TyG and HOMA-IR: correlation and agreement in a heterogeneous hospital population

The triglyceride-glucose (TyG) index is a simple composite marker proposed as a surrogate of insulin resistance and marker of cardiometabolic risk. However, its relationship and agreement with the homeostasis model assessment of insulin resistance (HOMA-IR) in heterogeneous hospital populations remain incompletely characterized. We retrospectively analyzed laboratory records from Hôpitaux Universitaires Henri Mondor for patients with simultaneous fasting measurements of glucose, insulin, and triglycerides between 2007 and 2021. HOMA-IR was calculated as fasting glucose (mmol/L) × fasting insulin (μU/mL)/22.5, and TyG as ln [fasting triglycerides (mg/dL) × fasting glucose (mg/dL)]/2. Patients were classified using previously published thresholds of HOMA-IR ≥ 3.23 and TyG ≥ 4.72. Correlation was assessed using Spearman's rank correlation coefficient, and agreement between classifications using Cohen's kappa coefficient. A total of 1,921 records, the mean age was 52.7 ± 0.3 years and 42.9% were women. Mean HOMA-IR and TyG were 3.46 ± 0.11 and 4.57 ± 0.01, respectively. HOMA-IR and TyG were moderately correlated (rho = 0.497, p < 0.0001). The two indices yielded the same classification in 68.9% of patients (k = 0.375; 95%CI; 0.334-0.416), indicating limited-to-moderate agreement. HOMA-IR and TyG were moderately correlated but showed limited categorical agreement, indicating that they should not be considered interchangeable. They may therefore be considered complementary markers, with TyG potentially providing broader information on cardiometabolic risk.