Using guidelines from Davis et al [14], we developed an EER for third-year medical students with three main goals: (1) to introduce fundamental principles of research methodology and terminology, (2) to promote collaborative problem-solving through teamwork, and (3) to foster more positive attitudes toward research training. The design team included 4 medical doctors, 1 pharmacist, and 1 medical resident, drawing on interdisciplinary expertise in clinical medicine, public health, and pedagogy.

A diagnostic accuracy study was deliberately chosen as the pedagogical framework as this type of research offers an accessible and clinically meaningful entry point for undergraduate students without previous exposure to research methods. It aligns closely with diagnostic reasoning processes familiar to most learners while also serving as a structured introduction to key methodological concepts—such as reference standards, population selection, and diagnostic performance metrics (eg, sensitivity and specificity)—that remain challenging even for many practicing clinicians [5,6]. By anchoring the learning objectives in a framework that is both practical and conceptually rich, this approach facilitates early engagement with evidence-based thinking without requiring advanced statistical background.

Learning objectives were defined a priori based on recurrent misconceptions observed in upper-year students and the collective teaching experience of the faculty involved. By the end of the session and debriefing, students were expected to (1) identify the key components of a diagnostic accuracy study protocol, including relevant methodological tools; (2) recognize the characteristics of an ideal reference standard, including performance, cost, invasiveness, and availability; (3) understand the structure of a study population that includes both individuals with and without the condition of interest; (4) construct and interpret a contingency table from diagnostic test and gold standard results; and (5) critically assess the strengths and limitations of a diagnostic test based on its performance metrics and intended clinical use.

The escape room scenario centered on a fictional outbreak of a zombie virus disease. Students were tasked with identifying an effective diagnostic test to detect infected individuals and locate an antidote. Working in teams, they had 60 minutes to examine the research notes of a mysterious scientist who had nearly discovered a high-performing test. To succeed, students needed to solve 6 sequential puzzles, each aligned with one of the predefined learning objectives (Figure 1).

The six puzzles included (1) classifying keywords relevant to diagnostic study protocols (study protocol), (2) selecting the most appropriate gold standard test from a series of candidates (gold standard), (3) identifying a valid study population using fictional application letters (study population), (4) analyzing mock laboratory test results to construct a contingency table (contingency table), (5) calculating diagnostic accuracy metrics based on the table (test metrics), and (6) interpreting these metrics to choose the best test in context (metric appraisal). The terms in parentheses will be used throughout the manuscript for clarity.

Each completed puzzle provided a clue, object, or code that allowed progression to the next stage of the game. A team succeeded when they completed all 6 puzzles within the allotted time, thereby unlocking the location of the antidote. The scenario was fully autonomous: game masters were available only to provide hints or intervene in case of technical difficulties upon request. The immersive experience included props such as a cryptex, mock test description cards, a book safe, invisible ink, a digital lockbox, a custom-designed computer program, role-play videos, QR codes, a spectrogram decryption smartphone app, and purpose-built board game–style cards (Figure 2). All materials were original creations funded by the Faculty of Medicine at the University of Lille, with a total cost of €1800 (approximately US $1925) for approximately 600 students.

Twelve 120-minute sessions were conducted over the course of the program. Each session included a 15-minute introductory briefing, 60 minutes of gameplay, and 45 minutes of structured debriefing. Sessions were held in a large open space of approximately 150 m², accommodating up to 45 students per session (Figure 3). Upon arrival, students were randomly assigned to teams of 4 to 6 using an algorithm triggered by swiping their university ID card.

Eight teams participated simultaneously in each session seated at individual tables spaced apart to allow for parallel problem-solving without interference. Although certain immersive elements were shared—such as a bookshelf, storage lockers with digital codes, a whiteboard, and a wall-mounted countdown timer—each team engaged independently with dedicated materials and instructions. All essential components were either prepacked in portable game bags (Figure 2) or duplicated in 8 identical sets to ensure autonomous progression. The briefing began with a short immersive video introducing the fictional mission and setting the narrative tone, aiming to engage students from the outset and foster a collaborative atmosphere (Multimedia Appendix 1).

Each session involved 9 facilitators: 1 game master assigned to each team and a session coordinator overseeing the entire room. Facilitators ensured smooth progression and conducted real-time formative assessments. After each puzzle, they rated the team’s clinical reasoning and collaborative dynamics using a tablet-based checklist with Likert scale items. This structure allowed students to work independently while maintaining consistent pedagogical oversight and documentation of performance. In total, 20 facilitators contributed across the sessions, representing a broad range of specialties including public health, radiology, nephrology, biology, biostatistics, pharmacology, pharmacy, pathology, family medicine, pediatrics, anesthesiology, and intensive care. All facilitators completed the escape room before student sessions to ensure fluency with the scenario and educational objectives.

To support scalability, all materials were designed for reuse. Game bags were reorganized after each session to allow for immediate reset. One month before implementation, the scenario was pilot-tested by 30 fourth-year medical students and 6 instructors whose feedback informed refinements to both content and logistics.

This monocentric prospective study was conducted between March 2023 and April 2023 at the Faculty of Medicine at the University of Lille All third-year medical students enrolled in the 2022‐2023 cohort were eligible to participate. The escape room was integrated into the mandatory curriculum as part of a course on research and critical appraisal and was a graded educational activity. Participation was required, and any student who missed a scheduled session was expected to provide a formal justification.

Before the session, participants provided demographic information (age range and sex) and reported previous experience with recreational escape games (“How many times have you participated in a recreational Escape Game?”). They also rated their comfort with critical appraisal of scientific articles using a 5-point Likert scale (“What is your level of comfort with critical appraisal of scientific articles?”) and submitted a free-text response identifying 3 words that best reflected their perception of critical appraisal (“Tell us the first three words that come to mind to describe your perception about critical appraisal of scientific articles”).

They then individually completed a knowledge questionnaire covering the 5 predefined learning objectives of the session. All questions were answered using “true” or “false” and grouped by objective. The same 5 learning objectives were reassessed during the debriefing immediately after the intervention using a different but equivalent set of “true”-or-“false” questions. A score out of 100 was calculated for each objective, and relative gain was defined as the absolute change between pre- and posttest scores divided by the pretest score. The complete list of questions is available in Multimedia Appendix 2. During the debriefing, students also completed an individual postintervention survey. They were asked to provide 3 words summarizing their perception of the escape room experience and to rate both the entertainment value and the pedagogical interest of the session on a scale from 0 to 10. They were again asked to rate their comfort with critical appraisal using the same 5-point Likert scale. In addition, they were asked the following: “Do you think an escape room is a suitable format for an introductory course on the critical appraisal of scientific articles?” Responses were collected using a 5-point Likert scale ranging from “Not at all suitable” to “Absolutely suitable.” Although “suitable” was not formally defined, the proximity of this question to those assessing the activity’s educational and entertainment value likely guided students to interpret it in terms of relevance, clarity, and pedagogical appropriateness. Finally, students were invited to leave open-ended comments about the session, including organizational aspects, perceived strengths, limitations, and suggestions for improvement.

This monocentric prospective study was conducted between March 2023 and April 2023 at the Faculty of Medicine at the University of Lille All third-year medical students enrolled in the 2022‐2023 cohort were eligible to participate. The escape room was integrated into the mandatory curriculum as part of a course on research and critical appraisal and was a graded educational activity. Participation was required, and any student who missed a scheduled session was expected to provide a formal justification.

The study protocol was reviewed and approved by the Institutional Review Board of the University of Lille in February 2023. The IRB determined that formal ethical approval was not required due to the pedagogical nature of the intervention and the exclusive use of anonymized data.

All procedures were conducted in accordance with the ethical standards of the institutional and national research committees and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.

Upon registration on the study platform, students completed a short presession questionnaire and provided individual informed consent, including authorization to analyze their anonymized responses for research purposes. The study also ensured the privacy and confidentiality of all participants. Data were collected and stored on a secure, password-protected server, and identifying details, such as names and email addresses, were immediately deidentified and replaced with unique participant IDs to protect anonymity. No personal health information was collected.

Participants were not compensated for their involvement in the study (Multimedia Appendix 2).

Descriptive statistics were used to summarize participant characteristics and questionnaire responses. Quantitative variables were reported as means and SDs or medians and IQRs depending on their distribution. Categorical variables were expressed as counts and percentages. To assess knowledge gains, pre- and postintervention scores were compared using paired t tests. Associations between participant characteristics (age, sex, previous experience with escape rooms, and self-reported ease with critical appraisal) and relative score improvement were explored using univariate linear regression models. All statistical tests were 2-sided, and a P value of <.05 was considered statistically significant. All analyses were conducted using the R software (version 4.3.1; R Foundation for Statistical Computing).

Students’ postintervention free-text comments were analyzed using the hierarchical classification method by Reinert [21], a lexical clustering approach based on word co-occurrence patterns. The analysis was conducted using the R Interface for Multidimensional Analysis of Texts and Questionnaires [22]. Comments were anonymized and preprocessed to remove low-frequency and noninformative words. The text corpus was segmented, and descending hierarchical classification was conducted to identify stable clusters of vocabulary. Factorial correspondence analysis was used to visualize the associations between lexical classes. The number of clusters retained was selected based on thematic coherence and interpretability. Two researchers (BLG and AH) independently reviewed and labeled each cluster using representative keywords and excerpts. Discrepancies were resolved through consensus. For reporting, all illustrative quotes were translated from French into English by the authors.

Of the 560 eligible third-year medical students, 530 (94.6%) participated in the escape room activity. Students who did not attend the escape room sessions were absent for justified personal or medical reasons. Of the 530 students who participated, 490 (92.5%) completed the preintervention questionnaire. Among them, 66.3% (325/490) were women, and 83.5% (409/490) were aged between 20 and 22 years. Most students reported limited or no previous experience with recreational escape rooms: 42% (206/490) had never participated in one, and 45.9% (225/490) had only done so once or twice. A substantial proportion expressed discomfort with critical appraisal of scientific literature: 18.8% (92/490) reported being “absolutely not comfortable,” and 33% (161/490) reported being “rather not comfortable” with the task (Table 1).

Divided into 96 teams across 12 sessions, students advanced through the escape room with a high degree of consistency. All teams completed the mission within the allotted time, with a mean duration of 53 (SD 4) minutes. Each of the 6 puzzles required approximately 10 minutes to solve. The fastest puzzle, focused on test metrics, was completed in an average of 6 minutes and 12 seconds (SD 2 min, 49 s), whereas the most time-consuming one, involving the construction of a contingency table, took 12 minutes and 21 seconds on average (SD 2 min, 51 s; (Table 2). Team progression was largely synchronous, with most groups working on the same puzzle simultaneously (Figure 4).

A total of 85.3% (452/530) of the students completed the postintervention questionnaire. Student performance improved markedly following the intervention. The average score increased from 62/100 (SD 1) before the session to 82/100 (SD 2) afterward, representing a 32% gain (SD 5%; P<.001). Average subscores improved across all 5 learning objectives: from 58 (SD 3) to 71 (SD 3) for study protocol, from 74 (SD 2) to 89 (SD 4) for gold standard, from 43 (SD 1) to 83 (SD 3) for study population, from 71 (SD 2) to 88 (SD 5) for contingency table, and from 72 (SD 2) to 79 (SD 5) for metric appraisal (P<.001 in all cases; Figure 5). No significant associations were observed between score improvement and student characteristics (sex, age, previous escape room experience, or initial self-reported ease with critical appraisal), as shown in Table 3.

Feedback was strongly favorable. Students rated the session highly in terms of enjoyment (mean 9.1/10, SD 1.1) and educational value (mean 8.2/10, SD 1.5). Most participants (393/452, 86.9%) reported feeling more comfortable with the appraisal of diagnostic accuracy studies after the session. Regarding the overall suitability of escape rooms for introducing research training, 79% (357/452) rated the format as “absolutely suitable” on a 5-point Likert scale (Figure 6). The term “suitable” was not explicitly defined but followed questions about entertainment and educational value, likely guiding interpretation in terms of relevance and clarity.

Students’ 3-word perceptions of research training shifted notably. Before the session, most descriptors were negative or reflected uncertainty (eg, “difficult,” “tedious,” “unknown,” or “complicated”). Afterward, the most frequent terms were markedly more positive—“engaging,” “entertaining,” “original,” and “educational” (Multimedia Appendix 3).

Thematic analysis of students’ free-text comments using the hierarchical classification by Reinert [21] identified 6 primary thematic clusters, reflecting engagement, teamwork, pedagogical value, conceptual understanding, and overall organization (Figure 7 [21]).