Two measurements suffice: waist-to-height ratio combined with TG/HDL ratio identifies occult insulin resistance as effectively as complex indices in 247 apparently healthy adults

Background Insulin resistance (IR) precedes type 2 diabetes and cardiovascular disease, yet fasting insulin remains impractical for population screening. Whether complex composite indices or bioimpedance outperform simple measurements for IR detection has not been established. Purpose To identify the simplest surrogate markers detecting occult IR (HOMA-IR ≥2.5) by comparing 21 candidates from three domains in apparently healthy adults. Methods We analysed 247 adults (132 female; age 44.3±11.1 years; no diabetes) from the STELA preventive cardiology registry at a tertiary hospital in the Czech Republic. Twenty-one surrogates were evaluated across blood biomarkers (TG/HDL, TyG, HbA1c, uric acid, ALT, hsCRP, ApoB, glucose), bioimpedance (visceral fat area, body fat%, skeletal muscle), and anthropometry/composites (BMI, waist, WHtR, BRI, ABSI, LAP, VAI). TG/HDL used mmol/L values (×2.29 for mg/dL equivalents). ROC analysis with DeLong pairwise comparisons (Holm–Bonferroni correction), domain-specific multivariate logistic regression with VIF assessment, and BMI-stratified subgroup analyses were performed. Results IR prevalence was 36.8% (91/247), rising from 14.5% (BMI <25) to 79.5% (BMI ≥30). Top univariate markers: LAP (AUC 0.808), TG/HDL (0.800), waist (0.797), WHtR (0.792), BMI (0.788). No pairwise DeLong comparison was significant after correction (0/36 tests) — simple markers were indistinguishable from complex indices. Bioimpedance visceral fat (AUC 0.763) did not outperform a tape measure (waist 0.797). The parsimonious model combining WHtR + TG/HDL (Table) achieved the highest collinearity-free AUC (0.848; R²=0.432; all VIF <1.5): WHtR OR 3.12/SD (95% CI 2.03–4.79; p<0.001) and TG/HDL OR 2.47/SD (1.48–4.11; p<0.001). This matched or exceeded bioimpedance (0.824), four-variable blood (0.813), and composite-index (0.828) models. As a cascade (WHtR ≥0.53 then TG/HDL ≥0.52 mmol/L-based [≥1.2 mg/dL-based]), specificity was 84.0%, PPV 70.9%, reducing insulin testing by 65%. Within BMI strata, anthropometric AUCs collapsed (waist 0.59–0.67) while TG/HDL persisted (0.67–0.84; Figure). Conclusions Waist-to-height ratio plus TG/HDL ratio — a tape measure and a routine lipid panel — matches complex multi-domain models for IR detection (AUC 0.848). Both are available in standard European preventive examinations. Within BMI categories, only blood markers retain discrimination, indicating IR screening requires metabolic, not merely anthropometric, assessment. External validation is warranted.