Predicting suicide attempts in early-onset major depressive disorder: A nomogram-based approach

Area of Science:

• Psychiatry
• Clinical Psychology
• Biomedical Data Science

Background:

• Early-onset major depressive disorder (MDD) is associated with severe illness, chronicity, and elevated suicide risk.
• Developing predictive models for suicidal attempts (SA) in this population is crucial for timely intervention.
• Retrospective analysis of clinical data aims to identify risk factors for SA in early-onset MDD.

Purpose of the Study:

• To construct and validate a preliminary predictive model for suicidal attempts (SA) in patients with early-onset major depressive disorder (MDD).
• To identify key demographic, clinical, and biochemical predictors associated with SA in this patient group.

Main Methods:

• Retrospective analysis of early-onset MDD patients (onset age ≤25 years) between 2013-2023.
• Utilized multivariable logistic regression, 10-fold cross-validation, ROC curve, and DCA to build and assess the predictive model.
• Included variables such as demographics, clinical history, alcohol/tobacco use, and biochemical markers (folate, ACTH, homocysteine).

Main Results:

• Significant differences found between suicide attempters (MDD-S) and non-attempters (MDD-N) in alcohol use, tobacco use, education, folate levels, and treatment modalities.
• A nomogram incorporating sex, occupation, education, marital status, tobacco/alcohol use, ACTH, and folate levels was developed.
• The model demonstrated moderate discriminative performance (C-index=0.734, AUC=0.734), with good clinical utility confirmed by DCA.

Conclusions:

• A predictive model for SA risk in early-onset MDD patients was successfully developed using clinical and biological data.
• The model exhibits moderate predictive accuracy and clinical usefulness.
• Further validation in larger, diverse populations is recommended to enhance generalizability.