Development of a Proactive Screening Tool for Mental Health Problems in Pharmacy Students

Abstract

Objective: To develop a tool for mental health problems screening in pharmacy students and to determine cut-off for identifying high risk group for the problems whose active surveillance is needed. Methods: The study design was a prognostic prediction research. The subjects were 591 pharmacy students studying in the 1^st to 5^th academic years. The primary outcome was mental health problems as measured by the Thai General Health Questionnaire-28 (Thai GHQ-28). The study developed the model for risk score from secondary data with the following variables: academic year, gender, being recipient of a scholarship, desire to change the faculty, and self-esteem.  Selection of variables was based on statistical significance of their association with the outcome, area under the receiver operating characteristic curve (AuROC) and feasibility in practice. Multiple logistic regression was performed for model development. Results: Low self-esteem was the most powerful predictor (OR: 7.85 [3.34, 18.46]). Discrimination power of the model was acceptable (AuROC: 0.75 [0.71, 0.79]). Risk prediction was precise because graph of risk derived from the model fitted to the observed data. Internal validation by bootstrap resampling revealed that AuROC was consistent with the observed data. The cut-point score of 21/22 from total score of 112 provided the highest discrimination power (AuROC: 0.68 [0.64, 0.72]). Conclusion: The model showed acceptable discrimination power and precision for mental health problems screening. It could be used to proactively screen for mental health problems and identify the high-risk students who need special attention.