What Predicts Student Depression Better Than Academic Discipline?

Academic discipline can be associated with student depression, but it may tell us far less than we assume. In our cross-sectional study of 602 university students in Lahore, prior depression treatment showed the largest adjusted association with PHQ-9 scores, while female gender and shorter sleep duration were also independently associated with greater depressive symptoms. Whether a student was studying medicine or another discipline mattered statistically, but the overall regression models explained only a small proportion of the differences between students.

By Taimoor Asghar

This distinction matters because university mental-health conversations often begin by categorizing students according to what they study. Medical students are described as being under exceptional pressure; engineering students face demanding workloads; business students worry about employment; and students in other disciplines have their own academic and economic uncertainties. These experiences may be real, but academic labels are imperfect summaries of the circumstances that influence a person’s mental health.

Our recently published BMC Psychology study provides a useful example. Rather than finding that medical students had the highest depression scores, we observed somewhat higher symptom levels among non-medical students. More importantly for the question in this article, several individual-level variables were associated with depression independently of discipline, and academic discipline by itself was clearly not a comprehensive explanation of why students differed.

What did our study actually find?

The study included 602 undergraduate university students in Lahore, Pakistan: 424 medical students and 178 non-medical students. Participants completed the Patient Health Questionnaire-9 (PHQ-9) and the Depression Anxiety Stress Scales-21 (DASS-21). We examined group differences and then used multivariable regression models to evaluate whether discipline and several participant characteristics were independently associated with symptom scores.

The results challenged the simple assumption that studying medicine automatically identifies the group with the greatest depressive burden. Median PHQ-9 scores were 9 among medical students and 10 among non-medical students. After adjustment, non-medical students still had higher PHQ-9 scores, with an estimated difference of approximately 1.43 points. The discipline difference was statistically significant.

But statistical significance is only part of the story.

The models had low R-squared values, ranging from approximately 0.029 to 0.041. In practical terms, the measured predictors accounted for only a small fraction of the overall variability in psychological symptoms. That result should discourage attempts to explain student depression using one demographic or academic category.

The full study, including its psychometric, item-response and symptom-network analyses, is available in our BMC Psychology article on the PHQ-9 and DASS-21 among medical and non-medical university students.

So what predicted depression better than academic discipline?

Three findings deserve particular attention: prior depression treatment, gender and sleep duration. They should not be interpreted as a complete predictive model of depression, and because our study was cross-sectional, they cannot establish causal direction. Nevertheless, they show why looking beyond a student’s degree programme can produce a more informative picture.

1. Prior depression treatment showed the strongest association

Among the variables included in the PHQ-9 regression model, a history of previous treatment for depression had the largest estimated association. Students reporting prior depression treatment scored approximately 4.15 points higher on the PHQ-9 after adjustment for the other variables in the model.

That difference was substantially larger than the adjusted difference associated with academic discipline.

The finding is clinically intuitive, but its interpretation requires care. Previous depression treatment is not necessarily a cause of current depressive symptoms. It is more reasonably understood as a marker that a person has experienced sufficient depressive difficulties in the past to receive treatment. Depression can recur, residual symptoms can persist, and some individuals undergoing treatment may still be symptomatic.

In other words, mental-health history contains information that a degree label cannot provide.

Two students may sit in the same lecture hall, take the same examinations and experience the same curriculum while having very different histories of psychological vulnerability. One may never have experienced a clinically significant depressive episode, while another may have previously required professional treatment. Classifying both simply as “medical students” or “business students” removes that distinction.

This is one reason student mental-health assessment should not rely primarily on assumptions about which faculties are supposedly high risk. Prior psychological history may be considerably more relevant to an individual’s current symptom burden.

2. Gender was consistently associated with psychological symptoms

Female gender was associated with higher symptom scores across all of the psychological domains examined in our adjusted analyses, including PHQ-9 depression and the DASS-21 depression, anxiety and stress scales. These associations were statistically significant in each model.

This consistency is notable because it suggests that the relationship was not restricted to a single measurement instrument. However, the result should not be interpreted as evidence that gender itself biologically determines whether a student becomes depressed.

Gender-related differences in psychological symptoms can reflect a complex mixture of biological, social, cultural, economic and behavioral influences. These may include differences in exposure to stressors, social expectations, safety concerns, family responsibilities, coping patterns, willingness to disclose symptoms and access to support. Our cross-sectional dataset was not designed to disentangle all of these pathways.

The practical lesson is narrower: if universities organize mental-health resources only around academic faculties, they may overlook characteristics that cut across every faculty.

A female student studying medicine and a female student studying another subject may share important experiences that are not captured by their academic discipline. Likewise, substantial variation exists within gender groups, so gender should inform population-level planning rather than become another simplistic label used to predict an individual’s mental health.

3. Sleep duration was associated with PHQ-9 scores

Sleep also stood out. Each additional reported hour of sleep was associated with approximately 0.37 points lower PHQ-9 scores in the adjusted model.

The direction of this association is plausible, but the cross-sectional design is especially important here. We cannot conclude from these data that increasing sleep by exactly one hour would cause a student’s PHQ-9 score to fall by 0.37 points.

The relationship between sleep and depression can run in both directions. Depressive symptoms can interfere with sleep, while inadequate or disturbed sleep may contribute to worse emotional functioning. Sleep can also be affected by examination schedules, employment, commuting, family demands, digital-media habits, physical illness, anxiety and other factors.

There is another important measurement issue: sleep disturbance is itself one of the nine symptoms assessed by the PHQ-9. Consequently, some statistical association between sleep duration and PHQ-9 scores is conceptually unsurprising. This does not make the finding meaningless, but it is another reason not to interpret the coefficient as a simple causal effect.

For universities, however, sleep is potentially more actionable than academic discipline. Institutions cannot reasonably change every student’s field of study to improve mental health. They can examine timetabling, excessive scheduling, overnight academic expectations, examination practices and campus education about healthy sleep.

Academic discipline mattered, but not as much as the stereotype suggests

It would be incorrect to say that academic discipline had no association with depression in our study. It did.

Non-medical students had higher PHQ-9 scores than medical students in both the unadjusted comparison and adjusted analysis. They also had higher adjusted DASS-21 depression and anxiety scores. The difference in DASS-21 stress, however, did not reach conventional statistical significance after adjustment.

These findings are useful precisely because they complicate a familiar narrative. Medical education is legitimately demanding, and there is extensive concern about the mental health of medical students. Nothing in our data argues against providing strong mental-health support to medical students.

What our data do challenge is the assumption that medical students should automatically be treated as the benchmark high-risk group while students elsewhere on campus receive less attention.

In this sample, the direction of the observed difference was the opposite of that stereotype.

Furthermore, a difference of roughly one to one-and-a-half PHQ-9 points between disciplines should not be confused with a dramatic separation of two psychologically distinct populations. Students within each discipline differed substantially from one another. The overlap between the groups is therefore more important than a simple ranking of their average or median scores might imply.

Why a low R-squared is one of the most useful findings

Regression results often attract attention when individual coefficients have small p-values. Yet the model-level result may be more revealing.

Our models explained only about 3% to 4% of the variance in symptom scores. That means most of the differences between students remained unexplained by the variables included in these models.

This does not mean the observed associations are false or useless. It means they are incomplete.

Student depression is unlikely to be adequately explained by a short list of demographic variables. Factors that were not fully captured in this study could plausibly contribute to individual differences, including:

• financial strain and economic insecurity;
• relationship difficulties and family conflict;
• social isolation or lack of support;
• academic performance pressure;
• uncertainty about employment after graduation;
• chronic physical illness;
• adverse life events;
• housing and commuting conditions;
• substance use;
• personality and coping styles;
• loneliness;
• exposure to discrimination or harassment;
• quality rather than simply duration of sleep;
• existing anxiety or other psychological conditions;
• and differences in access to professional support.

These factors were not all tested in our model, so the study cannot determine which of them would explain the remaining variance. They illustrate why depression should be approached as a multidimensional outcome rather than something that can be inferred from a student’s faculty.

Prediction is not the same as causation

The word “predictor” can easily create misunderstanding in observational research.

In a regression model, a predictor is a variable used to statistically account for variation in an outcome. Calling gender, discipline, sleep duration or previous treatment a predictor does not necessarily mean that it causes depression. Nor does it mean that the variable can accurately forecast which specific student will develop depression in the future.

Our study measured exposure characteristics and psychological symptoms at one point in time. That cross-sectional design establishes associations, not temporal sequences.

For example, shorter sleep could contribute to worse depressive symptoms, depression could cause sleep disruption, or both could be influenced by another factor. Similarly, previous treatment is likely to reflect earlier mental-health difficulties rather than functioning as a harmful exposure that produces future depression.

Longitudinal research following students over time would be better suited to examining which factors precede changes in depressive symptoms and how those relationships evolve during university education.

Individual symptoms may also matter more than group labels

Another component of our study reinforces the argument for looking beyond broad categories.

We did not analyze only total questionnaire scores. Item response theory and symptom-network methods were also used to investigate individual symptoms. Within the PHQ-9 analyses, items involving self-worth, concentration, depressed mood and appetite were among those providing relatively high information. In the broader symptom network, self-worth, concentration and downheartedness emerged as central symptoms.

These findings should be interpreted cautiously. Network centrality does not prove that treating a central symptom will automatically improve the entire symptom system, and some network stability estimates in the study were moderate rather than exceptionally high.

Nevertheless, the analyses illustrate an important conceptual point: two students with identical PHQ-9 totals can reach the same score through different combinations of symptoms.

One may mainly experience poor concentration, low self-worth and low mood. Another may report prominent sleep disturbance, fatigue and appetite changes. A third may have difficulties across several domains.

A faculty label tells us almost nothing about that symptom-level heterogeneity.

What should universities do differently?

If academic discipline explains only a small part of depressive symptom differences, campus mental-health systems should avoid allocating attention solely according to assumptions about which courses are most stressful.

Use campus-wide screening and support

Mental-health resources should be visible and accessible to students throughout the university rather than concentrated exclusively in traditionally high-pressure faculties. Medical students may need targeted services, but targeted programmes should complement rather than replace institution-wide support.

Make previous mental-health difficulties relevant to support pathways

Students with a history of depression or other significant psychological difficulties may benefit from easy routes back into professional care when symptoms recur. This does not mean universities should indiscriminately collect sensitive medical histories. It means support systems should make continuity of care possible when students voluntarily seek help.

Treat sleep as part of student well-being

Universities cannot control students’ sleep, but institutional practices can influence it. Scheduling, examination clustering, early-morning requirements and expectations of constant academic availability deserve consideration when designing healthier learning environments.

Avoid using demographics as diagnostic shortcuts

Group-level associations are useful for planning services, not diagnosing individuals. Being female, studying a particular subject or sleeping fewer hours does not establish that an individual has depression. Validated screening tools, clinical assessment when indicated and direct communication with students remain necessary.

What should researchers measure next?

The low explanatory power of our models points toward a productive next question: what additional variables account for meaningful differences in student depression?

Future studies could examine financial stress, perceived academic pressure, social support, loneliness, employment expectations, family relationships, adverse experiences, sleep quality, physical activity, digital behavior and access to mental-health care alongside conventional demographic characteristics.

Longitudinal designs would be particularly valuable. Following students from entry into university through examinations, clinical placements and graduation could help distinguish relatively stable vulnerability factors from temporary stress responses. Repeated symptom measurements could also reveal whether academic disciplines differ in trajectories rather than simply average scores at one time point.

There is also a case for studying interactions. Academic discipline might matter more under certain circumstances. Financial insecurity, for instance, could have different consequences depending on expected employment opportunities. Sleep deprivation may interact with examination periods. Gender-related experiences may vary considerably across institutions and disciplines.

These possibilities cannot be resolved by simply comparing mean depression scores between two faculties.

The larger lesson: ask about the student, not just the degree

The most useful interpretation of our findings is not that prior treatment, gender or sleep constitutes a perfect alternative prediction model. It does not. The low R-squared values make that clear.

Instead, the study demonstrates the limitations of academic stereotypes.

Academic discipline showed a measurable association with symptoms, but a student’s psychological history had a larger association with PHQ-9 scores in our adjusted model. Gender was consistently associated with multiple symptom domains, and sleep duration was independently related to PHQ-9 scores. Meanwhile, most variation between individuals remained unexplained.

That combination of findings points toward a more realistic framework for student mental health: risk is distributed across campuses, individual histories matter, daily-life circumstances matter, symptoms are heterogeneous, and no academic category provides a reliable shortcut to understanding an individual student.

Universities therefore should not ask only, “Which faculty has the worst mental health?” A better set of questions is: Which students are currently struggling? What individual and environmental factors are associated with their symptoms? Who has experienced mental-health difficulties before? Are students sleeping adequately? What barriers prevent them from seeking help? And are support services accessible regardless of academic discipline?

Those questions move student mental health away from stereotypes and toward evidence-informed prevention, screening and care.

Frequently asked questions

Did academic discipline predict depression in the study?

Yes. Non-medical students had higher PHQ-9 scores than medical students, including after adjustment for other variables in the regression model. However, discipline explained only part of the picture, and the overall model accounted for a small proportion of variability in depressive symptoms.

What was the strongest predictor of PHQ-9 scores?

Among the variables included in the model, previous treatment for depression had the largest adjusted association with PHQ-9 scores, corresponding to approximately 4.15 additional points. This should be interpreted as an association and a marker of prior mental-health difficulty, not evidence that treatment causes depression.

Was sleep associated with student depression?

Yes. Each additional reported hour of sleep was associated with approximately 0.37 points lower PHQ-9 scores after adjustment. Because the study was cross-sectional, however, it cannot determine whether shorter sleep contributed to depression, depression affected sleep, or both occurred together.

Were medical students more depressed than non-medical students?

No in this particular sample. Non-medical students had a median PHQ-9 score of 10 compared with 9 among medical students, and the adjusted analysis also indicated higher depressive symptoms among non-medical students. The finding should not automatically be generalized to every university or country.

Can these findings be used to identify whether an individual student has depression?

No. Group-level statistical associations cannot diagnose an individual. The PHQ-9 is a screening and symptom-severity instrument rather than a standalone diagnostic evaluation, and students with significant or persistent symptoms should receive appropriate professional assessment.

Medical disclaimer: This article discusses research findings for educational purposes and is not personal medical advice or a diagnostic assessment. Depression and other mental-health conditions require individualized evaluation. Anyone experiencing persistent depressive symptoms, substantial functional impairment, thoughts of self-harm or other serious mental-health concerns should seek appropriate professional care.