Development of a Questionnaire-based Post-Disaster Cardiovascular Disease Risk Score Incorporating Psychological Distress: The Fukushima Health Management Survey

Authors:
Preprint
Publisher website
Google Scholar
Abstract
Background: Large-scale disasters may increase cardiovascular disease (CVD) risk through psychological distress, disrupted healthcare access, and lifestyle changes, but existing CVD risk scores may not adequately capture disaster-related vulnerability.Methods: We analysed data from 23,452 participants aged 30–79 years in the Fukushima Health Management Survey after the 2011 Great East Japan Earthquake and Fukushima Daiichi nuclear accident. Incident CVD, defined as stroke or myocardial infarction, was monitored during follow-up. LASSO logistic regression and XGBoost were used only as exploratory screening tools, and predictors selected by both approaches were incorporated into a Cox proportional hazards model to develop the Questionnaire-based Post-Disaster Cardiovascular Risk Score Incorporating Psychological Distress (Q-DISC). Performance was assessed using C-statistics, net reclassification improvement (NRI), integrated discrimination improvement (IDI), and calibration analyses.Findings: During a median follow-up of 8.0 years, 1,368 participants developed incident CVD. Psychological distress assessed using the Kessler-6 Psychological Distress Scale (K6) was selected by both screening approaches and was independently associated with incident CVD. Adding K6 improved discrimination, NRI, and IDI across established CVD risk models. In the test dataset, Q-DISC showed acceptable discrimination (C-statistic, 0.770 [95% CI 0.745–0.800]) and calibration, with performance comparable to established laboratory- and non-laboratory-based models.Interpretation: Psychological distress provided prognostic information beyond conventional cardiovascular risk factors after a large-scale disaster. Q-DISC may support preliminary questionnaire-based cardiovascular risk stratification when laboratory testing or physical measurements are unavailable. External validation and, if necessary, recalibration are required before clinical implementation.