Abstract
Research on the efficacy of the non-invasive fatty liver index (FLI) in excluding or diagnosing non-alcoholic fatty liver disease (NAFLD) remains limited, but it has been utilized in population research to stratify risk. We evaluated the efficacy of FLI in identifying NAFLD and the frequency of NAFLD in each FLI class by a meta-analysis. Methods. No results were found until January 2021 across all four databases (CRD42021231367). We considered original articles that reported FLI performance using CT, MRI, or ultrasound as the reference standard. Using 30 and 60 as cut-offs, the number of participants with NAFLD in FLI classes <30, 30-60, and ≥60, as well as the number of subjects categorized as true positive/negative, were extracted. The data was pooled using a random-effects model. End result. Fatty liver disease did not have any secondary causes in any of the ten investigations that involved 27,221 participants. There found a prevalence of NAFLD of 14%, 42%, and 67% in the three FLI groups, respectively. For the lower cut-off, the following metrics were measured: sensitivity (81%), specificity (65%), positive predictive value (84%), negative predictive value (65%), likelihood ratio for positive results (2.3), and diagnostic odds ratio (7.8). For the higher cut-off, the corresponding metrics were 44%, 90%, 67%, 76%, 4.3, 0.6, and 7.3. Studies that used ultrasound as opposed to other imaging modalities generally revealed comparable performance. Final thoughts. FLI was able to adequately stratify the risk of NAFLD. On the other hand, it failed to definitively exclude or diagnose this disease, and the evidence of discriminatory performance was minimal.