Our New BMC Psychology Study Challenges a Common Assumption About Medical Students

A common assumption in university mental health research is that medical students are likely to experience greater psychological distress than students in other academic disciplines. Our newly published BMC Psychology study suggests that this assumption is too simple. In a sample of 602 university students in Lahore, Pakistan, non-medical students reported higher depressive and anxiety symptom scores than medical students, while the difference in stress was not statistically significant after adjustment. Just as importantly, academic discipline explained only a small fraction of the overall variation in mental health symptoms.

By Taimoor Asghar

The finding does not mean that medical students have no mental health burden. Nor does it establish that non-medical education causes depression or anxiety. Instead, it highlights a broader point: universities may miss important groups if mental health strategies are built around the expectation that one academic discipline automatically carries the greatest psychological risk.

Why the assumption about medical students is understandable

Medical education has many characteristics that can reasonably create concern about student well-being. Students face demanding examinations, large volumes of material, clinical responsibilities, uncertainty about future careers, competitive environments, exposure to illness and death, and substantial expectations about professional performance. For these reasons, medical students have received considerable attention in the mental health literature.

That attention is valuable. The problem begins when a legitimate concern becomes an untested comparison: if medical training is stressful, it can be tempting to assume that medical students must therefore be more depressed, anxious, or stressed than students studying other subjects.

Those are not equivalent statements. A group can experience substantial distress without having the highest average symptom scores in a university population. Students in engineering, business, social sciences, humanities, computing, or other disciplines may face different combinations of academic pressure, financial insecurity, employment uncertainty, family expectations, social isolation, disrupted sleep, or limited access to support.

The question therefore should not be, “Are medical students distressed?” They clearly can be. A more informative question is, “How do mental health symptoms compare across disciplines when different groups are measured using the same instruments and analysed within the same study?”

What our BMC Psychology study examined

Our study, Psychometric properties and symptom profiles of the PHQ-9 and DASS-21 among medical and non-medical university students, was published in BMC Psychology on August 12, 2026. It included 602 undergraduate students from universities in Lahore, Pakistan: 424 medical students and 178 non-medical students.

The study had two related aims. First, we compared depressive and other psychological symptom scores between medical and non-medical students. Second, we examined how two widely used questionnaires, the Patient Health Questionnaire-9 (PHQ-9) and the 21-item Depression Anxiety Stress Scales (DASS-21), performed psychometrically in this student population.

The analysis went beyond comparing simple averages. It included descriptive statistics, Wilcoxon rank-sum testing, multivariable linear regression, confirmatory factor analysis, item response theory, symptom network analysis, reliability assessment, stability testing, and measurement invariance analysis.

This matters because a comparison between groups becomes more informative when researchers also ask whether the measurement tools behave sufficiently similarly across those groups.

The central finding: non-medical students had higher symptom scores

The most immediately striking result was that non-medical students had higher PHQ-9 depression scores than medical students. The median PHQ-9 score was 10 among non-medical students and 9 among medical students.

The difference remained present in adjusted regression models. Non-medical students had higher scores on the PHQ-9, the DASS-21 Depression subscale, and the DASS-21 Anxiety subscale. The adjusted difference for DASS-21 Stress did not reach conventional statistical significance.

These results challenge the expectation that medical students would necessarily emerge as the group with greater psychological symptom severity.

However, the interpretation requires caution. A statistically significant difference between two groups is not the same as evidence that academic discipline is the dominant cause of their mental health. The regression models in our study had low R-squared values, ranging from approximately 0.029 to 0.041. In practical terms, the variables included in those models explained only a modest proportion of the differences in symptom scores.

That is one of the most useful messages from the study. Academic major may be associated with mental health scores, but it is far from a complete explanation.

A one-point difference should not become a stereotype in reverse

It would be a mistake to replace one oversimplification with another. Our findings should not be interpreted as proof that non-medical students are universally more depressed than medical students.

The PHQ-9 median difference in this sample was one point. The adjusted analyses supported differences across several outcomes, but statistical significance needs to be interpreted alongside effect size, context, uncertainty, and clinical relevance.

The study was cross-sectional, meaning exposure and outcome information were collected at one period rather than following students through time. Consequently, it cannot establish that studying a particular subject caused the observed mental health differences.

The participants also came from universities in Lahore, Pakistan. Educational systems, tuition structures, employment prospects, social expectations, accommodation arrangements, student support services, cultural norms, and academic environments vary across countries and institutions. The results should therefore not be automatically generalized to every university population.

The appropriate conclusion is narrower and more useful: in this Pakistani student sample, the assumption that medical students would have worse depression and anxiety scores was not supported by the observed data.

Academic discipline explained less than we might expect

Perhaps the broader lesson is that mental health cannot be understood well by looking at academic discipline alone.

Students within the same degree programme can have very different experiences. One may have supportive relationships, stable finances, adequate sleep, manageable academic expectations, and easy access to healthcare. Another student in the same programme may experience chronic sleep loss, financial pressure, family conflict, social isolation, uncertainty about employment, or a previous history of mental health treatment.

Grouping both students under a label such as “medical” or “non-medical” inevitably loses information.

Our analysis illustrates this problem. Female gender was associated with higher symptom scores across the measured domains. Each additional reported hour of sleep was associated with a lower PHQ-9 score, and previous treatment for depression was a particularly strong predictor of higher current PHQ-9 symptoms.

These associations do not establish causal relationships either, but they demonstrate why a single academic label is unlikely to explain the complexity of student mental health.

Sleep may deserve more attention in student mental health research

The relationship between sleep and depressive symptoms was another noteworthy finding. In our adjusted PHQ-9 model, each additional hour of sleep was associated with a 0.37-point lower PHQ-9 score.

This is an observational association and should not be interpreted as evidence that simply adding a specific number of sleep hours will produce a predictable reduction in depression scores. Sleep and mental health can influence one another, and both can be affected by third variables such as workload, physical illness, substance use, social circumstances, medication, or existing psychological problems.

Nevertheless, sleep is a useful example of why universities should think beyond disciplinary categories. Irregular schedules, late-night studying, long commutes, part-time employment, screen use, examination periods, and social demands may affect students across faculties.

If universities focus mental health support primarily on traditionally high-risk academic programmes, they may overlook common modifiable pressures operating across the entire student population.

The study also tested whether the PHQ-9 and DASS-21 worked adequately

Comparing mental health scores is only meaningful if researchers have reasonable confidence in the questionnaires being used. For that reason, our study included a detailed psychometric evaluation rather than relying only on total scores.

Internal consistency estimates were acceptable, with Cronbach’s alpha values ranging from 0.82 to 0.88 across the scales examined. Measurement invariance across medical and non-medical students was also supported, strengthening the interpretation that observed group comparisons were not simply the result of obvious measurement differences between disciplines.

The DASS-21 confirmatory factor analysis showed adequate overall model fit. However, the latent correlations between some DASS-21 domains were extremely high. The Depression-Stress correlation was 0.939, while the Anxiety-Stress correlation was 0.949.

Those correlations deserve attention because the DASS-21 is conceptually divided into depression, anxiety, and stress domains. When latent factors correlate so strongly, questions arise about how distinctly those constructs are functioning in a particular population.

This does not mean that the DASS-21 is unusable. It means researchers should avoid treating subscale labels as automatically independent psychological realities. Psychometric structure can vary across populations, languages, cultures, and contexts, and scale interpretation benefits from validation rather than assumption.

Some depression symptoms carried more information than others

Another part of the study used item response theory to examine the PHQ-9 at the symptom level. Items relating to self-worth, concentration, feeling down, and appetite showed relatively high discrimination in our analysis, while the loss-of-interest item showed the lowest discrimination parameter.

This illustrates a limitation of thinking about depression only as a single total score. Two students can receive the same PHQ-9 total while endorsing very different combinations of symptoms.

One student may predominantly report sleep disturbance, low energy, and concentration difficulty. Another may report depressed mood, low self-worth, appetite changes, and loss of interest. Their total score can look similar even though their experiences are not identical.

For researchers, item-level analysis may help identify which symptoms provide the most information within a particular population. For clinicians, however, statistical centrality or discrimination should never be treated as a substitute for a complete assessment. A symptom that is less informative statistically can still be highly important for an individual patient.

What the symptom network suggested

We also examined relationships among symptoms using network analysis. Self-worth, concentration, and downheartedness emerged among the more central symptoms in the estimated network.

Network analysis offers a different perspective from traditional scale scoring. Instead of assuming that symptoms are simply interchangeable indicators of one hidden condition, a network framework examines how symptoms may be related to one another.

For example, difficulty concentrating could interfere with academic performance, which could contribute to feelings of inadequacy, which might coexist with depressed mood. Such examples are conceptually plausible, but a cross-sectional symptom network cannot establish that one symptom actually caused another.

Our network stability coefficients ranged from 0.31 to 0.44, so the symptom-network findings should be interpreted with appropriate restraint. They are useful for generating hypotheses and identifying potentially informative symptom patterns, but they are not evidence for a treatment sequence or causal pathway.

Why the findings matter for universities

The practical implication is not that universities should reduce attention to medical student mental health. Medical students continue to deserve accessible and appropriately designed mental health support.

The implication is that support should not stop there.

University-wide mental health strategies may be more appropriate than programmes based mainly on assumptions about which faculty is supposed to be under the greatest pressure. A broader strategy could include:

• accessible counselling and psychological support across faculties;
• clear referral pathways for students with persistent or severe symptoms;
• attention to sleep, academic workload, financial strain, and social isolation;
• mental health literacy that reaches both medical and non-medical programmes;
• staff training to recognize students who may need professional support;
• confidential services that minimize stigma around help-seeking;
• regular evaluation of student well-being using locally appropriate and validated measures.

Importantly, screening scores should not be used to diagnose students automatically. Instruments such as the PHQ-9 and DASS-21 can help researchers characterize symptom burden and can contribute to clinical assessment, but diagnosis requires broader professional evaluation.

Why local evidence from Pakistan is valuable

Student mental health evidence is not automatically transferable between countries. University life in Pakistan can involve distinctive academic, cultural, financial, and family circumstances. Students may live with family or away from home, attend public or private institutions, face different employment prospects, experience varying levels of academic competition, and encounter different attitudes toward psychological care.

Cross-disciplinary evidence from Pakistan is therefore useful because it tests assumptions within the population in which they are being applied.

The findings also demonstrate why local psychometric assessment matters. A questionnaire that performs well in one population may not reproduce precisely the same factor structure, item characteristics, or symptom relationships somewhere else.

Rather than simply importing interpretations from studies conducted elsewhere, researchers can examine whether common instruments behave appropriately in their own student populations.

What this study cannot tell us

Several limitations are important when interpreting the results.

First, the cross-sectional design prevents conclusions about causality or changes over time. We cannot determine whether academic discipline produced differences in symptoms or whether other characteristics of the participants explain the association.

Second, the sample was drawn from universities in Lahore and should not be considered representative of every university student in Pakistan.

Third, the study relied on self-report questionnaires. These instruments measure reported symptoms and are not equivalent to structured psychiatric diagnoses.

Fourth, although several relevant variables were included in regression models, many possible influences on mental health were not captured. Factors such as detailed socioeconomic circumstances, family relationships, academic workload, employment, social support, personality, physical illness, substance use, adverse experiences, and institutional environments could contribute to symptom differences.

Finally, the relatively low variance explained by the regression models is itself a reminder that much of the individual variation in student mental health remained unaccounted for.

The larger lesson: test assumptions instead of designing policy around them

The most useful conclusion from our BMC Psychology study is not that one academic group should replace another at the top of a mental health risk hierarchy.

It is that the hierarchy itself may be the wrong starting point.

Medical students can experience substantial depression, anxiety, stress, burnout, and other psychological difficulties. At the same time, students outside medicine can experience comparable or greater symptom burdens. Which group appears worse can depend on the population, outcome, measurement instrument, institutional environment, sampling strategy, and analytical approach.

Universities therefore need data that reflect their own students rather than relying entirely on disciplinary stereotypes.

For researchers, the study also reinforces the value of looking beyond total scores. Reliability, factor structure, measurement invariance, item response characteristics, symptom networks, effect sizes, and explained variance all contribute information that a simple comparison of averages cannot provide.

For university leaders, the message is simpler: student mental health is a campus-wide issue. Support systems should be sufficiently broad to reach students whose difficulties do not fit the group we traditionally expect to be at highest risk.

Key takeaway

Our study found higher depressive and anxiety symptom scores among non-medical students than medical students in this Lahore sample, but academic discipline explained only a small portion of the total variation in psychological symptoms. The result challenges a common assumption without replacing it with a new universal rule. The more defensible conclusion is that universities should evaluate mental health across the whole student population and avoid assuming that degree programme alone tells us who is struggling most.

Frequently asked questions

Did the study find that medical students had better mental health?

The study found lower average depressive and anxiety symptom scores among medical students than non-medical students in this particular sample. That should not be interpreted as proof that medical students generally have good mental health or that non-medical students are always at greater risk.

Were non-medical students more depressed?

Non-medical students had a median PHQ-9 score of 10 compared with 9 among medical students, and the adjusted analysis also showed higher PHQ-9 scores among non-medical students. Because the study was cross-sectional, it cannot determine why the difference occurred or whether academic discipline caused it.

Did academic discipline explain most of the differences?

No. The regression models had relatively low R-squared values, indicating that only a small proportion of the variation in psychological symptom scores was explained by the variables included in the models. Student mental health is likely influenced by many factors beyond academic discipline.

Were the PHQ-9 and DASS-21 reliable in the study?

Both instruments showed acceptable internal consistency, with Cronbach’s alpha values between 0.82 and 0.88. Measurement invariance across academic discipline was supported. The DASS-21 showed adequate overall factor-model fit, although very high correlations between its latent domains raised questions about how distinctly depression, anxiety, and stress were separated in this population.

Can PHQ-9 or DASS-21 scores diagnose depression or anxiety?

No. These questionnaires measure symptoms and can support research, screening, or clinical assessment, but they are not substitutes for a comprehensive evaluation by an appropriately qualified healthcare professional.

Medical disclaimer: This article is intended for educational and research communication purposes only. It does not provide individual medical or psychiatric advice, diagnosis, or treatment. Anyone experiencing persistent psychological distress, thoughts of self-harm, or other concerning mental health symptoms should seek assessment from an appropriately qualified healthcare professional or urgent local care when necessary.