Clinical reasoning—the gathering and integration of clinical information combined with medical knowledge to generate a diagnosis and treatment plan—is a complex and challenging endeavor requiring extensive practice to reach proficiency [1, 2]. Even among physicians with many years of experience, diagnostic errors continue to be a problem, accounting for approximately 10% of patient deaths and contributing to other issues, such as delays in diagnosis and treatment and medication errors [3, 4].
Given the need to enhance clinical reasoning proficiency, there has been increased attention on learning methods to optimize these abilities. Common approaches include lectures, case-based learning, clinical case discussions, workplace learning, and simulation-based learning [5]. Simulation-based formats, which include virtual patients, pre-recorded videos (i.e., vignettes depicting a doctor-patient encounter [6]), and live scenarios (i.e., structured narrative embedded within a simulated clinical setting) [7, 8] have increased in popularity over the years. Their popularity has grown, in part, because they closely mirror authentic, clinical settings and patient-provider interactions [6], afford opportunities to practice myriad clinical activities in different contexts [9], and enable extensive opportunities for reflection [10, 11].
Although some researchers have examined the individual effects of traditional (e.g., paper cases) and simulation learning environments [12, 13], very few have examined the relative effectiveness of such approaches for enhancing clinical reasoning abilities [14]. Further, learning effectiveness research has typically focused on performance outcomes (e.g., diagnoses and direct observation in clinical or simulated settings) rather than the processes and overall experiences of medical professionals during clinical activities. Given these gaps, we experimentally examined the differential effects of two simulation learning environments (i.e., video and live scenario) across performance outcomes as well as the task-specific perceptions, cognitive reactions, and reflective judgments of medical professionals during clinical reasoning.
Clinical reasoning as complex and situated
Although clinical reasoning is often conceptualized as an end product, Ilgen, Eva, and Regehr argue that it can also be viewed as a complex, dynamic, and often uncertain process of meaning making [15]. They argue that the skillful deployment and completion of clinical reasoning tasks shift according to the case and context, painting a complex and situation-specific (situated) picture of clinical reasoning [15]. Beyond the complexity of the clinical reasoning tasks themselves, there is a developing literature on contextual factors—common features of clinical practice (e.g., patient frustration, interruptions, and language barriers) that typically are not used to establish the correct diagnosis [16,17,18]. Based on recent research [19, 20] and the theoretical proposition that knowing is bound to activity, social norms, environment, and cultural factors [21], the presence of contextual factors can lead physicians to think about and react to different aspects of a case. Differences in situation-specific perceptions and the metacognitive reactions to contextual factors can greatly alter the quality or accuracy of physicians’ diagnostic and management reasoning [18, 22].
Clinical reasoning and simulation-based learning environments
A variety of learning environments have been used to teach and assess clinical reasoning abilities and often emphasize differences in what is learned. For example, case-based learning and virtual patients emphasize the development of cognitive processes (i.e., interpretation of findings and hypothesis generation), whereas morbidity and mortality rounds and small group coaching place more of an emphasis on metacognition (i.e., monitoring and reflecting on one’s own thought processes) and educational strategies [23]. While all such approaches can support both cognitive and metacognitive skills to some degree, simulation-based learning environments are particularly well suited to address both [10, 11, 24]. Moreover, several studies highlight how post-simulation reflection can support participants’ clinical reasoning as they consider the meaning of their actions and experiences and scrutinize personal assumptions [25, 26].
All simulation environments overlap in terms of participant experiences. When comparing live scenarios and video case formats, both situate the clinical encounter in a fictitious, yet realistic setting depicting a provider-patient interaction [9, 27]. They also emphasize a sequential approach to presenting information (i.e., starting with a greeting, followed by a patient interview) and encourage participants to identify relevant clinical information, identify hypotheses, and solve a clinical problem [27, 28]. However, video cases and live scenarios can be distinguished in terms of duration, efficiency, and complexity of social interactions.
Video cases are quite popular, in part, because of their efficiency and accessibility. Participants are asked to view a pre-recorded provider-patient encounter that has a fixed and often short delivery time. The sequence of case content (e.g., interview, physician exam maneuvers, and lab results [27]) is pre-determined, so participants cannot influence aspects of the encounter. Conversely, live scenario-based simulations are more complicated and difficult to use, in part, because of the need for specially trained individuals (e.g., standardized patients, and simulationists) and the significant time required for design and implementation [29, 30]. Live scenarios also tend to be more intensive in that participants need to engage in complex, clinical activities (e.g., structured interventions such as focused assessment) while concurrently determining optimal ways to sequence these activities, an experience characterized by high levels of autonomy, agency, and cognitive demands [7]. Live scenarios can also be more unpredictable in terms of the duration of the patient encounter and the nature of the physician or patient responses [7].
These structural distinctions are not perfunctory, as they have the potential to influence the nature of the clinical reasoning processes used by medical professionals as well as their subjective reactions. Further, although researchers have examined the influence of different simulation approaches used to teach and evaluate clinical reasoning, such as live scenarios and videos, systematic and direct comparisons of these approaches remain limited [9, 14, 31,32,33]. Broadly speaking, the literature is mixed regarding the relative superiority of any given approach. For example, while Durning and colleagues reported no differences in clinical reasoning performance across standardized patient case, video case, and paper case formats [34], LaRochelle and colleagues observed that standardized patient cases and video cases were superior to paper cases, but only for certain subject areas [14].
Assessing processes during clinical reasoning
Early efforts to examine clinical reasoning processes emphasized behavioral observations and think-aloud protocols [35,36,37]. This early research helped establish a foundation for understanding the types of actions comprising the clinical reasoning process, such as interviewing, physical assessment, and testing hypotheses. While think-aloud protocols continue to be used within medical education [38], there have been recent attempts to apply unique analytic approaches, such as linguistic analysis, to interpret think-aloud data [20, 39]. One promising tool for understanding the process of clinical reasoning is automated coding of linguistic markers of cognitive processing using the Linguistic Inquiry and Word Count (LIWC) software [40, 41]. One set of LIWC markers is related to cognitive activity along with six dimensions: insight (e.g., think and know), cause (e.g., because and effect), discrepancy (e.g., should and would), tentativeness (e.g., maybe and perhaps), certainty (e.g., always and never), and differentiation (e.g., but and else) [42]. Frequency of these “cognitive processing” words corresponds with higher mental effort and greater focus on tasks like discerning, determining causal relations, and differentiating [43].
Self-regulated learning (SRL) microanalytic protocols have also been used to assess medical professionals’ cognitive and regulatory processes (e.g., planning, monitoring, and evaluative judgments during clinical reasoning) [38, 44,45,46,47]. These assessment protocols consist of contextualized questions directly targeting specific regulatory processes (e.g., monitoring and adaptive inferences) that are administered as individuals complete a target activity. Grounded in a social-cognitive perspective that SRL is a dynamic, three-phase cyclical process (i.e., forethought, performance, and reflection), SRL microanalytic protocols are able to assess how individuals strategically approach a task and set goals (i.e., forethought phase), control and monitor task completion (i.e., performance phase), and evaluate and reflect on performance (i.e., self-reflection phase) [46, 48].