Development and Validation of a Nomogram for Estimating the Risk of Suicide Tendencies Among Chinese Middle School Students: A Cross-Sectional Study

Objective: This study aims to develop and validate a risk estimation model for identifying suicidal tendencies among middle school students. The effectiveness of the model is evaluated, offering insights for preventing and managing student suicides in educational institutions. Methods: This study employed a cross-sectional design. From December 2018 to January 2019, a total of 12,798 middle school students from all 18 public schools in an urban district of Suzhou were surveyed. After data cleaning, 12,063 valid questionnaires were included and randomly divided into a training set (n=8,444) and a validation set (n=3,619) in a 7:3 ratio for model development and internal validation, respectively. Predictors were selected through univariate analysis and LASSO regression, with independent associated factors subsequently identified by multivariable logistic regression. Based on these factors, a nomogram risk estimation model was constructed using R software. To assess generalizability, external validation was performed using data from 6,262 valid questionnaires collected from 11 public middle schools in Changshu, Suzhou, in 2023. Results: The nomogram incorporated nine selected factors: trouble asking for help, parents' marital relationship, gender, school bullying, nightmares, depressive mood (PHQ02), sleep disturbance (PHQ03), feelings of worthlessness (PHQ06), and psychomotor changes (PHQ08). The model demonstrated good discrimination in the internal validation set area under the curve (AUC) 0.807 (95% CI [0.790, 0.824]) and in a temporal external validation cohort AUC 0.764 (95% CI [0.751, 0.778]). Calibration was satisfactory internally but required adjustment in the external cohort. Conclusion: This study developed and validated a multidimensional nomogram that effectively discriminates middle school students at risk of suicide, providing a framework for initial risk stratification. For application in new settings, local calibration of the model's risk estimates is mandatory. This tool holds potential to aid early identification in school and primary care contexts.