Key Differences Between Maddrey Score and MELD Score
The Maddrey Discriminant Function (mDF) and MELD score differ fundamentally in their components, clinical applications, and prognostic timeframes: mDF uses only bilirubin and prothrombin time to predict 28-day mortality specifically in alcoholic hepatitis, while MELD incorporates bilirubin, INR, and creatinine to predict 90-day mortality in end-stage liver disease of any etiology.
Component Variables
The two scores are built from different laboratory parameters 1:
Maddrey Score (mDF):
- Bilirubin (mg/dL)
- Prothrombin time (PT) - specifically the difference between patient's PT and control PT
- Formula: mDF = 4.6 × (Patient's PT - Control PT) + Total bilirubin
MELD Score:
- Bilirubin (mg/dL)
- INR (International Normalized Ratio)
- Creatinine (mg/dL)
- Formula: 3.8 × log(bilirubin) + 11.2 × log(INR) + 9.6 × log(creatinine) + 6.4
The critical distinction is that MELD includes renal function (creatinine) while mDF does not 1, 2.
Clinical Context and Application
Maddrey Score:
- Developed specifically for alcoholic hepatitis in 1978, modified in 1989 1
- Remains the most widely used score in clinical practice and trials for alcoholic hepatitis 1
- Cut-off of ≥32 identifies severe alcoholic hepatitis requiring specific therapy 1
- Predicts 28-day (1-month) mortality 2
MELD Score:
- Originally developed for patients undergoing TIPS procedures 3
- Well-validated prognostic score for cirrhosis of any etiology 1
- Used by UNOS for liver transplant allocation 4
- Predicts 90-day (3-month) mortality 1, 2, 3
- In alcoholic hepatitis specifically, MELD >20 indicates high 90-day mortality risk 1
Prognostic Performance
Recent evidence reveals important differences in predictive accuracy:
The most recent and highest quality study (2022, global multicenter with 2,581 patients) demonstrated that MELD is superior to mDF for predicting short-term mortality in alcoholic hepatitis 5. The area under the ROC curve for 28-day mortality was 0.776 for MELD versus only 0.701 for mDF, and for 90-day mortality was 0.773 for MELD versus 0.709 for mDF. This study explicitly concluded that "the mDF score should no longer be used to assess AH's prognosis" 5.
However, this contradicts some earlier studies showing mDF superiority 6, 7, while other studies found them equivalent 8, 9. The 2022 global study is the largest and most geographically diverse, making it the most reliable evidence.
Practical Limitations
Maddrey Score Limitations:
- Functions essentially as a categorical classifier once the threshold of 32 is exceeded 2
- Does not specify degree of risk beyond the cut-off 2
- Does not account for renal dysfunction, which significantly impacts prognosis 4
MELD Score Limitations:
- Creatinine can be unreliable in cirrhosis patients 4
- Not validated specifically for non-transplant candidates 4
- May favor patients with renal dysfunction in transplant allocation 4
Current Guideline Recommendations
The 2018 EASL guidelines and 2010 Hepatology guidelines recommend using both scores complementarily 1, 2:
- Use mDF (≥32) to identify severe alcoholic hepatitis requiring immediate therapy
- Use serial MELD calculations to evaluate disease trajectory over time 2
The 2017 AGA expert review specifies that mDF >32 predicts high 28-day mortality, while MELD >20 predicts high 90-day mortality 10.
Clinical Algorithm
For patients with suspected alcoholic hepatitis:
- Calculate both scores on admission 2
- Use mDF ≥32 as the threshold for initiating corticosteroid therapy (if no contraindications) 1
- Use MELD >20 to identify patients at high risk of 90-day mortality who may need more aggressive monitoring or transplant evaluation 1
- Recalculate MELD serially to track disease progression or response to therapy 2
Common Pitfalls
- Do not use the original 1978 Maddrey formula—only the 1989 modified version is validated 1, 2
- Do not apply MELD cut-offs from transplant allocation (typically ≥15) to alcoholic hepatitis—use the higher threshold of >20 1
- Do not rely solely on mDF given recent evidence of MELD's superior performance 5
- Do not forget that both scores have limitations in patients with renal dysfunction or those requiring renal replacement therapy 4