Abstract
BACKGROUND: Despite the high suicide rates among patients with end-stage kidney disease (ESKD), there is no suicide prediction model specifically designed for this vulnerable population. Herein, we aimed to develop and validate a novel suicide risk score for ESKD patients. METHODS: We analyzed data from the National Health Insurance Service (NHIS) of South Korea, including 251 819 patients aged above 18 years diagnosed with ESKD between 2007 and 2022 in South Korea. The mean follow-up duration was 6.6 years. The cohort was randomly divided into derivation (70%) and validation (30%) sets. Using multivariate Cox proportional hazard regression, key variables were incorporated to develop the suicide risk score, which was converted into a 48-point scoring system, which is composed of easily identifiable clinical parameters. RESULTS: Among 176 273 patients in the derivation cohort, 1126 (0.64%) patients committed suicide. The suicide risk score demonstrated moderate discrimination in both the derivation (C-statistic, 0.694) and validation (C-statistic, 0.709) cohorts, with good calibration. In the validation cohort, patients scoring below 16, 17-32 and 33-48 had predicted 10-year suicide risk of 0.2%, 1.2% and 7.7%, respectively, while the observed 10-year risk were 0.3%, 0.8% and 3.9%. These findings highlight the model's ability to effectively stratify risk using routinely available clinical data. CONCLUSIONS: The suicide risk score is a significant advancement in suicide risk prediction for ESKD patients. It is based on simple, routinely collected clinical indicators and provides an actionable tool for risk stratification and early intervention in daily practice.