Flipped Team Learning Improves Residents’ Clinical Reasoning in Trial

A 12-week randomized controlled trial of 56 general practice residents found that replacing the specified weekly lecture-based teaching strategy with a flipped classroom combined with team-based learning produced substantially higher clinical-reasoning scores than traditional instruction. Participants in the active-learning group averaged 14.68 out of 16 on the Lasater Clinical Judgment Rubric immediately after the intervention, compared with 11.21 among residents receiving lectures. The difference remained evident four weeks later, when the respective means were 13.96 and 10.82; both comparisons had p-values below 0.001. The single-center study, conducted at the First People’s Hospital of Huzhou in China from September 2023 to February 2026, measured educational outcomes among postgraduate residents rather than effects on patient health.

The trial compared two teaching strategies delivered during the same 12-week general practice rotation at First Affiliated Hospital of Huzhou Normal University and the First People’s Hospital of Huzhou. Each group received 24 instructional hours covering six clinical cases: hypertension, community-acquired pneumonia, type 2 diabetes, heart failure, chronic obstructive pulmonary disease and acute kidney injury. The flipped classroom combined with team-based learning, abbreviated FC-TBL, moved foundational instruction into short prerecorded videos and quizzes completed before class. Contact time was then devoted to individual and team readiness tests, team-based application of clinical cases, structured feedback and reflection. The comparison group received instructor-centered lectures on the same content, followed by a clinical demonstration, but no student practice, team activities, readiness tests, application exercise, structured feedback or reflection log.

Residents assigned to FC-TBL improved by 6.79 points on the clinical reasoning rubric from baseline to the immediate post-test, compared with an increase of 3.17 points in the lecture group. The authors calculated an effect size of 2.59 for the difference in improvement and reported post-hoc power above 0.99. The trial had been powered before enrollment for 28 participants per group, based on an assumed moderate between-group effect and a target of 80% power at a two-sided significance level of 0.05. All 56 randomized residents completed the intervention and follow-up, and baseline characteristics—including age, sex, undergraduate education, National Medical Licensing Examination scores and clinical reasoning—did not differ significantly between the groups. The absence of a baseline imbalance and the retention of a group difference at four weeks support the observed association, although the trial did not separate the effects of the flipped classroom and team-based learning components.

The active-learning group also received higher ratings on workplace performance assessments conducted immediately after the intervention. Its mean Mini-CEX score exceeded that of the lecture group in all seven domains: medical interviewing, physical examination, clinical reasoning, organization and efficiency, doctor-patient communication, humanistic care and overall clinical performance. For example, clinical reasoning averaged 8.04 versus 6.07, while overall clinical performance averaged 7.86 versus 6.07 on nine-point scales. The highest Mini-CEX score for the FC-TBL group was 8.04, whereas the highest reported mean for the lecture group was 6.32. Similarly, residents trained with FC-TBL scored higher in every measured Direct Observation of Procedural Skills domain, covering indications and contraindications, informed consent, preparation, technical execution, communication, professionalism and overall procedural proficiency. Overall procedural proficiency averaged 7.86 in the FC-TBL group and 6.04 in the lecture group, while technical execution averaged 7.71 and 5.89, respectively. All reported differences across these Mini-CEX and DOPS measures were significant at p-values below 0.001.

Questionnaires showed a parallel shift in residents’ reported learning experience. Mean teamwork scores were 86.42 in the FC-TBL group and 72.15 in the lecture group immediately after training, while self-directed learning scores were 85.73 and 70.88 and learning-engagement scores were 87.15 and 71.43. These advantages persisted at the four-week assessment: teamwork averaged 84.97 versus 71.06, self-directed learning 84.21 versus 70.14, and learning engagement 85.68 versus 70.92, with all differences reported at p-values below 0.001. Teaching satisfaction was also higher across all four measured dimensions: residents gave the active model mean scores of 88.57 for teaching-mode design, 87.86 for stimulation of learning interest, 86.79 for improvement of the learning process and 89.29 for overall satisfaction, compared with 72.14, 70.54, 71.07 and 73.57 in the lecture group. These results describe the response to the combined instructional strategy; they do not identify which component produced the difference.

The approach carried an immediate cost in perceived pressure. FC-TBL residents reported higher stress immediately after the intervention, averaging 62.34 on a 100-point scale, compared with 54.17 in the lecture group, a difference reported at p < 0.001. By the four-week follow-up, the difference had disappeared: mean stress was 51.28 and 50.96, respectively, with p = 0.892. The trial therefore measured a temporary increase in reported stress associated with the new model rather than evidence of persistent psychological harm. The authors considered orientation for active-learning demands a possible implementation consideration, but that suggestion was an implication of the study rather than an intervention that was separately tested.

Within FC-TBL sessions, the investigators also compared an individual readiness assurance test, or iRAT, with a subsequent team version, the tRAT. Residents first completed ten multiple-choice questions alone, then worked in teams of five or six to discuss the same questions, reach consensus and submit answers with immediate feedback. The individual test averaged 7.82 correct answers, while the team test averaged 9.21, a mean difference of 1.39 with p < 0.001. This within-group comparison showed higher test performance after team discussion, but it did not isolate team discussion from the immediate feedback provided by the scratch-card system. Nor did it determine which part of the combined FC-TBL program was chiefly responsible for the larger improvements observed against lecture-based learning. The result is therefore compatible with the authors’ proposed explanation that discussion and peer explanation can surface misunderstandings, but it is not a component-level test of that mechanism.

The practical distinction lay in how the groups used the same limited contact time. A representative hypertension session began with a 12-minute video and a five-question online quiz completed three to four days before class, followed by an in-class individual readiness test. Teams then analyzed a 65-year-old patient with uncontrolled hypertension, poor medication adherence, type 2 diabetes, leg edema, low potassium and reduced estimated glomerular filtration rate. Their task was to identify key problems, generate a differential diagnosis, revise treatment, plan patient education and arrange follow-up. Team presentations and an instructor-led debriefing then compared their responses with evidence-based guidelines before residents completed a brief structured reflection log. In contrast, the lecture group received the same case content and total contact time through a 90-minute instructor lecture and a 30-minute clinical demonstration, with questions permitted at the end. The comparison therefore tested a complete change in instructional strategy across the 12-week course, not the addition of occasional team activities to a few otherwise unchanged lectures.

Despite the size of the reported differences, several limitations constrain how far the findings can be generalized. The trial involved only 56 residents at one hospital, all of whom completed it, so the results may not transfer to other institutions, larger cohorts or different residency structures. Residents and instructors could not be blinded to their teaching method, although independent assessors were blinded for performance evaluations, leaving some potential for behavioral bias. Follow-up lasted only four weeks, which cannot establish durable changes in clinical skills, independent practice or patient outcomes. The Lasater Clinical Judgment Rubric depends partly on observer ratings and may be more subjective than standardized patient encounters or objective structured examinations, even though two trained blinded assessors scored performances independently. The teaching-satisfaction questionnaire was adapted for this context and may require further psychometric validation. Larger multicenter trials with longer follow-up, comparisons that separate the two instructional components and cost analyses would be needed to establish the model’s broader effectiveness and mechanisms.

Subject of Research: Randomized trial of flipped classroom and team-based learning for general practice residency training

Article Title: Effectiveness of a flipped classroom combined with team-based learning in standardized training of general practice residents: a randomized controlled trial

Article References: Xiao, M., Wang, Y., Gao, Y., Shi, S., & Qian, C. (2026). Effectiveness of a flipped classroom combined with team-based learning in standardized training of general practice residents: a randomized controlled trial. Frontiers in Medicine, 13, Article 1931590. https://doi.org/10.3389/fmed.2026.1931590

Image Credits: AI Generated

DOI: 10.3389/fmed.2026.1931590

Keywords: medical education, clinical reasoning, flipped classroom, team-based learning, residency training, randomized controlled trial, general practice, active learning, workplace assessment, Effectiveness, flipped, classroom