Comparison of eight insulin-independent indices for identifying insulin resistance in Chinese young and middle-aged adults without diabetes: favorable performance of METS-IR and TyG-BMI

Background Homeostasis model assessment of insulin resistance (HOMA-IR) mainly reflects basal IR, whereas dynamic measures such as the Matsuda index are less practical. Methods This cross-sectional study evaluated eight non-insulin-based metabolic indices [the metabolic score for insulin resistance [METS-IR], triglyceride-glucose body mass index [TyG-BMI], TyG, TyG-WC, LAP, TG/HDL-C, VAI, CMI] as IR surrogates in 710 Chinese adults without diabetes (aged 20–60 years). IR was separately defined using two criteria (1): HOMA-IR above the 75th percentile, and (2) Matsuda index below the 25th percentile, in normal-weight participants with normal glucose tolerance. Correlations, restricted cubic splines, and receiver operating characteristic (ROC) analyses were performed, including separate sex- and 10-year age-stratified subgroup analyses. Results METS-IR and TyG-BMI showed moderate correlations with both HOMA-IR and the Matsuda index. METS-IR, TyG, TyG-BMI, and TyG-WC were linearly associated with IR. In ROC analyses, overall AUCs for METS-IR and TyG-BMI were 0.776 and 0.770 for HOMA-IR-defined IR and 0.768 and 0.765 for Matsuda index-defined IR, indicating moderate discrimination. At cut-offs of 35.54 for METS-IR and 211.37 for TyG-BMI (which were consistent across both IR definitions), sensitivities were 66.1–68.9%, and specificities were 75.1–78.4%. Under the HOMA-IR definition, METS-IR had a significantly larger AUC than six indices, but not TyG-BMI. Under the Matsuda definition, its AUC did not differ from those of TyG-BMI, TyG-WC, LAP, or CMI. Rankings varied across subgroups. Conclusion METS-IR and TyG-BMI demonstrated comparable and favorable performance, serving as practical, insulin-independent initial screening markers for identifying elevated IR risk in Chinese young and middle-aged adults without diabetes. Rather than replacing insulin measurements, these indices offer valuable risk stratification tools in primary care or epidemiological settings where insulin assays are routinely unavailable.