; Kahneman et al., 2021). In their systematic review, Gandolfi et al. (2025) examine how clinical guidelines affect healthcare professionals' behavior and related outcomes. Analysis of 67 studies shows that guidelines improved behavior in fewer than half of cases, with prescriptions being the most studied area and doctors the primary focus. The findings underscore the complexity of interventions and the need for further research to design more effective strategies for implementing guidelines in practice. Behavioral mechanisms, in contrast, are the internal cognitive and mental processes through which individuals, including healthcare professionals, make decisions. These include social influences, beliefs, heuristics, and other mental processes that shape the choices people – professionals included – make (Gigerenzer et al., 2022; Kahneman, 2011; Kahneman et al., 2021; Thaler and Sunstein, 2008, 2021). Some interventions, such as audit and feedback, can operate through multiple underlying mechanisms, including social comparison, self-regulation, and feedback salience. Together, interventions and mechanisms represent the behavioral factors that drive decision-making.

The remainder of this section outlines a few non-exhaustive theoretical frameworks rooted in the broader field of behavioral sciences, which offer valuable insights for improving the management of prescribing decisions. Firstly, the Knowledge, Attitude, and Practice Framework posits that knowledge influences attitudes, which in turn shape practices or behaviors. In prescribing, this framework is used, for instance, to understand how familiarity with guidelines or clinical evidence impacts a physician's attitudes toward treatment options and their eventual prescribing decisions (Zarei et al., 2024). Then, the Capability, Opportunity, Motivation, and Behavior Model (Michie et al., 2011) integrates three core elements and provides a comprehensive lens for examining the multidimensional factors influencing prescribing practices (West and Michie, 2020): capability is conceived as the knowledge and skills required to perform a job or a task; opportunity encompasses external factors that enable or hinder the job or task, such as time constraints or access to resources; and motivation refers to the psychological processes driving behavior, including conscious decision-making and habitual patterns. Thirdly, the Theoretical Domains Framework is an amalgamation of multiple theories of behavior change (Cane et al., 2012), categorized into domains such as knowledge, skills, social influences, environmental context, and beliefs about consequences. It is used to identify barriers and enablers of behavior change, making it a tool for analyzing prescribing behaviors and designing targeted interventions (Atkins et al., 2017). Lastly, the Theory of Planned Behavior explains behavior as driven by three key components (Hassan et al., 2016): attitudes toward the behavior; subjective norms in the form of perceived social pressures, and perceived behavioral control, namely the belief in one's ability to perform the behavior. In the context of prescribing, it helps explore how individual beliefs, peer influence, and perceived barriers affect, for example, the likelihood of adhering to prescribing guidelines.

Other scholarship streams in the realm of improving prescription decisions by leveraging the micro-level perspective highlight how managers can design better decision-making environments. Towards this end, Patel et al. (2020a, b) propose the intervention ladder as a framework for leveraging nudges with different degrees of intrusiveness to improve healthcare decision-making. The nudges intervention ladder entails the following, from bottom to top: doing nothing, providing information, framing information, prompting implementation intentions, enabling choice, and guiding choice through default options. As a wise choice architect goes up the ladder, nudges tend to become more effective in achieving the desired conduct. Another effort to derive managerial toolkits informed by behavioral science evidence is that of Münscher and colleagues (2016), who formulated a taxonomy of choice architecture techniques targeting decision information, decision structure, and decision assistance.

The paper proposes a systematic literature review on the behavioral factors – interventions or mechanisms – behind prescribing decisions. In the Miner–Prospector Continuum (Breslin and Gatrell, 2023), our review lies between these two ends: it seeks to identify conceptual gaps in how behavioral factors have been applied to prescribing choices, categorize and problematize the literature, and transfer theories from behavioral science across different prescribing contexts to suggest a forward-looking research agenda for decision-making in management. This midpoint position aligns with the implementation of many steps in the guidelines for bibliometric–systematic literature reviews defined by Marzi et al. (2025). While performing a bibliometric analysis – which “spans five main methods: citation analysis, co-citation analysis, bibliographic coupling, co-author analysis, and co-occurrence (co-word) analysis” (Marzi et al., 2025, p. 83) – falls outside the scope of the current endeavor, our review aims to pinpoint the frequency of behavioral factors in prescribing by healthcare professionals and follows their steps 1 through 5 and step 10.a.I, i.e. the provision of a research agenda. Our systematic literature review qualifies as a structured domain-based effort (Paul and Criado, 2020), combining systematic procedures with a narrative framework to classify the literature and outline a forward-looking research agenda (Paul and Criado, 2020).

As to the search query definition and database selection, the search strategy was designed to capture studies that specifically address behavioral factors influencing prescription decisions. The search was restricted to the article title as far as prescribers and prescribing are concerned, whereas the search was open to all article sections as far as the behavioral factor is concerned. The search strategy was designed to capture a broad spectrum of research involving healthcare professionals such as physicians, clinicians, general practitioners, and non-medical prescribers. The primary focus was on studies examining aspects of prescribing, particularly within the context of behavioral science, nudging, choice architecture, biases, or noise. To balance the recency of the literature with the feasibility of synthesis, the search was limited to articles published within the past six years. This timeframe was considered sufficiently broad to capture a wide body of relevant research while remaining focused and analytically manageable. We acknowledge that earlier foundational studies have been instrumental in shaping the domain; however, our intention was not to revisit earlier frameworks but to assess how contemporary research has evolved. Studies published prior to this period were therefore not screened. Only peer-reviewed journal articles and reviews were retained, in line with the aim of synthesizing validated scientific evidence. Further refinements restricted the results to publications in English. The search string was as follows: TITLE ( (physician * OR clinician * OR doctor * OR practitioner * OR gp * OR hospitalist * OR prescriber * ) AND (prescri * ) ) AND (behavio * AND science OR nudge OR architecture OR bias * OR noise) AND PUBYEAR >2018 AND PUBYEAR <2026 AND (LIMIT-TO (SRCTYPE,“j”) ) AND (LIMIT-TO ( DOCTYPE, “ar”) OR LIMIT-TO (DOCTYPE,“re”) ) AND (LIMIT-TO ( LANGUAGE, “English”) ). The database selected for this review was Scopus, which provides the broadest interdisciplinary coverage of peer-reviewed scientific journals, a stable citation architecture, straightforward identification of articles and reviews in reputable outlets, and reproducibility of complex search queries (Taheri Moghadam et al., 2021). Cross-checks conducted on Web of Science and PubMed to test robustness yielded an almost perfect overlap, with Scopus retrieving a larger number of relevant studies. This suggests that relying on Scopus alone did not materially affect the comprehensiveness of the review. The literature search was initially conducted in June 2024 and then updated on September 17, 2025, leading to the identification of 777 primary studies. Moving to data cleaning and export, we retained studies that explicitly examined behavioral factors – that is, interventions or mechanisms – related to prescribing decisions by any type of healthcare professional. The selection was limited to primary studies focusing on prescriptions for human patients. We excluded study protocols or work addressing, for example, patient perspectives, general health recommendations, over-the-counter product advice, prescriptions for non-humans, or veterinarian prescribers. Two teams of authors independently screened half of the articles each, discussing any uncertainties regarding inclusion until reaching full agreement. This process resulted in the inclusion of 359 primary articles, marking the completion of data consolidation as defined in methodological checkpoint one by Marzi and colleagues (2025). Figure 1 summarizes this process following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.

Each included study was classified according to behavioral factor (intervention or mechanism) influencing prescribing, theoretical framework of reference (when explicitly mentioned in the abstract), research design, type of outcome (propensity versus actual prescribing behavior), type of prescriber, and geographic area. Primary studies were coded through content analysis of abstracts and full texts until information saturation was reached. Behavioral factors and research design allowed for multiple classifications and required more interpretative judgment from the authors. To ensure consistency ex-ante, a coding protocol with illustrative and non-exhaustive examples of behavioral factors and research designs was shared among the authors. For the remaining variables, exact information from the primary studies was recorded, following standard ex-post aggregation rather than interpretative coding. The coders conducted a pilot round on around fifty randomly selected studies to refine their catalog and ensure shared understanding. Following this, classification was carried out by two teams of authors who cross-checked each other's decisions. Any doubts or inconsistencies were resolved through multiple rounds of discussion until consensus was achieved. A third team of authors, not involved in the initial coding, verified the relevance and coherence of the codes applied. The resulting dataset provided the basis for identifying a forward-looking research agenda on decision-making in management, potentially extending beyond prescription choices.

Our search strategy led to the identification of 359 primary studies. Table 1 shows the distribution of articles by factor investigated to better understand professionals' prescribing decisions. Those factors can be conceived as either behavioral mechanisms or behavioral interventions. In particular, about 20% investigate multiple factors, about 16% of primary studies focus on training (e.g. Majumder et al., 2020), another 16% social influences (e.g. Stivers and Timmermans, 2021), about 11% guidelines (e.g. Wyse et al., 2019), about 8% beliefs (e.g. Manoharan et al., 2022), about 8% attitudes (e.g. Louie et al., 2019), and about 5% audit and feedback (e.g. Ivers et al., 2020). Among the primary studies that tackle more than one behavioral factor in prescribing, to be noted is the article of Liu et al. (2019) with the observation of forty-five drivers. Other studies explore more than one behavioral factor as prescribing is concerned (e.g. Anderson et al., 2024; Baazaoui et al., 2024; Chan et al., 2024). Then, the following behavioral factors are investigated in less than 5% of the primary studies, in decreasing share order: noise (e.g. Liu et al., 2021), incentives (e.g. Murayama and Marshall, 2024), fatigue (e.g. Christensen et al., 2022), accessibility (e.g. Alshehri et al., 2023), social norms (e.g. Lussier et al., 2025), stereotypes (e.g. McIntyre et al., 2021), target (e.g. Buchmueller et al., 2020), risk aversion (e.g. Szymczak et al., 2020), overconfidence (e.g. Menon et al., 2020), status quo (e.g. Dahlhausen et al., 2021), and visibility (Li et al., 2022). This list of behavioral factors seems to mirror the most recent debates within the behavioral sciences, whereby more or less rational and irrational levers coexist and do not exclude each other. Incentives, targets, audit and feedback can be encompassed within the range of more rational interventions and heuristics that are useful to make good decisions (Guercini and Milanesi, 2020; Thaler and Sunstein, 2008, 2021). Training can certainly be viewed as a tool to reduce irrationality and enhance rational decision-making. In fact, one possible source of decision variability in healthcare can be attributed to skills among medical doctors. Such differences in skills among prescribers “can explain 44% of the variation in diagnostic decisions” (Kahneman et al., 2021, p. 256). Other mechanisms, such as status quo and risk aversion, are native to the scholarship of cognitive biases and the predictably irrational aspect of decision making (Kahneman, 2011). Lately, fatigue and noise speak to the most recent advancements in decision-making sciences, namely the variability among subjects judging based on the same evidence (Kahneman et al., 2021).

For example, as to training, Majumder et al. (2020) made a review that aims to discuss the importance of the promotion of antimicrobial stewardship in medical and allied health professional curricula. Moving to social influences, Stivers and Timmermans (2021) investigate patient pressure and physicians' antibiotic prescribing. Then, Manoharan et al. (2022) investigate beliefs by collecting data by interview to fourteen physicians on cannabis prescription. For guidelines, Wyse et al. (2019) explored clinicians' strategies to identify and describe best practices for managing prescription opioid misuse. Moving to attitude, Louie et al. (2019) investigate the concerns of primary care physicians toward prescribing buprenorphine. With regards to audit and feedback, Ivers et al. (2020) focus on the impact of feedback interventions on physicians' antipsychotic prescribing in nursing homes. In the case of incentives, Murayama and Marshall (2024) find that industry non-research payments were significantly associated with increased prescriptions. Regarding stereotypes, McIntyre et al. (2021) explore the relationship between prescriber and patient gender on the prescription of benzodiazepines. Meanwhile, studies on the use of nudging interventions (Adusumalli et al., 2023; Ruetsch et al., 2021), electronic decision-support systems (Ferreira et al., 2023; Lim et al., 2020; Liu et al., 2019), and adherence to clinical guidelines demonstrate that small yet targeted changes in clinical workflows can substantially improve prescribing quality.

As to theoretical frameworks, Table 2 shows that only about 24% of primary studies mentioned one explicitly in the abstract. Within those, the Knowledge, Attitude, and Practice framework is the most used (n = 33) (Mittal et al., 2023), although being broad in scope and without specific hypotheses that can be empirically tested. The Theoretical Domains Framework (n = 17) (Acampora et al., 2023) and the Theory of Planned Behavior (n = 6) (Bhardwaj et al., 2023) follow. Mixed or hybrid frameworks (n = 11) signal attempts to integrate multiple perspectives (Chan et al., 2024). Interestingly, our synthesis finds that about 76% of primary studies lacked an explicit theoretical model.

As to the research design and typology of outcomes, observational questionnaires using actual prescription is the most frequent combination among our primary studies (Table 3). Considering methodologies alone, quantitative observational studies dominate, accounting for about 39% of the total. Qualitative inquiries are the second most common design, accounting for about 24%, followed by longitudinal studies (10%), systematic literature reviews (9%), administrative data (8%), and randomized controlled trials (6%). The remainder of primary studies (about 5%) employs a mixed-methods design (e.g. Baazaoui et al., 2024; Beaulieu-Jones et al., 2023; Huo et al., 2023; Metusela et al., 2023). As to the outcome variable, about 71% of studies reviewed employ actual prescription decisions (Lauffenburger et al., 2024), while the remaining 29% analyses respondents' propensity to prescribe (Lysen et al., 2024).

Lastly, Table 4 shows that European countries lead the research efforts, representing about 21% of the studies (e.g. Lansink et al., 2024), followed by North America with about 18% of cases (e.g. Soucy et al., 2024) and Asia with about 16% of the instances (e.g. Abedi et al., 2024). Five studies focus on multiple geographic areas (e.g. Choshen-Hillel et al., 2022), and the geographic information is not available for about 30% of primary studies.

This systematic review reveals several nuances of the behavioral factors – either mechanisms or interventions – shaping prescribing decisions, with implications for healthcare management. Training emerged as one of the most frequently investigated interventions, underscoring the need for continuous professional development programs that equip physicians with up-to-date knowledge and decision-making tools, in line with insights provided by Kahneman et al. (2021). The function of training that teaches the skills needed or the rules to follow to make certain decisions can represent an additional and distinct element compared to traditional nudges, falling within a type of ecological rationality and wisdom in the management of decision-making processes (Gigerenzer, 2014; Gigerenzer et al., 2022). At the same time, core cognitive biases that have long been demonstrated by behavioral science scholarship – such as overconfidence, stereotypes, loss aversion, and status quo (e.g. Battaglio et al., 2019; Belle et al., 2024a; Kahneman, 2011) – are less frequently studied in the context of prescribing decisions. The same applies to the latest behavioral science developments in the domain of noise and fatigue (Belle et al., 2024a; Kahneman et al., 2016). The effects of interventions such as training, guidelines implementation, incentives or audit and feedback mechanisms are easier to assess experimentally or through panel and pre–post designs. Training aims at enhancing professional competencies through various teaching methodologies. Guidelines are designed to standardize care and implement clinical advances (Gandolfi et al., 2025). Incentives are conventionally used to discourage undesirable and reward desirable behaviors. Audit and feedback mechanisms are based on command-and-control approaches. For instance, different training methodologies may vary in their effectiveness at improving competencies and reducing errors. Interestingly, behavioral sciences emphasize that sufficiently strong incentives or bans can effectively change behavior (Guercini and Milanesi, 2020; Thaler and Sunstein, 2008). In contrast, mechanisms such as social norms, beliefs and overconfidence, or stereotypes are far more difficult to study empirically in prescribing choices. Social norms reflect the desire to conform to others' behaviors or perceived expectations or, more broadly, emerge from responding to various forms of isomorphic pressure. Beliefs and overconfidence nurture confidence in one's judgments or abilities in a way that exceed actual accuracy. Stereotyping involves simplified assumptions about individuals based on group membership. These mechanisms are challenging to manipulate experimentally and difficult to detect in observational and administrative data. Similarly, identifying noise, or unwanted variability, and fatigue requires both high-quality administrative data linking professionals to prescribing decisions and managerial expertise to interpret patterns of outcomes and their effects. Certain models or frameworks – notably decision noise (Kahneman et al., 2016, 2021) – occupy a leading edge of innovation, marking emerging frontiers that the research community has only begun to explore.

The list of behavioral factors – either mechanism or intervention – is quite long, signaling the need to move beyond and deeper into the mental process that nurtures decisions (Gigerenzer et al., 2022; Kahneman, 2011). That said, our systematic review shows that the effectiveness of interventions such as training is more pronounced when combined with other interventions – such as guidelines and feedback – or leverage cognitive mechanisms influencing behavior (Sijbom et al., 2023). Social influences and social norms, then, highlight the pivotal role of peer comparisons, team dynamics, and patient-interactions in shaping professionals' decisions. For instance, prescribers may conform to perceived norms within their environment, whether through formal comparisons or perceived pressures, even when those deviate from evidence-based practices (e.g. Belle and Cantarelli, 2021; Wang et al., 2023). Cognitive biases, including beliefs, overconfidence and status quo effect, further extend the list of behavioral mechanisms in the management of prescriptions, although remaining a niche. Additionally, visibility and loss aversion were among the least studied, pointing to potential gaps in the literature. These underexplored areas may offer valuable opportunities for future research, especially given their relevance to high-stakes decision-making in healthcare. Taken together, these findings support the idea that professional experts are not always rational decision-makers and may nurture suboptimal outcomes.

Secondly, regarding the theories of reference in the reviewed studies, only a minority explicitly ground their work in frameworks. Those frameworks tend to be broad and make up a quite long list. In this respect, an improvement of management practices and processes aimed at reducing avoidable variation in prescriptions made by healthcare professionals might be represented by sticking to the core key elements of the behavioral science and theories that identifies testable hypotheses, featuring heuristics and cognitive biases (Kahneman, 2011), noise (Kahneman et al., 2021) and the efficacy of nudge-inspired interventions (Thaler and Sunstein, 2008, 2021).

Thirdly, with regard to the research designs and settings, observational surveys are the most used method, reflecting a focus on descriptive analysis rather than randomized designs. While these studies provide essential insights into existing behaviors, they often lack the causal chain required to identify effective interventions. Future research could address this gap by implementing interventions through random assignment, when technically feasible and ethically acceptable. Additionally, future work should leverage more mixed methods in search of results that triangulate (Mele and Belardinelli, 2019). The concentration of studies in Europe and North America raises questions about the generalizability of findings. Tackling the same research questions in other geographic contexts is certainly a necessity.

The results of this study should be considered, considering several limitations, many of which are typical of research of this kind. First, the review includes only studies published in English, which may have led to the exclusion of valuable contributions from non-English sources. The decision to focus on recent literature enhances the contemporary relevance of the findings but may come at the cost of overlooking earlier, foundational work. In addition, relying on peer-reviewed publications introduces a degree of publication bias: while peer review helps ensure methodological rigor, it may also limit the diversity of perspectives included. Furthermore, categorizing studies by behavioral factor and research design involved the most subjective judgment. Although classifications were made through a standard protocol and full consensus among coders, it is possible that different coders would interpret and label the material differently. The lack of information on healthcare system characteristics or regulatory environments in our dataset prevents us from discussing nuances in studying the same factor across different institutional contexts. This affects the generalizability of our findings, particularly for countries and geographical areas that are underrepresented or not represented in our work. Moreover, future work could complement our systematic review with bibliometric analyses that investigate network or co-occurrence to further explore conceptual linkages and thematic evolution in the domain of behavioral factors and prescribing. Lastly, the mostly descriptive nature of our work – qualifying as a structured domain-based effort (Paul and Criado, 2020) with a midpoint position on the Miner–Prospector Continuum (Breslin and Gatrell, 2023) and partial adoption of the guidelines by Marzi et al. (2025) – limit our ability to reduce the wide set of cognitive factors into a smaller number of generative ones.

The findings and discussion of our work offer tangible implications for the management of decision-making, providing actionable insights for improving prescribing practices as well as other types of professional judgments. First, managers should recognize the wide range of behavioral factors available to enhance decision quality. These tools span from interventions such as educational initiatives and training programs to structural or command-and-control mechanisms. Effective management requires a deep understanding of the cognitive factors that predictably influence decisions—from well-documented biases to more recent evidence on undesired variability, or decision noise. A superficial understanding of such concepts is insufficient for developing actionable management plans; rather, it is essential to achieve a comprehensive and operational understanding of their underlying principles, interactions, and implications to effectively translate theoretical knowledge into evidence-based management strategies. Notably, many successful interventions combine multiple behavioral factors. For example, training programs can move beyond information transfer to explicitly address bounded rationality by incorporating lectures and simulations on bias recognition and noise audits (Kahneman et al., 2021). Executive courses can also serve as testbeds for behavioral interventions and mechanisms, enabling organizations to pilot solutions and gather qualitative insights. Similarly, principles from choice architecture—such as the use of default options or decision prompts—can be embedded into workflows to encourage evidence-based actions while preserving professional autonomy. Second, managers aiming to improve decision quality should not overlook the powerful impact of removing barriers to sound judgment (Cantarelli, 2023; Thaler and Sunstein, 2021). Identifying and eliminating frictions, procedural obstacles, and sludge that hinder good decisions can be as valuable as introducing facilitators. Reducing administrative burdens is particularly relevant considering evidence showing that fatigue worsens judgments (Kahneman et al., 2021). Third, systematic experimentation should become a routine managerial practice. The methodological foundation of behavioral science – experimentation, particularly through randomized controlled trials – offers a rigorous means to assess the causal impact of interventions such as nudges, communication formats, or procedural reforms. Embedding experimentation within managerial units allows organizations to iteratively refine their decision-making processes. Triangulating these quantitative insights with qualitative feedback from professionals and well-recorded administrative data ensures that improvements are both effective and contextually appropriate.

Based on this review, Table 5 outlines a research agenda that is relevant for management scholars and can extend beyond prescribing. First, Kahneman and colleagues (2021) demonstrate that organizations are highly susceptible to both bias and noise, as both types of errors occur whenever judgments are made. This underscores the value of developing empirical frameworks to disentangle the relative contributions of bias versus noise across different decision domains. Research could also examine which components dominate in specific contexts, and how this informs theoretical models of managerial interventions. Third, advancing the measurement of decision noise remains a critical frontier. Following up on the work of Belle et al., 2024b, scholars can investigate the relative merits of different approaches to measure both between-subject variability (level noise) and within-subject variability (pattern noise).