부산대학교 도서관

Introduction: Existing models predicting right ventricular failure (RVF) after durable left ventricular assist device (LVAD) support are limited due to lack of multicenter validation, absence of intraoperative characteristics, and marginal predictive power. We sought to derive and validate a risk model to predict post-LVAD RVF.Methods: Advanced heart failure (HF) patients (N=798) requiring continuous-flow LVAD were enrolled at the Utah Transplant Affiliated Hospitals (n=477), Inova Heart & Vascular Institute (n=183), and Henry Ford Medical Center (n=138). Baseline clinical and intraoperative characteristics were recorded. The primary outcome was RVF incidence, defined as the need for right VAD (RVAD) or intravenous inotropes for >14 days. Bootstrap imputation and lasso variable selection were used to derive a predictive model which was then validated. A risk calculator was developed classifying patients into risk groups, and survival was compared.Results: Patients were predominantly white (72%), males (84%), aged 56±13 years. Patients in the RVF and non-RVF groups were comparable in terms of sex, age, and LVAD indication, while RVF patients more commonly had a history of systemic hypertension, non-ischemic cardiomyopathy, lower INTERMACS profiles, and more commonly required inotropic or temporary circulatory support pre-LVAD. Overall, 193 (24.2%) patients developed RVF with 109 (56.5%) requiring inotropes and 84 (43.5%) an RVAD. Multivariable predictors for RVF are shown in the Figure and achieved a c-statistic of 0.74 (95% CI: 0.70-0.78). Inclusion of intraoperative characteristics did not improve model performance. Cumulative survival was higher in the minimal risk group compared to low, moderate, and high.Conclusions: The STOP-RVF calculator effectively stratifies the risk for RVF after LVAD support by implementing routinely collected clinical data. It could impact patient selection and peri-operative management of advanced HF patients.