Development of a Point-based Model for Poor Academic Performance in Japanese University Students Based on Lifestyle Risk Factors: A Prospective Cohort Study

Abstract

Aims: Multiple lifestyle behaviors have been linked with poor academic performance (AP) in undergraduate students. It is unknown whether a risk prediction model that incorporates a combination of different lifestyle behaviors would predict the risk of poor AP. Study design: a population-based prospective study Methods: A total of 2,282 eligible participants (men: 66%) aged 18 to 22 years were followed up for four years. Poor AP was defined as a grade point average (GPA) of less than 2.0. Cox proportional hazards models were used to identify lifestyle behaviors associated with the development of poor AP, and a score was created from predictors in the final model. Results: During a median follow-up period of four years, 440 (29.2%) men and 111 (14.3%) women developed poor AP. The final model for poor AP in men included low frequency of breakfast intake, dietary intake ups and downs, shorter weekday study time, longer weekday video game time, doing exercise almost every day, self-rated irregular lifestyle. In women, alcohol intake once or more per week, low frequency of breakfast intake, dietary intake ups and downs were included in the final model for poor AP. The risk prediction models showed good discriminatory abilities (C statistics: 0.73, 95% confidence interval [CI]: 0.67, 0.78 for women; and 0.70, 95% CI: 0.66, 0.74 for men). Conclusions: This risk prediction model incorporating multiple unhealthy lifestyle behaviors predicts poor AP with good discrimination ability.