Development and validation of a practical prognostic nomogram for evaluating inpatient mortality of cirrhotic patients with acute variceal hemorrhage.

INTRODUCTION AND

OBJECTIVES

Cirrhotic patients with acute variceal hemorrhage (AVH) have high short-term mortality. Established prognostic scores are seldom applicable clinically, partially because they need external validation or contain subjective variables. We aimed to develop and validate a practical prognostic nomogram based on objective predictors to predict prognosis for cirrhotic patients with AVH.

PATIENTS AND

METHODS

We enrolled 308 AVH patients with cirrhosis from our center as the derivation cohort to develop a new nomogram using logistic regression and validated it in cohorts of patients from Medical Information Mart for Intensive Care (MIMIC) III (n = 247) and IV (n = 302).

RESULTS

International normalized ratio (INR), albumin (ALB) and estimated glomerular filtration rate (eGFR) were identified as predictors for inpatient mortality and a nomogram was constructed based on them. The nomogram discriminated well in both derivation and MIMIC-III/-IV validation cohorts with the area under the receiver operating characteristic curves (AUROCs) of 0.846 and 0.859/0.833, respectively and showed a better agreement between expected and observed outcomes (Hosmer-Lemeshow tests, all comparisons, P > 0.05) than other scores in all cohorts. Our nomogram had the lowest Brier scores (0.082/0.114/0.119 in training/MIMIC-III/MIMIC-IV) and highest R2 (0.367/0.393/0.346 in training/MIMIC-III/MIMIC-IV) compared to the recalibrated model for end-stage liver disease (MELD), MELD-hepatic encephalopathy (MELD-HE) and cirrhosis acute gastrointestinal bleeding (CAGIB) scores in all cohorts.

CONCLUSIONS

We developed a practical prognostic nomogram using easily verified indicators available in initial patient evaluation, which may serve as a reliable tool to accurately predict inpatient mortality for cirrhotic patients with AVH.