Abstract
BACKGROUND AND AIM: Hospice is underutilized in the management of patients with end-stage liver disease and may improve the patient experience at the end of life. This study aims to create a novel prognostic scale to accurately predict 6-month mortality to more comprehensively facilitate hospice referral. METHODS: Sociodemographic, clinical, and laboratory variables associated with mortality from the United Network for Organ Sharing database were tested in univariate analysis followed by multivariate analyses with four predictor groups: Demographics, Diagnoses, Complexities, and Laboratory studies to develop the hospice in end-stage liver disease prognostic scale (HELP) scale (70% sample, N = 13 516) followed with replication in a 30% (N = 5792) internal validation sample. RESULTS: Only the predictor groups of Complexities and Laboratory studies met the c-statistic threshold of 0.70 for inclusion in the multivariate analyses. Backward elimination in the final logistic regression and validated weighted transformation procedure resulted in: HELP scale = (functional status × 11) + (ascites × 3) + (SBP × 3) + (HE × 4) + (dialysis × 5) + (TIPS × -3) + (albumin × -3) + (MELD-Na ≥ 21 × 20). HELP scale had a strong predictive value for six-month mortality with Area under the Receiver Operating Curve (AUROC) 0.816 and replicated in the validation sample. CONCLUSION: HELP scale is a novel prognostic score utilizing the strength of model of end-stage liver disease-sodium (MELD-Na), along with clinical factors, for a more nuanced assessment of six-month mortality. This scale can provide an individualized approach in opening discussions of hospice referral and may be better accepted by patients and providers given its contextualization of important clinical factors.