Development of a Fatty Liver Index Based on Real-World Data

Abstract

Background

The aim of this study was to develop and validate a fatty liver index based on laboratory data (FLI-L) and a fatty liver index based on both physical examination and laboratory data (FLI-PL) in the hope of providing a more convenient, accurate, and quantitative method for the diagnosis of fatty liver disease.

Methods

The study included data for 12,391 patients obtained from the Third Medical Center of Chinese PLA General Hospital. FLI-L and FLI-PL were developed using binary logistic regression analysis. The diagnostic performance of the FLI-L and FLI-PL was evaluated using the area under the receiver-operating characteristic curve (AUC-ROC) with sensitivity, specificity, and positive and negative likelihood ratios. FLI-L and FLI-PL were subsequently validated in 3170 patients from the same hospital.

Results

The AUC-ROC for FLI-L was 0.876 with a cut-off value of 55.03. Sensitivity was 81.35 and specificity was 78.28, with an accuracy of 79.99% for discriminating between patients with and without fatty liver disease. The AUC-ROC for FLI-PL was 0.902 with a cut-off value of 20.51. Sensitivity was 85.10 and specificity was 79.64. FLI-PL classified 91.65% of patients correctly.

Conclusion

FLI-L and FLI-PL is used for simple and accurate quantitative diagnosis of fatty liver disease. This study provides evidence to support the use of this index in clinical management.