DEVELOPMENT AND PROSPECTIVE VALIDATION OF A CLINICAL INDEX TO PREDICTSURVIVAL IN AMBULATORY PATIENTS REFERRED FOR CARDIAC TRANSPLANT EVALUATION

Background Risk stratification of patients with end-stage congestive h eart failure is a critical component of the transplant candidate selec tion process. Accurate identification of individuals most likely to su rvive without a transplant would facilitate more efficient use of scar ce donor organs. Methods and Results Multivariable proportional hazard s survival models were developed with the use of data on 80 clinical c haracteristics from 268 ambulatory patients with advanced heart failur e (derivation sample). Invasive and noninvasive models (with and witho ut catheterization-derived data) were constructed. A prognostic score was determined for each patient from each model. Stratum-specific like lihood ratios were used to develop three prognostic-score risk groups. The models were prospectively validated on 199 similar patients (vali dation sample) by calculation of the area under the receiver operating characteristic curve for 1-year event-free survival, the censored c-i ndex for event-free survival, and comparison of event-free survival cu rves for prognostic-score risk strata. Outcome events were defined as urgent transplant or death without transplant. The noninvasive model p erformed well in both samples, and increased performance was not attai ned by the addition of catheterization-derived variables. Prognostic-s core risk groups derived from the noninvasive model in the derivation sample effectively stratified the risk of an outcome event in both sam ples (1-year event-free survival for derivation and validation samples , respectively: low risk, 93% and 88%; medium risk, 72% and 60%; high risk, 43% and 35%). Conclusions Selection of candidates for cardiac tr ansplantation may be improved by use of this noninvasive risk-stratifi cation model.