Fatty liver index calculator

The formula

The index comes from a logistic regression model of four numbers, multiplied by 100 so it runs from 0 to 100 (Bedogni 2006):

L = 0.953 × ln(triglycerides, mg/dL) + 0.139 × BMI (kg/m²) + 0.718 × ln(GGT, U/L) + 0.053 × waist (cm) - 15.745

FLI = e^L ÷ (1 + e^L) × 100

Here ln is the natural log. If your lab reports triglycerides in mmol/L, we multiply by 88.57 to get mg/dL; a waist in inches is multiplied by 2.54. For example, triglycerides of 141 mg/dL, a BMI of 29.5, GGT of 31 U/L and a waist of 98 cm (the median values of the people with fatty liver in the study (Bedogni 2006)) give L = 0.73 and an index of about 68.

The researchers started with 13 candidate measures, including age, alcohol intake, ALT, AST, glucose, insulin and skinfolds, and kept the four that predicted fatty liver best and are measured routinely (Bedogni 2006). Waist size made the largest contribution, followed by BMI, triglycerides and GGT (Bedogni 2006). The final model had an accuracy (area under the curve) of 0.84 (Bedogni 2006).

The cut points: under 30 and 60 or more

In plain words: of the people in the study who had fatty liver on ultrasound, 87% scored 30 or more, so a score under 30 made fatty liver less likely. Of those without it, 86% scored under 60, so a score of 60 or more made it more likely (Bedogni 2006). These numbers come from the same group the index was built in, which tends to make a test look better than it will elsewhere.

A research index, with limits

The authors suggested the index could help doctors choose who to send for an ultrasound and who might gain from more lifestyle advice, and help researchers pick people for studies. They also wrote that it is very important to validate it in other populations before it is used in practice (Bedogni 2006). They named two limits of their study: only 58% of the town's residents took part, and ultrasound cannot tell plain fat from fat with inflammation (steatohepatitis) (Bedogni 2006).

Ultrasound itself misses smaller amounts of fat, especially in people with obesity, and a liver that looks normal on ultrasound can still have inflammation or scarring (AASLD 2023). Because the index was built to match ultrasound findings, it shares those blind spots.

Fat is not scarring: FIB-4 comes first

The question that matters most in fatty liver disease is whether the liver is scarred (fibrosis) (AASLD 2023). The fatty liver index does not answer it. The American Association for the Study of Liver Diseases recommends FIB-4 as the first test for advanced scarring (AASLD 2023). If you have AST, ALT and a platelet count, use the FIB-4 calculator. The NAFLD fibrosis score is a second scarring score for people who already have fatty liver disease.

Who should think about this

Fatty liver disease is common in people with overweight, type 2 diabetes or high blood pressure, and usually has no symptoms (NIDDK). NIDDK estimates that about 24% of US adults have it (NIDDK). Read fatty liver disease (MASLD), explained for the background. Several of the index's inputs also track insulin resistance: the TyG index calculator uses fasting triglycerides and glucose, and the waist-to-height ratio looks at waist size on its own, and the lipid accumulation product combines waist size with fasting triglycerides. The hepatic steatosis index is another research index for liver fat, built in Korea from ALT, AST, BMI, sex and diabetes. A high index is a reason to talk to your doctor, not a diagnosis.