Real World Performance of the Model for End-Stage Liver Disease Score Across Different Races and Ethnicities

Area of Science:

• Hepatology
• Transplant medicine
• Health equity

Background:

• The Model for End-Stage Liver Disease (MELD) score is crucial for liver disease severity assessment and organ allocation in the U.S.
• Existing literature lacks comprehensive analysis of MELD-Na and MELD 3.0 performance across diverse racial and ethnic populations.
• Understanding potential score disparities is vital for equitable liver transplant candidacy evaluation.

Purpose of the Study:

• To evaluate the performance of MELD-Na and MELD 3.0 scores in predicting mortality or liver transplant (LT) across different racial and ethnic groups.
• To identify any disparities in the discriminative and calibration abilities of these scores among Asian, White, African American, and Hispanic patients with cirrhosis.
• To compare score performance between two distinct healthcare datasets (MIMIC-IV and Emory University).

Main Methods:

• Retrospective cohort study utilizing data from MIMIC-IV and Emory University databases.
• Inclusion of patients diagnosed with cirrhosis across specified racial and ethnic categories.
• Analysis of MELD score accuracy using Area Under the Receiver Operating Characteristic Curve (AUROC) for discrimination and calibration plots for calibration.

Main Results:

• No significant differences in discriminative ability for in-hospital death or LT were observed across racial/ethnic groups in either dataset (p > 0.05).
• A minor influence on 90-day outcomes was noted in a small Asian patient subgroup within the MIMIC-IV data.
• Both MELD scores tended to overpredict outcomes at higher values, with MELD-Na showing greater overprediction in African Americans within MIMIC-IV.

Conclusions:

• MELD-Na and MELD 3.0 demonstrate consistent discriminative performance across major racial and ethnic groups.
• Outcome prediction was superior in the Emory cohort compared to MIMIC-IV, suggesting site-specific influences and potential relevance of the Emory cohort's characteristics to MELD score derivation populations.
• Observed disparities in cirrhosis care are more likely attributable to socioeconomic factors than inherent biases in MELD score performance.