Uncertainty in Decision-Making in Medicine: A Scoping Review and Thematic Analysis of Conceptual Models

Abstract

Purpose

The practice of medicine is rarely straightforward. Data used to facilitate medical decision-making may be conflicting, ambiguous, or scarce, and providing optimal care requires balancing clinicians’ expertise and available evidence with patients’ preferences. To explore uncertainty in decision-making across disciplines, the authors performed a scoping review and thematic analysis of the literature to formulate a model describing the decision-making process in medicine under uncertain conditions.

Method

In 2016, the authors performed a comprehensive search of key databases using a combination of keywords and controlled vocabulary. They identified and reviewed 3,398 records. After applying their inclusion and exclusion criteria to the titles and abstracts then full texts, 19 articles were selected. The authors applied a qualitative thematic analysis to these articles, using codes to extract themes related to uncertainty in decision-making.

Results

The 19 articles spanned 6 fields of study and 5 disciplines within the health sciences. The thematic analysis revealed 6 main themes: recognition of uncertainty, classification of uncertainty, stakeholder perspectives, knowledge acquisition, decision-making approach, and evaluation of the decision-making process.

Conclusions

Based on the themes that emerged from their thematic analysis of the literature characterizing the effects of uncertainty and ambiguity on the decision-making process, the authors developed a framework depicting the interplay between these themes with a visual representation of the decision-making process under uncertain conditions. Future research includes further development and validation of this framework to inform medical school curricula.

The core predicament of medicine - the thing that makes being a patient so wrenching, being a doctor so difficult, and being a part of society that pays the bills they run up so vexing - is uncertainty … Medicine’s ground state is uncertainty. And wisdom - for both the patients and doctors - is defined by how one copes with it.

— Atul Gawande¹

The practice of medicine is rarely straightforward. Medical decision-making is fraught with the potential for bias, and the optimal evaluation and management plan for a patient may vary by individual or system factors, such as the patient’s degree of pain and suffering as well as the cost of the care.² Data used to facilitate decision-making may be conflicting, ambiguous, or scarce, and providing optimal care requires balancing clinicians’ expertise and the available evidence with patients’ preferences. Collectively, researchers attribute these challenges to the uncertainty inherent in medicine.³

Uncertainty affects the clinical decision-making process, which has been described as a complex processing of knowledge and experience, using both critical thinking skills and intuition, while recognizing the inherent bias that can exist in medical decisions.^2,4,5 Sources of uncertainty include the complexity of clinical information, the probability of particular outcomes, and individual clinician characteristics, such as tolerance for ambiguity⁶ or an individual’s ability to cope with complexity, risk, and uncertainty.⁷ A low tolerance for ambiguity has been associated with increased resource utilization and higher rates of burnout in practicing physicians. Furthermore, it has been linked to poorer attitudes toward the underserved, decreased leadership capacity, psychological distress, and subspecialty choice in medical students.^8–11

In medical education, managing uncertainty and tolerance for ambiguity are often taught indirectly through observation, role modeling, and informal curricular experiences.^3,12 While these informal experiences may be beneficial, the medical education community now recognizes the need to formally address uncertainty as part of training.¹³ Many articles have been devoted to the teaching of clinical decision-making and reasoning as well as shared decision-making.^5,14–16 However, the existing body of literature has not described models for connecting concepts of uncertainty directly to the clinical or shared decision-making process. Thus, it has been a challenge to develop formal curricula teaching these skills.

The purpose of this study was to explore uncertainty in decision-making across disciplines to formulate a model describing the role of uncertainty in medicine at any point in the decision-making process (i.e., during diagnosis, formulating treatment options). Specifically, we sought to describe how the literature characterizes the effects of uncertainty and ambiguity on the decision-making process by performing a scoping review and thematic analysis of relevant conceptual frameworks, models, and taxonomies. With so many documented connections between clinicians’ tolerance for ambiguity and their behaviors and attitudes, we chose to focus on ambiguity as the source of uncertainty in medicine.¹⁷ We believed that, by systematically exploring and describing the literature, we would better understand the role of uncertainty in decision-making and be able to develop a novel, comprehensive framework for managing uncertainty to inform future curriculum development efforts.

Method

We conducted a scoping review following the methodology outlined by Arksey and O’Malley.¹⁸ A scoping review is a type of literature review that maps the extent of research on a topic within a specific field(s). In comparison to systematic reviews, scoping reviews allow for a broader exploration of a topic area. We decided that this methodology was ideal for characterizing the scope of the existing research related to uncertainty in the decision-making process.

Our strategy for identifying and selecting articles involved a three-phase process: (1) searching the relevant databases to identify candidate articles for inclusion, (2) reviewing the titles and abstracts of the candidate articles identified during the databases search, and (3) reviewing the full texts of each article that advanced past the title and abstract screening. Details of this process and the outcomes of each phase are described below and in Figure 1.

Figure 1.

Article search and selection process for a scoping review of the literature on conceptual models of uncertainty in decision-making, 2016.

Study identification

One member of the research team (J.W.C.), a research librarian, conducted a search of the following databases: OVID/Medline, PsycINFO, Cumulative Index to Nursing and Allied Health Literature (CINAHL), ERIC, and ABI Inform, for English language publications from inception of the database until November 4, 2016. The search used a combination of keywords and controlled vocabulary for the concepts of “decision-making,” “uncertainty,” and “conceptual model” and was piloted in Medline and subsequently adapted to each database. A total of 3,398 records were identified through the databases search and exported into RefWorks (ProQuest), a reference management software. After we removed duplicates, 3,203 publications remained.

Study selection

We screened the 3,203 publications for inclusion by applying two criteria to the title and abstract of each article: The article must (1) discuss decision-making under ambiguous or uncertain conditions and (2) present an original framework or the expansion of an existing framework into a new setting, regardless of the focus of the model (i.e., prescriptive, conceptual, naturalistic, diagnostic, etc.). Members of the research team (M.A.H., D.DG., J.W.C.) piloted this screening process with 100 articles to build a common understanding of how to apply the inclusion criteria before dividing the remaining articles into 3 sets of approximately 1,000 manuscripts each.

One primary reviewer read each set of articles to determine if they met the 2 inclusion criteria; the other team members acted as secondary reviewers. Articles were labeled as meeting the criteria, not meeting the criteria, or possibly meeting the criteria. Each primary reviewer presented the articles from her or his set with preliminary inclusion recommendations to the 2 secondary reviewers. Consensus regarding inclusion or exclusion was determined for each article. If there was disagreement, we used majority voting to determine whether to pass each article on for full-text review. At the conclusion of this phase, 89 articles were selected for full-text review. An additional 29 articles were identified from the reference lists of these articles, resulting in a total of 118 articles advancing to the next phase of review.

We next conducted a second review by applying the same inclusion criteria to the full texts of these 118 articles. We again divided the articles into 3 sets. Each set was reviewed by one primary reviewer who determined whether each article met one or both inclusion criteria. The articles were then presented to the 2 secondary reviewers to reach consensus on whether the article met the inclusion criteria and fit within the scope of our research question. After the full-text review, we excluded 99 articles for one or more reasons, including involving only a mathematical or statistical model, being solely concerned with diagnostic uncertainty, or focusing on economic forecasting in uncertain conditions. We selected the remaining 19 articles for inclusion in our scoping review.

Data extraction

After the full-text review, we extracted the following information from the included articles using a standardized data collection form: citation information, objective of the article, field of study, type of contribution (model, framework, or taxonomy), description of a figure if applicable, and implications of the findings relevant to our research question. For the empirical studies, we also collected the following information: hypothesis, methodology, study population, and results. We chose not to evaluate the methodological quality of individual studies as the aim of our review was to describe the extent of the literature in this area rather than comment on its rigor.^18,19 We used a spreadsheet to organize the summary information describing each study.

Data collation and analysis

After compiling this initial description of the included articles, we conducted a thematic analysis using grounded theory.²⁰ Thematic analysis, a foundational tool for conducting qualitative research, allowed us to take a flexible approach in developing a rich, detailed interpretation of the included articles.²¹ We adhered to the 6-phase process delineated by Braun and Clarke in our methodology.²² This approach has been recommended over weighting individual articles for their contributions to allow for a more comprehensive representation of the topic.²³ Three authors (M.A.H., D.DG., J.W.C.) reread the articles and each assigned initial codes, extracting representative text to support the coding. We then reviewed all codes as a group and came to consensus on differences to achieve a single, final set of codes. We collated these codes to develop applicable themes and identify relevant subthemes. Using this information, we created a single comprehensive definition for each theme.

Results

The primary characteristics and contributions of the 19 articles identified during our scoping review and included in our thematic analysis are summarized in Table 1.^24–42 These articles spanned 6 fields of study and included 5 disciplines. Approximately half (9/19, 47%) of the articles came from the health sciences literature. The vast majority (16/19, 84%) were published in the most recent 2 decades. More than half (10/19, 52%) provided conceptual frameworks, while 5 (26%) provided either quantitative or qualitative data, all from interventional or observational studies. Table 2 provides details about these 5 studies.

Table 1.

Summary of 19 Articles Included in a Scoping Review of the Literature on Conceptual Models of Uncertainty in Decision-Making, 2016

First author (year)	Field	Contribution	Description
Boschetti (2011)24	Management	Taxonomy	The author proposes a visual of the decision-making process that classifies the types of knowledge and uncertainty created throughout the process, modeling how the (1) amount of uncertainty, (2) level of awareness of uncertainty, and (3) multiple interpretations or perceptions of the same problem all affect decision-making in complex settings.
Boyd (2008)25	Health sciences: dentistry	Taxonomy	The author adapts King and Kitchener’s reflective stages48 to explore and define the stages of reflective thinking and judgement in dental students through coded journals and interviews. The author observed a growth in the development of reflective thinking and judgement over time (see Table 2).
Brand (2006)26	Psychology	Conceptual framework	The authors present a visual that builds on previous frameworks to further delineate how cognitive functions differ when making decisions under risk (cognitive strategies and emotional biases) as compared to under ambiguity (emotional biases).
Brugnach (2008)27	Management	Taxonomy	The authors propose a relational approach to uncertainty analysis that considers the multiple and sometimes conflicting framing of problems in the context of the socio-technical-environmental system in which uncertainty is identified. They also suggest implications for dealing with uncertainty in water management, describing specific strategies for dealing with unpredictability, incomplete knowledge, and multiple knowledge frames.
Charles (1997)28	Health sciences: medicine	Taxonomy	The authors describe the shared decision-making process in terms of 4 key characteristics: (1) physician and patient involvement, (2) information sharing from both parties, (3) both parties take steps to come to consensus, and (4) an agreement is reached on treatment.
Corliss (1995)29	Health sciences: medicine	Conceptual framework	The author presents a visual of the clinical decision-making process with 4 major components: (1) clinician’s knowledge base, (2) taking a case history, (3) clinical testing, (4) managing the patient, and 2 central elements for the various feedback loops (information and a decision point). The model uses an overall hypothetico-deductive logic, predominantly involving hypothesis formation and testing, that is ultimately dependent on a “treatment threshold” to make a clinical decision.
Cristancho (2016)30	Health sciences: medicine	Conceptual framework	The authors present a visual representation of intraoperative decision-making with a reconciliation cycle at the core of an iterative process of gaining and transforming information in real time and refining it by expert use (see Table 2).
Dinur (2011)31	Management	Taxonomy	The author defines the concepts of common and “uncommon” sense with the errors that can happen in decision-making. Specifically, the author proposes a set of “O errors” which occur during low uncertainty when using common sense to make decisions and “M errors” which occur when low certainty (high uncertainty) exists but mechanistic approaches are taken.
Falzer (2009)32	Health sciences: medicine	Conceptual framework	The authors propose a 3-step model of evidence-based decision-making based on contextual strategies they observed trainees using in various case vignettes (see Table 2). The 3 steps are to (1) recognize the decision scenario, (2) apply a contextual strategy if available, and (3) apply a more complex strategy if needed.
Han (2013)33	Health sciences: medicine	Taxonomy	The author synthesizes available research to propose a visual taxonomy, describing uncertainty in terms of the (1) source (probability, ambiguity, and complexity), (2) issue (scientific, practical, and personal), and (3) locus (patient versus clinician).
McCullough (2013)34	Ethics	Taxonomy	The author presents the professional medical ethics model of decision making, based on a comprehensive exploration of ethics concepts and terms. The model is meant to improve the quality of medical decisions by reducing the uncontrolled variation that can result from the non-deliberative decisions of patients and clinicians in response to conditions of clinical uncertainty.
McKenna (2005)35	Management	Conceptual framework	The authors present a comprehensive visual that illustrates the decision-making process both as a cycle of action, discovery, and choice, and as a cycle of craft, art, and science.
McKenzie (2009)36	Management	Conceptual framework	The authors present a visual of the decision-making process in uncertainty-ambiguity-contradictory settings, describing how conventional and unconventional thinking capacities play a role in this process and refining it using expert review (see Table 2).
Menard (2012)37	Health sciences: medicine	Conceptual framework	The authors present a visual model of the collaborative decision-making process for managing uncertainty in oncology treatment.
Mumford (2008)38	Ethics	Educational model / curriculum	The authors developed an ethics training course based on a proposed sense-making model of ethical decision-making as represented in a visual diagram. They developed a curriculum based on this model and, when assessed in doctoral students, results showed that gains in ethical decision-making were maintained over time (see Table 2).
Politi (2011)39	Health sciences: medicine	Conceptual framework	The authors present a visual representation of a model incorporating both the physician’s and patient’s cognitive and communicative capabilities to manage uncertainty and achieve a “shared mind” in medical decision-making.
Thompson (2001)40	Health sciences: nursing	Conceptual framework	The authors build on Katz’s model of uncertainty49 with Baumann and colleagues’ concepts of micro and macro (un)certainty50 to propose a conceptual framework for dealing with uncertainty in nursing practice. They describe 3 strategic approaches involving rationality, bounded rationality, and intuition.
Tversky (1974)41	Psychology	Theory	The authors describe 3 heuristics in the decision-making process under conditions of uncertainty: representativeness, availability, and adjustment and anchoring. They expand on the errors and bias resulting from these heuristics.
Whitney (2003)42	Psychology	Conceptual framework	The author presents a visual representation of a decision plane used to map medical decisions based on predefined levels of importance and uncertainty. The model identifies zones in which patients or physicians have decision priority, where priorities are shared, and where there is room for conflict based on decision characteristics.

Table 2.

Summary of 5 Empirical Studies Included in a Scoping Review of the Literature on Conceptual Models of Uncertainty in Decision-Making, 2016

First author (year)	Methodology	Measures	Outcomes
Boyd (2008)25	Qualitative: Coded interviews and journaling over a year of the clinical curriculum	Use of King and Kitchener’s reflective judgement model of intellectual development48	Demonstrated growth in reflective judgement
Cristancho (2016)30	Qualitative: Semi-structured interviews using a constructivist grounded theory approach with a constant comparison analysis of follow-up interviews	Perceived challenges in specific cases	Updated model informed by experienced surgeons’ perspectives using the reconciliation cycle in challenging intraoperative decisions
Falzer (2009)32	Qualitative: Within-subjects design in which vignettes were constructed and participants (21 residents from a psychiatry program) were asked to indicate whether they endorsed specific recommendations in each vignette	Vignettes developed by manipulating contingent and antecedent variables	Participants used consistent decision strategies in responding to clearly positive or negative forecasts. When patient forecasts were more ambiguous, the strategies for decision-making became more complex and inconsistent.
McKenzie (2009)36	Qualitative: Interviews with 6 participants identified as experts in the field	Expert review of the model	Face validation of proposed model
Mumford (2008)38	Quantitative: A standard pre-post study design using a previously validated assessment tool for measuring competency in ethical decision-making before and after the curriculum	Ethical decision-making test scores and participant reactions to the training course	Significant increase in test scores on the post-test that was maintained at 6 months for those followed up; high ratings for participant satisfaction

We conducted our thematic analysis in an open-ended manner to capture all elements of decision-making. While other authors have described the focus of decision-making at the individual, dyadic, or team levels, we purposefully sought to review the literature outside of these individual paradigms to formulate a comprehensive analysis of the decision-making process.^5,14,43 Our analysis revealed 6 primary themes or aspects of uncertainty in decision-making: the recognition of uncertainty in a decision, the classification of uncertainty, the consideration of stakeholder perspectives, the role of knowledge acquisition, the choice of a particular decision-making approach, and the ongoing evaluation of the decision-making process. We describe each of these themes and the corresponding subthemes below and in Table 3.

Table 3.

Results of a Thematic Analysis of 19 Articles Included in a Scoping Review of the Literature on Conceptual Models of Uncertainty in Decision-Making, 2016

Themes and subthemes	Definition
Recognition of uncertainty	Identifying the potential for uncertainty in a decision and recognizing that it must be addressed in the decision-making process
Classification of uncertainty	Categorizing the uncertainty in a decision by type and magnitude to gain a better understanding of how to address it
Source	Categorizing uncertainty by source or origin (i.e., complexity versus ambiguity)
Magnitude	Categorizing uncertainty by level or gravity (i.e., high versus low)
Stakeholder perspectives	Recognizing all those affected by a decision and the relevance of various viewpoints in making the decision
Recognition	Being aware that multiple perspectives exist
Gathering	Collecting all relevant viewpoints
Negotiation	Deliberating between viewpoints to balance any potential conflict
Knowledge acquisition	Gaining knowledge through experience and education or identifying the information or evidence needed to make the decision
Decision-making approach	Identifying the concepts, strategies, methods, and tools used to make the decision
Intuitive	Drawing on one’s ability to identify patterns based on knowledge developed from past experiences and expertise in the domain
Protocol-driven	Relying on a structured set of rules or decision aids to make the decision
Team-based	Incorporating multidisciplinary perspectives into making the decision
Shared	Involving those affected in making the decision
Evaluation of the decision-making process	Appraising the decision-making process at each step
Assessment	Evaluating the quality of information gained through recognition, classification, consideration of perspectives, and knowledge acquisition
Synthesis	Considering all available information, perspectives, and potential outcomes in context to make the decision
Reflection	Considering the consequences and probabilities of various decision options (both individually and with stakeholders)

Recognition of uncertainty

One overarching theme we found was the discussion of how clinicians recognize uncertainty. Articles that discussed this theme focused on awareness as a spectrum, the pitfalls if uncertainty is not recognized, and the influence of recognition on the decision-making process. Boschetti proposed a spectrum of awareness based on 4 quadrants created by 2 axes: uncertain to certain and unaware to aware. For the decision maker, understanding whether you fall into the certain and aware quadrant of “known knowns” or the uncertain and unaware quadrant of “unknown unknowns” can have drastic implications for your ability to make a sound decision.²⁴ Other authors highlighted the pitfalls of not recognizing uncertainty, including failing to identify the correct problem and asking the wrong question.³⁵ Finally, other authors addressed the effects of recognizing uncertainty on decisions, including using it to initiate information gathering or as an impetus to select a specific decision approach.^27,30,32

Classification of uncertainty

The second theme we identified was the classification of the uncertainty in a decision. Han synthesized the literature to describe uncertainty in medicine in terms of source, issue (i.e., practical, scientific, and personal), and locus (i.e., patient versus clinician). He further distinguished between probability, complexity, and ambiguity as the sources of uncertainty surrounding the potential outcomes of a treatment decision. For example, when the probability of success for a treatment is neither very high nor very low, it creates uncertainty in the outcome. Complexity can influence uncertainty when there are multiple different factors affecting the outcome of a treatment decision, such as comorbidities or confounding patient demographics. Lastly, ambiguity arises from a lack of evidence or from disagreement among experts in the outcome of a clinical decision.³³ Additional articles discussed the classification of uncertainty not by source but by level or magnitude. These authors suggested that the level of uncertainty in a situation should determine which party (i.e., clinician or patient) assumes the role of decision maker.^31,42

Stakeholder perspectives

Many of the articles we included in our review addressed the significance of considering various stakeholder perspectives in the decision-making process. The term stakeholder refers to the individuals and/or groups of individuals (e.g., patients, physicians, health systems, communities, etc.) affected by the decision-making process. A clinician’s ability to consider the opinions of key stakeholders can affect her or his ability to manage uncertainty. McKenzie and colleagues described this process as the ability to “acknowledge and hold contradictions until one finds a position that transcends the tensions. Transcendence produces a solution that resonates across the meaning systems of a broad constituency of stakeholders.”³⁶

The steps to consider the viewpoints of key stakeholders included recognizing the various possible viewpoints, gathering such viewpoints, and negotiating between them to determine the appropriate course of action. Recognizing that different stakeholder perspectives exist at all is an important first step. As described by Brugnach and colleagues, differing perspectives on a decision may exist for a number of reasons, such as different interpretations of the same information or conflicting views regarding how a specific situation should be managed.²⁷ After recognizing that other perspectives exist, clinicians must communicate those viewpoints to gather perspective and determine if there is conflict. Finally, clinicians must negotiate between the various stakeholders to come to a consensus. One author described this process as straightforward, incorporating patient preferences into a decision.³⁴ Several other authors discussed clinicians establishing a shared mind set regarding the decision so they could incorporate patient preferences into future action plans or goal setting and as a prerequisite specific to shared decision-making.^28,39 Menard and colleagues described a complex, cyclical process consisting of repeated interactions between patients and physicians to share new information and continually update perspectives until a decision is reached.³⁷

Knowledge acquisition

Another theme that emerged focused on the knowledge clinicians gained from and about the context, people, and problem to make a decision. Several authors covered this theme, describing it as the gaining of knowledge through experience and education. They further suggested that finding this evidence is central to arriving at a decision. For example, knowledge acquisition was discussed in terms of how well individuals are able to search for information or identify incomplete knowledge in their problem space.⁴¹ In addition, how individuals both understand and misunderstand the problem has implications for their knowledge base and decision-making.³⁵ Moreover, we discovered research that discussed how the activity of reflection adds to an individual’s knowledge base by creating new ways of thinking about a problem.²⁵

Decision-making approach

The most prominent theme in our analysis was clinicians’ overall approach to decision-making, defined as the concepts, strategies, methods, and tools used to reach a decision point and the context in which they are chosen. For example, several of the articles addressed the concept of time as it refers to purposefully managing the relative urgency in which a decision must be made. Cristancho and colleagues described a model for making intraoperative decisions for which time is a constraint. The clinician(s) must choose to make the decision with limited or incomplete information or delay a decision with potentially negative consequences.³⁰

We further identified 4 subthemes that characterized clinicians’ overall approach to decision-making in terms of the strategies they use: intuitive, protocol-driven, team-based, or shared. The intuitive approach is defined as drawing on one’s ability to identify patterns based on knowledge developed from past experiences and expertise within a domain.⁴⁴ For example, Thompson and Dowding discussed how nurses use intuition when confronted with a lack of clarity around decision-making for patient care activities.⁴⁰ Similarly, Mumford and colleagues described a “sense making” approach to decision-making that relies on the development of mental models, or cognitive representations of a situation that intertwine the structure and meaning of the situation, and the many different ways the situation may occur.^38,45 In contrast, the protocol-driven approach to decision-making relies on a structured set of rules or decision aids, such as decision trees, which illustrate possible choices and decision analysis models.⁴⁰

The team-based decision-making approach incorporates the perspectives of the physician, patient, and other members of the multidisciplinary health care team. Menard and colleagues described this approach as a collaborative meeting among the members of the health care team to establish their priorities for care in addition to those of the patient.³⁷ In contrast, the shared decision-making approach focuses only on the dynamics between the clinician and the patient. For example, Charles and colleagues described the characteristics of shared decision-making as involving 2 participants, physician and patient, who come to a consensus and share information.²⁸

Evaluation of the decision-making process

The final theme focuses on the metacognitive exercise clinicians undertake during the decision-making process. The decision-maker actively considers how she or he has assessed, synthesized, and reflected on the information gathered during the previous stages of decision-making. This assessment is made to determine the quality of the information gathered. If the quality of the information is positive, it is then synthesized to make a judgement. Corliss described the need for physicians to assess and synthesize their knowledge base, the patient’s case history, and information from clinical testing as well as less well-defined information, such as social factors, ethics, and politics.²⁹ Next, reflection pertains to the consideration of the consequences of the decision and the probabilities of various outcomes by all parties involved. Reflective thinking, through activities such as journaling, is one strategy for processing information in an uncertain clinical situation and guiding adaptation in the decision-making process.^25,38 Collectively then, assessment, synthesis, and reflection become tools to evaluate how a clinician adapts information in complex decision-making, leading to her or him choosing an approach to manage the uncertainty and arrive at a decision.^24,29

Discussion

In this scoping review, we set out to describe how the literature across disciplines characterizes the effects of uncertainty and ambiguity on the decision-making process. We reviewed more than 3,000 publications for possible inclusion and selected 19 articles from various fields of research that met our inclusion criteria. The included articles provided several models for conceptualizing uncertainty in decision-making, and our analysis highlighted 6 themes that described important components of this process. We used these themes to develop a framework with each theme representing a stage in the decision-making process (see Figure 2). We propose that this framework could be used to make medical decisions, whether in clinical reasoning during diagnosis, evaluation, or choosing a treatment option by clinicians, the health care team, or with patients.

Figure 2.

Framework for making decisions under uncertain conditions. Included in this framework are the 6 themes from a thematic analysis of 19 articles identified during a scoping review of the literature on uncertainty in decision-making. Four themes are listed as strategies to reduce uncertainty. The decision-maker uses these strategies along with an ongoing evaluation process of assessing, synthesizing, and reflecting on information obtained. Various decision-making approaches aide the process when uncertainty persists and no clear decision exists. This framework can be used to make medical decisions when uncertainty is present, whether in clinical reasoning during diagnosis, evaluation, or choosing a treatment option by the clinician, health care team, or with patients. Of note, if the uncertainty is naturally removed from the decision (e.g., the patient’s condition changes resulting in one treatment option becoming preferable over another), the framework would no longer apply.

Consider the hypothetical clinician who is deciding between two treatment options. Suppose the clinician has two options (A and B) with similar efficacy outcomes but variable side effect profiles. Treatment option A has the potential for severe acute side effects, but the known probability of these side effects is lower. Treatment option B has the potential for less severe side effects, but they are chronic in nature and with a higher probability of occurring. The clinician must help the patient decide between the two treatment options as they are mutually exclusive.

Using the framework presented in Figure 2, the clinician would first recognize the uncertainty in the situation, as neither option is clearly better than the other. Next, the clinician would attempt to understand what makes the situation uncertain: Is it the ambiguity of a wide probability range or insufficient evidence? Or complexity based on the number of potential side effects? Once the uncertainty is identified and classified, the clinician would determine all stakeholders in the decision (i.e., family, patient, and provider) and gather their perspectives on both treatment options. Following their input, the clinician may need to seek more knowledge to clarify and further inform the decision.

At any or all these steps, the clinician must assess the situation, synthesize the information, and reflect on the potential outcomes, in an ongoing evaluation of the decision-making process. The culmination of these steps might reduce the uncertainty in the situation to a point where one option becomes clearly better than the other, and there is no longer uncertainty. However, despite going through this process, some uncertainty may persist, forcing the clinician to consider various approaches to help guide the final decision. These approaches may include the use of tools, such as decision aides, decision analysis calculations, or decision trees to illustrate the probability of various outcomes and side effects, or the use of team-based or shared decision-making strategies. If the patient’s condition changes, resulting in one treatment option becoming preferable over the other, or the uncertainty is otherwise naturally removed from the decision, the framework would no longer apply.

During our analysis of the articles included in our scoping review, we discovered that each article emphasized one or more elements of the decision-making process described in Table 3 and Figure 2. However, no single article combined these components into an all-encompassing process map for managing uncertainty in decision-making. We attempted to fill this gap by proposing the framework in Figure 2, which is a complex iterative and recursive process describing how the themes we extracted work together.

This process originates in a clinician’s ability to recognize uncertainty. It further asks the decision-maker to continually increase her or his understanding of the uncertainty and priorities involved in making the decision before ultimately arriving at that decision through a purposefully chosen approach or by effectively reducing the uncertainty in the situation so the decision is clear. We propose this framework to improve clinicians’ understanding of how medical decisions are made under uncertain conditions and to act as a guide to managing that uncertainty. The framework was developed to be generalizable to any situation with uncertainty regardless of source, degree, or context, including time. Moreover, it gives the decision-maker distinct steps to consider to ensure that a comprehensive and deliberate process is used.

Limitations

Due to the nature of our methodology, there are some limitations to our study. As is common with scoping reviews, our inclusion and exclusion criteria were broad to allow for the data to come from any context.¹⁸ However, all decisions were made by consensus and in discussion to ensure common understanding and consistency in the application of the criteria among reviewers. Next, our search included specific terminology to identify articles that presented a model or framework. Therefore, articles that discussed uncertainty and the decision-making process but did not explicitly discuss a model or framework may not have been captured. We also excluded non-English language articles as well as specific formats, such as book chapters, which may have limited representation from disciplines in which the journal article is not the main mode of scholarly discourse.

Future research

There are several avenues for future research. First, much of the research in this area considers a clinician’s ability to manage uncertainty as a stable skill. However, the ability to manage uncertainty evolves over time, either through natural evolution or through formal education and/or training.^46,47 Future research should determine if a clinician’s ability to manage uncertainty may be influenced by her or his experience. Second, the study of decision-making in uncertain situations in health care offers an opportunity to examine the influence of the environment and culture on the decision-making process. Are certain providers or specialists better at managing uncertainty? What factors (e.g., personal attributes, team structure, institutional/departmental culture, etc.) contribute to this ability?

Next, future research should explore the role of the curriculum in teaching uncertainty in decision-making. We believe that managing uncertainty should be explicitly taught in the curriculum rather than treated as an overlooked part of the daily routine of providers, as it often is now. Researchers then should develop an assessment tool to measure trainees’ learning. Lastly, researchers should develop and validate a more robust framework than the one we presented here, to delineate the psychological mechanisms that are relevant to managing uncertainty in decision-making in the context of health care.

Conclusions

Through this scoping review, we explored how uncertainty in decision-making is represented in the health sciences literature and in other fields of research. We found that a single model describing this process in its entirety was noticeably absent from the literature. To fill this gap, we created a framework for making decisions under uncertain conditions, but further development and validation of that framework are needed to inform medical school curricula.