Predicting Right Ventricular Failure Following Left Ventricular Assist Device Support: An INTERMACS Validation Study
Recommended Citation
Demertzis ZD, Ijaz N, Taleb I, Kyriakopoulos C, Peruri A, Chaudhary K, Dranow E, Alharethi R, Selzman C, Stehlik J, Koliopoulou A, Tang D, Shah P, Drakos S, Cowger J. Predicting Right Ventricular Failure Following Left Ventricular Assist Device Support: An INTERMACS Validation Study. 2022; :S53.
Document Type
Conference Proceeding
Publication Date
4-1-2022
Abstract
Introduction: : The Utah Right Ventricular Failure (RVF) Risk score is a 7-factor predictive model (African American, history of HTN, INTERMACS profile 1-2, Na <130mEq/L, BUN >35mg/dL, PA pulse pressure <36mmHg and RA/PCWP >0.5) that achieved a high discriminative performance (AUC 0.73) for predicting RVF in a three-center study of patients undergoing left ventricular assist device (LVAD) implantation and was validated in an external two-center cohort (AUC 0.71). Objective: : Validate a modified Utah RVF (mRVF) Risk score on a larger U.S. geographical scale. Methods: : INTERMACS data were requested for patients (age ≥19 years) undergoing isolated continuous flow LVAD implant between January 1, 2012 to through December 31, 2017. A modified (mRVF) score was computed without “history of hypertension” due to omission of variable on INTERMACS database. We excluded cases that were missing any of the 7-factor predictor variables. Primary outcome was early RVF, defined as the need for RVAD (BiVAD) support. Secondary outcomes were 90-day all-cause mortality. Multivariable logistic regression analysis was used to validate the predictive model. Results: : The study included 12550 (59.9%) INTERMACS cases that had all variables of the novel predictive model available. Of those cases, incidence of RVF was 3.37% with a median [25th, 75th] predictive mRVF score of 5 [3.38, 6.62]. In the patient population who did not have RVF post-LVAD implantation, their median [25th, 75th] predictive mRVF score was 4 [2.22, 5.78]. The novel predictive model had modest accuracy for predicting the risk of RVF post-LVAD implantation (AUC 0.678). The modified Utah RVF score was poorly predictive of 90-day mortality (area under the ROC curve 0.571). Conclusion: : The modified Utah RVF Risk score offered modest accuracy when applied to a large real world LVAD registry. Further studies examining the impact of unanticipated intra-operative and post-operative events on risk stratification are necessary.
First Page
S53
