The National Health and Nutrition Examination Survey (NHANES), a major initiative by the Centers for Disease Control and Prevention (CDC), aims to evaluate the health and nutritional status of the U.S. population. Our study utilized data from 10 NHANES cycles, spanning 1999 to 2018, including a total of 101,316 participants. After excluding participants with missing data on the albumin creatinine ratio (ACR), total cholesterol, HDL (required for the NHHR formula), pregnant women, and individuals under 18 years of age, the final cohort comprised 41,225 subjects. The inclusion and exclusion process is illustrated in Figure 1 .

The independent variable in this study was NHHR, calculated as non-HDL-C divided by HDL-C (10). Non-HDL-C is derived by subtracting HDL-C from total cholesterol (TC). Enzyme tests for TC and HDL-C levels were performed using an automated biochemical analyzer. The study used the Roche Cobas 6000 and Roche Modular P chemistry analyzers to measure TC concentrations. These data were obtained directly from the NHANES database.

The dependent variable in this study is the albumin-to-creatinine ratio (ACR). Blood and urine samples from NHANES participants were collected at standardized mobile screening centres. Urinary albumin and creatinine levels were measured using solid-phase fluorescence immunoassay and the modified Jaffe kinetic method with a single-spot urine sample. While modernized for large-scale population studies, this method is rooted in the same scientific principles as the radioimmunoassay approach described by Chavers et al. (11), a historically recognized gold standard for albumin quantification. These measurements were directly retrieved from the NHANES dataset, which utilized automated biochemical analyzers (e.g., Roche Cobas 6000 and Roche Modular P chemistry analyzer) to perform the tests. The ACR was calculated by dividing the urinary albumin concentration (mg) by the urinary creatinine concentration (g). Significant proteinuria was defined as an ACR greater than 300 mg/g (12). In our analysis, macroalbuminuria was treated as an outcome variable.

Covariates potentially influencing the association between NHHR and proteinuria were included in our study (13, 14). These covariates comprised sex (male/female), age (<60 years/≥60 years), ethnicity (Mexican American/Other Hispanic/Non-Hispanic White/Non-Hispanic Black/Other), education level (less than high school/high school or equivalent/college graduate or above), marital status (married/widowed/divorced or separated/never married/living with partner), poverty-to-income ratio (PIR: <1.3, 1.3–3.49, ≥3.5), smoking status (yes/no), alcohol consumption (yes/no), physical activity (yes/no), energy intake(<1500,>=1500 <2500,>=2500kcal), body mass index (BMI: <25, >=25, <30, ≥30 kg/m²), hypertension (yes/no), diabetes mellitus (yes/no), estimated glomerular filtration rate (eGFR: <60/≥60 mL/min/1.73 m²), alanine aminotransferase (ALT, IU/L), aspartate aminotransferase (AST, IU/L), triglycerides (mg/dL), albumin (ALB, g/L), and blood uric acid (UA, mg/dL). Smoking status was defined as having smoked at least 100 cigarettes in a lifetime. Alcohol consumption was defined as having consumed at least 12 alcoholic drinks per year or alcohol ≥3 times in the past 12 months. Diabetes was defined as fasting blood glucose ≥7.0 mmol/L, HbA1c ≥6.5%, having been diagnosed with diabetes by a doctor, currently taking insulin, or using medication to lower blood sugar. Hypertension was defined as systolic blood pressure ≥130 mmHg, diastolic blood pressure ≥80 mmHg, having been diagnosed with high blood pressure by a doctor or currently taking prescription medication for hypertension. Serum creatinine (Scr) was measured using the Jaffe rate method and calibrated by standardized isotope dilution mass spectrometry. eGFR was calculated using the CKD-EPI equation, accounting for participant sex, race, age, and serum creatinine levels (15). Detailed measurement procedures for these variables are publicly available at www.cdc.gov/nchs/nhanes/.

All statistical analyses followed Centers for Disease Control and Prevention (CDC) guidelines, utilizing NHANES sampling weights and accounting for the complex multistage cluster design. Continuous variables were presented as means with standard errors (SE), while categorical variables were expressed as proportions. Weighted Student’s t-tests (for continuous variables) and weighted chi-square tests (for categorical variables) were applied to compare differences across NHHR tertiles. The associations between NHHR and proteinuria were analyzed using three multivariate logistic regression models. Model 1 was unadjusted for covariates. Model 2 adjusted for sex, age, and race. Model 3 adjusted for sex, age, race, education level, marital status, PIR, physical activity, eGFR, ALT, AST, triglycerides, serum uric acid, BMI, hypertension, diabetes, alcohol consumption, and smoking status. Notably, NHHR underwent natural logarithmic (Ln) transformations for regression analysis due to skewed distributions (as shown in Supplementary Figure 1 ). Subgroup analyses of the NHHR-proteinuria association were conducted using stratification factors such as sex (male/female), age (<60/≥60 years), PIR (<1.3/1.3–3.5/>3.5), BMI (<25/>=25, <30/>=30kg/m²), hypertension (yes/no), diabetes (yes/no), smoking (yes/no), alcohol consumption (yes/no), eGFR (<60/≥60 mL/min/1.73 m²), energy intake (<1500/>=1500, <2500/>=2500kcal), and physical activity (yes/no). These stratification factors were also considered prespecified potential effect modifiers. Interaction terms were included to assess the heterogeneity of associations across subgroups. Finally, we performed sensitivity analyses; Specifically, we excluded people who used drugs with effects on fat metabolism (statins, steroid hormones, diuretics, beta-blockers) in the last 30 days before performing multivariate logistic regressions. Continuous variables are interpolated using the median, and categorical variables are interpolated using the plurality. All analyses were performed using R version 3.4.3 (http://www.R-project.org, The R Foundation) and Empower software 2.0 version 0 (www.empowerstats.com; X&Y Solutions, Inc., Boston, MA). The significance level was set at p<0.05.

41,225 participants met the inclusion and exclusion criteria, with 49.97% being male and 50.03% female. The racial distribution included 45.98% non-Hispanic white, 17.17% Mexican American, 20.08% non-Hispanic black, and 26.39% from other racial/ethnic groups. All participants’ mean ACR and NHHR values were 45.52 mg/g and 3.01, respectively. All clinical characteristics of the participants are presented in Table 1 , All clinical characteristics of the participants are presented in Table 1 , categorized by the third quartile of NHHR. Significant differences were observed in age, sex, race, BMI, PIR, smoking status, education, and marital status (P < 0.05). Individuals in the highest NHHR tertile were predominantly male and non-Hispanic white. Additionally, those with higher NHHR were more likely to be married and have higher rates of alcohol consumption, elevated triglycerides, and comorbidities such as hypertension, obesity, and elevated ACR.

The association between NHHR and macroalbuminuria is presented in Table 2 . In the unadjusted model, each 1-unit increase in Ln-NHHR was associated with a 51% higher prevalence of macroalbuminuria, and this positive association remained significant after adjusting for all covariates (OR = 1.34; 95% CI, 1.13–1.59, p<0.0001). Using tertile 1 as a reference, all three models demonstrated a positive association between NHHR and macroalbuminuria at the T3 level [[Model 1: OR (95% CI) 1.51 (1.28–1.78); Model 2: OR (95% CI) 1.57 (1.33–1.87); Model 3: OR (95% CI) 1.33 (1.10–1.62)], with p for trend <0.05 in each model.

A smoothed curve fit was applied to explore the nonlinear relationship between NHHR and macroalbuminuria in greater detail. The results indicated a positive association between NHHR and the prevalence of macroalbuminuria ( Figure 2 ). Further subgroup analysis based on BMI (<25, 25–29.9, ≥30kg/m²) revealed that this relationship was more pronounced in the BMI ≥30 kg/m² group ( Figure 3 ).

Our subgroup analyses revealed that the association between NHHR and macroalbuminuria was inconsistent. Specifically, the correlation was not statistically significant (p > 0.05) for participants with BMI < 25kg/m², BMI between 25 and 30kg/m², PIR < 1.3, those who did not consume alcohol, and were not diabetic. Additionally, in the BMI subgroups, the odds ratios were: OR = 1.03 (95% CI, 0.74–1.44, p = 0.8593) for BMI < 25kg/m², OR = 1.22 (95% CI, 0.90–1.66, p = 0.1988) for BMI between 25 and 30kg/m², and OR = 1.89 (95% CI, 1.44–2.47, p < 0.0001) for BMI ≥ 30kg/m². A significant difference was observed between groups, with p for interaction = 0.0111. In contrast, sex, age, PIR, eGFR, hypertension, diabetes mellitus, energy intake, smoking, alcohol consumption, and work activity showed no significant effect on this relationship ( Table 3 , p > 0.05 for all interactions).

To address the potential confounding effects of medication use, we conducted a sensitivity analysis by excluding participants who reported using certain prescription drugs that may influence lipid metabolism within one month prior to the NHANES interview. The exclusion criteria were based on the “Prescription Medication” questionnaire, which systematically records medication use. Participants using corticosteroids, diuretics, β-blockers, or statins were excluded. Additionally, information on statin use was further supplemented by the BPQ100D questionnaire. Based on these criteria, we excluded 577 participants using corticosteroids, 5,681 using diuretics, 2,367 using β-blockers, and 3,413 using statins (with some participants using multiple medications). Therefore, the final sample size for the sensitivity analysis was 29,187 participants. To ensure the robustness of our findings, we also performed a multivariate regression analysis. The results from the sensitivity analysis were consistent with the direction of the main analysis ( Supplementary Table 1 ).

Furthermore, the original dataset included lipid-lowering treatment records for 8,925 participants (Yes/No), and we conducted an interaction test specifically for these 8,925 participants. The results indicated that lipid-lowering medication use did not affect the association between NHHR and macroalbuminuria, and no significant interaction was observed ( Table 3 ).