COVID-19 Curriculum Disruptions and Future Preparedness in Iranian Medical Education: A National Mixed-Methods Study

Abstract

Background

COVID-19 disrupted clinical training, assessment, and student engagement in medical education worldwide. We aimed to identify the curriculum challenges faced by Iranian medical education during the pandemic and the modifications that stakeholders consider necessary for future public health emergencies.

Methods

In this sequential exploratory mixed-methods study, a literature review and 14 semi-structured interviews informed a 39-item questionnaire, which was distributed online to students, trainees, and faculty of Iranian medical schools.

Results

The survey yielded 1095 valid responses (mean age 29.8±5.3 years; 61.5% women) from 58 medical schools. The lowest-rated experience items concerned compensation for lost clinical exposure through simulation and virtual patients (1.59±0.90 on a 5-point scale), regular quizzes or short assignments to monitor learning (1.60±0.98), timely feedback on assignments (1.73±1.06), and valid alternatives for assessing practical skills (1.80±1.07), whereas personal access to the internet and equipment scored highest (3.57±1.07). Between-institution differences were statistically significant but small in five of seven domains (η²H=0.018-0.053). All nine forward-looking reform items were endorsed (3.92±1.03 to 4.42±0.71). The assessment domain showed poor reliability, and the seven-domain confirmatory model fitted inadequately.

Conclusions

Difficulty concentrated in the components of training that depend on physical presence and supervision. Recalled experience differed between institutions, although this design cannot attribute that variation to institutional characteristics. Endorsement of reform was near-uniform, but because the items were positively keyed, it indicates the direction of stakeholder priorities rather than their intensity or feasibility. The questionnaire domains served as a reporting framework rather than validated scales and require revision before reuse.

Introduction

The COVID-19 pandemic revealed systemic weaknesses in global medical education. In a brief timeframe, medical schools were mandated to substitute traditional in-person training with emergency remote teaching, halt or alter clinical rotations, rethink examinations, and safeguard students, professors, patients, and healthcare workers from infection risks.^1,2 The implementation of these measures was essential for public health; however, they significantly disrupted the fundamental apprenticeship model of medical education, particularly in bedside instruction, clinical reasoning during patient interactions, procedural training, and the development of professional identity.³

While pandemics are not unprecedented in medical education, COVID-19 was distinct from prior crises in terms of magnitude, duration, and systemic repercussions. Previous outbreaks, including SARS and MERS, have illustrated that pandemics can disrupt clerkships, electives, examinations, and residency preparation.^4–7 COVID-19, however, caused extensive, prolonged disruption, impacting both pre-clinical and clinical teaching concurrently and compelling institutions to rapidly implement digital systems. A global survey of higher education institutions revealed significant regional disparities in the capacity to shift to online education, suggesting that the pandemic’s effects were influenced not only by educational elements but also by infrastructure, resources, governance, and institutional preparedness.^8,9 Consequently, future curricula must encompass disaster and pandemic preparedness, infection prevention, digital health and telemedicine, interprofessional collaboration, ethical decision-making, communication amid uncertainty, and appropriate student involvement in emergency response.¹⁰ Evidence indicates that disaster and pandemic preparedness training may enhance medical students’ knowledge, attitudes, and perceived readiness;¹¹ nevertheless, this training has not been regularly integrated as a fundamental component of undergraduate medical curricula.

Research from Iran during the pandemic identified infrastructural, managerial, educational, and individual barriers to virtual medical education, including limited interaction and challenges adapting practical components to online formats,^12,13 issues with engagement, and the challenge of substituting practical and clinical learning for online formats.^14,15 However, most of this evidence is drawn from single institutions or specific stakeholder groups, limiting national-level conclusions. To our knowledge, no prior Iranian study has combined qualitative theme generation with a nationally distributed, multi-institutional survey spanning the full continuum from pre-clinical students to attending faculty.

Despite increasing attention to COVID-19’s impact on Iranian medical education, important gaps remain. First, the existing literature is predominantly from high-income countries, and evidence from LMICs, where infrastructure, governance, and resource capacity differ substantially, is limited.^3,9 Second, faculty digital unpreparedness has been identified as a pervasive challenge in the region,¹⁶ yet it has received insufficient attention in Iranian curriculum reform discussions. Third, clinical examination and practical assessment adaptations during the pandemic have been internationally documented,¹⁷ but whether Iranian medical universities adopted adequate and equitable compensatory strategies remains unknown. Fourth, the psychological burden on medical students during the pandemic is well-established globally,¹⁸ yet mental health considerations have rarely been formally integrated into curriculum reform proposals in Iran. Fifth, evidence-based best practices for online teaching in health professions education continue to evolve,^19,20 and their applicability to the Iranian national context has not been systematically evaluated.

It is imperative to address these gaps, as future pandemics, disasters, and systemic shocks are likely, and medical schools cannot depend solely on reactive interventions. A comprehensive awareness of curricular deficiencies and necessary revisions can guide the development of a more robust, equitable, and contextually relevant structure of medical education in Iran. Thus, in the current study, we sought to identify the curriculum challenges faced in Iranian medical education during the pandemic era and necessary modifications to enhance the medical curriculum for future public health emergencies, while maintaining educational quality, learner safety, and social accountability.

Material and Methods

We conducted a multi-center mixed-methods study to identify the challenges in the curriculum and necessary modifications in Iranian medical education arising from the COVID-19 pandemic. The research employed a sequential exploratory strategy. During the qualitative phase, published evidence and empirical data from semi-structured interviews were utilized to develop an initial framework of difficulties and reform alternatives. During the quantitative step, these candidate items were structured into a questionnaire and assessed through a nationally adapted survey of qualified stakeholders from Iranian medical schools across the country.

This design was chosen because it addressed both an explanatory inquiry into the curriculum problems encountered during the pandemic and a prioritization inquiry into which reforms should be deemed significant and viable for future implementation. The qualitative component offered contextual depth and item development, and the quantitative component provided a nationwide multi-institutional survey of perceived significance, feasibility, and consensus.

Study Setting and Population

The research was conducted in Iran and focused on medical education during the pivotal period of the COVID-19 pandemic, from March 2020 to March 2022, when medical institutions faced significant interruptions to in-person instruction, clinical rotations, evaluations, and educational administration.

Eligible participants were individuals with firsthand experience in medical education in Iranian medical schools during this timeframe. The target population comprised undergraduate medical students in pre-clinical and clinical stages, interns, residents, fellowship trainees, and attendings engaged in medical education. During the qualitative interview phase, we intentionally targeted faculty members and educational stakeholders with clinical expertise who were directly involved in undergraduate medical training during the pandemic. During the national survey phase, eligibility was broadened to encompass individuals who were medical students, trainees, or faculty members of an Iranian medical school during the pandemic.

Phase I: Literature Review and Preliminary Framework Development

The initial phase involved a review of the national and international literature on medical education during COVID-19 and prior infectious disease pandemics. This was a narrative, not a systematic, review intended to generate candidate items. We searched PubMed, Scopus, Web of Science, and Google Scholar from inception to the end of 2025 using terms combining COVID-19 or pandemic with medical education, curriculum, online or virtual learning, clinical training, and assessment, supplemented by reference-list screening and policy documents from administrators. Two authors (F.R. and M.A.) screened and extracted the material, and no formal quality appraisal was performed. Extracted content was organized into the categories listed below. The extracted data encompassed documented educational obstacles, implemented adjustments, curriculum deficiencies, changes in assessment, digital infrastructure issues, interruptions in clinical training, psychological and ethical issues, and suggested improvements.

Phase II: Qualitative Exploration and Semi-Structured Interviews

During the second phase, we conducted semi-structured interviews with key informants who had firsthand experience in medical education during the COVID-19 pandemic. Participants were chosen through purposive sampling to ensure diversity in educational role, clinical discipline, institutional background, and experience with undergraduate medical education. Faculty participants qualified if they had at least 3 years of teaching or clinical education experience and had engaged in medical education during the pandemic. Sampling continued until thematic saturation was reached. After each interview, the research team updated a structured log comparing newly emergent codes against the accumulated codebook; saturation was judged to be reached when successive interviews yielded no further novel codes. Recurrence of the principal concepts was observed after the twelfth interview, and two additional interviews (13^th and 14th) were conducted and logged with no new codes emerging, confirming saturation.

The interview guide was established based on the literature review and subsequently refined through discussion within the research team. Interviews examined perceived disruptions in theoretical instruction, clinical rotations, bedside education, evaluation, student-faculty interaction, digital infrastructure, student well-being, ethical dilemmas, and proposed curricular improvements for future pandemic readiness. The lead investigator (F.R.) conducted interviews after obtaining informed consent. Each interview lasted roughly 30 to 45 minutes, contingent upon participant availability and the profundity of the topic. Interviews were audio-recorded with participant consent. Field notes were recorded during and immediately following interviews to provide contextual observations and analytical insights. 14 semi-structured interviews were conducted with individuals with direct experience in medical education during the COVID-19 pandemic. The interviewees comprised 11 clinical or educational faculty members and 3 students or graduates engaged in medical education. Faculty participants encompassed pediatrics, obstetrics and gynecology, surgery, cardiology, gastrointestinal, pulmonology, community medicine, infectious diseases, endocrinology, and rheumatology. The interviews generated a framework for curriculum challenges and reform priorities, which was then transformed into the national survey instrument.

Audio recordings were transcribed verbatim, and a first coding framework was established based on literature-derived categories. However, additional codes were inductively incorporated when participants articulated difficulties not encompassed by the initial framework. Coding involved iterative reading, identification of meaning units, assignment of codes, establishment of categories, and refinement of overarching themes. Discrepancies in coding and categorization were reconciled by deliberation within the research team. Interview transcripts were coded by F.R. manually in Persian. Coding was not performed independently in duplicate, and no inter-coder agreement statistic was computed; credibility relied on iterative team discussion and the saturation log described above. Codes were grouped into 9 subthemes within 8 main themes (Table S1).

Phase III: Questionnaire Development and National Survey

The literature study findings and qualitative interviews were synthesized to create a structured questionnaire that targets key curriculum issues and reform goals. Items were categorized into seven domains: theoretical education, clinical and skills-based education, assessment and examinations, digital infrastructure and access, faculty-student interaction, psychological and ethical issues, and future curriculum policy and preparedness.

The final questionnaire comprised demographic items and 39 substantive items across these seven domains. Substantive items were evaluated with five-point Likert-type response alternatives. Anchors varied depending on item content, ranging from “strongly agree” to “strongly disagree”, “very high” to “very low”, or percentage-based categories reflecting the degree of practice occurrence. A concluding open-ended question enabled participants to share further experiences or suggestions. The full wording of all items, their domain assignment, and the scoring convention applied to each are provided in Table S2.

Prior to nationwide dissemination, the draft questionnaire was evaluated by the study team and subject-matter experts in medical education and health policy to examine content relevance, clarity, redundancy, and face validity; these experts contributed content-area expertise rather than formal training in survey design or psychometrics. The review involved three experts in medical education and health policy, in two rounds. Their comments were incorporated by consensus within the study team. No content validity index was calculated, and no formal pilot test was conducted before national distribution; validity and reliability were examined empirically in the full sample (Cronbach’s alpha, exploratory and confirmatory factor analysis; see Data Process and Statistical Analysis and Results). Methodological aspects of scale development, including factor structure, were subsequently examined empirically by the study team through the exploratory and confirmatory factor analyses reported below.

The questionnaire was conducted online via a secure web-based form on the “Porsline” online survey platform (https://porsline.ir/). The survey link was sent via professional, academic, and student networks inside medical schools throughout Iran. We used daily reminder messages for about 2 weeks to increase response rates. Participation was voluntary and anonymous, except for optional contact information respondents could choose to submit. Because recruitment used open distribution channels rather than a defined sampling frame, the size of the invited population could not be determined, and a conventional response rate cannot be calculated. Participation and completion rates are therefore reported in accordance with CHERRIES.²¹ A prospective sample-size or power calculation was not performed, as the study was descriptive and exploratory rather than structured around a single pre-specified hypothesis test. Sample-size adequacy was instead judged against established guidance for the planned multivariable analyses: the complete-case sample available for exploratory and confirmatory factor analysis (n=708) corresponds to a subjects-to-item ratio of approximately 18:1 for the 39-item instrument, above commonly cited minimum ratios for stable factor recovery.

Ethical Considerations

The Ethics Committee of Tehran University of Medical Sciences approved the study, with the ethics code IR.TUMS.SPH.REC.1404.135. All participants were informed of the study’s objectives, the voluntary nature of their participation, confidentiality protocols, and their right to withdraw at any time. Prior to data collection for interviews, written informed consent was secured, and authorization was sought for audio recording. Completion and submission of the online questionnaire constituted informed consent prior to answering the questions. Identification details were not mandatory for participation in the survey. Interview recordings and transcripts were securely preserved and accessible only to the research team. Identifiers were removed prior to analysis, and results were presented in aggregate form to safeguard participant confidentiality. Participants were informed that anonymized excerpts of their interview or free-text responses might be quoted in publications, and their consent (written for interviews, electronic for the survey) included agreement to publication of anonymized responses and direct quotations. Quotations are attributed only by participant group (faculty member or student), and survey comments were translated into English and stripped of identifying details.

Data Process and Statistical Analysis

Data were exported from the online survey platform and screened for completeness, duplicate submissions, logical inconsistencies, and potential spam responses. Duplicate submissions were flagged by the Porsline platform using IP-address and device-fingerprint matching and were removed prior to analysis. Because this method cannot fully distinguish a genuine repeat submission from independent, legitimate submissions made from a shared device (for example, a shared computer laboratory or a household device used by more than one respondent), a small number of true responses may have been misclassified as duplicates; we note this as a limitation of the duplicate-detection procedure. Questionnaires with unusually uniform answers (such as patterned or implausibly uniform responses) or evidence of invalid responding (such as implausibly short completion times defined as less than five minutes) were excluded from the final analysis.

Continuous variables are presented as mean±standard deviation (SD), whereas categorical variables are presented as frequencies and percentages. Items with negative wording were reverse-coded where appropriate so that higher scores consistently represented more favorable educational experiences or stronger endorsement of curriculum reform. Six items (Q24-Q29) offered a “do not know/did not participate” response. These were treated as missing rather than as scale values, and domain scores were computed only for respondents who answered every item within a domain. The denominators for digital infrastructure and access (n=1,008) and for faculty-student interaction (n=1,018) are therefore smaller than the analytic sample. Items Q7-Q12 were displayed only to respondents with clinical-stage experience and were summarised for the 808 respondents to whom they applied. Reliability of the full instrument and of each a priori domain was estimated with Cronbach’s alpha, and item performance within domains was examined through corrected item-total correlations. The dimensional structure of the 39-item instrument was then assessed. Sampling adequacy was verified with the Kaiser-Meyer-Olkin (KMO) measure and Bartlett’s test of sphericity. Factor retention was guided by eigenvalues greater than 1, scree-plot inspection, and Horn’s parallel analysis using 500 simulated datasets. Exploratory factor analysis was performed using principal-axis factoring with direct-oblimin rotation; absolute loadings ≥0.30 were considered meaningful. The pre-specified correlated seven-domain model was evaluated using maximum-likelihood confirmatory factor analysis. Model fit was assessed using χ², CFI, TLI, RMSEA with 90% confidence intervals, and SRMR; standardized loadings <0.40 were considered weak. For institutions with more than 20 respondents (12 of 58 institutions; n=641), between-institution differences in domain and item scores were explored using the Kruskal–Wallis test, with the rank-based eta-squared (η²H) as the effect-size measure, and Cronbach’s alpha was computed separately within each of these institutions (Tables S4–6); item-level P values were adjusted for multiplicity using the Benjamini-Hochberg procedure. The remaining 46 institutions had samples too small to support stable estimation of either group differences or institution-specific reliability and were not compared individually. Analyses were conducted using SPSS v.27 and Python v.3.12. All analyses were descriptive; domain-level P values and confidence intervals are reported without adjustment for multiple comparisons.

Results

Qualitative Findings

Fourteen semi-structured interviews were conducted with clinical and educational faculty members (n=11) and students or graduates active in medical education (n=3). The faculty participants represented a range of clinical and academic disciplines. Recurrence of the principal concepts was observed after the twelfth interview, and two additional interviews were conducted to confirm conceptual saturation.

The first theme was theoretical education disruption. Participants reported an abrupt switch from face-to-face to online education, inconsistent instructor and university preparation, varied teaching quality, and difficulties staying engaged. However, recorded sessions and repeated access to educational content were considered helpful for future courses. Another major subject was clinical education disruption. Clinical abilities were thought to have suffered from fewer clinical rotations, patient contact, bedside teaching, and direct observation and feedback. Participants suggested simulation, virtual patients, telemedicine, scenario-based learning, and instructional videos as supplements to supervised bedside education.

Assessment, digital infrastructure, and faculty-student interaction were among other themes. Participants questioned the authenticity, fairness, and security of online exams and noted the challenge of measuring practical ability remotely. Randomized questions, timed and staged exams, online oral exams, video-based skills assessments, and competency-oriented assessments were suggested. Additionally, both students and institutions reported unequal access to reliable internet, equipment, educational platforms, and technical support. Participants also reported fewer online class questions, discussions, comments, and active engagement.

The remaining themes covered psychological support, professional ethics, student roles in health emergencies, and curricular readiness. Anxiety, academic uncertainty, repeated schedule changes, digital weariness, and academic progression concerns were reported. They stressed systematic psychological-support procedures, flexible scheduling, and educational overload management. There was also confusion about students’ rights, responsibilities, safety, and clinical engagement during public-health situations. Overall, participants supported a formal educational-continuity plan with competency-based education, structured simulation and virtual-patient activities, faculty development, transparent assessment procedures, and clearly defined student participation policies during future pandemics. The themes, subthemes, conceptual codes, and illustrative quotations derived from the interviews are presented in Table S1.

Participants Characteristics

1095 valid replies were recorded and used in the quantitative analysis. The mean age of responders was 29.76±5.33 years (range: 22–63 years). Among the respondents, 673 were female (61.5%), and 422 were male (38.5%). The majority of individuals were single (n=743, 67.9%), whilst 352 (32.1%) were married. Participants were from 58 universities throughout Iran. The predominant groups originated from Tehran University of Medical Sciences (n=159, 14.5%), Shahid Beheshti University of Medical Sciences (n=73, 6.7%), Iran University of Medical Sciences (n=64, 5.8%), Shiraz University of Medical Sciences (n=61, 5.6%), and Tabriz University of Medical Sciences (n=58, 5.3%). During the pandemic, reported educational or professional roles comprised basic sciences students (n=255, 23.3%), physiopathology students (n=309, 28.2%), clerkship students (n=409, 37.4%), interns (n=416, 38.0%), residents (n=80, 7.3%), fellows (n=7, 0.6%), and attending faculty members (n=44, 4.0%) (Table 1). Participants were permitted to report multiple roles during the pandemic; role categories were not considered mutually exclusive. Figure 1 demonstrates the participants’ flow through the national survey.

Table 1.

National Participants’ Characteristics

Variable	Subgroup	n (%) or Mean±SD (range)
Age (range)		29.76±5.33(22–63)
Sex	Female	673 (61.5)
	Male	422 (38.5)
Marital status	Single	743 (67.9)
	Married	352 (32.1)
Educational Role	Basic Sciences Student	255 (23.3)
	Physiopathology Student	309 (28.2)
	Clerkship	409 (37.4)
	Internship	416 (38.0)
	Residency	80 (7.3)
	Fellowship	7 (0.6)
	Attending	44 (4.0)
Universities represented		58
Open-ended question respondents		131 (12.0)

Notes: Educational roles can overlap due to the time span of the COVID-19 pandemic.

Figure 1.

Flow of participants through the national survey.

One hundred thirty-one respondents submitted an optional open-ended final question in the survey. These remarks were preserved for qualitative contextualization of survey results and are described in Table S3. The remarks pertained to experiences with online education, clinical exposure, examination integrity, educational workload, readiness for future disruptions, and the questionnaire format.

Questionnaire Reliability and Internal Consistency

The full 39-item instrument showed acceptable internal consistency among the 708 participants with complete responses to all items, with Cronbach’s alpha coefficients of 0.821 for raw scores and 0.817 for standardized scores. Deleting individual items did not materially improve the overall coefficient; the largest increase occurred after deletion of Q14 (α=0.827). Domain-level analyses used complete cases available within each domain. Future curriculum policy and preparedness showed the strongest internal consistency (α=0.822; n=1,095), followed by faculty-student interaction (α=0.760; n=1,018), psychological and ethical considerations (α=0.743; n=1,095), and theoretical education (α=0.739; n=1,095). Clinical and skills-based education (α=0.681; n=808) and digital infrastructure and access (α=0.679; n=1,008) showed modest internal consistency, whereas assessment and examinations performed poorly (α=0.435; n=1,095). Q6 had a corrected item-total correlation of 0.105 within the theoretical-education domain, and its deletion increased the coefficient from 0.739 to 0.803. Q19 had a negative corrected item-total correlation within the assessment domain (−0.119), and its deletion increased the coefficient from 0.435 to 0.565. Q14-Q18 also had corrected item-total correlations below 0.30 within the assessment domain.

The complete-case correlation matrix was suitable for factor analysis (KMO=0.861; Bartlett’s test of sphericity χ²(741)=8566.5, P<0.001). Although ten eigenvalues exceeded 1.0, parallel analysis using 500 normally distributed simulated datasets supported retention of seven factors. The first seven observed eigenvalues exceeded both the corresponding simulated means and 95^th percentiles. The eighth observed eigenvalue did not exceed either reference value (observed=1.132; simulated mean=1.233; simulated 95^th percentile=1.256). The seven-factor principal-axis solution with direct-oblimin rotation accounted for 40.4% of the total variance before rotation.

The first factor was principally defined by theoretical-education items Q1-Q5; Q6 did not load meaningfully on any factor. The second factor combined all psychological, ethical, and future-policy items (Q31-Q39). The remaining factors principally represented remote-examination format and misconduct (Q14, Q15, and Q20), interactive teaching and faculty-student interaction (Q25-Q30), digital infrastructure and access (Q21-Q23), disruption of clinical education (Q7, Q8, Q10, and Q12), and compensatory approaches to reduced clinical exposure (principally Q9 and Q11, with weak alignment of Q18). Several items showed weak or structurally inconsistent loadings, particularly Q6, Q13, Q16-Q19, and Q24.

The pre-specified correlated seven-domain confirmatory model showed inadequate overall fit (χ²(681)=2426.5, P<0.001; CFI=0.782; TLI=0.762; RMSEA=0.060, 90% CI 0.058–0.063; SRMR=0.089). Seven items had absolute standardized loadings below 0.40 on their designated factors: Q6, Q13, Q16-Q19, and Q24. Thus, the empirical structure only partially supported the original content-domain framework, with the greatest inconsistency occurring in the assessment and examinations domain.

Perception of Challenges, Future Policies and Preparedness

Item-level descriptive statistics are presented in Table 2. Among the items describing experience during the pandemic, the lowest means concerned compensation for lost clinical exposure: effective use of simulation-based training and virtual patients (Q9, 1.59±0.90) and the use of valid alternative methods for assessing practical skills, such as online OSCEs or submitted procedural videos (Q18, 1.80±1.07). Routine formative practices were rated similarly low, including regular quizzes or short assignments to monitor learning (Q27, 1.60±0.98) and timely, constructive feedback on assignments (Q28, 1.73±1.06). The highest-rated experience item was personal access to the internet and to the equipment needed for online classes (Q23, 3.57±1.07), followed by the perceived comparability of examination difficulty with pre-pandemic examinations (Q17, 3.14±1.08) and the existence of a plan for continuing theoretical teaching (Q1, 3.11±1.14).

Table 2.

Descriptive Characteristics of Survey Items

Number	Domain	Item	N	Mean ± SD
1	Theoretical education	During the COVID-19 pandemic, there was a clear plan for the continuation of theoretical classes.	1095	3.11 ± 1.14
2	Theoretical education	The quality of theoretical content delivered through online classes was acceptable.	1095	2.77 ± 1.10
3	Theoretical education	The intended learning objectives of theoretical courses were adequately covered according to the curriculum.	1095	2.89 ± 1.08
4	Theoretical education	Updated educational resources appropriate for the pandemic conditions were provided.	1095	2.84 ± 1.11
5	Theoretical education	What percentage of scheduled classes were conducted live online (rather than using pre-recorded materials)?	1095	2.93 ± 1.34
6	Theoretical education	During the COVID-19 pandemic, my stress and anxiety related to academic and educational issues were high.	1095	2.73 ± 1.22
7	Clinical/skills-based education	Adequate opportunities were provided for participation in clinical rotations appropriate to my level of training.	808	2.38 ± 1.13
8	Clinical/skills-based education	Bedside learning and participation in teaching rounds noticeably decreased during the pandemic.	808	1.85 ± 0.95
9	Clinical/skills-based education	To compensate for reduced clinical exposure, to what extent were simulation-based training and virtual patients used effectively?	808	1.59 ± 0.90
10	Clinical/skills-based education	During the COVID-19 pandemic, education on common non-COVID diseases received less attention in clinical settings.	808	2.06 ± 1.03
11	Clinical/skills-based education	Telemedicine was used as a structured educational opportunity for clinical training.	808	2.44 ± 1.09
12	Clinical/skills-based education	Despite virtual education, the effectiveness of my learning from courses and clinical rotations met my expectations.	808	2.54 ± 1.08
13	Assessment and examinations	Online assessment methods were transparent and fair.	1095	2.36 ± 1.06
14	Assessment and examinations	What percentage of online examinations were conducted remotely from students’ homes?	1095	2.21 ± 1.47
15	Assessment and examinations	What percentage of online examinations consisted of multiple-choice questions?	1095	1.59 ± 1.09
16	Assessment and examinations	The difficulty level of examinations conducted during the pandemic was consistent with the material that had been taught.	1095	3.06 ± 1.12
17	Assessment and examinations	The difficulty level of examinations conducted during the pandemic was comparable to similar examinations administered before the pandemic.	1095	3.14 ± 1.08
18	Assessment and examinations	To what extent were valid alternative methods (eg, online OSCEs or submission of procedural videos) used to assess practical skills?	1095	1.80 ± 1.07
19	Assessment and examinations	In what percentage of examinations were measures implemented to prevent or reduce academic misconduct (eg, randomized questions, randomized answer choices, disabling backtracking)?	1095	3.06 ± 1.52
20	Assessment and examinations	Despite implementing various preventive measures, cheating and academic misconduct during examinations were more common than under normal circumstances.	1095	1.81 ± 0.97
21	Digital infrastructure and access	The university’s technical infrastructure (platforms, servers, and technical support) was adequate for online teaching.	1095	2.85 ± 1.17
22	Digital infrastructure and access	Internet connectivity (stability and reliability) was acceptable during most online sessions.	1095	2.93 ± 1.14
23	Digital infrastructure and access	I had sufficient personal access to the internet and the necessary equipment to participate effectively in online classes.	1095	3.57 ± 1.07
24	Digital infrastructure and access	In what percentage of classes were sessions systematically recorded and made available for later access?	1025	2.35 ± 1.41
25	Digital infrastructure and access	In what percentage of classes were interactive tools (eg, chat, discussion rooms, polls) used effectively?	1039	2.27 ± 1.27
26	Faculty-student interaction	For what proportion of online classes was there a clear opportunity for questions and answers during or after the session?	1050	2.29 ± 1.32
27	Faculty-student interaction	For what proportion of classes were regular quizzes or short assignments used to monitor learning?	1046	1.60 ± 0.98
28	Faculty-student interaction	For what proportion of assignments did instructors provide timely and constructive feedback?	1038	1.73 ± 1.06
29	Faculty-student interaction	What percentage of students attended online classes and participated actively?	1052	2.47 ± 1.26
30	Faculty-student interaction	The quality of faculty–student interaction (encouraging participation, responding to questions, etc) was satisfactory.	1095	2.64 ± 1.00
31	Psychological and ethical issues	How necessary do you consider incorporating structured psychological support protocols (eg, screening, referral, group sessions) into future curricula for pandemic preparedness?	1095	3.92 ± 1.03
32	Psychological and ethical issues	During future pandemics, how necessary is flexible academic scheduling and workload management to reduce students’ psychological stress?	1095	4.29 ± 0.81
33	Psychological and ethical issues	Professional ethics topics related to pandemics (eg, fairness in resource allocation, patient dignity, privacy in online education) should be systematically incorporated into the curriculum.	1095	4.19 ± 0.83
34	Psychological and ethical issues	Standards for mental health support and supportive faculty–student interactions should be developed and taught.	1095	4.34 ± 0.73
35	Future curriculum policy and preparedness	A formal faculty- and university-level plan is necessary to ensure educational continuity during future pandemics.	1095	4.42 ± 0.71
36	Future curriculum policy and preparedness	To maintain curricular balance, the curriculum should ensure adequate coverage of common non-pandemic diseases.	1095	4.41 ± 0.69
37	Future curriculum policy and preparedness	Competency-based education (emphasizing the acquisition and demonstration of competencies rather than simply completing coursework) would improve students’ preparedness for future pandemics.	1095	4.29 ± 0.76
38	Future curriculum policy and preparedness	Clear policies defining students’ roles during public health emergencies and pandemics (responsibilities, rights, and support) should be developed and communicated in advance.	1095	4.34 ± 0.73
39	Future curriculum policy and preparedness	How necessary is the structured integration of simulation-based education and virtual patients as a permanent component of the curriculum during disruptions to clinical education?	1095	4.22 ± 0.86

Notes: Items are abbreviated. Full wording, domain assignment and scoring conventions, including reverse-scored items, are given in Table S2.

Dispersion was widest for items asking respondents to estimate how often an institutional practice occurred rather than to express a judgement: the proportion of examinations in which measures against academic misconduct were applied (Q19, SD 1.52), the proportion of examinations taken remotely from home (Q14, SD 1.47), the proportion of sessions recorded and made available afterwards (Q24, SD 1.41), the proportion of classes delivered live rather than pre-recorded (Q5, SD 1.34), and the availability of a defined question-and-answer mechanism (Q26, SD 1.32). This pattern is consistent with the psychometric findings reported above, in which items using percentage-based response bands aligned poorly with agreement items placed in the same content domain.

Domain scores stratified by institution are presented in Table 3. Among the twelve institutions contributing more than 20 respondents, mean theoretical education scores ranged from 14.13±3.95 at Gilan University of Medical Sciences (2.36 per item) to 19.10±4.39 at Tehran University of Medical Sciences (3.18). Clinical and skills-based education ranged from 11.65±3.18 at Urmia University of Medical Sciences (1.94) to 14.05±3.82 at Tehran University of Medical Sciences (2.34). Assessment and examinations ranged from 17.38±4.09 at Islamic Azad University of Tehran (2.17) to 20.61±5.02 at Isfahan University of Medical Sciences (2.58). Digital infrastructure and access ranged from 11.29±3.12 at Gilan University of Medical Sciences (2.26 per item) to 15.15±3.79 at Shiraz University of Medical Sciences (3.03), and faculty-student interaction ranged from 7.95±2.92 at Gilan University of Medical Sciences (1.59) to 12.27±4.44 at Shiraz University of Medical Sciences (2.45). In each of these five experience domains, the 95% CIs of the highest- and lowest-scoring institutions did not overlap, although for clinical and skills-based education the separation was marginal.

Table 3.

Scores Stratified by Universities

University	N	Theoretical Education	Clinical Education	Assessment and Examination	Digital Infrastructure and Access	Faculty-Student Interaction	Psychology and Ethics	Future Curriculum Policy and Preparedness
Tehran	159	19.10±4.39 [18.41-19.79]	14.05±3.82 [13.32-14.77]	19.58±4.42 [18.89-20.28]	15.02±3.65 [14.42-15.62]	11.60±4.26 [10.90-12.31]	16.31±2.64 [15.89-16.72]	21.36±2.87 [20.91-21.81]
Shahid Beheshti	73	16.78±4.48 [15.74-17.83]	12.28±3.48 [11.38-13.18]	18.53±4.01 [17.60-19.47]	14.14±3.85 [13.23-15.05]	10.19±3.25 [9.41-10.97]	16.85±2.65 [16.23-17.47]	21.67±3.10 [20.95-22.39]
Iran	64	16.98±4.67 [15.82-18.15]	12.33±4.31 [11.19-13.48]	19.14±4.22 [18.09-20.19]	14.59±4.13 [13.50-15.67]	10.93±4.25 [9.81-12.05]	17.25±2.80 [16.55-17.95]	22.09±3.29 [21.27-22.92]
Shiraz	61	18.48±4.56 [17.31-19.64]	13.56±3.95 [12.34-14.77]	19.51±4.12 [18.45-20.56]	15.15±3.79 [14.17-16.14]	12.27±4.44 [11.11-13.43]	16.75±2.65 [16.08-17.43]	21.56±2.98 [20.80-22.32]
Tabriz	58	16.43±4.08 [15.36-17.50]	12.26±3.88 [10.99-13.54]	17.64±3.90 [16.61-18.66]	12.47±3.57 [11.51-13.44]	8.67±3.10 [7.84-9.49]	16.29±2.97 [15.51-17.07]	21.72±3.00 [20.94-22.51]
Azad Tehran	48	18.69±4.26 [17.45-19.93]	12.08±3.79 [10.52-13.64]	17.38±4.09 [16.19-18.56]	13.74±3.90 [12.58-14.90]	10.72±4.14 [9.49-11.95]	17.06±2.37 [16.37-17.75]	22.19±2.51 [21.46-22.92]
Mashhad	47	17.28±4.38 [15.99-18.56]	12.71±3.90 [11.50-13.93]	18.89±3.47 [17.88-19.91]	14.80±4.37 [13.43-16.18]	10.44±3.74 [9.32-11.57]	16.77±2.46 [16.04-17.49]	21.77±3.10 [20.86-22.68]
Isfahan	38	17.53±5.02 [15.88-19.18]	11.80±3.89 [10.35-13.25]	20.61±5.02 [18.95-22.26]	14.32±4.13 [12.88-15.77]	10.92±4.97 [9.23-12.60]	17.39±2.52 [16.57-18.22]	22.50±2.87 [21.56-23.44]
Mazandaran	26	17.85±3.77 [16.32-19.37]	11.76±2.32 [10.71-12.82]	19.77±3.49 [18.36-21.18]	13.22±4.36 [11.33-15.10]	11.00±5.06 [8.81-13.19]	17.46±1.92 [16.68-18.24]	22.73±2.27 [21.81-23.65]
Gilan	23	14.13±3.95 [12.42-15.84]	11.94±2.30 [10.76-13.13]	19.17±4.22 [17.35-21.00]	11.29±3.12 [9.87-12.70]	7.95±2.92 [6.62-9.28]	16.35±3.35 [14.90-17.80]	22.17±2.55 [21.07-23.28]
Urmia	22	15.68±4.62 [13.63-17.73]	11.65±3.18 [10.01-13.28]	19.95±4.43 [17.99-21.92]	13.00±3.35 [11.39-14.61]	10.35±3.03 [8.93-11.77]	16.23±3.05 [14.87-17.58]	21.64±3.13 [20.25-23.02]
Golestan	22	18.00±4.05 [16.21-19.79]	13.41±2.43 [12.16-14.66]	19.50±3.81 [17.81-21.19]	11.95±4.81 [9.76-14.14]	9.86±3.14 [8.47–11.25]	16.05±2.55 [14.91-17.18]	21.14±2.44 [20.06-22.22]
Others	454	16.67±4.66 [16.24-17.10]	12.97±3.96 [12.54-13.40]	18.96±4.34 [18.56-19.36]	13.84±4.09 [13.44-14.23]	10.80±3.97 [10.42–11.19]	16.83±2.42 [16.61-17.05]	21.58±2.82 [21.32-21.84]
Total	1095	17.26±4.62 [16.99-17.54]	12.86±3.85 [12.60-13.13]	19.03±4.27 [18.78-19.28]	14.01±4.03 [13.76-14.26]	10.75±4.05 [10.50–11.00]	16.75±2.57 [16.60-16.90]	21.69±2.88 [21.52-21.86]

Notes: All values are presented as mean±SD [95% CI]. Universities are ordered by number of respondents; those contributing more than 20 respondents are listed individually. “Others” pools the 46 institutions contributing 20 or fewer respondents together with 2 respondents who did not specify an institution, and is presented for completeness rather than as a comparison group. N is the total number of respondents per row. Three domains have smaller denominators. Clinical education items were displayed only to respondents with clinical-stage experience (Tehran 110, Shahid Beheshti 60, Iran 57, Shiraz 43, Tabriz 38, Azad Tehran 25, Mashhad 42, Isfahan 30, Mazandaran 21, Gilan 17, Urmia 17, Golestan 17, Others 331, Total 808). Six items offered a “do not know/did not participate” response that was treated as missing, so digital infrastructure and access (Tehran 147, Shahid Beheshti 71, Iran 58, Shiraz 59, Tabriz 55, Azad Tehran 46, Mashhad 41, Isfahan 34, Mazandaran 23, Gilan 21, Urmia 19, Golestan 21, Others 413, Total 1008) and faculty–student interaction (Tehran 144, Shahid Beheshti 69, Iran 58, Shiraz 59, Tabriz 57, Azad Tehran 46, Mashhad 45, Isfahan 36, Mazandaran 23, Gilan 21, Urmia 20, Golestan 22, Others 418, Total 1018) are based on complete responders within each domain. Confidence intervals are descriptive, and no adjustment for multiple comparisons was applied.

Restricting the comparison to the 12 institutions with more than 20 respondents, Kruskal–Wallis tests indicated statistically significant but small-magnitude between-institution differences in five of seven domains (Theoretical Education, Clinical Education, Assessment, Digital Access, and Faculty-Student Interaction; all P<0.05, η²H=0.018-0.053; Table S4), with no significant difference for the Psychological Impact or Policy domains (both P>0.10). These effect sizes indicate that institutional membership accounted for approximately 2–5% of the variance in ranked domain scores, a real but modest source of heterogeneity relative to within-institution variability. At item level, 20 of 39 items differed across institutions at P<0.05 and 17 after Benjamini–Hochberg adjustment (η²H up to 0.076); among the nine forward-looking items, only Q32 remained significant after adjustment (Table S4). Domain-level Cronbach’s alpha computed separately within these 12 institutions ranged widely (for instance, 0.15–0.76 for Clinical Education and 0.10–0.55 for Assessment) without systematic improvement over the pooled estimates, a pattern consistent with sampling variability at these sample sizes rather than a genuine institution-specific reliability advantage.

The two forward-looking domains behaved differently from the five experience domains. Psychological and ethical preparedness averaged 16.75±2.57 of 20 (4.19 per item) and future curriculum policy and preparedness averaged 21.69±2.88 of 25 (4.34 per item). Institutional means were closely grouped in both, from 16.05±2.55 at Golestan University of Medical Sciences to 17.46±1.92 at Mazandaran University of Medical Sciences for psychological and ethical preparedness, and from 21.14±2.44 at Golestan University of Medical Sciences to 22.73±2.27 at Mazandaran University of Medical Sciences for curriculum policy and preparedness, with overlapping confidence intervals between the extreme institutions in each case. All nine constituent items had means above 3.9 of 5 (Table 2), the highest being the need for a formal faculty- and university-level plan to ensure educational continuity (Q35, 4.42±0.71) and the need to preserve adequate curricular coverage of common non-pandemic conditions (Q36, 4.41±0.69).

Discussion

This national mixed-methods study documented how students, trainees, and faculty across Iranian medical schools retrospectively perceived the educational disruption of the COVID-19 pandemic, and which reforms they considered necessary for future emergencies. The lowest ratings clustered in areas requiring physical presence and supervision, particularly compensatory strategies for lost clinical exposure, alternative assessment of practical skills, and formative interaction with teachers. Ratings of institutional provision varied across universities, although the design does not permit that variation to be attributed to institutional characteristics.

An important methodological finding concerned the instrument itself. Although the seven content domains were derived from the literature and from interviews with stakeholders, the empirical structure only partially reproduced them. Internal consistency was adequate for six domains but poor for assessment and examinations, and the confirmatory model showed inadequate fit. Two features of the item set appear to explain this. First, the assessment domain combined conceptually distinct constructs, since perceptions of fairness and difficulty were pooled with reports of examination format and frequency of misconduct-prevention measures, which need not covary within an institution. A university that used home-based multiple-choice examinations with randomized items may have been perceived as either fair or unfair, so agreement items and frequency items were not measuring a common latent variable. Second, several items mixed response metrics, combining Likert agreement with percentage bands, which weakens covariance among items in the same domain. These findings suggest that the domains function better as an organizing framework for reporting than as psychometric scales, and that a revised instrument should separate perceptual judgments from reports of educational practice. This is not an isolated problem. Questionnaire development in medical education has long been described as variable in rigor, and reviews of instruments used to evaluate learning environments have found that psychometric evaluation is frequently incomplete or methodologically weak.^22,23

Given this measurement performance, the findings below are interpreted primarily at item level, and domain scores are treated as descriptive summaries rather than as validated scale scores. Endorsement of the forward-looking items was uniformly high, with item means ranging from 3.92 to 4.42 out of 5 and comparatively small dispersion. Respondents therefore did not treat pandemic preparedness as a marginal addition to the curriculum. Two cautions apply to this pattern. All nine reform items were positively keyed, which invites acquiescent responding, and their content is difficult to disagree with in the abstract, since few respondents would oppose mental health support or educational continuity planning in principle. The small variance across institutions is also partly a ceiling artifact, because items with means above 4.2 have limited room to differ. The strength of this endorsement should therefore be read as evidence of direction rather than of intensity or of feasibility, which this study did not measure. Support for this interpretation comes from the free-text responses themselves. Among the 131 optional comments, at least one respondent explicitly characterized the final section of the questionnaire as leading and observed that most people would agree with its statements. Another noted the likelihood of recall bias. Incorporating such comments rather than setting them aside is consistent with the exploratory purpose of the open-ended item, and it tempers the interpretation of the reform subscales.

The findings are consistent with international reports showing that COVID-19 disrupts the apprenticeship model of medical education by limiting bedside teaching, clinical rotations, procedural exposure, and patient-based learning.² Rose describes the early pandemic as a moment in which medical schools must balance learner safety, patient safety, and educational continuity.²⁴ This tension is also visible in the present findings. Similarly, Lucey and Johnston argue that the pandemic accelerates changes that had already been emerging in medical education, including digital learning, flexible progression, and new approaches to clinical exposure.²⁵ Our results extend these observations by showing how these disruptions appear across a national medical education system rather than within a single institution or isolated program.

The results also align with the literature on e-learning and digital transformation. Systematic reviews and cross-sectional studies report that online learning helps preserve educational continuity but does not fully replace clinical skill acquisition, bedside interaction, or professional socialization.^26,27 Iranian studies similarly describe infrastructural limitations, variable readiness, and reduced engagement during virtual education.¹⁵ The present study supports these findings but adds that digital infrastructure and faculty-student interaction vary between universities, suggesting that national policy alone is insufficient unless institutional implementation capacity is also addressed.

Some studies present a more favorable view of post-pandemic e-learning. For example, research from Iran shows that some medical students express willingness to continue e-learning after the pandemic, particularly when satisfaction, usability, and information quality are adequate.²⁸ Studies on online assessment also suggest that remote examinations can be implemented with careful planning and technical support.^29,30 These findings do not necessarily contradict the current results. Rather, they show that digital education can be acceptable and useful when designed deliberately. The differences probably reflect context, infrastructure, assessment stakes, faculty preparation, and whether online tools are used as emergency substitutes or as planned components of a blended curriculum.

One reading of the inter-institutional pattern is that the pandemic functioned as a stress test of pre-existing educational capacity. This interpretation should be advanced cautiously. Perceptions of infrastructure and perceptions of educational continuity were reported by the same individuals within the same instrument, so their association is subject to common-method bias, and no objective institutional data on platforms, bandwidth, faculty development, or class size were collected. What the data support is narrower: respondents at different institutions recalled different educational experiences of the same national event, and the domains most affected were those requiring physical presence and supervision. Theoretical content transferred to remote delivery more readily than clinical competence, which depends on patient availability, supervision, infection-control policy, and repeated opportunities for observation and graded responsibility.

The high endorsement of future curriculum policy and preparedness indicates that stakeholders do not view pandemic reform as a temporary response. They appear to recognize preparedness as a core educational outcome. This interpretation is supported by the literature, which shows that disaster and pandemic preparedness training can improve medical students’ knowledge, attitudes, and perceived readiness.¹¹ It is also compatible with broader calls for health professions education to include social accountability, digital health, interprofessional collaboration, ethics, and adaptive competency-based training.¹⁰

Several stakeholder-informed priorities follow from these findings, offered as candidates for curriculum planning rather than as interventions whose effectiveness this study demonstrates. We link each to the items that support it and do not rank them, because respondents were not asked to rank or cost the measures and all nine reform items fell within a narrow band (3.92–4.42). First, the highest endorsement was for a formal faculty- and university-level continuity plan (Q35, 4.42±0.71), alongside clear policies defining student roles in emergencies (Q38, 4.34±0.73) and systematic inclusion of pandemic-related ethics (Q33, 4.19±0.83); together, these point toward a structured preparedness strand with defined learner participation. Second, alternative assessment of practical skills, regular formative quizzes, and timely feedback were among the lowest-rated experience items (Q18, 1.80±1.07; Q27, 1.60±0.98; Q28, 1.73±1.06), which suggests a case for explicit minimum expectations for formative interaction and practical-skills assessment, although the survey did not ask about national standards as such. Third, simulation and virtual patients were rarely used effectively in practice (Q9, 1.59±0.90) and were endorsed as a permanent component of the curriculum during disruptions to clinical education (Q39, 4.22±0.86); the survey did not compare them with other options.^31,32 Fourth, endorsement of mental-health support was high for developing standards and flexible scheduling (Q34, 4.34±0.73; Q32, 4.29±0.81), but structured psychological support protocols received the lowest mean of the nine reform items (Q31, 3.92±1.03), so support for specific mechanisms appears less uniform than support for the principle. Because all nine items were positively keyed, these means indicate direction rather than relative priority.^33,34 Whether any of these measures can be implemented, and at what cost, was not evaluated.

Strengths, Limitations, and Future Recommendations

The study has several strengths. It draws on 1095 respondents from 58 medical schools. It spans the full range of educational roles, from pre-clinical students to attending faculty, which is unusual in the Iranian literature on pandemic-era medical education and allows the national picture to be described rather than inferred from single institutions. The instrument was not imported but built from stakeholder interviews, so the items reflect concerns raised by the people who experienced the disruption. Full item wording, domain assignment, and scoring conventions are reported (Table S2), participant flow is documented in accordance with CHERRIES,²¹ and the psychometric behavior of the instrument was formally tested rather than assumed. Dissenting free-text responses, including one that questioned the framing of the reform items, were retained and used to qualify our own interpretation.

The most consequential limitation is measurement. Although the seven domains were derived from literature and interviews, only part of that structure was reproduced empirically. Internal consistency failed in the assessment domain, and the confirmatory model fitted poorly. Content validity rested on expert review without a quantified content validity index or a formal pilot, and Q6 illustrates the cost: an item about academic stress and anxiety was placed within the theoretical education domain at the design stage, where it could not plausibly cohere, and it duly failed to load. Two further methodological choices constrain what the factor analyses can support. The exploratory and confirmatory analyses were performed on the same 708 complete cases, so the confirmatory result is not an independent cross-validation. Estimation also treated five-point Likert responses as continuous, whereas polychoric correlations with robust weighted least squares are better suited to ordinal indicators and can yield different loadings and fit.³⁵ Domain scores should therefore be read as descriptive summaries and not as validated scale scores.

Sampling imposes a second set of constraints. Recruitment used open professional and student networks, so no sampling frame existed; the invited population is unknown, and a conventional response rate cannot be reported. Self-selection plausibly favored respondents with more reliable connectivity and stronger views about the pandemic. Missing data compound this. Complete responses to all 39 items were available for 708 of 1095 respondents, and those 708 differed from the remainder: they were older (30.88±5.41 vs 27.71±4.51 years), more often women (64.8% vs 55.3%), and reported more favorable assessment and policy scores. The factor-analytic subsample is thus not interchangeable with the analytic sample, and the structural findings should be treated as provisional. Institutional composition varied as well. The proportion of respondents reporting residency ranged from none at some universities to roughly one in six at others, and clerkship representation ranged from about a quarter to nearly half. No adjustment was made for this, so differences between institutional means partly reflect who answered rather than what was provided. Confidence intervals are descriptive, were not adjusted for multiple comparisons, and are wide where respondent numbers were small, which is why 46 institutions were pooled.

Response style deserves particular caution given the endorsement pattern. All nine forward-looking items were positively keyed, with no reverse-keyed counterparts, no forced-choice format, and no ranking task, so acquiescence and social desirability cannot be separated from genuine agreement, and means above 4.2 leave little room for variance. Perceptions of infrastructure and perceptions of educational continuity were also elicited from the same respondents within a single instrument, which invites common-method variance and inflates the apparent coherence between them.³⁶ No objective institutional data on platforms, bandwidth, class size, or faculty development were collected against which self-report could be checked.

Design and context add further caveats. Data were gathered retrospectively, well after the period respondents were asked to recall, which introduces recall bias and probably favors salient or emotionally marked experiences. Fieldwork coincided with national internet restrictions and shutdowns linked to war and protests, which disrupted recruitment and may have shaped who was reachable. On the qualitative side, interviews were conducted by a single investigator, coding disagreements were resolved by team discussion without independent double coding or a reported agreement statistic, and no member checking was undertaken, so transferability rather than generalisability is the appropriate standard for that phase.

Furthermore, since 46 of the 58 participating institutions contributed 20 or fewer respondents (several fewer than 10), institution-specific psychometric estimates for those institutions were unstable and are not reported individually beyond the exploratory comparison above; this constrains our ability to determine whether measurement properties are uniform across sites and argues against disaggregating the psychometric evaluation by institution with the present data.

Several priorities follow for future work. The instrument should be rebuilt so that perceptual judgments and reports of educational practice sit in separate scales, with balanced keying, quantified content validity, cognitive interviewing and a pilot before national use, and it should then be revalidated on split samples using estimation appropriate to ordinal data.^22,35 Linking respondent perceptions to objective institutional indicators would allow the stress-test interpretation to be tested rather than proposed. Once larger per-institution samples are available, multilevel models with institution as a random effect, adjusted for educational role, age and sex, would partition between- and within-institution variance and separate compositional from institutional differences more precisely than the present pooled design allows; the same data would allow reliability to be re-estimated within institutions to test whether measurement properties are uniform across sites. Prospective or repeated-measures designs would remove the dependence on recall that constrains the present findings. Most importantly, stated support for reform is not evidence that reform is achievable. Studies that examine cost, faculty capacity, clinical placement availability, and the educational outcomes of preparedness curricula are the necessary next step, since this study establishes what stakeholders want and not what a system can deliver.

Conclusion

Participants from Iranian medical schools recalled the COVID-19 period as most damaging to the components of medical training that cannot be relocated off-site. The lowest ratings concerned compensatory use of simulation and virtual patients, valid alternatives for assessing practical skills, and routine formative interaction with teachers, whereas personal access to connectivity and equipment was rated comparatively well. Recalled experience differed among respondents from different universities. That variation cannot be traced to institutional characteristics within this design. It should be read as a description of how a single national event was experienced across the system rather than as a ranking of institutional performance.

Endorsement of the forward-looking items was high and varied little between institutions. Because these items were positively keyed and difficult to oppose in principle, and because their means approached the top of the response scale, the finding establishes the direction of stakeholder priorities and not their intensity, their relative ranking, or their feasibility. Read with that caution, the results point toward protecting clinical learning when services are disrupted, planning for skills assessment when in-person examination is not possible, embedding student support within educational continuity planning, and defining learner roles before the next emergency rather than during it. Whether such measures are achievable within existing faculty, infrastructure, and clinical capacity is a question this study raises but cannot answer. The instrument developed here organized these questions usefully. However, it did not perform as a set of validated scales, and its revision is a prerequisite for measuring preparedness rather than describing intent. Larger per-institution samples analyzed with multilevel models, a revised and revalidated instrument, and linkage to objective institutional indicators are the logical next steps.

Data Sharing Statement

The data supporting the findings of this study are presented within the article and its Supplementary File. The Supplementary File comprises Table S1 (themes, subthemes, conceptual codes and illustrative quotations from the qualitative phase), Table S2 (full wording, domain assignment and scoring convention for all 39 items, including the direction of reverse-scored items), Table S3 (anonymized English translations of the 131 optional free-text responses, classified by thematic domain), and Tables S4–6 (between-institution comparisons and within-institution reliability). De-identified participant-level data are not publicly available because of confidentiality considerations.

Ethics Approval

The study protocol received approval from the Ethics Committee of Tehran University of Medical Sciences under ethics code IR.TUMS.SPH.REC.1404.135. All protocols were carried out in compliance with the Declaration of Helsinki. Prior to the interviews, written informed consent was obtained on paper, and an electronic consent statement was obtained, which respondents had to confirm before the questionnaire loaded.

Consent for Participation

All participants provided informed consent to participate in the current study prior to completing the questionnaire.

Consent for Publication

All participants consented to the publication of anonymized responses and direct quotations. No identifying information is reported.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

Amirhossein Takian reports Leadership or fiduciary roles from World Health Summint, outside the submitted work. The authors report no other conflicts of interest in this work.