Our study population was obtained from the NHANES database. NHANES, carried out by the National Center for Health Statistics (NCHS) under the Centers for Disease Control and Prevention (CDC) adopts a stratified, multistage, randomized sampling framework to represent the U.S. population at a national level. Participants underwent questionnaires, clinical examinations, and laboratory analyses. The study protocol was approved by the NCHS Ethics Review Board, and all participants gave written informed consent (https://www.cdc.gov/nchs/nhanes/about/erb.html). The analysis used data from 10 NHANES cycles conducted between 1999–2000 and 2017-2018. Exclusions applied to those under 20 years old or missing METS-VF or the urinary albumin-to-creatinine ratio (UACR) data, resulting in 22514 eligible participants ( Figure 1 ).

METS-VF was the exposure variable in this study. METS-VF= 4.466 + 0.011 × [Ln (METS-IR)] ³ + 3.239 × [Ln (WHtR)] ³ + 0.319 × (sex) + 0.594 × [Ln (age)] (male=1, female=0) (14). METS-IR= Ln [(2 × fasting plasma glucose) + (triglycerides)] × (body mass index)/[Ln (high-density lipoprotein cholesterol)] (22). On the other hand, albuminuria (UACR ≥30 mg/g) was the outcome variable in this study. UACR was calculated as urinary albumin divided by urinary creatinine. Urinary albumin was measured using solid-phase fluorescence immunoassay, and urinary creatinine was assessed via the modified Jaffe kinetic method. Detailed measurement methods are provided at: https://wwwn.cdc.gov/nchs/nhanes.

Potential covariates included age, sex, ethnicity, marital status, poverty-income ratio (PIR), education level, smoking history, hypertension, diabetes, CVDs, body mass index (BMI), waist circumference (WC), height, systolic blood pressure (SBP), diastolic blood pressure (DBP), fasting plasma glucose (FPG), hemoglobin A1c (HbA1c), triglycerides (TG), total cholesterol (TC), high-density lipoprotein cholesterol (HDL-c), low-density lipoprotein cholesterol (LDL-c), and the estimated glomerular filtration rate (eGFR). Ethnicity categories included Mexican American, Other Hispanic, Non-Hispanic White, Non-Hispanic Black, and individuals of Other Race. Education was categorized as less than high school, high school, or college or above. Smoking status included current and former smokers. BMI is calculated as weight (kg)/height² (m²) and classified as <25 kg/m², 25–30 kg/m², and ≥30 kg/m². The eGFR was estimated using the Chronic Kidney Disease Epidemiology Collaboration equation (23). Hypertension was defined as SBP ≥140 mmHg, DBP ≥90 mmHg, self-reported history, or antihypertensive medication use. Diabetes was defined as FPG ≥126 mg/dL, HbA1c ≥6.5%, self-reported history, or hypoglycemic medication use. Self-reported heart attack, stroke, heart failure, coronary artery disease, or angina was used to determine the presence of CVDs. Detailed measurements are provided at: https://wwwn.cdc.gov/nchs/nhanes.

To align with the guidelines, the analysis utilized descriptive statistics weighted by population(https://wwwn.cdc.gov/nchs/nhanes/analyticguidelines.aspx). Continuous variables were reported as median (interquartile range), and categorical variables as counts (weighted percentages). Missing data were assumed to be missing at random and imputed using the random forest algorithm. Group comparisons used Kruskal-Wallis and chi-square tests. Logistic regression models assessed the association between METS-VF and albuminuria risk, with three adjustment levels: Model 1 (unadjusted), Model 2 (adjusted for age group (<60/≥60 years), sex, ethnicity, marital status, PIR, education level, and smoking history), and Model 3 (Model 2 + hypertension, diabetes, CVDs, BMI group (<25/25-30/≥30kg/m²), TC, LDL-c, and eGFR). Restricted cubic splines (RCS) were employed to examine potential nonlinearity. The optimal threshold value was determined through maximum likelihood estimation. Risk differentials across the identified threshold were then quantified using a segmented regression approach. Receiver operating characteristic (ROC) curves were conducted with the `pROC` package and decision curve analyses (DCA) with the `rmda` package in R to compare METS−VF with other indicators for classification accuracy and clinical utility, based on default parameters. Stratified analyses were conducted by age, sex, ethnicity, BMI, hypertension, diabetes, CVDs, and eGFR. Mediation analyses using the Sobel test were conducted to evaluate whether HbA1c, blood pressure (SBP, DBP), oxidative stress (gamma−glutamyl transferase [GGT], serum uric acid [SUA]), and inflammation (white blood cell count [WBC], systemic immune−inflammation index [SII], neutrophil−to−lymphocyte ratio [NLR]) mediated the relationship between METS-VF and albuminuria risk (24). Statistical analyses used R software (version 4.2.0), with P<0.05 considered significant.

The analysis included 22514 adults (median age: 49 years; interquartile range: 34–64 years) with ethnic distribution: 4057 (weighted: 8.50%) Mexican American, 1918 (weighted: 5.43%) Other Hispanic, 10042 (weighted: 68.25%) Non-Hispanic White, 4429 (weighted: 10.46%) Non-Hispanic Black, and 2068 (weighted: 7.37%) Other Race ( Table 1 ). Albuminuria patients tended to be older and exhibited unfavorable socioeconomic and lifestyle characteristics, including lower PIR, less education, being unmarried, and smoking (P<0.001). They also had shorter stature, lower HDL-c, LDL-c, and eGFR, but higher BMI, WC, SBP, DBP, FPG, HbA1c, TG, and METS-VF levels (P<0.001). Additionally, they showed a higher prevalence of hypertension, diabetes, and CVDs (P<0.001). Participants stratified by METS-VF quartiles (Q1–Q4) revealed that higher quartile groups were older, predominantly male, and more likely to have lower PIR, less education, be married, and smoke ( Table 2 ) (P<0.001). They also had higher BMI, WC, FPG, HbA1c, SBP, DBP, TG, TC, and LDL-c levels, lower HDL-c and eGFR levels, and a significantly higher prevalence of hypertension, diabetes, and CVDs compared to Q1 (P<0.001). Notably, albuminuria prevalence increased with METS-VF levels (P<0.001) ( Figure 2 ).

Table 3 presents the logistic regression analysis results for the association between METS-V, stratified into quartiles (Q1-Q4), and the risk of albuminuria. In the unadjusted model, with Q1 as reference, the odds ratios (ORs) for Q2, Q3, and Q4 were 1.332 (95%confidence interval [CI]:1.156-1.535), 2.119 (95%CI:1.857-2.418), and 4.113 (95%CI:3.635-4.652), respectively. In Model 2, adjusted for age group, sex, ethnicity, marital status, PIR, education level, and smoking history, the ORs for Q1, Q2, Q3, and Q4 were 1.000 (reference), 1.194 (95%CI:1.033-1.380), 1.635 (95%CI: 1.424-1.878), and 2.685 (95%CI:2.341-3.079), respectively. When fully adjusting for potential confounding factors in Model 3, the ORs also maintained a comparable trend. In this fully adjusted model, the ORs for Q2, Q3, and Q4 were 1.167 (95%CI:0.991-1.374), 1.452 (95%CI:1.204-1.750), and 1.984 (95%CI:1.594-2.471), respectively, when compared to Q1 as the reference. Additionally, the analysis of METS-VF as a continuous variable also demonstrated a positive association with albuminuria risk (OR = 1.406, 95%CI:1.243-1.590) after adjusting for confounding factors. RCS analysis indicated a nonlinear, J-shaped relationship ( Figure 3 ) (P for nonlinearity<0.001), with a threshold of 6.128 ( Table 4 ). Below this threshold, the OR was 0.663 (95%CI:0.540-0.814); above it, the OR rose to 2.118 (95%CI:1.804-2.486).

Figure 4 displays ROC and DCA results. The area under the curve (AUC) values for METS-VF, WHtR, METS-IR, BMI, and WC were 66.0%, 61.5%, 58.3%, 55.5%, and 58.9%, respectively, indicating METS-VF’s superior discriminative ability for albuminuria risk. Additionally, DCA analysis also showed that the METS-VF model provided higher net benefit across a wider threshold probability range, suggesting greater clinical utility.

Based on stratified analyses, higher METS−VF was positively associated with albuminuria across age, sex, ethnicity, BMI, hypertension, diabetes, CVDs, and eGFR subgroups ( Figure 5 ). By age, the ORs were 1.370 (95%CI:1.207-1.556) for <60 years and 1.547 (1.295-1.847) for ≥60 years (P for interaction = 0.141). By sex, the ORs were 1.102 (0.927-1.310) in females and 1.871 (1.533-2.284) in males (P for interaction < 0.001). By ethnicity, the ORs were 1.635 (1.315-2.032) in Mexican Americans, 1.325 (1.148-1.529) in non−Hispanic Whites, 1.366 (1.156-1.613) in non−Hispanic Blacks, 1.535 (1.147-2.054) in other Hispanics, and 1.595 (1.250-2.035) in other races (P for interaction = 0.218). By BMI, the ORs were 1.187 (1.036-1.360) for <25 kg/m², 2.201 (1.742-2.781) for 25–30 kg/m², and 1.710 (1.360-2.150) for ≥30 kg/m² (P for interaction < 0.001). By hypertension status, the ORs were 1.309 (1.147-1.493) without hypertension and 1.612 (1.376-1.889) with hypertension (P for interaction = 0.006). By diabetes status, the ORs were 1.387 (1.224-1.571) without diabetes and 1.542 (1.251-1.902) with diabetes (P for interaction = 0.282). By CVDs, the ORs were 1.399 (1.236-1.585) without CVDs and 1.473 (1.169-1.856) with CVDs (P for interaction = 0.641). By eGFR, the ORs were 1.800 (1.24-2.607) for <60 ml/min/1.73 m² and 1.451 (1.270-1.658) for ≥60 ml/min/1.73 m² (P for interaction = 0.279). Overall, associations were directionally consistent, with significant effect modification by sex, BMI, and hypertension, indicating stronger associations for males, individuals with BMI ≥25 kg/m², and those with hypertension (P for interaction < 0.05). To assess the mediating role of HbA1c, SBP, DBP, oxidative stress (GGT and SUA), and inflammation (WBC, SII, and NLR), we conducted a Sobel test, which confirmed a significant indirect effect of METS-VF on albuminuria through HbA1c (18.43%), SBP (28.81%), DBP (9.87%), GGT (11.61%), SUA (13.12%), WBC (5.84%), SII (4.10%), and NLR (10.32%) ( Figure 6 ) (P<0.001).