Abstract 10680: Risk-Prediction of in-hospital Mortality After Heart Transplantation Using Recipient and Donor Variables at Transplant
Background: A quantitative model to predict the risk of early post-transplant mortality after heart transplant (HT) has not been validated. We sought to derive and validate a risk-prediction model for post-transplant in-hospital mortality in HT recipients using variables available at transplant.
Methods: We derived the model using data in all recipients >18 years old who underwent their first HT in the US during 01/07-06/09 (n=4248). We validated it internally using bootstrapping (200 random samples, n=4248 each) and externally in an independent cohort of HT recipients transplanted in the US during 07/09-10/10 (n=2346).
Results: Post-transplant in-hospital mortality was 4.7% in the derivation cohort. The recipient-based model had 6 variables (Figure): age (>65 yrs), diagnosis (ischemic, restrictive/hypertophic, congenital heart disease vs. dilated/valvar cardiomyopathy), mechanical support (ECMO, total artificial heart, BIVAD, LVAD vs. none), ventilator support, GFR (30-59, <30, dialysis vs. >60 ml/min/1.73 m2) and serum bilirubin (1.0-2.5, >2.5 vs. <1.0 mg/dl). Model discrimination and calibration in the derivation cohort (C-statistic 0.722, Hosmer-Lomeshow [HL] P value 0.56, Figure) and during internal validation (C-statistic 0.721) were acceptable. Model performance in the independent cohort was reasonable (predicted mortality 4.9%, actual mortality 4.3%, R2= 0.95, c-statistic 0.68, HL P value = 0.32). The addition of donor age (>55 yrs, 40-55 yrs vs. <40 yrs) and ischemic time over 4.5 hrs to the model improved model performance in both derivation (C-statistic 0.742, HL P-value 0.67) and validation (C-statistic 0.686, HL P-value 0.34) cohorts.
Conclusions: This risk-prediction model using 6 baseline recipient variables has potential utility during decision-making for HT candidacy and in counseling patients regarding their post-transplant prognosis. The model with additional donor variables may be useful during donor selection.
- © 2011 by American Heart Association, Inc.