Exploring Clinical Lessons Learned by Experienced Hospitalists from Diagnostic Errors and Successes.

BACKGROUND: Diagnostic errors cause significant patient harm. The clinician's ultimate goal is to achieve diagnostic excellence in order to serve patients safely. This can be accomplished by learning from both errors and successes in patient care. However, the extent to which clinicians grow and navigate diagnostic errors and successes in patient care is poorly understood. Clinically experienced hospitalists, who have cared for numerous acutely ill patients, should have great insights from their successes and mistakes to inform others striving for excellence in patient care. OBJECTIVE: To identify and characterize clinical lessons learned by experienced hospitalists from diagnostic errors and successes. DESIGN: A semi-structured interview guide was used to collect qualitative data from hospitalists at five independently administered hospitals in the Mid-Atlantic area from February to June 2022. PARTICIPANTS: 12 academic and 12 community-based hospitalists with ≥ 5 years of clinical experience. APPROACH: A constructivist qualitative approach was used and "reflexive thematic analysis" of interview transcripts was conducted to identify themes and patterns of meaning across the dataset. RESULTS: Five themes were generated from the data based on clinical lessons learned by hospitalists from diagnostic errors and successes. The ideas included appreciating excellence in clinical reasoning as a core skill, connecting with patients and other members of the health care team to be able to tap into their insights, reflecting on the diagnostic process, committing to growth, and prioritizing self-care. CONCLUSIONS: The study identifies key lessons learned from the errors and successes encountered in patient care by clinically experienced hospitalists. These findings may prove helpful for individuals and groups that are authentically committed to moving along the continuum from diagnostic competence towards excellence.

Teaching medical students to apply deliberate reflection.

PURPOSE: Deliberate reflection on initial diagnosis has been found to repair diagnostic errors. We investigated the effectiveness of teaching students to use deliberate reflection on future cases and whether their usage would depend on their perception of case difficulty. METHOD: One-hundred-nineteen medical students solved cases either with deliberate-reflection or without instructions to reflect. One week later, all participants solved six cases, each with two equally likely diagnoses, but some symptoms in the case were associated with only one of the diagnoses (discriminating features). Participants provided one diagnosis and subsequently wrote down everything they remembered from it. After the first three cases, they were told that the next three would be difficult cases. Reflection was measured by the proportion of discriminating features recalled (overall; related to their provided diagnosis; related to alternative diagnosis). RESULTS: The deliberate-reflection condition recalled more features for the alternative diagnosis than the control condition (p = .013) regardless of described difficulty. They also recalled more features related to their provided diagnosis on the first three cases (p = .004), but on the last three cases (described as difficult), there was no difference. CONCLUSION: Learning deliberate reflection helped students engage in more reflective reasoning when solving future cases.

Learning deliberate reflection in medical diagnosis: does learning-by-teaching help?

Deliberate reflection has been found to foster diagnostic accuracy on complex cases or under circumstances that tend to induce cognitive bias. However, it is unclear whether the procedure can also be learned and thereby autonomously applied when diagnosing future cases without instructions to reflect. We investigated whether general practice residents would learn the deliberate reflection procedure through 'learning-by-teaching' and apply it to diagnose new cases. The study was a two-phase experiment. In the learning phase, 56 general-practice residents were randomly assigned to one of two conditions. They either (1) studied examples of deliberate reflection and then explained the procedure to a fictitious peer on video; or (2) solved cases without reflection (control). In the test phase, one to three weeks later, all participants diagnosed new cases while thinking aloud. The analysis of the test phase showed no significant differences between the conditions on any of the outcome measures (diagnostic accuracy, p = .263; time to diagnose, p = .598; mental effort ratings, p = .544; confidence ratings, p = .710; proportion of contradiction units (i.e. measure of deliberate reflection), p = .544). In contrast to findings on learning-by-teaching from other domains, teaching deliberate reflection to a fictitious peer, did not increase reflective reasoning when diagnosing future cases. Potential explanations that future research might address are that either residents in the experimental condition did not apply the learned deliberate reflection procedure in the test phase, or residents in the control condition also engaged in reflection.

Identifying and prioritizing educational content from a malpractice claims database for clinical reasoning education in the vocational training of general practitioners.

Diagnostic reasoning is an important topic in General Practitioners' (GPs) vocational training. Interestingly, research has paid little attention to the content of the cases used in clinical reasoning education. Malpractice claims of diagnostic errors represent cases that impact patients and that reflect potential knowledge gaps and contextual factors. With this study, we aimed to identify and prioritize educational content from a malpractice claims database in order to improve clinical reasoning education in GP training. With input from various experts in clinical reasoning and diagnostic error, we defined five priority criteria that reflect educational relevance. Fifty unique medical conditions from a malpractice claims database were scored on those priority criteria by stakeholders in clinical reasoning education in 2021. Subsequently, we calculated the mean total priority score for each condition. Mean total priority score (min 5-max 25) for all fifty diagnoses was 17,11 with a range from 13,89 to 19,61. We identified and described the fifteen highest scoring diseases (with priority scores ranging from 18,17 to 19,61). The prioritized conditions involved complex common (e.g., cardiovascular diseases, renal insufficiency and cancer), complex rare (e.g., endocarditis, ectopic pregnancy, testicular torsion) and more straightforward common conditions (e.g., tendon rupture/injury, eye infection). The claim cases often demonstrated atypical presentations or complex contextual factors. Including those malpractice cases in GP vocational training could enrich the illness scripts of diseases that are at high risk of errors, which may reduce diagnostic error and related patient harm.

Deliberate practice of diagnostic clinical reasoning reveals low performance and improvement of diagnostic justification in pre-clerkship students.

PURPOSE: Diagnostic errors are a large burden on patient safety and improving clinical reasoning (CR) education could contribute to reducing these errors. To this end, calls have been made to implement CR training as early as the first year of medical school. However, much is still unknown about pre-clerkship students' reasoning processes. The current study aimed to observe how pre-clerkship students use clinical information during the diagnostic process. METHODS: In a prospective observational study, pre-clerkship medical students completed 10-11 self-directed online simulated CR diagnostic cases. CR skills assessed included: creation of the differential diagnosis (Ddx), diagnostic justification (DxJ), ordering investigations, and identifying the most probable diagnosis. Student performances were compared to expert-created scorecards and students received detailed individualized formative feedback for every case. RESULTS: 121 of 133 (91%) first- and second-year medical students consented to the research project. Students scored much lower for DxJ compared to scores obtained for creation of the Ddx, ordering tests, and identifying the correct diagnosis, (30-48% lower, p < 0.001). Specifically, students underutilized physical exam data (p < 0.001) and underutilized data that decreased the probability of incorrect diagnoses (p < 0.001). We observed that DxJ scores increased 40% after 10-11 practice cases (p < 0.001). CONCLUSIONS: We implemented deliberate practice with formative feedback for CR starting in the first year of medical school. Students underperformed in DxJ, particularly with analyzing the physical exam data and pertinent negative data. We observed significant improvement in DxJ performance with increased practice.

Do malpractice claim clinical case vignettes enhance diagnostic accuracy and acceptance in clinical reasoning education during GP training?

BACKGROUND: Using malpractice claims cases as vignettes is a promising approach for improving clinical reasoning education (CRE), as malpractice claims can provide a variety of content- and context-rich examples. However, the effect on learning of adding information about a malpractice claim, which may evoke a deeper emotional response, is not yet clear. This study examined whether knowing that a diagnostic error resulted in a malpractice claim affects diagnostic accuracy and self-reported confidence in the diagnosis of future cases. Moreover, suitability of using erroneous cases with and without a malpractice claim for CRE, as judged by participants, was evaluated. METHODS: In the first session of this two-phased, within-subjects experiment, 81 first-year residents of general practice (GP) were exposed to both erroneous cases with (M) and erroneous cases without (NM) malpractice claim information, derived from a malpractice claims database. Participants rated suitability of the cases for CRE on a five-point Likert scale. In the second session, one week later, participants solved four different cases with the same diagnoses. Diagnostic accuracy was measured with three questions, scored on a 0-1 scale: (1) What is your next step? (2) What is your differential diagnosis? (3) What is your most probable diagnosis and what is your level of certainty on this? Both subjective suitability and diagnostic accuracy scores were compared between the versions (M and NM) using repeated measures ANOVA. RESULTS: There were no differences in diagnostic accuracy parameters (M vs. NM next step: 0.79 vs. 0.77, p = 0.505; differential diagnosis 0.68 vs. 0.75, p = 0.072; most probable diagnosis 0.52 vs. 0.57, p = 0.216) and self-reported confidence (53.7% vs. 55.8% p = 0.390) of diagnoses previously seen with or without malpractice claim information. Subjective suitability- and complexity scores for the two versions were similar (suitability: 3.68 vs. 3.84, p = 0.568; complexity 3.71 vs. 3.88, p = 0.218) and significantly increased for higher education levels for both versions. CONCLUSION: The similar diagnostic accuracy rates between cases studied with or without malpractice claim information suggests both versions are equally effective for CRE in GP training. Residents judged both case versions to be similarly suitable for CRE; both were considered more suitable for advanced than for novice learners.

Insights into diagnostic errors in endocrinology: a prospective, case-based, international study.

BACKGROUND: Diagnostic errors in internal medicine are common. While cognitive errors have previously been identified to be the most common contributor to errors, very little is known about errors in specific fields of internal medicine such as endocrinology. This prospective, multicenter study focused on better understanding the causes of diagnostic errors made by general practitioners and internal specialists in the area of endocrinology. METHODS: From August 2019 until January 2020, 24 physicians completed five endocrine cases on an online platform that simulated the diagnostic process. After each case, the participants had to state and explain why they chose their assumed diagnosis. The data gathering process as well as the participants' explanations were quantitatively and qualitatively analyzed to determine the causes of the errors. The diagnostic processes in correctly and incorrectly solved cases were compared. RESULTS: Seven different causes of diagnostic error were identified, the most frequent being misidentification (mistaking one diagnosis with a related one or with more frequent and similar diseases) in 23% of the cases. Other causes were faulty context generation (21%) and premature closure (17%). The diagnostic confidence did not differ between correctly and incorrectly solved cases (median 8 out of 10, p = 0.24). However, in incorrectly solved cases, physicians spent less time on the technical findings (such as lab results, imaging) (median 250 s versus 199 s, p < 0.049). CONCLUSIONS: The causes for errors in endocrine case scenarios are similar to the causes in other fields of internal medicine. Spending more time on technical findings might prevent misdiagnoses in everyday clinical practice.

Improving medical residents' self-assessment of their diagnostic accuracy: does feedback help?

When physicians do not estimate their diagnostic accuracy correctly, i.e. show inaccurate diagnostic calibration, diagnostic errors or overtesting can occur. A previous study showed that physicians' diagnostic calibration for easy cases improved, after they received feedback on their previous diagnoses. We investigated whether diagnostic calibration would also improve from this feedback when cases were more difficult. Sixty-nine general-practice residents were randomly assigned to one of two conditions. In the feedback condition, they diagnosed a case, rated their confidence in their diagnosis, their invested mental effort, and case complexity, and then were shown the correct diagnosis (feedback). This was repeated for 12 cases. Participants in the control condition did the same without receiving feedback. We analysed calibration in terms of (1) absolute accuracy (absolute difference between diagnostic accuracy and confidence), and (2) bias (confidence minus diagnostic calibration). There was no difference between the conditions in the measurements of calibration (absolute accuracy, p = .204; bias, p = .176). Post-hoc analyses showed that on correctly diagnosed cases (on which participants are either accurate or underconfident), calibration in the feedback condition was less accurate than in the control condition, p = .013. This study shows that feedback on diagnostic performance did not improve physicians' calibration for more difficult cases. One explanation could be that participants were confronted with their mistakes and thereafter lowered their confidence ratings even if cases were diagnosed correctly. This shows how difficult it is to improve diagnostic calibration, which is important to prevent diagnostic errors or maltreatment.

Advancing Diagnostic Safety Research: Results of a Systematic Research Priority Setting Exercise.

BACKGROUND: Diagnostic errors are a major source of preventable harm but the science of reducing them remains underdeveloped. OBJECTIVE: To identify and prioritize research questions to advance the field of diagnostic safety in the next 5 years. PARTICIPANTS: Ninety-seven researchers and 42 stakeholders were involved in the identification of the research priorities. DESIGN: We used systematic prioritization methods based on the Child Health and Nutrition Research Initiative (CHNRI) methodology. We first invited a large international group of expert researchers in various disciplines to submit research questions while considering five prioritization criteria: (1) usefulness, (2) answerability, (3) effectiveness, (4) potential for translation, and (5) maximal potential for effect on diagnostic safety. After consolidation, these questions were prioritized at an in-person expert meeting in April 2019. Top-ranked questions were subsequently reprioritized through scoring on the five prioritization criteria using an online questionnaire. We also invited non-research stakeholders to assign weights to the five criteria and then used these weights to adjust the final prioritization score for each question. KEY RESULTS: Of the 207 invited researchers, 97 researchers responded and 78 submitted 333 research questions which were then consolidated. Expert meeting participants (n = 21) discussed questions in different breakout sessions and prioritized 50, which were subsequently reduced to the top 20 using the online questionnaire. The top 20 questions addressed mostly system factors (e.g., implementation and evaluation of information technologies), teamwork factors (e.g., role of nurses and other health professionals in the diagnostic process), and strategies to engage patients in the diagnostic process. CONCLUSIONS: Top research priorities for advancing diagnostic safety in the short-term include strengthening systems and teams and engaging patients to support diagnosis. High-priority areas identified using these systematic methods can inform an actionable research agenda for reducing preventable diagnostic harm.

Specific Disease Knowledge as Predictor of Susceptibility to Availability Bias in Diagnostic Reasoning: a Randomized Controlled Experiment.

BACKGROUND: Bias in reasoning rather than knowledge gaps has been identified as the origin of most diagnostic errors. However, the role of knowledge in counteracting bias is unclear. OBJECTIVE: To examine whether knowledge of discriminating features (findings that discriminate between look-alike diseases) predicts susceptibility to bias. DESIGN: Three-phase randomized experiment. Phase 1 (bias-inducing): Participants were exposed to a set of clinical cases (either hepatitis-IBD or AMI-encephalopathy). Phase 2 (diagnosis): All participants diagnosed the same cases; 4 resembled hepatitis-IBD, 4 AMI-encephalopathy (but all with different diagnoses). Availability bias was expected in the 4 cases similar to those encountered in phase 1. Phase 3 (knowledge evaluation): For each disease, participants decided (max. 2 s) which of 24 findings was associated with the disease. Accuracy of decisions on discriminating features, taken as a measure of knowledge, was expected to predict susceptibility to bias. PARTICIPANTS: Internal medicine residents at Erasmus MC, Netherlands. MAIN MEASURES: The frequency with which higher-knowledge and lower-knowledge physicians gave biased diagnoses based on phase 1 exposure (range 0-4). Time to diagnose was also measured. KEY RESULTS: Sixty-two physicians participated. Higher-knowledge physicians yielded to availability bias less often than lower-knowledge physicians (0.35 vs 0.97; p = 0.001; difference, 0.62 [95% CI, 0.28-0.95]). Whereas lower-knowledge physicians tended to make more of these errors on subjected-to-bias than on not-subjected-to-bias cases (p = 0.06; difference, 0.35 [CI, - 0.02-0.73]), higher-knowledge physicians resisted the bias (p = 0.28). Both groups spent more time to diagnose subjected-to-bias than not-subjected-to-bias cases (p = 0.04), without differences between groups. CONCLUSIONS: Knowledge of features that discriminate between look-alike diseases reduced susceptibility to bias in a simulated setting. Reflecting further may be required to overcome bias, but succeeding depends on having the appropriate knowledge. Future research should examine whether the findings apply to real practice and to more experienced physicians.

Promotion of knowledge transfer and retention in year 2 medical students using an online training exercise.

It was recently shown that novice medical students could be trained to demonstrate the speed-to-diagnosis and diagnostic accuracy typical of System-1-type reasoning. However, the effectiveness of this training can only be fully evaluated when considering the extent to which knowledge transfer and long-term retention occur as a result, the former of which is known to be notoriously difficult to achieve. This study aimed to investigate whether knowledge learned during an online training exercise for chest X-ray diagnosis promoted either knowledge transfer or retention, or both. Second year medical students were presented with, and trained to recognise the features of four chest X-ray conditions. Subsequently, they were shown the four trained-for cases again as well as different representations of the same conditions varying in the number of common elements and asked to provide a diagnosis, to test for near-transfer (four cases) and far-transfer (four cases) of knowledge. They were also shown four completely new conditions to diagnose. Two weeks later they were asked to diagnose the 16 aforementioned cases again to assess for knowledge retention. Dependent variables were diagnostic accuracy and time-to-diagnosis. Thirty-six students volunteered. Trained-for cases were diagnosed most accurately and with most speed (mean score = 3.75/4, mean time = 4.95 s). When assessing knowledge transfer, participants were able to diagnose near-transfer cases more accurately (mean score = 2.08/4, mean time = 15.77 s) than far-transfer cases (mean score = 1.31/4, mean time = 18.80 s), which showed similar results to those conditions previously unseen (mean score = 0.72/4, mean time = 19.46 s). Retention tests showed a similar pattern but accuracy scores were lower overall. This study demonstrates that it is possible to successfully promote knowledge transfer and retention in Year 2 medical students, using an online training exercise involving diagnosis of chest X-rays, and is one of the few studies to provide evidence of actual knowledge transfer.

Evidence supporting dual-process theory of medical diagnosis: a functional near-infrared spectroscopy study.

PURPOSE: The objective of this study was to determine the extent to which the dual-process theory of medical diagnosis enjoys neuroscientific support. To that end, the study explored whether neurological correlates of system-2 thinking could be located in the brain. It was hypothesised that system-2 thinking could be observed as the activation of the prefrontal cortex. METHOD: An experimental paradigm was applied that consisted of a learning and a test phase. During the learning phase, 22 medical students were trained in diagnosing chest X-rays. Four of these eight cases were presented repeatedly, to develop a high level of expertise for these cases. During the test phase, all eight cases were presented and the participants' prefrontal cortex was scanned using functional near-infrared spectroscopy. Response time and diagnostic accuracy were recorded as behavioural indicators. RESULTS: The results revealed that participants' diagnostic accuracy in the test phase was significantly higher for the trained cases as compared with the untrained cases (F[1, 21] = 138.80, p < 0.001, η2  = 0.87). Also, their response time was significantly shorter for these cases (F[1, 21] = 18.12, p < 0.001, η2  = 0.46). Finally, the results revealed that only for the untrained cases, could a significant activation of the anterolateral prefrontal cortex be observed (F[1, 21] = 21.00, p < 0.01, η2  = 0.34). CONCLUSION: The fact that only untrained cases triggered higher levels of blood oxygenation in the prefrontal cortex is an indication that system-2 thinking is a cognitive process distinct from system 1. Implications of these findings for the validity of the dual-process theory are discussed.

Correction to: Debiasing versus knowledge retrieval checklists to reduce diagnostic error in ECG interpretation.

Due to an unfortunate turn of events, Fig. 3 was omitted from the original publication.

Debiasing versus knowledge retrieval checklists to reduce diagnostic error in ECG interpretation.

There is an ongoing debate regarding the cause of diagnostic errors. One view is that errors result from unconscious application of cognitive heuristics; the alternative is that errors are a consequence of knowledge deficits. The objective of this study was to compare the effectiveness of checklists that (a) identify and address cognitive biases or (b) promote knowledge retrieval, as a means to reduce errors in ECG interpretation. Novice postgraduate year (PGY) 1 emergency medicine and internal medicine residents (n = 40) and experienced cardiology fellows (PGY 4-6) (n = 21) were randomly allocated to three conditions: a debiasing checklist, a content (knowledge) checklist, or control (no checklist) to be used while interpreting 20 ECGs. Half of the ECGs were deliberately engineered to predispose to bias. Diagnostic performance under either checklist intervention was not significantly better than the control. As expected, more errors occurred when cases were designed to induce bias (F = 96.9, p < 0.0001). There was no significant interaction between the instructional condition and level of learner. Checklists attempting to help learners identify cognitive bias or mobilize domain-specific knowledge did not have an overall effect in reducing diagnostic errors in ECG interpretation, although they may help novices. Even when cognitive biases are deliberately inserted in cases, cognitive debiasing checklists did not improve participants' performance.

Inducing System-1-type diagnostic reasoning in second-year medical students within 15 minutes.

Purpose: Diagnostic reasoning literature debates the significance of "dual-process theory" and the importance of its constituent types of thinking: System-1and System-2. This experimental study aimed to determine whether novice medical students could be trained to utilize System-1 thinking when making diagnoses based on chest X-rays. Method: Second-year medical students were recruited and presented with a series of eight online chest X-rays cases. Participants were shown half of the cases repeatedly during a training phase and the other half only twice. During the final test phase, they were shown all eight cases, providing a diagnosis as a free text answer. Dependent variables were diagnostic accuracy and response time. Results: Thirty-two students participated. During the test phase, students responses were significantly more accurate and faster for cases which had been seen repeatedly during the training phase (mean score = 3.56/4, mean time = 2.34 s) compared with cases which had been seen only twice (mean score = 1.59/4, mean time = 7.50 s). Conclusion: This study demonstrates that it is possible to induce in novice students the speed-to-diagnosis and diagnostic accuracy typical of System-1-type reasoning. The full experimental design and the chest X-rays used may provide new opportunities to explore some of the issues surrounding dual-process theory.

Diagnostic errors by medical students: results of a prospective qualitative study.

BACKGROUND: Diagnostic errors occur frequently in daily clinical practice and put patients' safety at risk. There is an urgent need to improve education on clinical reasoning to reduce diagnostic errors. However, little is known about diagnostic errors of medical students. In this study, the nature of the causes of diagnostic errors made by medical students was analyzed. METHODS: In June 2016, 88 medical students worked on eight cases with the chief complaint dyspnea in a laboratory setting using an electronic learning platform, in summary 704 processed cases. The diagnostic steps of the students were tracked and analyzed. Furthermore, after each case the participants stated their presumed diagnosis and explained why they came to their diagnostic conclusion. The content of these explanations was analyzed qualitatively. RESULTS: Based on the diagnostic data gathering process and the students' explanations, eight different causes could be identified of which the lack of diagnostic skills (24%) and inadequate knowledge base (16%) were the most common. Other causes that often contributed to a diagnostic error were faulty context generation (15%) and premature closure (10%). The causes of misdiagnosis varied per case. CONCLUSIONS: Inadequate skills/knowledge and faulty context generation are the major problems in students' clinical reasoning process. These findings are valuable for improving medical education and thus reducing the frequency of diagnostic errors in students' later everyday clinical practice.

Relationship between non-technical skills and technical performance during cardiopulmonary resuscitation: does stress have an influence?

BACKGROUND: Non-technical skills, such as task management, leadership, situational awareness, communication and decision-making refer to cognitive, behavioural and social skills that contribute to safe and efficient team performance. The importance of these skills during cardiopulmonary resuscitation (CPR) is increasingly emphasised. Nonetheless, the relationship between non-technical skills and technical performance is poorly understood. We hypothesise that non-technical skills become increasingly important under stressful conditions when individuals are distracted from their tasks, and investigated the relationship between non-technical and technical skills under control conditions and when external stressors are present. METHODS: In this simulator-based randomised cross-over study, 30 anaesthesiologists and anaesthesia residents from the VU University Medical Center, Amsterdam, the Netherlands, participated in two different CPR scenarios in random order. In one scenario, external stressors (radio noise and a distractive scripted family member) were added, while the other scenario without stressors served as control condition. Non-technical performance of the team leader and technical performance of the team were measured using the 'Anaesthetists' Non-technical Skill' score and a recently developed technical skills score. Analysis of variance and Pearson correlation coefficients were used for statistical analyses. RESULTS: Non-technical performance declined when external stressors were present (adjusted mean difference 3.9 points, 95% CI 2.4 to 5.5 points). A significant correlation between non-technical and technical performance scores was observed when external stressors were present (r=0.67, 95% CI 0.40 to 0.83, p<0.001), while no evidence for such a relationship was observed under control conditions (r=0.15, 95% CI -0.22 to 0.49, p=0.42). This was equally true for all individual domains of the non-technical performance score (task management, team working, situation awareness, decision-making). CONCLUSIONS: During CPR with external stressors, the team's technical performance is related to the non-technical skills of the team leader. This may have important implications for training of CPR teams.

Characteristics differentiating problem representation synthesis between novices and experts.

BACKGROUND: Formulating a thoughtful problem representation (PR) is fundamental to sound clinical reasoning and an essential component of medical education. Aside from basic structural recommendations, little consensus exists on what characterizes high-quality PRs. OBJECTIVES: To elucidate characteristics that distinguish PRs created by experts and novices. METHODS: Early internal medicine residents (novices) and inpatient teaching faculty (experts) from two academic medical centers were given two written clinical vignettes and were instructed to write a PR and three-item differential diagnosis for each. Deductive content analysis described the characteristics comprising PRs. An initial codebook of characteristics was refined iteratively. The primary outcome was differences in characteristic frequencies between groups. The secondary outcome was characteristics correlating with diagnostic accuracy. Mixed-effects regression with random effects modeling compared case-level outcomes by group. RESULTS: Overall, 167 PRs were analyzed from 30 novices and 54 experts. Experts included 0.8 fewer comorbidities (p < .01) and 0.6 more examination findings (p = .01) than novices on average. Experts were less likely to include irrelevant comorbidities (odds ratio [OR] = 0.4, 95% confidence interval [CI] = 0.2-0.8) or a diagnosis (OR = 0.3, 95% CI = 0.1-0.8) compared with novices. Experts encapsulated clinical data into higher-order terms (e.g., sepsis) than novices (p < .01) while including similar numbers of semantic qualifiers (SQs). Regardless of expertise level, PRs following a three-part structure (e.g., demographics, temporal course, and clinical syndrome) and including temporal SQs were associated with diagnostic accuracy (p < .01). CONCLUSIONS: Compared with novices, expert PRs include less irrelevant data and synthesize information into higher-order concepts. Future studies should determine whether targeted educational interventions for PRs improve diagnostic accuracy.