Abstract
Mental health disorders often peak during late adolescence and early adulthood, yet university students frequently delay seeking help. This study establishes a data-driven framework for identifying undergraduate students at high risk for mental health problems using non-clinical indicators like sleep, diet, and study engagement. Academic staff should detect students who meet these criteria at the earliest and provide mental health support to reduce the duration of untreated illness (DUI) among undergraduate university students.
Introduction
University students are a vulnerable population for mental health issues, yet they often suffer from a long Duration of Untreated Illness (DUI) due to stigma or lack of self-awareness. Delayed intervention is linked to poorer clinical outcomes and higher dropout rates. While clinical assessments are the gold standard, they are often inaccessible in a general academic setting. This research, published in Frontiers in Psychiatry, seeks to bridge this gap by identifying "indirect" risk factors—lifestyle habits and school adaptation—that are easier for academic staff to monitor but still highly predictive of a student's mental state.
Methods and Study Population
The researchers conducted a cross-sectional study of first-year students at Osaka University during the COVID-19 pandemic (Summer 2020). From an initial pool of 3,294 respondents, 1,644 students were selected for final analysis based on complete data and regular affiliation.
Students were categorized as "high-risk" if they reported both subjective depression and fatigue. The study utilized the Utrecht Work Engagement Scale for Students (UWES-S-J) to measure study engagement (vigor, dedication, and absorption) alongside self-reported sleep duration and meal frequency. To determine exact numerical thresholds (cut-off points), the team employed Receiver Operating Characteristic (ROC) curves and the Youden index.
Key Results and Predictive Criteria
The analysis revealed that a staggering 30.1% of the sample met the criteria for the high-risk group. The ROC analysis yielded three specific criteria for identifying these students:
- Study Engagement: A UWES-S-J score of < 37.5 points (out of 84).
- Sleep: < 6 hours of sleep on weekdays.
- Nutrition: < 2.5 meals per day on average.
As shown in the ROC curves, the UWES-S-J score was the strongest predictor, with an Area Under the Curve (AUC) of 0.631. Interestingly, while weekday sleep duration was a significant predictor, holiday sleep was not, suggesting that the inability to maintain a healthy schedule under academic pressure is a primary red flag.
Discussion and Practical Application
The authors intentionally chose cut-off points with high sensitivity (73%–82%). In an early-detection context, it is preferable to "over-flag" students for a check-in than to miss those who are truly struggling. Study engagement serves as a powerful proxy for deteriorating mental health; a drop in "vigor" or "dedication" to studies often precedes clinical crisis.
Limitations
Because the study is cross-sectional, it cannot determine causality (e.g., whether poor sleep causes depression or vice versa). Additionally, the data was collected during the COVID-19 pandemic, which may have influenced the prevalence of mental health issues and altered normal study habits. Future work may involve using "sensing technology" like wearables to automate the collection of these metrics for real-time monitoring.
Conclusion
By establishing specific cut-off values for accessible lifestyle metrics, this study provides academic staff with a practical toolkit to identify students who may need support. Early intervention based on these non-clinical signs can significantly reduce DUI and improve the long-term well-being of undergraduate students.
Summary updated