Abstract
Background
Fatty liver disease is frequently identified during routine health examinations, but practical strategies for risk stratification in adults with thyroid dysfunction remain limited. This study evaluated metabolic dysfunction-associated fatty liver disease (MAFLD) prevalence across thyroid function categories and developed health examination-based risk stratification models.Methods
This retrospective cross-sectional study included 46,311 adults who underwent routine health examinations from November 2019 to February 2025. Participants were classified as euthyroid, hypothyroid, or hyperthyroid. MAFLD was defined as ultrasound-confirmed hepatic steatosis plus overweight/obesity, diabetes, or metabolic dysregulation. Logistic regression and restricted cubic spline analyses assessed thyroid-related associations. Least absolute shrinkage and selection operator regression and multivariable logistic regression were used to develop models for adults with hypothyroidism or hyperthyroidism. Discrimination, calibration, and utility for post-examination risk stratification were evaluated.Results
Among 46,311 participants, 42,978 were euthyroid, 2,645 had hypothyroidism, and 688 had hyperthyroidism. MAFLD prevalence was 33.0%, 35.5%, and 19.9%, respectively. FT3 was positively associated with MAFLD, whereas FT4 was inversely associated; TSH was not independently associated after full adjustment. The hypothyroid model retained age, BMI, ALT, TG, and HDL-C, with AUCs of 0.906 and 0.901 in the training and validation sets. The hyperthyroid model retained age, BMI, FT4, ALT, and HDL-C, with AUCs of 0.918 and 0.932. Both models showed good calibration and potential utility for post-examination risk stratification.Conclusions
Routine health examination indicators can stratify fatty liver risk among adults with thyroid dysfunction and may help identify individuals requiring targeted liver-metabolic follow-up. External validation and prospective evaluation are needed.