The Association Between Online Education and Mental Health Among Medical Students During the COVID‐19 Pandemic: A Cross‐Sectional Study

ABSTRACT

Background and Aims

This study aimed to examine associations between mental health symptoms and several pandemic‐related experiences specifically online education, social isolation, financial concerns, and health anxieties among medical students during the coronavirus disease‐2019 (COVID‐19) pandemic. Due to the cross‐sectional design and the absence of a pre‐pandemic or contemporaneous in‐person comparison group, causal relationships cannot be established, and the associations of online education cannot be isolated from the broader pandemic context.

Methods

This cross‐sectional study was conducted at Tehran University of Medical Sciences during the COVID‐19 pandemic. The sampling frame comprised all basic sciences and physiopathology medical students (N = 1247). A simple random sample of 400 students was drawn; 330 completed the online questionnaire (response rate: 82.5%). The questionnaire included the Patient Health Questionnaire (PHQ‐9), Perceived Stress Scale (PSS‐10), and researcher‐developed items on pandemic‐related experiences. Data were analyzed using SPSS‐24. Potential selection and non‐response biases are discussed in the limitations.

Results

330 medical students participated in the study, of whom 137 (41.5%) were female and 193 (58.5%) were male. Clinical screening revealed that 26.5% of participants had minimal or no symptoms, while 46.8% had mild depression, 17.3% had moderate depression, and 9.4% had severe depression. For analytical comparison, when combining minimal and mild severities, 73.3% presented with minimal/mild depression. Additionally, 4.2% had mild stress, 91.5% had moderate stress, and 4.2% had severe anxiety.

Conclusion

Several factors, including a history of psychiatric illness, lack of private study space, fear of social isolation, health‐related anxiety, dissatisfaction with virtual classes, poor internet access, and financial concerns, were significantly associated with depression and/or perceived stress among medical students during the COVID‐19 pandemic. Given the cross‐sectional design and the absence of multivariable adjustment, these findings should be interpreted strictly as descriptive, unadjusted associations rather than independent predictors. The results highlight exploratory areas that may warrant further confirmatory investigation.

Abbreviations

LMS

learning management systems

PHQ‐9

patient health questionnaire

PSS‐10

perceived stress scale

1. Introduction

During the coronavirus disease‐2019 (COVID‐19) pandemic, educational institutions worldwide implemented infection‐control measures to sustain academic continuity, primarily through campus closures, physical distancing, and the rapid adoption of virtual education [1]. In Iran, schools and universities suspended in‐person operations in March 2020 [2], transitioning to virtual education. Here, virtual education is defined as a formal instructional delivery mode where all course content and interactions are mediated through electronic systems and digital curricula, replacing traditional face‐to‐face classrooms [3]. This transition relied on online platforms specifically digital infrastructures such as learning management systems (LMS), video conferencing software (e.g., Zoom, Skyroom), and social media applications to host educational content and facilitate synchronous or asynchronous communication [4]. Throughout the pandemic, these platforms served as the primary environments for academic instruction and student development [5].

Mental health, defined as a state of well‐being that enables individuals to cope with life stressors, realize their potential, learn and work productively, and contribute to their communities [6], was severely challenged during this period. Emerging evidence indicates that the pandemic had widespread adverse psychological effects across various demographic groups. Notably, research among Iranian populations reveals that medical students experienced significantly higher levels of stress, anxiety, and depression compared to both healthcare professionals and the general public [7]. This psychological burden has been linked to online learning challenges and pandemic‐related anxiety, which collectively compromised academic performance, reduced attention span, impaired concentration, and diminished study motivation [8].

Beyond the loss of physical social interactions, the structural demands of virtual learning environments further impacted students. The pressure to maintain academic performance was associated with heightened anxiety, compounded by the difficulty of concentrating while studying at home [9]. For some students, participating in video calls exacerbated personal anxieties, while others struggled to acquire the training necessary for academic success. Furthermore, prolonged online engagement led to cognitive exhaustion for both students and instructors a phenomenon termed “fatigue magnification” [10]. This mental exhaustion is largely attributed to the cognitive load of processing continuous video‐mediated interactions, which differs significantly from natural, face‐to‐face communication [11].

Consequently, distance‐learning students often report higher levels of isolation and lower self‐confidence than their peers in traditional classrooms [12]. For medical students, these challenges were compounded by the suspension of lectures, clinical rotations, and critical examinations [13]. Although healthcare professionals frequently experience psychological distress during crises [14], medical students in their clinical training years are particularly vulnerable to severe psychological morbidity [15]. Recommended coping strategies to mitigate this distress include limiting media exposure, maintaining sleep hygiene, engaging in regular physical exercise, eating a balanced diet, and resuming routine academic or professional roles [16].

Despite the widespread shift to virtual education and online learning platforms, limited research has investigated their utilization and psychological impact within the Iranian context. Given the ongoing challenges of the COVID‐19 pandemic, its documented threat to student well‐being, and the reduction in face‐to‐face interpersonal communication, it is crucial to evaluate the mental health of students utilizing these digital platforms [17]. Identifying the specific predictors of psychological distress in this population is essential for designing and implementing targeted supportive interventions [18].

To address this gap, this study aimed to investigate the relationship between mental health status and the utilization of online educational platforms among medical students at Tehran University of Medical Sciences during the COVID‐19 pandemic.

2. Materials and Methods

This was a cross‐sectional study conducted entirely during the COVID‐19 pandemic when all medical education had shifted to online platforms. Consequently, there was no contemporaneous group of students receiving in‐person instruction, nor were pre‐pandemic mental health data available for this population. As such, this study cannot isolate the effects of online education from the multitude of concurrent pandemic‐related stressors (e.g., fear of infection, social isolation, economic disruption, disrupted clinical training, uncertainty about professional futures). All reported associations reflect correlations at a single time point and should be interpreted as descriptive and hypothesis‐generating rather than causal or analytically comparative regarding online education specifically.

2.1. Sampling Frame and Simple Random Sampling

The sampling frame consisted of all medical students enrolled in the basic sciences and physiopathology stages (first 4 years) of the general medicine program at Tehran University of Medical Sciences (TUMS) during the September–October 2021 semester. A complete list of eligible students (N = 1247) was obtained from the university registrar's office. Each student was assigned a unique sequential number, and a simple random sample of 400 students was drawn using a computer‐generated random number table (Random Allocation Software, version 2.0). This sample size exceeded our target of 330 to account for anticipated non‐response and incomplete questionnaires.

2.2. Recruitment and Data Collection

Selected students were contacted via both university email and SMS text message. Up to three contact attempts were made: an initial invitation (Day 1), a reminder (Day 7), and a second reminder (Day 14). The invitation included a study explanation, a link to the online questionnaire (hosted on Porsline, an Iranian secure survey platform), and an electronic informed consent form. Participation was voluntary, and no incentives were provided.

2.3. Response Rate and Non‐Response Bias Assessment

Of the 400 students invited, 330 completed the full questionnaire (response rate: 82.5%). Among non‐respondents (n = 70), 42 did not respond to any contact attempt, 18 opened the invitation but did not complete the questionnaire, and 10 declined participation. To assess non‐response bias, we compared respondents and non‐respondents on available demographic data (age, gender, educational year) from the university registrar. No significant differences were found (gender: p = 0.67; age: p = 0.42; educational year: p = 0.73). However, data on mental health status, internet access quality, or study engagement were not available for non‐respondents, so residual bias on these unmeasured variables cannot be ruled out:

$$N=\frac{Z^{2}P(1-P)}{d^2}=\frac{3.84(0.31)(0.69)}{{(0.05)}^2}=330$$

2.4. Validation and Piloting

The researcher‐developed questionnaire contained two multi‐item domains: Virtual Learning Experiences (8 items) and Pandemic‐Related Anxieties (6 items). Items were scored using a five‐point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). To ensure content validity, the initial item pool was reviewed by a panel of five experts (two psychiatrists, two medical education specialists, and one epidemiologist). Items were refined for clarity and relevance to the Iranian medical education context. Face validity and feasibility were subsequently assessed in a pilot study involving 20 medical students who were not included in the final sample. Feedback from the pilot resulted in the simplification of technical terminology related to learning management systems and virtual education platforms. Internal consistency reliability was evaluated using Cronbach's alpha, with a predefined threshold of α ≥ 0.70 considered acceptable. The Virtual Learning Experiences domain demonstrated an alpha coefficient of 0.74, while the Pandemic‐Related Anxieties domain demonstrated an alpha coefficient of 0.78, indicating acceptable internal consistency. Test–retest reliability was not assessed because the survey was administered anonymously at a single time point during an evolving pandemic context, precluding repeat measurement of the same participants. Consequently, the temporal stability of the researcher‐developed scales remains uncertain and should be evaluated in future validation studies.

2.5. Definition of the PHQ‐9 Scoring Bands

The primary outcome for depression severity was assessed using the PHQ‐9. Raw scores were interpreted using standard clinical cut‐offs: 0–4 indicates minimal or no depressive symptoms, 5–9 indicates mild depression, 10–14 indicates moderate depression, 15–19 indicates moderately severe depression, and 20–27 indicates severe depression. For cross‐tabulation and analytical efficiency in the bivariate analyses (including Table 1), the scoring bands were categorized into three tiers: Minimal/Mild Depression (scores 0–9, combining the minimal and mild bands), Moderate Depression (scores 10–14), and Severe Depression (scores ≥ 15, combining moderately severe and severe bands). This classification was applied consistently across all comparative tables to maintain sufficient cell sizes for statistical testing.

Table 1.

Demographic characteristics based on PHQ‐9 (depression) and PSS‐10 (stress).

Demographic	Category	Condition	Minimal/Mild depression (0–9)	Moderate depression	Severe depression	Mean	Total (n, %)	p‐value
Gender	Female	Depression	95 (69.3%)	27 (19.7%)	15 (10.9%)	10.85	137 (41.5%)	0.384
		Stress	8 (5.8%)	119 (86.9%)	10 (7.3%)	20.75	137 (41.5%)	0.028*
	Male	Depression	147 (76.2%)	30 (15.5%)	16 (8.3%)	10.18	193 (58.5%)
		Stress	6 (3.1%)	183 (94.8%)	4 (2.1%)	19.67	193 (58.5%)
Education level	Science	Depression	190 (75.5%)	37 (14.7%)	25 (9.9%)	10.39	252 (77.3%)	0.044
		Stress	9 (3.6%)	237 (94.0%)	6 (2.4%)	19.94	252 (77.3%)	0.003*
	Physiopathology	Depression	49 (66.2%)	20 (27.0%)	5 (6.8%)	10.62	74 (22.7%)
		Stress	5 (6.8%)	61 (82.4%)	8 (10.8%)	20.58	74 (22.7%)
Marital status	Single	Depression	229 (74.8%)	51 (16.7%)	26 (8.5%)	10.33	306 (92.7%)	0.055
		Stress	11 (3.6%)	284 (92.8%)	11 (3.6%)	20.16	306 (92.7%)	0.011*
	Married	Depression	13 (54.2%)	6 (25.0%)	5 (20.8%)	12.08	24 (7.3%)
		Stress	3 (12.5%)	18 (75.0%)	3 (12.5%)	19.58	24 (7.3%)
Living situation	Alone	Depression	22 (71.0%)	6 (19.4%)	3 (9.7%)	11.29	31 (9.4%)	0.197
		Stress	2 (6.5%)	27 (87.1%)	2 (6.5%)	19.35	31 (9.4%)	0.197
	Dormitory	Depression	33 (71.7%)	10 (21.7%)	3 (6.5%)	10.41	46 (13.9%)
		Stress	3 (6.5%)	43 (93.5%)	0 (0.0%)	19.26	46 (13.9%)
	Parents	Depression	178 (75.7%)	34 (14.5%)	23 (9.8%)	10.80	235 (71.2%)
		Stress	7 (3.0%)	218 (92.8%)	10 (4.3%)	20.35	235 (71.2%)
	Others	Depression	9 (50.0%)	7 (38.9%)	2 (11.1%)	14.17	18 (5.5%)
		Stress	2 (11.1%)	14 (77.8%)	2 (11.1%)	20.67	18 (5.5%)
Economic status	Low	Depression	4 (36.4%)	4 (36.4%)	3 (27.3%)	14.91	11 (3.3%)	0.068
		Stress	0 (0.0%)	10 (90.9%)	1 (9.1%)	23.27	11 (3.3%)	0.451
	Medium	Depression	148 (74.0%)	33 (16.5%)	19 (9.5%)	10.50	200 (60.6%)
		Stress	6 (3.0%)	186 (93.0%)	8 (4.0%)	20.10	200 (60.6%)
	High	Depression	90 (75.6%)	20 (16.8%)	9 (7.6%)	9.97	119 (36.1%)
		Stress	8 (6.7%)	106 (89.1%)	5 (4.2%)	19.86	119 (36.1%)
Nationality	Iranian	Depression	233 (73.5%)	53 (16.7%)	31 (9.8%)	10.47	317 (96.1%)	0.255
		Stress	12 (3.8%)	293 (92.4%)	12 (3.8%)	20.18	317 (96.1%)	0.013*
	Non‐Iranian	Depression	9 (69.2%)	4 (30.8%)	0 (0.0%)	10.23	13 (3.9%)
		Stress	2 (15.4%)	9 (69.2%)	2 (15.4%)	18.69	13 (3.9%)
Total		Depression	242 (73.5%)	57 (17.3%)	31 (9.4%)	10.46	330	—
		Stress	14 (4.2%)	302 (91.5%)	14 (4.2%)	20.12	330

Note: All analyses presented are bivariate (unadjusted). No multivariate adjustment for potential confounders (e.g., gender, educational level, marital status, history of psychiatric illness, family communication patterns, financial concerns, health anxiety, fear of isolation) has been performed. Associations may therefore reflect residual confounding and should be interpreted as descriptive and hypothesis‐generating rather than as estimates of independent or causal effects. Because more than 40 statistical tests were conducted across these tables without multiplicity corrections, the familywise Type I error rate is severely inflated (approx 0.87). Individual p‐values should be treated strictly as descriptive measures of sample differences, and attention should be paid to the effect sizes described in the text.

^*Significance symbol.

2.6. Sample Size Calculation

The sample size was initially calculated based on the study's original analytic objective of comparing PHQ‐9 and PSS‐10 scores as continuous outcomes between independent groups using two‐sided t‐tests. However, during analysis, depression and perceived stress were also categorized into ordinal severity groups, and most reported associations were evaluated using chi‐square tests. With a final sample size of 330 participants, the study retained adequate power to detect moderate associations in contingency‐table analyses (approximately Cramér's V ≥ 0.15–0.20 at α = 0.05). Nevertheless, the study was not specifically powered for every individual subgroup comparison reported, and therefore all findings should be interpreted as exploratory and hypothesis‐generating rather than confirmatory. Using a two‐sided independent t‐test with 80% power (β = 0.20) and a significance level of 5% (α = 0.05), the required sample size to detect a small‐to‐moderate effect size (Cohen's d = 0.3) is approximately 176 participants per group, or 352 participants total. Cohen's d = 0.3 was selected as the minimal clinically meaningful difference for psychological outcome measures such as the PHQ‐9 and PSS‐10, based on conventions in mental health research (Cohen, 2000) [19].

Accounting for a potential 10% incomplete response rate, a target sample of 330 participants was set. The final sample of 330 participants therefore provides 80% power to detect effect sizes of d ≥ 0.3 in primary between‐group comparisons. For categorical outcomes analyzed with chi‐square tests, this sample size provides adequate power to detect absolute differences of approximately 15–20 percentage points between groups (e.g., 30% vs. 50%) at α = 0.05.

Given the exploratory nature of this cross‐sectional study and the number of comparisons performed (exceeding 40 tests across Tables 1, 2, 3), we did not adjust the significance level for multiple comparisons. Consequently, the familywise Type I error rate under the global null approaches 0.87, and several significant results are expected by chance alone. Therefore, all p‐values must be interpreted descriptively as hypothesis‐generating rather than confirmatory. To assess the practical significance of these associations independently of p‐values, effect size estimates are provided in the text, using Cramér's V for categorical contingency tables and Cohen's d for continuous metric comparisons. All reported associations require replication.

Table 2.

Relationship of depression and stress with history of psychiatric illness and relationships with family/friends.

Questions	Category	Condition	Mild (PHQ‐9: 5–9)	Moderate (PHQ‐9: 10–14)	Severe (PHQ‐9: ≥ 15)	Mean	Total	p value
Do you have a history of psychiatric issues and use of psychiatric medications?	Yes	Depression	50 (54.9%)	30 (33.0%)	11 (12.1%)	13.57	91 (27.6%)	0.000*
		Stress	7 (7.7%)	78 (85.7%)	6 (6.6%)	20.45	91 (27.6%)	0.060
	No	Depression	192 (80.3%)	27 (11.3%)	20 (8.4%)	9.28	239 (72.4%)
		Stress	7 (7.7%)	224 (93.7%)	8 (3.3%)	19.99	239 (72.4%)
On average, how many hours a day do you communicate with your friends?	1 h or less	Depression	99 (77.3%)	16 (12.5%)	13 (10.2%)	9.72	128 (39.1%)	0.165
		Stress	5 (3.9%)	121 (94.5%)	2 (1.6%)	19.73	128 (39.1%)	0.165
	More than 1 h	Depression	141 (70.9%)	41 (20.6%)	17 (8.5%)	10.91	199 (60.9%)
		Stress	9 (4.5%)	178 (89.4%)	12 (6.0%)	20.33	199 (60.9%)
Through which method do you mostly communicate with friends?	Virtual	Depression	195 (71.7%)	52 (19.1%)	25 (9.2%)	10.69	272 (82.4%)	0.158
		Stress	12 (4.4%)	248 (91.2%)	12 (4.4%)	20.13	272 (82.4%)	0.892
	In person	Depression	47 (81.0%)	5 (8.6%)	6 (10.3%)	9.40	58 (17.6%)
		Stress	2 (3.4%)	54 (93.1%)	2 (3.4%)	20.07	58 (17.6%)
On average, how many hours a day do you communicate with family members?	1 h or less	Depression	50 (71.4%)	14 (20.0%)	6 (8.6%)	10.42	70 (21.2%)	0.782
		Stress	5 (7.1%)	64 (91.4%)	1 (1.4%)	19.68	70 (21.2%)	0.181
	More than 1 h	Depression	195 (73.8%)	43 (16.5%)	25 (9.6%)	10.47	260 (78.7%)
		Stress	9 (3.5%)	238 (91.5%)	13 (5.0%)	—	—

Note: All analyses presented are bivariate (unadjusted). No multivariate adjustment for potential confounders (e.g., gender, educational level, marital status, history of psychiatric illness, family communication patterns, financial concerns, health anxiety, fear of isolation) has been performed. Associations may therefore reflect residual confounding and should be interpreted as descriptive and hypothesis‐generating rather than as estimates of independent or causal effects. Because more than 40 statistical tests were conducted across these tables without multiplicity corrections, the familywise Type I error rate is severely inflated (approx 0.87). Individual p‐values should be treated strictly as descriptive measures of sample differences, and attention should be paid to the effect sizes described in the text.

^*PHQ‐9 categorization: mild (5–9), moderate (10–14), severe (combined moderately severe [15–19] and severe 20–27])

Table 3.

Association of depression and stress with social and online education factors.

Domain /question	Category	Depression severity	Moderate (n = 57)	Severe (n = 31)	p‐value	Stress severity	Moderate (n = 302)	Severe (n = 14)	p‐value
		Mild (n = 242)				Mild (n = 14)
Family communication
How do you mostly communicate with family members?	Virtual	215 (77.3%)	34 (12.2%)	29 (10.4%)	< 0.001	8 (2.9%)	262 (94.2%)	8 (2.9%)	< 0.001
	In‐person	25 (51.0%)	23 (46.9%)	1 (2.0%)		6 (12.2%)	37 (75.5%)	6 (12.2%)
Perceived support
How much support do you receive from those around you?	Low	24 (46.2%)	17 (32.7%)	11 (21.2%)	< 0.001	1 (1.9%)	45 (86.5%)	6 (11.5%)	0.037
	Medium	126 (72.4%)	34 (19.5%)	14 (8.0%)		8 (4.6%)	159 (91.4%)	7 (4.0%)
	High	92 (88.5%)	6 (5.8%)	6 (5.8%)		5 (4.8%)	98 (94.2%)	1 (1.0%)
Virtual class quality
Satisfaction with virtual class quality	Low	128 (67.4%)	42 (22.1%)	20 (10.5%)	0.046	9 (4.7%)	173 (91.1%)	8 (4.2%)	0.966
	Medium	68 (80.0%)	11 (12.9%)	6 (7.1%)		3 (3.5%)	79 (92.9%)	3 (3.5%)
	High	46 (83.6%)	4 (7.3%)	5 (9.1%)		2 (3.6%)	50 (90.9%)	3 (5.5%)
Interest in virtual classes	Low	152 (77.2%)	27 (13.7%)	18 (9.1%)	0.018	6 (3.0%)	188 (95.4%)	3 (1.5%)	0.019
	No idea	60 (64.5%)	26 (28.0%)	7 (7.5%)		5 (5.4%)	81 (87.1%)	7 (7.5%)
	High	30 (75.0%)	4 (10.0%)	6 (15.0%)		3 (7.5%)	33 (82.5%)	4 (10.0%)
Participation in virtual classes	Low	119 (74.8%)	27 (17.0%)	13 (8.2%)	0.103	8 (5.0%)	144 (90.6%)	7 (4.4%)	0.395
	No idea	40 (60.6%)	16 (24.2%)	10 (15.2%)		3 (4.5%)	58 (87.9%)	5 (7.6%)
	High	83 (79.0%)	14 (13.3%)	8 (7.6%)		3 (2.9%)	100 (95.2%)	2 (1.9%)
Motivation and skills
Worry about not learning necessary skills	Low	—	—	—	0.007	7 (7.5%)	82 (88.2%)	4 (4.3%)	0.038
	No idea	—	—	—		4 (4.8%)	73 (86.9%)	7 (8.3%)
	High	66 (64.3%)	21 (30.4%)	6 (5.4%)		3 (2.0%)	147 (96.1%)	3 (2.0%)
Motivation to participate in virtual classes	Low	58 (69.0%)	21 (25.0%)	5 (6.0%)	0.031	6 (3.5%)	166 (96.0%)	1 (0.6%)	< 0.001
	No idea	118 (83.3%)	15 (7.1%)	20 (9.5%)		8 (8.2%)	83 (85.6%)	6 (6.2%)
	High	136 (78.6%)	23 (13.3%)	14 (8.1%)		0 (0.0%)	53 (88.3%)	7 (11.7%)
IT skill level	Low	71 (73.2%)	19 (19.6%)	7 (7.2%)	0.014	4 (5.1%)	74 (94.9%)	0 (0.0%)	0.108
	No idea	35 (58.3%)	15 (25.0%)	10 (16.7%)		4 (3.4%)	111 (93.3%)	4 (3.4%)
	High	63 (80.8%)	5 (6.4%)	10 (12.8%)		6 (4.5%)	117 (88.0%)	10 (7.5%)
Infrastructure
Private space for virtual classes	Yes	91 (76.5%)	21 (17.6%)	7 (5.9%)	< 0.001	9 (4.4%)	191 (94.1%)	3 (1.5%)	0.007
	No	88 (66.2%)	31 (23.3%)	14 (10.5%)		5 (3.9%)	111 (87.4%)	11 (8.7%)
Type of internet used	WiFi	173 (85.2%)	14 (6.9%)	16 (7.9%)	0.011	4 (5.8%)	64 (92.8%)	1 (1.4%)	0.004
	Mobile Internet	69 (54.3%)	43 (33.9%)	15 (11.8%)		8 (7.8%)	85 (83.3%)	9 (8.8%)
	Both	53 (76.8%)	10 (14.5%)	6 (8.7%)		2 (1.3%)	153 (96.2%)	4 (2.5%)
Device used for virtual classes	Laptop	62 (60.8%)	28 (27.5%)	12 (11.8%)	0.060	3 (4.8%)	59 (93.7%)	1 (1.6%)	0.045
	Smartphone	127 (79.9%)	19 (11.9%)	13 (8.2%)		7 (8.9%)	66 (83.5%)	6 (7.6%)
	Both	52 (82.5%)	8 (12.7%)	3 (4.8%)		4 (2.2%)	174 (94.1%)	7 (3.8%)
Pandemic‐related worries
Worry about falling behind in studies	Low	49 (62.0%)	18 (22.8%)	12 (15.2%)	< 0.001	9 (7.1%)	116 (91.3%)	2 (1.6%)	< 0.001
	No idea	139 (75.1%)	31 (16.8%)	15 (8.1%)		5 (5.3%)	78 (83.0%)	11 (11.7%)
	High	105 (82.7%)	12 (9.4%)	10 (7.9%)		0 (0.0%)	108 (99.1%)	1 (0.9%)
Worry about financial issues	Low	55 (58.5%)	33 (35.1%)	6 (6.4%)	0.002	5 (3.2%)	151 (96.2%)	1 (0.6%)	0.007
	No idea	82 (75.2%)	12 (11.0%)	15 (13.8%)		7 (7.4%)	82 (86.3%)	6 (6.3%)
	High	130 (82.8%)	17 (10.8%)	10 (6.4%)		2 (2.7%)	66 (88.0%)	7 (9.3%)
Worry about isolation and losing relationships	Low	62 (65.3%)	25 (26.3%)	8 (8.4%)	< 0.001	8 (7.0%)	105 (92.1%)	1 (0.9%)	0.012
	No idea	48 (64.0%)	15 (20.0%)	12 (16.0%)		5 (5.8%)	74 (86.0%)	7 (8.1%)
	High	96 (84.2%)	11 (9.6%)	7 (6.1%)		1 (0.8%)	123 (94.6%)	6 (4.6%)
Worry about own health	Low	58 (67.4%)	26 (30.2%)	2 (2.3%)	0.004	6 (5.9%)	93 (92.1%)	2 (2.0%)	0.111
	No idea	88 (67.7%)	20 (15.4%)	22 (16.9%)		6 (7.2%)	72 (86.7%)	5 (6.0%)
	High	86 (85.1%)	7 (6.9%)	8 (7.9%)		2 (1.4%)	137 (93.8%)	7 (4.8%)
Worry about health of loved ones	Low	57 (68.7%)	21 (25.3%)	5 (6.0%)	0.002	3 (6.3%)	44 (91.7%)	1 (2.1%)	0.175
	No idea	99 (67.8%)	29 (19.9%)	18 (12.3%)		6 (8.5%)	61 (85.9%)	4 (5.6%)
	High	44 (91.7%)	3 (6.3%)	1 (2.1%)		5 (2.4%)	197 (93.4%)	9 (4.3%)

Note: Data are presented as n (row %). Percentages may not sum to 100% due to rounding. p‐values are from chi‐square tests comparing distribution across depression severity (mild/moderate/severe) or stress severity (mild/moderate/severe) categories. Bold p‐values indicate statistical significance (p < 0.05). Mean scores for depression and stress are omitted from this merged table for brevity but are available in the original tables. All analyses are bivariate (unadjusted); associations may reflect residual confounding and should be interpreted as hypothesis‐generating.

All analyses presented are bivariate (unadjusted). No multivariate adjustment for potential confounders (e.g., gender, educational level, marital status, history of psychiatric illness, family communication patterns, financial concerns, health anxiety, fear of isolation) has been performed. Associations may therefore reflect residual confounding and should be interpreted as descriptive and hypothesis‐generating rather than as estimates of independent or causal effects.

Because more than 40 statistical tests were conducted across these tables without multiplicity corrections, the familywise Type I error rate is severely inflated (approx 0.87). Individual p‐values should be treated strictly as descriptive measures of sample differences, and attention should be paid to the effect sizes described in the text.

This text discusses two information collection tools and how to utilize them. The first tool is the PHQ‐9 questionnaire, which is a screening and diagnostic tool for depression. It comprises 9 self‐report questions based on DSM‐IV criteria and takes approximately 3 to 5 min to answer. Dadfar et al. [20] found that this tool has acceptable validity and reliability (α = 0.86) for outpatients in Iran.

The second tool is the PSS‐10 questionnaire, which is a self‐report scale consisting of 10 questions that assess people's perceived stress levels. It has two parts, with the first part examining people's helplessness and the second part examining people's efficiency. The average time to complete this questionnaire is 3 to 5 min. According to Maroufizadeh's study, it has good validity and reliability (α = 0.865) in Iran [21].

In addition to these tools, a researcher‐made questionnaire is also used to gather demographic information, assess anxiety caused by COVID‐19, and identify problems associated with using the virtual classroom.

There are various methods for collecting data, including:

Utilizing Latin journals and books

Conducting searches using search engines and databases

Recording the information and data gathered from questionnaires

The methodology for data analysis involved in putting the collected data into SPSS‐24 software for result presentation. Descriptive statistics such as mean and standard deviation were used to describe quantitative variables while frequency and frequency percentage were used to describe qualitative variables. For analytical analysis, the chi‐square test was employed to examine the relationship between qualitative variables. Additionally, when dealing with normally distributed quantitative variables, the independent t‐test was used. If the distribution was non‐normal, the Mann–Whitney test was used for comparison. A significance level of 0.05 was used in all statistical tests.

Importantly, as this is a cross‐sectional study, all reported associations reflect correlations at a single time point. No causal relationships can be inferred from these data, and reverse causality (e.g., pre‐existing depression influencing perceptions of online education) cannot be ruled out. Furthermore, all analyses presented are bivariate (unadjusted). We did not perform multivariate modeling (e.g., multiple logistic or linear regression) to adjust for potential confounders such as gender, educational level, marital status, history of psychiatric illness, family communication patterns, financial concerns, health anxiety, or fear of social isolation. These variables are likely interrelated, and the reported associations may therefore reflect residual confounding. Consequently, our findings should be interpreted as descriptive and hypothesis‐generating, not as estimates of independent or causal effects. This study was approved by the Research Ethics Board of Tehran University of Medical Sciences (IR. TUMS. MEDICINE. REC.1400.620). Because the survey screened for sensitive mental health symptoms (depression and stress), an ethical ‘safety net’ was implemented. Upon completion of the questionnaire (or if a participant chose to withdraw), a closing page was displayed providing contact details for the university's Student Counseling Center and national mental health helplines. This ensured that students who may have been distressed by the screening process or who recognized a need for support had immediate access to professional referral resources.

We confirm that this study was designed, conducted, and reported in strict compliance with the STROBE checklist for cross‐sectional studies.

3. Results

Because all participants were engaged in online education during the study period and no comparison group (e.g., pre‐pandemic students or students receiving in‐person instruction during the pandemic) was available, all results are descriptive of associations within this online‐learning population. The absence of a comparison group means that the prevalence of depression and stress reported here cannot be attributed specifically to online education as distinct from the broader pandemic context. The results revealed that 26.5% of participants had minimal or no symptoms, while 46.8% had mild depression, 17.3% had moderate depression, and 9.4% had severe depression. Additionally, 4.2% reported mild perceived stress, 91.5% moderate perceived stress, and 4.2% severe perceived stress. When tracking demographic associations in Table 1, the minimal and mild categories were consolidated into a single ‘Minimal/Mild’ severity group representing 73.3% of the total sample (n = 242).

The study found notable associations between various demographic factors and anxiety and depression levels (Table 1). Specifically, gender was identified as a significant factor for perceived stress, with women exhibiting higher rates of moderate‐to‐severe stress than men, though the overall effect size was small (Cramér's V = 0.14, p = 0.028). Additionally, an association was found between the stage of education and mental health, with students in the basic sciences reporting higher stress levels than those in the physiopathology stage (Cramér's V = 0.19, p = 0.003).

Regarding technology‐related factors (Table 2), students who used mobile internet connections reported higher mean stress scores compared to those using WiFi (21.01 vs. 19.87), representing a small‐to‐medium effect size (Cohen's d = 0.28), though this difference did not reach statistical significance in this sample size. Similarly, students using smartphones as their primary device reported higher stress levels than those using laptops (20.45 vs. 19.96; Cohen's d = 0.11).

In Table 3, the strongest descriptive associations were observed between depression severity and structural learning infrastructure. Specifically, a lack of a private space for virtual classes (Cramér's V = 0.24, p < 0.001) and relying solely on mobile internet connections (Cramér's V = 0.28, p = 0.011) showed moderate practical significance, identifying these environmental factors as critical avenues for future targeted tracking.

Specifically, gender was identified as a significant factor, with women exhibiting higher levels of severe anxiety (Table 2). Additionally, a correlation was found between the level of education and anxiety and depression, with students in the basic science field reporting higher levels of both (Table 3). Marital status was also a significant factor, with married individuals experiencing more severe anxiety than unmarried individuals. Moreover, nationality was found to be linked to anxiety, with non‐Iranians displaying higher levels of severe anxiety (Table 1). Finally, the study identified a significant relationship between a history of illness and depression.

Individuals with moderate depression and anxiety tended to rely on virtual connections with their families more often. They exhibited little interest in attending virtual classes, but they typically had a designated private space for participation and access to Wi‐Fi and the Internet. Participants frequently utilized mobile phones and laptops to attend classes (Table 2).

Regarding technology‐related factors (Table 2), students who used mobile internet connections reported higher mean stress scores compared to those using WiFi (21.01 vs. 19.87), though this difference did not reach statistical significance. Similarly, students using smartphones as their primary device for online classes reported higher stress levels than those using laptops (20.45 vs. 19.96). These patterns, while not statistically significant, are clinically suggestive.

A noticeable correlation existed between anxiety and poor virtual learning skills. While most participants reported minimal anxiety related to financial concerns, those who did experience such anxieties tended to exhibit more severe anxiety symptoms. Additionally, participants who expressed greater worry about social isolation tended to experience more severe depression and anxiety symptoms, with a statistically significant relationship between the two (Table 3).

Given the exploratory nature of this study and the number of comparisons performed, p‐values are reported without adjustment for multiple testing and should be interpreted descriptively.

4. Discussion

4.1. Methodological Limitations and Design Constraints

Before interpreting the findings, it is essential to acknowledge a fundamental limitation of this study: its cross‐sectional design, which precludes the determination of temporal sequence or causal direction. For example, while we observed an association between dissatisfaction with virtual classes and depressive symptoms, this finding is equally compatible with two opposing interpretations: (1) features of online education exacerbated depressive symptoms, or (2) students with pre‐existing or emerging depression perceived their online learning environment more negatively. Because our data cannot distinguish between these trajectories, we frame all findings strictly as descriptive associations.

Similarly, the high prevalence of moderate stress (91.5%) must be interpreted with caution. Medical students consistently demonstrate elevated baseline stress relative to the general population, even under non‐pandemic conditions. While the COVID‐19 pandemic and the rapid transition to remote learning likely contributed to this burden, the absence of a pre‐pandemic or contemporaneous in‐person comparison cohort within this study prevents us from attributing this elevation solely to these factors. Crucially, these figures represent self‐reported symptom burden rather than clinical diagnoses of structured stress disorders.

4.2. Comparative Analysis of Depression and Stress Prevalence

The prevalence of moderate‐to‐severe depression in our sample (26.7%) aligns closely with pre‐pandemic estimates among Iranian medical students. Specifically, Jaafari et al. (2018) reported a 23.7% prevalence at the same institution (Tehran University of Medical Sciences) using the PHQ‐9 [22], while a 2019 national survey found a prevalence of 29.1% [23]. This stability suggests that the pandemic and the shift to online instruction did not substantially elevate depression rates beyond baseline levels in this population. However, this interpretation remains tentative due to potential variations in sampling methods and assessment timing across cohorts.

Conversely, the prevalence of moderate‐to‐severe perceived stress (95.7%) represents a marked deviation from pre‐pandemic norms. Previous PSS‐10 evaluations of Iranian medical students reported moderate‐to‐severe stress rates of 43.6% [24] and 47.2% [25]. Globally, a large‐scale meta‐analysis encompassing over 18,000 medical students reported a pooled stress prevalence of 48.3% [26]. Our finding of 95.7% is approximately double these baseline rates, indicating a severe, population‐wide escalation of psychological distress.

This heightened stress profile is consistent with other pandemic‐era studies of medical students in fully remote settings. For instance, Ganjoo et al. and Saraswathi et al. reported moderate‐to‐severe stress rates of 92.4% and 88.6% among Indian medical students, respectively [13, 27]. Similarly, a multicountry analysis spanning seven nations found that fully online cohorts fell at the upper end of a 78% to 94% stress prevalence range [28, 29].

4.3. Comparison With In‐Person and Hybrid Educational Modalities

Direct comparison with contemporaneous cohorts is limited because few medical schools maintained in‐person training during the pandemic [30]. However, institutions that preserved physical or hybrid instruction reported lower stress levels. For example, Kim et al. observed a moderate‐to‐severe stress prevalence of 54.2% in South Korea under a hybrid curriculum [28], while Hammarstedt et al. reported a prevalence of 41.7% in Sweden, where universities largely remained open [29]. While these cross‐national comparisons are inherently confounded by regional differences in pandemic severity, state restrictions, and cultural baseline metrics, they collectively suggest that fully online instruction was associated with a higher stress burden than hybrid or in‐person models [31].

4.4. Socio‐Environmental and Infrastructure Correlates of Distress

Our findings demonstrate that psychological distress during the pandemic was not merely a byproduct of viral apprehension, but was also deeply tied to the infrastructure of virtual learning [32]. While prior domestic literature highlighted general pandemic‐related anxiety [33], our data reveal that lack of private study space, poor internet quality, and limited digital literacy directly correlated with severe symptom clusters [34].

These environmental and socio‐economic vulnerabilities are consistent with broader literature. Dissatisfaction with virtual delivery, low motivation, and academic anxiety have been shown to exacerbate depressive symptoms, with financial instability and health anxiety further compounding severe presentation [25, 26, 27, 28, 29]. Previous studies have similarly linked high internet dependence to compromised psychological well‐being [30, 32], and identified gender, education level, and marital status as significant covariates of student depression [31].

Furthermore, our finding of an association between in‐person family communication and lower depression severity must be interpreted descriptively. Cross‐sectional data cannot determine whether in‐person interaction is inherently protective [35], or if students with better baseline mental health are more likely to reside in stable family environments. Students living alone or in dormitories who rely solely on virtual platforms may experience heightened isolation and financial strain due to their living arrangements rather than the communication medium itself [36]. Nonetheless, these dynamics align with studies showing that adolescent and young‐adult distress is significantly mediated by family communication, isolation fears, and social support systems [33, 34].

4.5. Educational Design and Policy Implications

The nearly universal stress prevalence (95.7%) in our sample indicates that elevated stress was a normative, population‐wide state rather than an isolated student experience. This divergence between high, acute stress and relatively stable depression rates suggests that institutional responses must go beyond individual clinical counseling [37]. Universities should implement universal, structural interventions such as academic workload adjustments, stress‐management integration, and subsidized technological access during periods of extended remote learning [38].

Finally, it is critical to acknowledge that “online education” is not a uniform intervention. During the pandemic, digital learning spanned a wide spectrum, from passive, asynchronous lectures to highly interactive synchronous sessions, virtual simulations, and immersive technologies [39]. Because we did not capture detailed data on specific instructional modalities, learner interaction levels, or simulated tool usage, our results should not be generalized to all virtual learning environments.

Emerging evidence in medical education indicates that highly interactive, experiential, and simulation‐based digital formats yield significantly more favorable psychological and academic outcomes than passive, screen‐based instruction. The distress documented in our sample likely reflects a combination of the emergency transition context, social isolation, technological barriers, and a lack of interactive design, rather than remote delivery in its optimized forms. Future research should evaluate whether structured, collaborative, or immersive virtual platforms can mitigate the psychological burden observed in this study.

5. Conclusion

This study highlights a substantial psychological burden among medical students during the COVID‐19 pandemic, characterized by a high prevalence of perceived stress (95.7% moderate‐to‐severe) and moderate‐to‐severe depressive symptoms (26.7%). Bivariate analyses revealed that these mental health challenges were descriptively associated with a history of psychiatric illness, lack of private study spaces, reliance on mobile internet connections, dissatisfaction with virtual classes, and pandemic‐related anxieties such as fear of social isolation and financial concerns. However, due to the cross‐sectional design, the exploratory nature of the analyses without multiplicity corrections, and the absence of multivariable adjustments or a contemporaneous in‐person comparison group, these correlations cannot establish causality or isolate the direct impact of online education from the broader pandemic context. These findings suggest that institutional support during periods of remote learning should integrate universal stress‐reduction initiatives with practical technological and environmental resources, while pointing to critical avenues that warrant further longitudinal, multivariable investigation.

Author Contributions

Nastaran Kazemirad: data curation, software, writing – review and editing, writing – original draft, funding acquisition, formal analysis. Atefeh Mohammad Jafari Dokesh: conceptualization, methodology, supervision, resources, project administration. Nahid Lorzadeh: investigation, validation, visualization, project administration, data curation, supervision. Roya Vaziri‐Harami: conceptualization, methodology, software, data curation, supervision, formal analysis, validation, investigation.

Funding

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Transparency Statement

The lead author Nastaran Kazemirad affirms that this manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned (and, if relevant, registered) have been explained.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.