About the MELD 3.0 (Model for End-Stage Liver Disease, 2021 revision)
MELD 3.0 is the current US liver transplant allocation score, adopted by UNOS in 2023. It keeps MELD-Na's four variables — bilirubin, INR, creatinine and sodium — and adds serum albumin plus 1.33 points for female sex, while lowering the creatinine cap from 4.0 to 3.0 mg/dL. It discriminates 90-day waiting-list mortality slightly better than MELD-Na (C-statistic 0.869 versus 0.862) and, more importantly, corrects a sex disparity: it reclassified a net 8.8% of patients who died to a higher MELD tier, particularly women.
Formula
MELD 3.0 = 1.33 (if female)
+ 4.56 × ln(bilirubin)
+ 0.82 × (137 − Na) − 0.24 × (137 − Na) × ln(bilirubin)
+ 9.09 × ln(INR)
+ 11.14 × ln(creatinine)
+ 1.85 × (3.5 − albumin) − 1.83 × (3.5 − albumin) × ln(creatinine)
+ 6- ln
- Natural logarithm, which is why bilirubin, INR and creatinine are all floored at 1.0.
- bilirubin
- mg/dL, floored at 1.0. Also appears in an interaction with sodium.
- Na
- Serum sodium in mEq/L, bounded to 125–137.
- INR
- Floored at 1.0.
- creatinine
- mg/dL, floored at 1.0 and capped at 3.0. Set to 3.0 after two or more dialysis sessions in the past week.
- albumin
- g/dL, bounded to 1.5–3.5. Also appears in an interaction with creatinine.
- 1.33 (female)
- Added for female sex, correcting the under-prioritisation of women under earlier MELD versions.
- Two interaction terms make this genuinely different in shape from MELD-Na: sodium with bilirubin, and albumin with creatinine. The effect of each variable therefore depends on the others, which is why MELD 3.0 cannot be derived from a MELD-Na score.
- The result is rounded and bounded to 6–40.
- The creatinine cap is 3.0 mg/dL, not 4.0. This is one of the easiest differences from MELD-Na to overlook.
- For SI units, divide bilirubin in µmol/L by 17.1, creatinine in µmol/L by 88.4, and albumin in g/L by 10 — or use the unit toggles on this page.
Interpreting the result
Read it exactly as you would MELD-Na — as a 90-day mortality estimate driving waiting-list priority — but do not compare a MELD 3.0 with a MELD-Na as though they were the same measurement. The scales coincide at 6 and 40 and the interaction terms mean the two diverge unpredictably in between, so a patient whose score 'went up' when a centre switched formulas has not necessarily deteriorated. As with all MELD variants, the trend across repeated measurements carries more information than any single value, and a low score does not exclude significant frailty, sarcopenia or refractory ascites, none of which appear in the equation.
| Score | Band | What it means | Action |
|---|---|---|---|
| ≤ 9 | Low severity | Low 90-day waiting-list mortality | Treat the underlying liver disease; no transplant-specific action from the score alone |
| 10–19 | Moderate severity | Intermediate 90-day mortality; rising waiting-list priority | Transplant evaluation; identify and treat precipitants |
| 20–29 | High severity | High 90-day mortality | Active transplant work-up; escalate level of care |
| 30–40 | Very high severity | Very high 90-day mortality without transplant | Urgent transplant assessment and critical care input |
Scroll the table sideways for every column.
What the MELD 3.0 needs (7 inputs)
- Total bilirubin (mg/dL)
- Floored at 1.0, as in every MELD variant. It also appears in an interaction term with sodium, so its effect on the score depends on how hyponatraemic the patient is.
- INR
- Floored at 1.0. Still distorted by warfarin and direct oral anticoagulants — MELD 3.0 did not solve that, and MELD-XI remains the alternative for anticoagulated patients.
- Serum creatinine (mg/dL)
- Floored at 1.0 and capped at 3.0 — lower than MELD-Na's 4.0. The reduced cap limits how much renal dysfunction can dominate the score, part of the same correction that addresses the sex disparity.
- Serum sodium (mEq/L)
- Bounded to 125–137. It appears both on its own and in an interaction with bilirubin, so the penalty for hyponatraemia is not a fixed number of points.
- Serum albumin (g/dL)
- New in MELD 3.0, bounded to 1.5–3.5. It enters both directly and in an interaction with creatinine. Note that recent albumin infusion — after large-volume paracentesis, for instance — will raise the measured value without any change in synthetic function.
- Sex
- Female sex adds 1.33 points. This is a deliberate equity correction, not a biological claim about disease severity: it compensates for the systematic under-prioritisation of women under earlier MELD versions.
- Dialysis in the past week
- Two or more sessions of haemodialysis, or 24 hours of continuous veno-venous haemodialysis, in the previous seven days. Sets creatinine to the 3.0 cap.
Units. Enter bilirubin and creatinine in mg/dL and albumin in g/dL, or switch the unit toggles to µmol/L and g/L and the calculator converts exactly (bilirubin ÷ 17.1, creatinine ÷ 88.4, albumin ÷ 10). Sodium is in mEq/L, numerically identical to mmol/L. Remember the creatinine cap here is 3.0 mg/dL, which is 265 µmol/L — lower than MELD-Na's 4.0 mg/dL.
What it returns
- MELD 3.0 score (6–40)
- An integer on the same 6–40 scale as MELD-Na, so the numbers look comparable — but they are not interconvertible, and the same patient can score differently under each.
- Estimated 90-day mortality band
- The severity band the score falls into. MELD 3.0 was fitted against 90-day waiting-list mortality specifically.
How it is calculated
Kim and colleagues refitted the MELD model on all US liver transplant waiting-list candidates from January 2016 to December 2018, testing candidate variables against 90-day mortality. Female sex and serum albumin were retained as independent predictors beyond the MELD-Na components, and two interaction terms were introduced because the effect of sodium proved to depend on bilirubin and the effect of albumin on creatinine. Lowering the creatinine cap to 3.0 was part of the same reweighting: creatinine had been carrying more of the score than its true prognostic contribution justified, and that over-weighting was one of the mechanisms by which women — who tend to have lower creatinine at equivalent renal function because of lower muscle mass — were disadvantaged.
Facts & figures
| Feature | MELD-Na | MELD 3.0 |
|---|---|---|
| Bilirubin, INR, creatinine, sodium | Yes | Yes |
| Serum albumin | No | Yes, bounded 1.5–3.5 g/dL |
| Female sex adjustment | None | +1.33 points |
| Creatinine cap | 4.0 mg/dL | 3.0 mg/dL |
| Interaction terms | Sodium × MELD | Sodium × bilirubin; albumin × creatinine |
| Score range | 6–40 | 6–40 |
| US allocation status | Superseded | Current since 2023 |
Scroll the table sideways for every column.
| Metric | Value |
|---|---|
| Derivation cohort | All US waiting-list candidates, Jan 2016 – Dec 2018 |
| C-statistic, MELD 3.0 | 0.869 |
| C-statistic, MELD-Na | 0.862 (P < .01) |
| Decedents reclassified to a higher tier | Net 8.8%, particularly women |
| Simulated waiting-list deaths | 7,788 with MELD 3.0 vs 7,850 with MELD-Na (P = .02) |
Scroll the table sideways for every column.
Evidence
Derivation — US national transplant registry
2021All adult liver transplant waiting-list candidates registered in the US national registry between January 2016 and December 2018, refitting the MELD model against 90-day mortality and testing additional candidate variables.
C-statistic 0.869 for MELD 3.0 against 0.862 for MELD-Na (P < .01). MELD 3.0 correctly reclassified a net 8.8% of decedents into a higher MELD tier, affording them a higher chance of transplantation, particularly among women.
Simulation of allocation impact
2021Simulation modelling of the same registry population, comparing waiting-list deaths under each formula.
Fewer waiting-list deaths under MELD 3.0 than MELD-Na — 7,788 versus 7,850 (P = .02) — alongside the stated aim of addressing the sex disparity in access.
External validation
2025MELD 3.0 has since been validated in independent cohorts outside the United States, including a German clinical cohort study assessing it alongside ReMELD-Na.
Broadly reproduces the discrimination reported in the derivation work; comparative performance against locally derived alternatives varies by cohort.
How it compares
MELD 3.0 vs MELD-Na
MELD 3.0 has superseded MELD-Na for US allocation and is the one to use for current listing decisions; MELD-Na remains relevant for comparability with older literature and in jurisdictions that have not adopted the update.
MELD 3.0 adds albumin and 1.33 points for female sex, lowers the creatinine cap from 4.0 to 3.0, and introduces interaction terms between sodium and bilirubin and between albumin and creatinine. Its discrimination advantage is modest (C-statistic 0.869 versus 0.862) but it reclassified a net 8.8% of decedents to a higher tier and reduced simulated waiting-list deaths. Because of the interaction terms the two scores are not interconvertible — you cannot adjust a MELD-Na to get a MELD 3.0.
MELD 3.0 vs Child-Turcotte-Pugh
Use MELD 3.0 for transplant priority and prognosis, Child-Pugh whenever guidance is written in classes — the two are not substitutes and cannot be converted.
MELD 3.0 is a fitted survival model on six objective variables plus sex; Child-Pugh is an unweighted five-item score containing two subjectively graded items. MELD 3.0 discriminates 90-day mortality far better. But drug labels, trial eligibility and procedural guidance are expressed as Child-Pugh class A/B/C, and no MELD score answers that question, so both are usually recorded in patients under assessment.
MELD 3.0 vs MELD-XI
Use MELD-XI instead when the patient is anticoagulated, since MELD 3.0 still includes INR and is therefore still inflated by warfarin or a direct oral anticoagulant.
MELD 3.0 addressed the sex disparity and added albumin but left the INR problem untouched. In an anticoagulated patient the INR reflects the prescription as much as the liver, and the score overstates severity. MELD-XI drops INR entirely and uses bilirubin and creatinine only; it discriminates less well in patients who are not anticoagulated, so it is a targeted substitution rather than a general improvement.
Pearls & pitfalls
- The creatinine cap is 3.0, not 4.0. Carrying MELD-Na's cap over is the commonest error when moving between the two.
- The 1.33 points for female sex is an equity correction for measurement bias, not a statement that female sex worsens liver disease. Expect to have to explain that.
- MELD 3.0 and MELD-Na are not interconvertible. The interaction terms mean the difference between them varies by patient, so a score change on switching formulas is not clinical deterioration.
- Recent albumin infusion inflates the albumin term. A patient given albumin after large-volume paracentesis will score better without being better.
- Albumin is bounded to 1.5–3.5 g/dL, so a profoundly low albumin contributes no more than 1.5 does.
- INR remains distorted by anticoagulation. MELD 3.0 did not fix this — use MELD-XI when the patient is anticoagulated.
- The 6 and 40 bounds are unchanged, so the score still cannot distinguish among the sickest patients.
- PELD, not MELD 3.0, applies below age 12.
Critical actions
- Confirm which formula your centre or registry is using before quoting a score, since MELD-Na and MELD 3.0 give different numbers for the same patient.
- Check whether albumin has been infused recently before trusting the score.
- Check whether the patient is anticoagulated before trusting an INR-driven score.
- Track the trend across repeated measurements rather than acting on a single value.
- Look for a precipitant whenever the score rises — infection, variceal bleeding, hepatorenal syndrome, drug injury.
- Assess frailty and sarcopenia separately; both affect transplant outcome and neither is in the equation.
- Do not use MELD 3.0 to judge the risk of a specific operation — use a surgical risk model that accounts for the procedure.
Why this score exists
MELD 3.0 is unusual among clinical scores in that its principal purpose was fairness rather than accuracy. The discrimination gain over MELD-Na is real but small — a C-statistic of 0.869 against 0.862 — and on its own would not have justified changing a national allocation policy. What did justify it was evidence that women were being systematically under-prioritised relative to their actual mortality, partly because creatinine understates renal impairment in people with less muscle mass. The authors' response was twofold: add variables that capture severity independently of muscle mass (albumin), and add an explicit sex term to correct the residual gap. The 1.33 points for female sex is therefore not a claim that women's liver disease is worse at equivalent laboratory values — it is a correction for a known measurement bias in the other variables.
About the creator
First author, 2021 derivation study
Led the development of MELD 3.0, adding sex and albumin and recalibrating the coefficients to correct the disadvantage women faced under MELD-Na.
Senior author
Co-led the analysis of national registry data underlying the revision.
Limitations
- Derived in a US waiting-list population between 2016 and 2018, so it reflects that era's referral patterns, aetiologies and practice.
- The discrimination gain over MELD-Na is small in absolute terms; the substantive advance is the equity correction, which is harder to observe in an individual patient.
- The explicit sex term requires explanation and can be misread as a biological claim rather than a correction for measurement bias.
- Albumin is confounded by infusion, sepsis and proteinuria, none of which reflect synthetic liver function.
- INR remains distorted by anticoagulation.
- The 6–40 bounds still compress the extremes, so the sickest candidates remain indistinguishable at 40.
- It omits frailty, sarcopenia and quality of life, all of which affect transplant outcome.
- It is not validated in acute liver failure and does not apply below age 12.
If you are the patient
MELD 3.0 is the score used in the United States to decide how urgently someone needs a liver transplant, and it replaced an older version in 2023. It is worked out from five blood tests — bilirubin, INR, creatinine, sodium and albumin — plus whether you are male or female. A higher number means more advanced liver disease and a higher priority on the transplant waiting list. The reason for the update is worth knowing: under the old version, women were being placed lower on the list than their actual risk justified, largely because one of the blood tests (creatinine) tends to read lower in people with less muscle, making kidney problems look milder than they are. MELD 3.0 corrects for that, which is why the calculation includes sex. Two other things: the number can go up and down, and the direction over time matters more than any single reading; and the score does not capture everything your team considers, such as your strength, nutrition and general fitness, which they assess separately.
Frequently asked questions
What is MELD 3.0?#
MELD 3.0 is the current US liver transplant allocation score, adopted by UNOS in 2023. It builds on MELD-Na by adding serum albumin and 1.33 points for female sex, and lowering the creatinine cap from 4.0 to 3.0 mg/dL. It was fitted against 90-day waiting-list mortality in US candidates from 2016 to 2018.
What is the difference between MELD 3.0 and MELD-Na?#
MELD 3.0 adds albumin and a 1.33-point female-sex adjustment, lowers the creatinine cap from 4.0 to 3.0 mg/dL, and introduces interaction terms between sodium and bilirubin and between albumin and creatinine. Because of those interactions the two are not interconvertible — the same patient can score differently under each, and the difference varies by patient.
Why does MELD 3.0 add points for female sex?#
To correct a documented disparity in access to transplantation. Women were being under-prioritised relative to their actual mortality, partly because creatinine understates renal impairment in people with lower muscle mass. The 1.33 points is a correction for that measurement bias, not a claim that female sex makes liver disease worse at equivalent laboratory values.
Is MELD 3.0 more accurate than MELD-Na?#
Slightly, and meaningfully fairer. The C-statistic for 90-day mortality was 0.869 against 0.862 for MELD-Na (P < .01) — a small absolute gain. The more consequential finding was that MELD 3.0 correctly reclassified a net 8.8% of patients who died into a higher tier, particularly women, and simulation showed fewer waiting-list deaths (7,788 versus 7,850).
What is the creatinine cap in MELD 3.0?#
3.0 mg/dL, lower than MELD-Na's 4.0 mg/dL. Carrying the old 4.0 cap across is one of the most common errors when switching between the two formulas. Two or more dialysis sessions in the previous week set creatinine to the cap.
Does MELD 3.0 include albumin?#
Yes — albumin is one of the two variables added over MELD-Na, bounded to 1.5–3.5 g/dL, and it also appears in an interaction term with creatinine. Be aware that recent albumin infusion, for example after large-volume paracentesis, raises the measured value without any change in synthetic liver function.
When did UNOS start using MELD 3.0?#
UNOS implemented MELD 3.0 for liver allocation in 2023, following its publication in Gastroenterology in 2021. MELD-Na had been in use since 2016, and the original MELD since 2002.
Can MELD 3.0 be used in children?#
No. MELD variants apply from age 12 upward; PELD, the Pediatric End-Stage Liver Disease score, is used for younger candidates and includes growth failure and age under one year rather than sodium and sex.
References
Original / primary reference
Preceding formulas
- Kim WR, Biggins SW, Kremers WK, et al. Hyponatremia and mortality among patients on the liver-transplant waiting list. N Engl J Med. 2008;359(10):1018-1026.
- Kamath PS, Wiesner RH, Malinchoc M, et al. A model to predict survival in patients with end-stage liver disease. Hepatology. 2001;33(2):464-470.