About the SAFE Score (Steatosis-Associated Fibrosis Estimator)
The SAFE (Steatosis-Associated Fibrosis Estimator) score combines seven routine values — age, BMI, diabetes status, AST, ALT, globulin and platelet count — to estimate the risk of significant fibrosis (stage F2 or higher) in metabolic (MASLD/NAFLD) liver disease. Below 0 the score rules out significant fibrosis with a negative predictive value of 88–92%; at or above 100 it rules it in; between 0 and 100 is indeterminate. It was designed for primary care and, unlike FIB-4, handles age as a continuous variable rather than through fixed thresholds, which is why it performs better than FIB-4 in the general population and at the extremes of age.
Formula
SAFE = 2.97·Age + 5.99·BMI + 154.85·ln(AST) − 58.23·ln(ALT) + 195.48·ln(globulin) − 141.61·ln(platelets) + 62.85·(diabetes) − 75- Age
- Years, weight 2.97.
- BMI
- kg/m², weight 5.99.
- AST
- U/L, natural log, weight 154.85.
- ALT
- U/L, natural log, weight −58.23.
- globulin
- g/dL (total protein − albumin), natural log, weight 195.48.
- platelets
- ×10⁹/L, natural log, weight −141.61.
- diabetes
- 1 if present, 0 if absent; weight 62.85.
- The AST, ALT, globulin and platelet terms are natural logarithms (ln); age and BMI enter linearly.
- Globulin is total serum protein minus albumin, in g/dL. If reported in g/L, divide by 10 first (or use the unit toggle).
- The coefficients are large because the score is scaled so that its decision points fall at 0 and 100 — the raw number is not a probability and not a stage.
Interpreting the result
Treat SAFE as a triage tool with a wide safe zone below 0. A negative score excludes significant fibrosis with a negative predictive value near 90% and, in population data, marks a group whose long-term survival matches people without any fatty liver — these patients can usually stay in primary care with metabolic management. A score of 100 or above marks high risk and should route the patient to specialist assessment and elastography. The 0–100 band is indeterminate and must be worked up further rather than treated as normal. The score's advantage over FIB-4 is concentrated in the young and the old and in shrinking that indeterminate group, so it is most useful precisely where FIB-4 is least reliable.
| Score | Band | What it means | Action |
|---|---|---|---|
| < 0 | Low risk | Significant fibrosis (≥ F2) ruled out — NPV 88–92%; 25-year survival like people without steatosis | Manage metabolic risk in primary care; reassess over time |
| 0–100 | Intermediate risk | Indeterminate — neither ruled in nor out | Proceed to a second-line test (elastography or an imaging/blood fibrosis score) |
| ≥ 100 | High risk | High probability of significant fibrosis; correlated with shorter survival (adjusted HR ≈ 1.5) | Refer for specialist fibrosis assessment |
What the SAFE Score needs (6 inputs)
- Age (years)
- Enters linearly with a positive weight. Unlike FIB-4, SAFE uses age as a continuous term rather than against fixed thresholds, which is the main reason it behaves better at the extremes of age.
- BMI (kg/m²)
- Body mass index, positive weight — the metabolic-load term. Along with diabetes it anchors the score in the metabolic phenotype that FIB-4 ignores.
- Diabetes (yes/no)
- A large positive contribution when present, reflecting how strongly type 2 diabetes raises the probability of significant fibrosis in fatty liver disease.
- AST and ALT (U/L)
- Both enter as natural logarithms — AST with a large positive weight, ALT with a negative one, so a rising AST:ALT pattern drives the score up, as in FIB-4 and the NAFLD Fibrosis Score.
- Globulin (g/dL)
- Total serum protein minus albumin, entering as a natural log with a large positive weight. A rising globulin is a marker of chronic liver injury. This is the one input not always on a basic panel.
- Platelet count (×10⁹/L)
- Enters as a natural log with a negative weight — the portal-hypertension marker, as in FIB-4. A falling platelet count raises the score.
What it returns
- SAFE score
- A continuous value on a scale where the decision points are 0 and 100. It is a risk estimate for significant fibrosis, not a fibrosis stage.
- Risk band
- Low (< 0), intermediate (0–100) or high (≥ 100). The intermediate band is an instruction to test further, not a reassuring result.
How it is calculated
SAFE is a regression model fitted against biopsy-defined significant fibrosis and then validated against liver stiffness and long-term mortality. It reuses the markers that make FIB-4 and the NAFLD Fibrosis Score work — the AST:ALT pattern and a falling platelet count — but adds the metabolic drivers (BMI, diabetes) and a marker of chronic injury (globulin), and, critically, keeps age as a continuous variable instead of binning it. Modelling age continuously is what removes FIB-4's age-related bias, because the score no longer flips a patient across a threshold on a birthday. The output is deliberately rescaled so the two operating points sit at round numbers, 0 for rule-out and 100 for rule-in.
Facts & figures
| SAFE | Band | Meaning |
|---|---|---|
| < 0 | Low risk | Rules out ≥ F2 (NPV 88–92%) |
| 0–100 | Intermediate | Second-line testing required |
| ≥ 100 | High risk | Rules in ≥ F2; refer |
Target is significant fibrosis (stage F2 or higher).
| Feature | SAFE | FIB-4 |
|---|---|---|
| Inputs | Age, BMI, DM, AST, ALT, globulin, platelets | Age, AST, ALT, platelets |
| Age handled as | Continuous variable | Fixed thresholds |
| General-population discrimination | Higher | Lower |
| Prognostic (long-term mortality) | Yes | Yes, but less well in the young |
SAFE was built to outperform FIB-4 specifically in primary-care and general populations.
Evidence
Derivation — biopsy-defined fibrosis
2023Developed against biopsy-confirmed fibrosis stage in metabolic liver disease cohorts, selecting age, BMI, diabetes, AST, ALT, globulin and platelets and scaling the model so its rule-out and rule-in points fell at 0 and 100.
At SAFE < 0, negative predictive values for ruling out ≥ F2 were 88% and 92% in the two testing sets. The score discriminated significant fibrosis better than FIB-4, particularly at the extremes of age.
General US population validation
2023Applied to the general adult US population using liver stiffness as the fibrosis reference, and to NHANES III with 25-year mortality follow-up.
Outperformed FIB-4 for identifying significant fibrosis and reclassified many FIB-4-indeterminate patients. SAFE < 0 marked survival comparable to people without steatosis; SAFE > 100 carried an adjusted hazard ratio of about 1.5 for mortality.
How it compares
SAFE Score vs FIB-4
SAFE was purpose-built to beat FIB-4 in primary care — it discriminates significant fibrosis better in the general population and at age extremes, and leaves fewer indeterminate results — at the cost of three extra inputs (BMI, diabetes, globulin).
FIB-4 is simpler and guideline-embedded, using only age, AST, ALT and platelets, but its fixed age thresholds cause systematic error in the young and old and it leaves about a third of patients indeterminate. SAFE models age continuously and adds metabolic and chronic-injury markers, which is why it reclassifies many FIB-4-indeterminate patients correctly. Where the two disagree at the extremes of age, SAFE is the more reliable; where simplicity and existing data matter, FIB-4 remains the pragmatic first step.
SAFE Score vs NAFLD Fibrosis Score
They share most inputs and both target advanced/significant fibrosis, but SAFE swaps albumin for globulin, keeps age continuous, and was validated for primary-care case-finding with a mortality link, making it the more modern general-population tool.
The NAFLD Fibrosis Score uses age, BMI, glucose/diabetes, AST, ALT, platelets and albumin; SAFE uses the same metabolic backbone but substitutes globulin for albumin and was calibrated and validated specifically for community case-finding, including 25-year survival data. In a specialist clinic either performs well; in unselected primary-care populations SAFE has the stronger validation.
SAFE Score vs Fibrotic NASH Index (FNI)
Different endpoints — SAFE estimates significant fibrosis (≥ F2), FNI estimates at-risk NASH (activity plus fibrosis) — so choose the score whose target matches the decision at hand.
SAFE answers 'is there significant fibrosis' for case-finding and referral; FNI answers 'is there the active, fibrotic NASH phenotype that treatment targets'. A patient can be SAFE-positive for fibrosis while their FNI is aimed at whether that fibrosis is accompanied by treatment-relevant activity. They are complementary rather than competing.
Pearls & pitfalls
- SAFE estimates the risk of significant fibrosis (≥ F2); it is not a fibrosis stage and does not diagnose steatohepatitis.
- Globulin is total protein minus albumin, in g/dL — it is the one input not always on a basic panel, and getting its units wrong (g/L versus g/dL) badly distorts a logarithmic term.
- Its edge over FIB-4 is greatest in the young and the old; in middle age the two agree more often.
- The 0–100 intermediate band still requires a second-line test — treating it as negative is the same mistake as ignoring FIB-4's indeterminate zone.
- The coefficients are large and the raw score is not a probability; read the band, not the magnitude.
- It shares FIB-4's vulnerability to non-hepatic causes of a low platelet count or a raised AST.
- It was derived and validated in metabolic liver disease and should not be applied to viral or alcohol-related disease without separate evidence.
Critical actions
- Confirm globulin is entered in g/dL (total protein − albumin); a units error here is the commonest way to get SAFE badly wrong.
- Act on the intermediate band with a second-line test rather than filing it as normal.
- At high risk, refer for specialist assessment and elastography rather than repeating bloods.
- Where FIB-4 and SAFE disagree in a young or old patient, weight SAFE, since that is where it was designed to be more reliable.
- Treat the metabolic drivers — weight, glycaemic control, cardiovascular risk — at every band; a low SAFE does not make metabolic disease benign.
Why this score exists
SAFE was built as a direct answer to a known failing of FIB-4 in the setting where most fatty liver disease is actually found — primary care. Its authors' central design decision was to stop treating age as a switch and model it continuously, which is what removes the systematic over- and under-calling FIB-4 shows at the ends of the age range. Adding BMI, diabetes and globulin brought in the metabolic and chronic-injury information FIB-4 lacks. The choice to rescale the output so the thresholds land at 0 and 100 is cosmetic but deliberate: it makes a large-coefficient regression easy to communicate as a simple three-band result.
About the creator
First author, 2022 derivation study
Derived the steatosis-associated fibrosis estimator to improve on FIB-4 in primary-care populations.
Senior author
Also senior author of the MELD 3.0 revision.
Limitations
- It estimates the risk of significant fibrosis, not a fibrosis stage, and does not diagnose steatohepatitis.
- It requires BMI, diabetes status and globulin in addition to the routine FIB-4 inputs, and globulin is not on every basic panel.
- The intermediate 0–100 band still needs second-line testing, so it does not by itself resolve every case.
- It shares FIB-4's sensitivity to non-hepatic causes of thrombocytopenia and transaminitis.
- The large-coefficient formula is not mentally computable, so it depends on a calculator and on correct input units.
- It was derived and validated in metabolic (MASLD/NAFLD) liver disease and should not be transferred to other liver diseases without dedicated evidence.
If you are the patient
The SAFE score is a way of estimating how likely it is that fatty liver disease has caused significant scarring in the liver, using information that is easy to collect: your age, your body mass index, whether you have diabetes, and four blood results (two liver enzymes, a protein called globulin, and your platelet count). It was designed for use in general practice, and it works better than the older FIB-4 score for younger and older people. A score below 0 means significant scarring is unlikely — reassuringly, people with a score this low tend to live just as long as people with no fatty liver at all — and you would usually be managed with lifestyle and metabolic care. A score of 100 or more means scarring is more likely and you would be referred to a specialist. A score in between means the test could not tell, and another test such as a liver scan is needed. The main thing your team checks is that the globulin value is entered in the right units, because that part of the calculation is sensitive to it.
Frequently asked questions
What is the SAFE score?#
The SAFE (Steatosis-Associated Fibrosis Estimator) score is a blood-and-clinical score that estimates the risk of significant liver fibrosis (stage F2 or higher) in metabolic fatty liver disease. It combines age, BMI, diabetes, AST, ALT, globulin and platelets, and was designed for first-line case-finding in primary care.
What SAFE score rules out significant fibrosis?#
A SAFE score below 0 is the rule-out result: significant fibrosis is unlikely, with a negative predictive value of 88–92%. In population data, people with a SAFE below 0 had 25-year survival comparable to people without any steatosis, underlining how reassuring a negative result is.
Is the SAFE score better than FIB-4?#
In primary-care and general populations, yes — SAFE discriminated significant fibrosis better than FIB-4 and reclassified many of FIB-4's indeterminate results, mainly because it models age continuously instead of using fixed thresholds. FIB-4 remains simpler (four inputs) and is embedded in guidelines, so it is still a reasonable first step, but SAFE is the more reliable of the two at the extremes of age.
What does a SAFE score above 100 mean?#
It falls in the high-risk band: significant fibrosis is likely and specialist assessment with elastography is warranted. A score above 100 also carried an adjusted hazard ratio of about 1.5 for long-term mortality in population data, so it flags both fibrosis risk and prognosis.
What units does the SAFE calculator need?#
Age in years, BMI in kg/m², AST and ALT in U/L, platelets in ×10⁹/L, and globulin in g/dL (globulin being total serum protein minus albumin). Because globulin and the enzymes enter as natural logarithms, using the wrong units — particularly g/L instead of g/dL for globulin — will produce a badly wrong score.
Does SAFE replace a FibroScan or biopsy?#
No. SAFE is a triage tool that excludes significant fibrosis cheaply in low-risk patients and flags higher-risk ones for further testing. Those with an intermediate or high score still need a second-line test such as elastography, and occasionally biopsy, to confirm and stage the fibrosis.
References
Original / primary reference
Clinical practice guidelines
- Berzigotti A, Tsochatzis E, Boursier J, et al. EASL Clinical Practice Guidelines on non-invasive tests for evaluation of liver disease severity and prognosis – 2021 update. J Hepatol. 2021;75(3):659-689.
- EASL-EASD-EASO Clinical Practice Guidelines on the management of metabolic dysfunction-associated steatotic liver disease (MASLD). J Hepatol. 2024;81(3):492-542.
- Kanwal F, Shubrook JH, Adams LA, et al. Clinical Care Pathway for the Risk Stratification and Management of Patients With Nonalcoholic Fatty Liver Disease. Gastroenterology. 2021;161(5):1657-1669.
Further reading
- Sterling RK, Lissen E, Clumeck N, et al. Development of a simple noninvasive index to predict significant fibrosis in patients with HIV/HCV coinfection (FIB-4). Hepatology. 2006;43(6):1317-1325.
- Tavaglione F, De Vincentis A, Jamialahmadi O, et al. Development and Validation of a Score for Fibrotic Nonalcoholic Steatohepatitis (FNI). Clin Gastroenterol Hepatol. 2023;21(6):1523-1532.e1.