This cross-sectional, multicenter analysis evaluated whether the triglyceride–glucose (TyG) index is associated with microalbuminuria in a large Chinese population and whether diabetes status modifies that association. The investigation used a secondary analysis of an existing, publicly available dataset to quantify the TyG–microalbuminuria relationship and explore interactions by diabetes.
Data were derived from the China Multicenter Longitudinal Study of Diabetes Cancer Risk Assessment as shared by Ye et al. After exclusions for primary kidney disease, ACEI/ARB use, extreme sleep durations, and incomplete covariate information, the analytic sample comprised 33,416 participants from eight regional centers. The original study had institutional review board approval and written informed consent; this secondary analysis used de-identified, public data and required no additional ethical approval.
Trained staff collected demographic, anthropometric, blood pressure, and lifestyle information. Fasting blood samples (≥8 hours) provided measurements of triglycerides, fasting blood glucose (FBG), lipids, creatinine, liver enzymes, and other routine biochemistry. Participants without diabetes underwent a 75 g oral glucose tolerance test; participants with diabetes underwent a 100 g test for postprandial glucose assessment. Morning urine samples were used to calculate urinary albumin-to-creatinine ratio (UACR).
Microalbuminuria was defined as UACR ≥ 30 mg/g. Diabetes was defined by FBG ≥ 7.0 mmol/L, 2-hour PBG ≥ 11.1 mmol/L, or documented prior diagnosis. Estimated glomerular filtration rate (eGFR) was calculated using the MDRD formula. The TyG index was calculated from fasting triglyceride and glucose values using the established ln-based formula reported in the source dataset.
Complete-case analysis was used because missingness was low (<2% for variables). The TyG index was analyzed as a continuous variable (per 1-unit increase) and by quartiles. Multivariable logistic regression provided odds ratios (ORs) and 95% confidence intervals (CIs) for microalbuminuria. Four incremental models adjusted sequentially for demographic, anthropometric, blood pressure, biochemical, sleep variables, and finally diabetes status. Collinearity was assessed via variance inflation factors and diagnostic procedures (residuals, leverage). Restricted cubic spline (RCS) models tested nonlinearity. Subgroup analyses and interaction testing included sex, age, smoking, alcohol, obesity, hypertension, and diabetes status. Multiple testing correction used the Benjamini–Hochberg method; additive interaction was assessed with RERI. E-values were computed to appraise unmeasured confounding.
Mean age was 57.7 ± 9.3 years; 33.2% were male. Overall, 22.8% had diabetes and 77.2% did not. The TyG index had a mean ≈ 8.8. Median UACR was 10 mg/g, and the prevalence of microalbuminuria was 14.5%. Higher TyG quartiles were associated with progressively higher proportions of microalbuminuria, chronic kidney disease, diabetes, and hypertension.
In unadjusted analyses, each 1-unit rise in TyG was associated with increased odds of microalbuminuria (unadjusted OR reported in source). After multivariable adjustment across four models, the association remained statistically significant. In the fully adjusted model (Model IV, which included demographics, anthropometrics, blood pressure, lipids, liver indices, eGFR, sleep variables, smoking/alcohol, and diabetes status), the OR per 1-unit higher TyG was 1.28 (95% CI 1.20–1.36; P < 0.001). Analyses by TyG quartile showed a monotonic increase in microalbuminuria risk across quartiles (P for trend < 0.001). RCS modeling demonstrated a linear positive association (P overall < 0.001; P for non-linearity = 0.187).
Stratified analyses showed the TyG–microalbuminuria association in both participants without diabetes (OR 1.13; 95% CI 1.04–1.23) and with diabetes (OR 1.46; 95% CI 1.32–1.60). The interaction between TyG and diabetes status was statistically significant (P < 0.001; FDR-corrected P = 0.008), indicating a stronger estimated effect of TyG on microalbuminuria among participants with diabetes. Additive interaction metrics (RERI) and dichotomized TyG comparisons (Q1–Q2 vs Q3–Q4) were used per the source to quantify joint effects.
The authors performed multiple sensitivity checks and subgroup analyses to evaluate robustness. Diagnostic procedures included multicollinearity testing (VIF), posterior predictive checks, residual analyses, and model-fit assessments. Results were described as robust to these checks in the source. Subgroup heterogeneity was assessed by likelihood ratio tests, and multiple testing correction was applied.
In this large, population-based sample, the TyG index was linearly associated with higher odds of microalbuminuria after comprehensive adjustment. The association existed in persons with and without diabetes but was stronger among those with diabetes. Given TyG’s simplicity—requiring only fasting triglyceride and glucose—these findings suggest it may serve as a readily available marker to identify individuals at elevated odds of albuminuria across glycemic status, supporting its potential utility in screening or risk stratification pending prospective validation.
As the analysis was cross-sectional, temporal and causal inferences cannot be made. The study relied on secondary, pre-existing data and did not pre-specify power calculations. The TyG–microalbuminuria association could be influenced by residual confounding, though E-values and sensitivity analyses were reported. Details beyond those reported in the source (for example, exact TyG formula coefficients, threshold values, or RERI numeric results) were not provided in the source text.
The TyG index showed a robust, linear association with microalbuminuria in a large Chinese cohort, with a stronger estimated effect among individuals with diabetes. The authors recommend longitudinal studies to determine temporality and to assess whether TyG improves early identification of individuals at risk for kidney injury and cardiovascular complications.