*Purpose: The MELD-Na score widely used for medical decision making but is not a good predictor of patient outcomes at low scores. Although portal hypertensive complications of cirrhosis are known to negatively affect patient outcomes, translation to MELD-Na points is not clear.

*Methods: HealthLNK, a population based cohort of the greater Chicago area containing de-duplicated electronic health record data from seven large health systems was linked to the state death registry and the national transplant database (UNOS). Patients were included if they did not have a diagnosis of hepatocellular carcinoma, had an ICD9 diagnosis code of cirrhosis with either ascites, variceal bleeding (VB), or encephalopathy (HE), and had at least one MELD-Na measurement during 2006-2012. Blinded death certificate review for cause of death was performed for patients who died during the study period. Multivariate competing risk analysis was performed to model liver related cause of death and incorporated age, gender, Insurance status, Elixhauser score, etiology of cirrhosis, ascites, HE, variceal bleeding, and peak MELD-Na score. The R package rms to was use to translate linear predictors from the model to MELD-Na scores.

*Results: During the study period, 4,637 patients met the inclusion criteria and contributed a total of 10,588 person years of follow up. The patients were an average of 55 years old, 40% female, 46% White, and 49% insured by Medicare/Medicaid with an average Elixhauser Comorbidity score of 6.8. The most common etiologies of cirrhosis were alcohol (56%) and Hepatitis C (35%). Patients had an average minimum MELD-Na of 15 and average peak of 22, 76% had ascites, 71% had HE, 5% of patients had variceal bleeding. 1,871 (40%) of patients died without transplantation, most commonly of a liver-related cause (n=1,248; 67%) and 487 (11%) underwent transplantation. Multivariate competing risk analysis demonstrated that ascites and HE conferred increased MELD-Na whereas VB conferred a decreased risk of bleeding (Table). A nomogram was developed based on this model (Figure).

*Conclusions: Risk equivalency can be translated to MELD-Na units, and can used as a nomogram calculator (Figure). Future work requires validation in different population cohort.

« Back to 2020 American Transplant Congress