Validity evidence supporting clinical skills assessment by artificial intelligence compared with trained clinician raters.

BACKGROUND: Artificial intelligence (AI) is becoming increasingly used in medical education, but our understanding of the validity of AI-based assessments (AIBA) as compared with traditional clinical expert-based assessments (EBA) is limited. In this study, the authors aimed to compare and contrast the validity evidence for the assessment of a complex clinical skill based on scores generated from an AI and trained clinical experts, respectively. METHODS: The study was conducted between September 2020 to October 2022. The authors used Kane's validity framework to prioritise and organise their evidence according to the four inferences: scoring, generalisation, extrapolation and implications. The context of the study was chorionic villus sampling performed within the simulated setting. AIBA and EBA were used to evaluate performances of experts, intermediates and novice based on video recordings. The clinical experts used a scoring instrument developed in a previous international consensus study. The AI used convolutional neural networks for capturing features on video recordings, motion tracking and eye movements to arrive at a final composite score. RESULTS: A total of 45 individuals participated in the study (22 novices, 12 intermediates and 11 experts). The authors demonstrated validity evidence for scoring, generalisation, extrapolation and implications for both EBA and AIBA. The plausibility of assumptions related to scoring, evidence of reproducibility and relation to different training levels was examined. Issues relating to construct underrepresentation, lack of explainability, and threats to robustness were identified as potential weak links in the AIBA validity argument compared with the EBA validity argument. CONCLUSION: There were weak links in the use of AIBA compared with EBA, mainly in their representation of the underlying construct but also regarding their explainability and ability to transfer to other datasets. However, combining AI and clinical expert-based assessments may offer complementary benefits, which is a promising subject for future research.

Data sharing and big data in health professions education: Ottawa consensus statement and recommendations for scholarship.

Changes in digital technology, increasing volume of data collection, and advances in methods have the potential to unleash the value of big data generated through the education of health professionals. Coupled with this potential are legitimate concerns about how data can be used or misused in ways that limit autonomy, equity, or harm stakeholders. This consensus statement is intended to address these issues by foregrounding the ethical imperatives for engaging with big data as well as the potential risks and challenges. Recognizing the wide and ever evolving scope of big data scholarship, we focus on foundational issues for framing and engaging in research. We ground our recommendations in the context of big data created through data sharing across and within the stages of the continuum of the education and training of health professionals. Ultimately, the goal of this statement is to support a culture of trust and quality for big data research to deliver on its promises for health professions education (HPE) and the health of society. Based on expert consensus and review of the literature, we report 19 recommendations in (1) framing scholarship and research through research, (2) considering unique ethical practices, (3) governance of data sharing collaborations that engage stakeholders, (4) data sharing processes best practices, (5) the importance of knowledge translation, and (6) advancing the quality of scholarship through multidisciplinary collaboration. The recommendations were modified and refined based on feedback from the 2022 Ottawa Conference attendees and subsequent public engagement. Adoption of these recommendations can help HPE scholars share data ethically and engage in high impact big data scholarship, which in turn can help the field meet the ultimate goal: high-quality education that leads to high-quality healthcare.

Simulation-based assessment of upper abdominal ultrasound skills.

BACKGROUND: Ultrasound is a safe and effective diagnostic tool used within several specialties. However, the quality of ultrasound scans relies on sufficiently skilled clinician operators. The aim of this study was to explore the validity of automated assessments of upper abdominal ultrasound skills using an ultrasound simulator. METHODS: Twenty five novices and five experts were recruited, all of whom completed an assessment program for the evaluation of upper abdominal ultrasound skills on a virtual reality simulator. The program included five modules that assessed different organ systems using automated simulator metrics. We used Messick's framework to explore the validity evidence of these simulator metrics to determine the contents of a final simulator test. We used the contrasting groups method to establish a pass/fail level for the final simulator test. RESULTS: Thirty seven out of 60 metrics were able to discriminate between novices and experts (p < 0.05). The median simulator score of the final simulator test including the metrics with validity evidence was 26.68% (range: 8.1-40.5%) for novices and 85.1% (range: 56.8-91.9%) for experts. The internal structure was assessed by Cronbach alpha (0.93) and intraclass correlation coefficient (0.89). The pass/fail level was determined to be 50.9%. This pass/fail criterion found no passing novices or failing experts. CONCLUSIONS: This study collected validity evidence for simulation-based assessment of upper abdominal ultrasound examinations, which is the first step toward competency-based training. Future studies may examine how competency-based training in the simulated setting translates into improvements in clinical performances.

Higher SARS-CoV-2 detection of oropharyngeal compared with nasopharyngeal or saliva specimen for molecular testing: a multicentre randomised comparative accuracy study.

BACKGROUND: Testing is critical for detecting SARS-CoV-2 infection, but the best sampling method remains unclear. OBJECTIVES: To determine whether nasopharyngeal swab (NPS), oropharyngeal swab (OPS) or saliva specimen collection has the highest detection rate for SARS-CoV-2 molecular testing. METHODS: We conducted a randomised clinical trial at two COVID-19 outpatient test centres where NPS, OPS and saliva specimens were collected by healthcare workers in different orders for reverse transcriptase PCR testing. The SARS-CoV-2 detection rate was calculated as the number positive by a specific sampling method divided by the number in which any of the three sampling methods was positive. As secondary outcomes, test-related discomfort was measured with an 11-point numeric scale and cost-effectiveness was calculated. RESULTS: Among 23 102 adults completing the trial, 381 (1.65%) were SARS-CoV-2 positive. The SARS-CoV-2 detection rate was higher for OPSs, 78.7% (95% CI 74.3 to 82.7), compared with NPSs, 72.7% (95% CI 67.9 to 77.1) (p=0.049) and compared with saliva sampling, 61.9% (95% CI 56.9 to 66.8) (p<0.001). The discomfort score was highest for NPSs, at 5.76 (SD, 2.52), followed by OPSs, at 3.16 (SD 3.16) and saliva samples, at 1.03 (SD 18.8), p<0.001 between all measurements. Saliva specimens were associated with the lowest cost, and the incremental costs per detected SARS-CoV-2 infection for NPSs and OPSs were US$3258 and US$1832, respectively. CONCLUSIONS: OPSs were associated with higher SARS-CoV-2 detection and lower test-related discomfort than NPSs for SARS-CoV-2 testing. Saliva sampling had the lowest SARS-CoV-2 detection but was the least costly strategy for mass testing. TRIAL REGISTRATION NUMBER: NCT04715607.

Curriculum and assessment tool for less invasive surfactant administration: an international Delphi consensus study.

BACKGROUND: Training and assessment of operator competence for the less invasive surfactant administration (LISA) procedure vary. This study aimed to obtain international expert consensus on LISA training (LISA curriculum (LISA-CUR)) and assessment (LISA assessment tool (LISA-AT)). METHODS: From February to July 2022, an international three-round Delphi process gathered opinions from LISA experts (researchers, curriculum developers, and clinical educators) on a list of items to be included in a LISA-CUR and LISA-AT (Round 1). The experts rated the importance of each item (Round 2). Items supported by more than 80% consensus were included. All experts were asked to approve or reject the final LISA-CUR and LISA-AT (Round 3). RESULTS: A total of 153 experts from 14 countries participated in Round 1, and the response rate for Rounds 2 and 3 was >80%. Round 1 identified 44 items for LISA-CUR and 22 for LISA-AT. Round 2 excluded 15 items for the LISA-CUR and 7 items for the LISA-AT. Round 3 resulted in a strong consensus (99-100%) for the final 29 items for the LISA-CUR and 15 items for the LISA-AT. CONCLUSIONS: This Delphi process established an international consensus on a training curriculum and content evidence for the assessment of LISA competence. IMPACT: This international consensus-based expert statement provides content on a curriculum for the less invasive surfactant administration procedure (LISA-CUR) that may be partnered with existing evidence-based strategies to optimize and standardize LISA training in the future. This international consensus-based expert statement also provides content on an assessment tool for the LISA procedure (LISA-AT) that can help to evaluate competence in LISA operators. The proposed LISA-AT enables standardized, continuous feedback and assessment until achieving proficiency.

Acquisition and usage of robotic surgical data for machine learning analysis.

BACKGROUND: The increasing use of robot-assisted surgery (RAS) has led to the need for new methods of assessing whether new surgeons are qualified to perform RAS, without the resource-demanding process of having expert surgeons do the assessment. Computer-based automation and artificial intelligence (AI) are seen as promising alternatives to expert-based surgical assessment. However, no standard protocols or methods for preparing data and implementing AI are available for clinicians. This may be among the reasons for the impediment to the use of AI in the clinical setting. METHOD: We tested our method on porcine models with both the da Vinci Si and the da Vinci Xi. We sought to capture raw video data from the surgical robots and 3D movement data from the surgeons and prepared the data for the use in AI by a structured guide to acquire and prepare video data using the following steps: 'Capturing image data from the surgical robot', 'Extracting event data', 'Capturing movement data of the surgeon', 'Annotation of image data'. RESULTS: 15 participant (11 novices and 4 experienced) performed 10 different intraabdominal RAS procedures. Using this method we captured 188 videos (94 from the surgical robot, and 94 corresponding movement videos of the surgeons' arms and hands). Event data, movement data, and labels were extracted from the raw material and prepared for use in AI. CONCLUSION: With our described methods, we could collect, prepare, and annotate images, events, and motion data from surgical robotic systems in preparation for its use in AI.

The fundamentals of Artificial Intelligence in medical education research: AMEE Guide No. 156.

The use of Artificial Intelligence (AI) in medical education has the potential to facilitate complicated tasks and improve efficiency. For example, AI could help automate assessment of written responses, or provide feedback on medical image interpretations with excellent reliability. While applications of AI in learning, instruction, and assessment are growing, further exploration is still required. There exist few conceptual or methodological guides for medical educators wishing to evaluate or engage in AI research. In this guide, we aim to: 1) describe practical considerations involved in reading and conducting studies in medical education using AI, 2) define basic terminology and 3) identify which medical education problems and data are ideally-suited for using AI.

Cost of simulation-based mastery learning for abdominal ultrasound.

BACKGROUND: Ultrasound is an essential diagnostic examination used in several medical specialties. However, the quality of ultrasound examinations is dependent on mastery of certain skills, which may be difficult and costly to attain in the clinical setting. This study aimed to explore mastery learning for trainees practicing general abdominal ultrasound using a virtual reality simulator and to evaluate the associated cost per student achieving the mastery learning level. METHODS: Trainees were instructed to train on a virtual reality ultrasound simulator until the attainment of a mastery learning level was established in a previous study. Automated simulator scores were used to track performances during each round of training, and these scores were recorded to determine learning curves. Finally, the costs of the training were evaluated using a micro-costing procedure. RESULTS: Twenty-one out of the 24 trainees managed to attain the predefined mastery level two times consecutively. The trainees completed their training with a median of 2h38min (range: 1h20min-4h30min) using a median of 7 attempts (range: 3-11 attempts) at the simulator test. The cost of training one trainee to the mastery level was estimated to be USD 638. CONCLUSION: Complete trainees can obtain mastery learning levels in general abdominal ultrasound examinations within 3 hours of training in the simulated setting and at an average cost of USD 638 per trainee. Future studies are needed to explore how the cost of simulation-based training is best balanced against the costs of clinical training.

Business as (un)usual: A qualitative study of clerkship experiences during a health crisis.

INTRODUCTION: During a health crisis, hospitals must prioritise activities and resources, which can compromise clerkship-based learning. We explored how health crises affect clinical clerkships using the COVID-19 pandemic as an example. METHODS: In a constructivist qualitative study, we conducted 22 semi-structured interviews with key stakeholders (i.e. medical students and doctors) from two teaching hospitals and 10 different departments. We used thematic analysis to investigate our data and used stakeholder theory as a sensitising concept. RESULTS: We identified three themes: (1) emotional triggers and reactions; (2) negotiation of legitimacy; and (3) building resilience. Our results suggest that the health crisis accentuated already existing problems in clerkships, such as students' feelings of low legitimacy, constant negotiation of roles, inconsistencies navigating rules and regulations and low levels of active participation. Medical students and doctors adapted to the new organisational demands by developing increased resilience. Students responded by reaching out for guidance and acceptance to remain relevant in the clinical clerkships. Doctors developed a behaviour of closing in and focused on managing themselves and their patients. This created tension between these two stakeholder groups. CONCLUSION: A health crisis can critically disrupt the hierarchical structure within the clinical clerkships and exacerbate existing conflicts between stakeholder groups. When medical students are not perceived as legitimate stakeholders in clinical clerkships during a health crisis, their attendance is perceived as unnecessary or even a nuisance. Despite increased student proactiveness and resilience, their roles inevitably shift from being doctors-to-be to students-to-be-managed.

The concept of errors in medical education: a scoping review.

The purpose of this scoping review was to explore how errors are conceptualized in medical education contexts by examining different error perspectives and practices. This review used a scoping methodology with a systematic search strategy to identify relevant studies, written in English, and published before January 2021. Four medical education journals (Medical Education, Advances in Health Science Education, Medical Teacher, and Academic Medicine) and four clinical journals (Journal of the American Medical Association, Journal of General Internal Medicine, Annals of Surgery, and British Medical Journal) were purposively selected. Data extraction was charted according to a data collection form. Of 1505 screened studies, 79 studies were included. Three overarching perspectives were identified: 'understanding errors') (n = 31), 'avoiding errors' (n = 25), 'learning from errors' (n = 23). Studies that aimed at'understanding errors' used qualitative methods (19/31, 61.3%) and took place in the clinical setting (19/31, 61.3%), whereas studies that aimed at 'avoiding errors' and 'learning from errors' used quantitative methods ('avoiding errors': 20/25, 80%, and 'learning from errors': 16/23, 69.6%, p = 0.007) and took place in pre-clinical (14/25, 56%) and simulated settings (10/23, 43.5%), respectively (p < 0.001). The three perspectives differed significantly in terms of inclusion of educational theory: 'Understanding errors' studies 16.1% (5/31),'avoiding errors' studies 48% (12/25), and 'learning from errors' studies 73.9% (17/23), p < 0.001. Errors in medical education and clinical practice are defined differently, which makes comparisons difficult. A uniform understanding is not necessarily a goal but improving transparency and clarity of how errors are currently conceptualized may improve our understanding of when, why, and how to use and learn from errors in the future.

Is two a crowd? A qualitative analysis of dyad learning in an OBGYN clinical clerkship.

INTRODUCTION: Dyad learning occurs when two students work together to acquire new skills and knowledge. Several studies have provided evidence to support the educational rationale for dyad learning in the controlled simulated setting. However, the role of dyad learning in the clinical setting remains uncertain. Unlike the simulated setting, learning in the clinical setting depends on a complex interplay between medical students, doctors, nurses and patients potentially making dyad learning less valuable in clerkships. The objective of this study was to explore how key stakeholders perceive the value of implementing dyad learning during medical students' clinical clerkships. METHODS: In a constructivist qualitative study, we conducted 51 semi-structured interviews with 36 key stakeholders involved in dyad learning, including 10 medical students, 12 doctors, five nurses and nine patients. Data were coded inductively using thematic analysis, then coded deductively using stakeholder theory as a theoretical framework. RESULTS: We found that stakeholders generally perceived the educational impact of dyad learning in the clinical setting similarly but disagreed on its value. Students emphasised that dyad learning made them participate more actively during patient encounters and patients did not mind having two students present. Doctors and nurses considered dyad learning disruptive to the balance between service and training and reported that it did not resonate with their perception of good patient care. CONCLUSION: Dyad learning enables students to be more active during their clinical clerkships, but it easily disrupts the balance between service and training. This disruption may be exacerbated by the shifted balance in priorities and values between different stakeholder groups, as well as by making implicit teaching obligations more explicit for supervising doctors and nurses. Consequently, implementing dyad learning may not be perceived as valuable by doctors and nurses in the clinical setting, regardless of its pedagogical rationale.

The myth of ivory tower versus practice-oriented research: A systematic review of randomised studies in medical education.

CONTEXT: A long-standing myth in medical education research is a divide between two different poles: research aiming to advance theory with little focus on practical applications ('ivory tower' research) and practically oriented research aiming to serve educators and decision-makers with little focus on advancing theory ('in-the-trenches' practice). We explored this myth in a sample of randomised medical education studies using Stokes' four-quadrant framework for the classification of research perspective. METHODS: We searched MEDLINE, Embase, CINAHL, PsychINFO, ERIC, Web of Science and Scopus for studies in medical education using a randomised design that were published between 1 January 2018 and 31 December 2018. We used Stokes' four-quadrant framework to categorise the studies according to their use of theory, concepts and their justification for practical use. We compared medical education research published in medical education journals and clinical journals. RESULTS: A total of 150 randomised studies were included in the analysis. The largest segment of studies (46.7%) was categorised as use-inspired basic research (Pasteur's Quadrant), closely followed by pure applied research (40.7%, Edison's Quadrant). Only a few studies were categorised as aiming to advance knowledge with no thought for practical educational application (2.0%, Bohr's Quadrant). The proportion of studies that included educational concepts and theory differed according to publication in clinical journals or medical education journals: 40.5% vs 71.8%, respectively, P < .001. There were no differences between journals with regard to the proportion of studies that included a practical educational or clinical rationale (P = .99). CONCLUSION: In a large sample of studies using randomised designs, we found no evidence to support the myth that medical education research divides between two singular poles represented by 'ivory tower research' and 'in-the-trenches practice'. We did confirm prevailing assumptions regarding an emphasis on non-theoretical medical education research in clinical journals.

The role of data science and machine learning in Health Professions Education: practical applications, theoretical contributions, and epistemic beliefs.

Data science is an inter-disciplinary field that uses computer-based algorithms and methods to gain insights from large and often complex datasets. Data science, which includes Artificial Intelligence techniques such as Machine Learning (ML), has been credited with the promise to transform Health Professions Education (HPE) by offering approaches to handle big (and often messy) data. To examine this promise, we conducted a critical review to explore: (1) published applications of data science and ML in HPE literature and (2) the potential role of data science and ML in shifting theoretical and epistemological perspectives in HPE research and practice. Existing data science studies in HPE are often not informed by theory, but rather oriented towards developing applications for specific problems, uses, and contexts. The most common areas currently being studied are procedural (e.g., computer-based tutoring or adaptive systems and assessment of technical skills). We found that epistemic beliefs informing the use of data science and ML in HPE poses a challenge for existing views on what constitutes objective knowledge and the role of human subjectivity for instruction and assessment. As a result, criticisms have emerged that the integration of data science in the field of HPE is in danger of becoming technically driven and narrowly focused in its approach to teaching, learning and assessment. Our findings suggest that researchers tend to formalize around the epistemological stance driven largely by traditions of a research paradigm. Future data science studies in HPE need to involve both education scientists and data scientists to ensure mutual advancements in the development of educational theory and practical applications. This may be one of the most important tasks in the integration of data science and ML in HPE research in the years to come.

Social ties between team members affect patient satisfaction: a data-driven approach to handling complex network analyses.

Research from outside the medical field suggests that social ties between team-members influence knowledge sharing, improve coordination, and facilitate task completion. However, the relative importance of social ties among team-members for patient satisfaction remains unknown. In this study, we explored the association between social ties within emergency teams performing simulated caesarean sections (CS) and patient-actor satisfaction. Two hundred seventy-two participants were allocated to 33 teams performing two emergency CSs in a simulated setting. We collected data on social ties between team-members, measured as affective, personal and professional ties. Ties were rated on 5-point Likert scales. In addition, participants' clinical experience, demographic data and their knowledge about team members' roles were surveyed. Perceived patient satisfaction was measured on a 5-point Likert scale. Data was analysed with a linear regression model using elastic net regularization. In total, 109 predictor variables were analysed including 84 related to social ties and 25 related to clinical experience, demographics and knowledge test scores. Of the 84 variables reflecting social ties, 34 (41%) had significant association with patient satisfaction, p < 0.01. By contrast, a significant association with patient satisfaction was found for only one (4%) of the 25 variables reflecting clinical experience, demographics and knowledge of team roles. Affective ties and personal ties were found to be far more important predictors in the statistical model than professional ties and predictors relating to clinical experience. Social ties between emergency team members may be important predictors of patient satisfaction. The results from this study help to enhance our conceptual understanding of social ties and their implications for team-dynamics. Our study challenges existing views of team-performance by placing emphasis on achieving collective competence through affective and personal social ties, rather than focusing on traditional measures of expertise.

Fundamentals of randomized designs: AMEE Guide No. 128.

This AMEE Guide summarizes fundamentals of a major experimental design option for medical education researchers: the randomised study. Medical education researchers face an overwhelming taxonomy of study design options; given the breadth of information on experimental design, the purpose of this Guide is to offer a resource for medical education researchers wishing to equip themselves with helpful information for when to match a study's objective and the use of randomised designs. Once a research question has been formulated study design is the cornerstone of the intricate, nested activities of any research project. Researchers negotiate many decisions in the pursuit of choosing an appropriate design approach; failure to do so can undermine a project's capacity to, for example, sufficiently test a hypothesis or theory. Written as an introduction, this Guide is intended for medical education researchers seeking to build on and synthesise the existing corpus of literature on experimental and quasi-experimental design approaches. While not comprehensive, presented are key concepts alongside relevant examples from the field of health professions education.

Social ties influence teamwork when managing clinical emergencies.

BACKGROUND: Our current understanding of medical team competence is traditionally influenced by an individualistic perspective focusing on individual team members' knowledge, skills as well as on effective communication within the team. However, team dynamics may influence team performance more than previously anticipated. In particular, recent studies in other academic disciplines suggest that social ties between team members may impact team dynamics but this has not been explored for medical teams. We aimed to explore intensive care staff's perceptions about teamwork and performance in clinical emergencies focusing particularly on the teams' social ties. METHODS: Semi-structured interviews were conducted with a purposive sample of intensive care staff. We used a thematic analysis approach to data interpretation. RESULTS: Thematic saturation was achieved after three group interviews and eight individual interviews. Findings demonstrated that social ties influenced teamwork by affecting the teams' ability to co-construct knowledge, coordinate tasks, the need for hierarchy, the degree to which they relied on explicit or implicit communication, as well as their ability to promote adaptive behavior. CONCLUSIONS: Social ties may be an important factor to consider and acknowledge in the design of future team training, as well as for work planning and scheduling of team activities during clinical practice. More research is needed into the causal effect of social ties on team performance and outcome.

Up or down? A randomized trial comparing image orientations during transvaginal ultrasound training.

INTRODUCTION: There is no international consensus on the orientation of transvaginal ultrasound images and no evidence exists to support the superiority of one image orientation over the other. The aim of this study was to compare learning curves and skills transfer in a group of novices randomized to top-down or bottom-up image orientation, and to determine whether individual preferences for image orientation affect learning and skills transfer. MATERIAL AND METHODS: 60 senior medical students, with no prior ultrasound experience, were randomized to orient the image top-down or bottom-up during training on an ultrasound simulator until attaining expert levels of performance. Participants then completed a transfer test involving a systematic ultrasound examination on a physical mannequin using real ultrasound equipment. Performance was assessed during the transfer test by two independent raters using the objective structured assessment of ultrasound skills (OSAUS) score and a global rating score. RESULTS: The bottom-up group reached the expert level with significantly fewer attempts than did the top-down group [median ± interquartile range: 4 ± 2 vs 5 ± 3] (U = 285.5, P = 0.014). The bottom-up group used less time to achieve the expert level (median ± interquartile range: 3 h 2 m ± 1 h 14 m vs 3 h 28 m ± 2 h 21 m) (U = 301.5, P = 0.029). The two groups performed similarly during the transfer test with respect to their OSAUS scores (top-down 56.7% vs bottom-up 53.2%, P = 0.13). The global rating scores were higher in the top-down group (top-down 57.1% vs bottom-up 50.0%, P = 0.02). CONCLUSIONS: Orientation of the images bottom-up rather than top-down, led to a steeper learning curve, but had little or no impact on the subsequent transfer of skills.

The Effects of Simulation-based Transvaginal Ultrasound Training on Quality and Efficiency of Care: A Multicenter Single-blind Randomized Trial.

OBJECTIVE: To explore the effect of adding simulation-based transvaginal ultrasound training to trainees' clinical training compared with only clinical training on quality of and efficiency of care. BACKGROUND: Simulation-based ultrasound training may be an effective adjunct to clinical training, but no studies have examined its effects on quality and efficiency of care. METHODS: Trainees from 4 University Hospitals in East Denmark were included (N = 54). Participants were randomized to either simulation-based ultrasound training and clinical training (intervention group, n = 28), or to clinical training only (control group, n = 26).The primary outcome was patient-reported discomfort during transvaginal ultrasound examinations performed by study participants. Secondary outcomes included patient-reported perceived safety and confidence in ultrasound provider. Finally, the need for trainee supervision or repeated patient examinations was recorded. RESULTS: In total, 1150 patient ratings were collected. The intervention was associated with a reduction of patient discomfort by 18.5% [95% confidence interval (CI), 10.7-25.5; P < 0.001), and with a 7.9% (95% CI, 0.5-14.7; P = 0.04) increase in perceived safety. The intervention group participants received 11.1% (95% CI, 2.5-18.9) higher scores on patients' confidence compared with control group participants (P = 0.01). When the number of days of clinical training was doubled, the odds for trainee supervision or repeated patient examination was reduced by 45.3% (95% CI, 33.5-55.1) and 19.8% (95% CI, 4.1-32.9) in the intervention and control group, respectively (P = 0.005). CONCLUSIONS: Simulation-based ultrasound training improved quality of care and reduced the need for repeated patient examination and trainee supervision.

Imperfect practice makes perfect: error management training improves transfer of learning.

CONTEXT: Traditionally, trainees are instructed to practise with as few errors as possible during simulation-based training. However, transfer of learning may improve if trainees are encouraged to commit errors. The aim of this study was to assess the effects of error management instructions compared with error avoidance instructions during simulation-based ultrasound training. METHODS: Medical students (n = 60) with no prior ultrasound experience were randomised to error management training (EMT) (n = 32) or error avoidance training (EAT) (n = 28). The EMT group was instructed to deliberately make errors during training. The EAT group was instructed to follow the simulator instructions and to commit as few errors as possible. Training consisted of 3 hours of simulation-based ultrasound training focusing on fetal weight estimation. Simulation-based tests were administered before and after training. Transfer tests were performed on real patients 7-10 days after the completion of training. Primary outcomes were transfer test performance scores and diagnostic accuracy. Secondary outcomes included performance scores and diagnostic accuracy during the simulation-based pre- and post-tests. RESULTS: A total of 56 participants completed the study. On the transfer test, EMT group participants attained higher performance scores (mean score: 67.7%, 95% confidence interval [CI]: 62.4-72.9%) than EAT group members (mean score: 51.7%, 95% CI: 45.8-57.6%) (p < 0.001; Cohen's d = 1.1, 95% CI: 0.5-1.7). There was a moderate improvement in diagnostic accuracy in the EMT group compared with the EAT group (16.7%, 95% CI: 10.2-23.3% weight deviation versus 26.6%, 95% CI: 16.5-36.7% weight deviation [p = 0.082; Cohen's d = 0.46, 95% CI: -0.06 to 1.0]). No significant interaction effects between group and performance improvements between the pre- and post-tests were found in either performance scores (p = 0.25) or diagnostic accuracy (p = 0.09). CONCLUSIONS: The provision of error management instructions during simulation-based training improves the transfer of learning to the clinical setting compared with error avoidance instructions. Rather than teaching to avoid errors, the use of errors for learning should be explored further in medical education theory and practice.

Practical trials in medical education: linking theory, practice and decision making.

CONTEXT: Concerns have been raised over the gap between education theory and practice and how research can contribute to inform decision makers on their choices and priorities. Little is known about how educational theories and research outcomes produced under optimal conditions in highly controlled settings generalise to the real-life education context. One way of bridging this gap is applying the concept of practical trials in medical education. In this paper we elaborate on characteristics of practical trials and based on examples from medical education we discuss the challenges, limitations and future directions for this kind of research. CURRENT STATE: Practical trials have the overall aim of informing decision makers. They are carried out in real-life settings and are characterised by (i) comparison of viable alternative education strategies, (ii) broad inclusion criteria regarding participants across several settings and (iii) multiple outcome measures with long-term follow-up to evaluate both benefits and risks. Questions posed by practical trials may be proactive in applying theory in the development of educational innovations or reactive to educational reforms and innovations. Non-inferiority or equivalence designs are recommended when comparing viable alternatives and the use of crossover designs, cluster randomisation or stepped wedge trial designs are feasible when studying implementations across several settings. Outcome measures may include variables related to learners, teachers, educational administration, quality of care, patient outcomes and cost. CONCLUSIONS: Practical trials in medical education may contribute to bridge the gap between education theory and practice and aid decision makers in making evidence-based choices and priorities. Conducting practical trials is not without challenges and rigorous design and methods must be applied. Of concern is that the practical focus may lead to failure to include a sound theoretical basis in the research questions and the interventions studied, and that authors fail to obtain informed consent from their participants.