Development and validation of the risk score for estimating suicide attempt in patients with major depressive disorder

• Published in: Social Psychiatry and Psychiatric Epidemiology Vol. 59; no. 6; pp. 1029 - 1037
• Main Authors: Huang, Zhi-Xin, Wang, Qizhang, Lei, Shasha, Zhang, Wenli, Huang, Ying, Zhang, Caiping, Zhang, Xiangyang
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01-06-2024; Springer Nature B.V
• Subjects: Adult; Calibration; China - epidemiology; Depressive Disorder, Major - diagnosis; Depressive Disorder, Major - psychology; Dynamic models; Epidemiology; Female; Humans; Machine Learning; Male; Medicine; Medicine & Public Health; Mental depression; Middle Aged; Nomograms; Peroxidase; Psychiatric Status Rating Scales; Psychiatry; Reproducibility of Results; Risk Assessment; Risk Factors; Suicide, Attempted - psychology; Suicide, Attempted - statistics & numerical data; Suicides & suicide attempts; Systolic pressure; China; Mood disorders; Psychotic disorder; Depression; Suicide; Anxiety disorders
• ISSN: 0933-7954; 1433-9285
• DOI: 10.1007/s00127-023-02572-3

Early identification of high-risk patients with Major depressive disorder (MDD) having suicide attempts (SAs) is essential for timely targeted and tailored psychological interventions and medications. This study aimed to develop and validate a web-based dynamic nomogram as a personalized predictor of SA in MDD patients. A dynamic nomogram was developed using data collected from 1718 patients in China. The dynamic model was established based on a machine learning-based regression technique in the training cohort. We validated the nomogram internally using 1000 bootstrap replications. The nomogram performance was assessed using estimates of discrimination (via the concordance index) and calibration (calibration plots). The nomogram incorporated five predictors, including Hamilton anxiety rating scale (odds ratio [OR]: 1.255), marital status (OR: 0.618), clinical global impressions (OR: 2.242), anti-thyroid peroxidase antibodies (OR: 1.002), and systolic pressure levels (OR: 1.037). The model demonstrated good overall discrimination (Harrell’s
C
-index = 0.823). Using decision curve analysis, this model also demonstrated good clinical applicability. An online web server was constructed (
https://odywong.shinyapps.io/PRSM/
) to facilitate the use of the nomogram. Based on these results, our study developed a nomogram to predict SA in MDD patients. The application of this nomogram may help for patients and clinicians to make decisions.