Association between fibrosis-4 index and all-cause mortality by hepatic steatosis category in United States adults: A retrospective cohort study

BACKGROUNDThe hepatic steatosis index (HSI) and fibrosis-4 index (FIB-4) are readily calculated from routinely collected clinical data; they share aminotransferase components, while FIB-4 is also strongly influenced by age and platelet count.We hypothesized that the association between FIB-4 and mortality would vary across established HSI screening categories and that combining the indices would provide limited incremental prognostic information. AIMTo evaluate associations between FIB-4 and mortality across HSI categories and assess the incremental predictive value of both indices. METHODSThis retrospective cohort study included adults aged ≥ 45 years who participated in the 1999-2018 National Health and Nutrition Examination Survey, with mortality follow-up through 2019.HSI was categorized as < 30, 30-36, or > 36, and elevated FIB-4 was defined using age-specific thresholds.HSI and FIB-4 were surrogate indices, and the primary cohort lacked imaging or histologic confirmation of steatotic liver disease or fibrosis.Mortality associations were evaluated using survey-weighted Cox models, restricted cubic splines, and period-specific models.Incremental prediction was evaluated using Harrell's C-index, time-dependent area under the receiver operating characteristic curve, Brier score, and apparent calibration.Missing-data robustness was assessed using multiple imputation. RESULTS / 40Among 27197 participants (6654 deaths), the fully adjusted complete-case model included 21695 participants (5079 deaths), excluding 5502 participants with missing covariate data.Fully adjusted hazard ratios for elevated vs low FIB-4 were 1.132 (95%CI: 0.931-1.376),1.350 (1.223-1.491),and 1.173 (1.063-1.295) in the HSI < 30, 30-36, and > 36 groups, respectively.The overall interaction was not statistically significant (P = 0.098).Both indices showed nonlinear associations with mortality.Period-specific estimates were not statistically significant beyond 10 years of follow-up, and effect modification was not consistent over time.The pooled interaction remained statistically nonsignificant after multiple imputation (P = 0.056).Adding both indices to the base clinical model increased Harrell's C-index by 0.00319 (95%CI: 0.00190-0.00447),whereas adding their interaction provided no further improvement. CONCLUSIONElevated FIB-4 was associated with mortality in HSI 30-36 and > 36, but not significantly in HSI < 30; the overall interaction was nonsignificant, and predictive improvement did not establish clinical utility.