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

BACKGROUND

Early-onset major depressive disorder (MDD) is a severe mental health condition that often presents with a more chronic and severe clinical course, along with an increased risk of suicide. To construct a preliminary model to estimate suicidal attempts (SA) risk using retrospective clinical data in patients with early-onset MDD.

METHODS

This retrospective study analyzed data from patients diagnosed with early-onset MDD between 2013 and 2023, with an onset age of ≤25 years. Variables including demographic details, clinical characteristics, and biochemical parameters, such as folate, adrenocorticotropic hormone (ACTH), and homocysteine (HCY) levels, were processed for analysis. Participants were categorized into two groups: those with suicide attempts (MDD-S) and those with non-suicidal attempts (MDD-N). A multivariable logistic regression model was applied to evaluate the relationship between independent variables and SA. The model was optimized using a 10-fold cross-validation technique, and its predictive performance was examined through Receiver Operating Characteristic (ROC) curve and decision curve analysis (DCA).

RESULTS

Significant differences were observed between the MDD-S and MDD-N groups across several factors, including alcohol consumption (12.9 % vs. 5.3 %, p = 0.004), tobacco use (21.1 % vs. 8.8 %, p < 0.001), education level (p < 0.001), and folate levels (p = 0.01). Clinically, the MDD-S group had higher rates of modified electroconvulsive therapy (63.2 % vs. 40.1 %, p < 0.001) and mood stabilizer use (41.6.1 % vs. 31.3 %, p = 0.02), indicating more severe depressive episodes. The final nomogram included key predictors such as sex, occupation, education level, marital status, tobacco use, alcohol consumption, ACTH and folate levels. The model demonstrated moderate discriminative performance, with a C-index of 0.734, which was consistent after bootstrap validation. ROC analysis showed an AUC of 0.734 (95 % CI: 0.685-0.775), and decision curve analysis confirmed a higher clinical benefit of the nomogram.

CONCLUSIONS

This study developed a model to estimate the risk of suicide attempts in early-onset MDD patients based on clinical and biological data. While the model demonstrated moderate discriminative ability, further validation in larger and more diverse populations is necessary.