The National Health and Nutrition Examination Survey (NHANES) is a population-based, nationally representative cross-sectional survey (https://www.cdc.gov/nchs/nhanes/about_nhanes.htm) conducted by the US Centers for Disease Control and Prevention. Informed consent was obtained from each participant. Our study included data from six cycles: 1999-2000, 2001-2002, 2003-2004, 2005-2006, 2007-2008, and 2009-2010. Participants under 18 years of age were excluded due to the lack of education, poverty-to-income ratio (PIR), and chronic disease data. After further excluding individuals under 20 years of age, those with missing weighted values, missing CKD assessment indicators, eGFR and urinary albumin-to-creatinine ratio (UACR), and missing data on CLR, a total of 13,862 respondents from the NHANES database (1999-2010) were included in this study ( Figure 1 ).

CRP levels and peripheral blood lymphocyte counts were measured using automated blood analysis equipment. CLR was calculated as follows (20):[CRP level (mg/dL)×100]/peripheral blood lymphocyte count (pcs/μl).

CKD is defined as fulfilling any of the following (21, 22): eGFR<60mL/min/1.73m² (2); UACR ≥30mg/g. The eGFR was calculated based on the CKD epidemiological collaboration equation (23): 142×min(Scr/κ, 1)^α×max(Scr/κ, 1)^-1.209×0.993^Age×1.012[if female], where κ = 0.7 (females) or 0.9 (males), a= -0.241(female) or 0.302(male), Scr = serum creatinine in mg/dL, The “min(Scr/κ, 1)” factor indicates the minimum of Scr/κ or 1.0 and “max(Scr/κ, 1)” indicates the maximum of Scr/κ or 1.0.

Four categories of covariates were included in this study, including socio-demographic information (gender, age, ethnicity, education, PIR, marital status), lifestyle behaviors (like smoking, body mass index [BMI], and alcohol status), chronic diseases associated with renal disease (hypertension, cancer, diabetes mellitus, anemia, and hyperuricemia), and biochemical indicators of bone metabolism (calcium and phosphorus). Participants aged 20 years or older were included, categorized by age as young (20-44 years), middle-aged (45-64 years), and elderly (≥65 years). Ethnicity categories included Mexican American, Other Hispanic, Non-Hispanic White, Non-Hispanic Black, and Other Race (including Multi-Racial). Marital status was classified as married/living with a partner, never married, and widowed/divorced. Educational status was categorized as below high school, high school graduate, and college or above. PIR was categorized as <1, 1-1.99, 2-3.99, and ≥4. Smoking status was categorized into three groups: current smokers (those who smoked every day or intermittently), former smokers (those who smoked more than 100 cigarettes in their lifetime but were currently non-smokers), and never smokers (those who smoked less than 100 cigarettes in their lifetime). BMI was categorized as low to normal (<25 kg/m²), overweight (25-30 kg/m²), and obese (≥30 kg/m²). Alcohol consumption was divided into three categories: nondrinkers (those who drank less than 12 drinks in their lifetime), low to moderate drinkers (those who drank 12 drinks in their lifetime but consumed less than 2 drinks per day for men and less than 1 drink per day for women in the past 12 months), and heavy drinkers (those who drank 12 drinks in their lifetime and consumed more than 2 drinks per day for men and more than 1 drink per day for women in the past 12 months). Hypertension and cancer were defined based on a diagnosis by a healthcare professional. Diabetes was diagnosed if any of the following criteria were met: a previous diagnosis of diabetes, fasting blood glucose ≥ 7 mmol/L, or glycated hemoglobin ≥ 6.5%. Anemia was diagnosed based on a hemoglobin concentration of less than 13 g/dl for men and less than 12 g/dl for women (24). Hyperuricemia was diagnosed based on a uric acid level of more than 7 mg/dL in men and more than 6 mg/dL in women (25).

The results of NHANE were collected through multiple sectors and there will inevitably be missing data during the survey, our study performed multiple imputations for data with a missing rate of no more than 10 percent (26). Given NHANE’s complex multistage sampling design, appropriate weighting was applied in our analyses. Group differences were assessed using t-tests or ANOVA. Continuous variables were presented as mean ± standard deviation, while categorical variables were reported as frequencies and weighted percentages. We investigated the association between CLR and CKD prevalence risk using weighted logistic regression models. Additionally, the relationship between CLR and markers of kidney injury in CKD was explored using weighted generalized linear regression models. Model 1 was unadjusted, while Model 2 adjusted for age, sex, race, education, PIR, marital status, smoking status, drinking status, and BMI. Model 3 was further adjusted for hypertension, cancer, diabetes, hyperuricemia, anemia, phosphorus, and calcium, based on Model 2. The non-linear relationships between CLR and CKD prevalence risk, eGFR level, and UACR level were examined using weighted restricted cubic splines (RCS). Additionally, the predictive ability of CLR for CKD was assessed using the receiver operating characteristic (ROC) curve and the calculation of the area under the curve (AUC). The optimal cutoff value (OCV) of CLR was determined based on the maximum Youden’s index.

We conducted stratified analyses using logistic and linear regression, stratifying by sex, age, race, education, PIR, marital status, smoking status, drinking status, BMI, hypertension, cancer, diabetes mellitus, hyperuricemia, and anemia classifications. To validate the robustness of the findings, we conducted four sensitivity analyses: First, stroke and cardiac disease were further included after adjusting for all covariates to control for the potential confounding effect of these systemic inflammation-related conditions. Second, unweighted model analyses were performed to assess whether the results were influenced by the weighted data processing. Third, CKD was redefined using stricter criteria (eGFR < 45 ml/min/1.73 m² or ACR ≥ 30 mg/g). Fourth, the use of prescribed medications was incorporated into the analysis, adjusting for covariates such as anti-inflammatory drugs, lipid-lowering drugs, glucose-lowering drugs, and antihypertensive medications, to control for the potential impact of medications on the association between CLR and CKD. All analyses were conducted using R software (version 4.4.0; www.r-project.org), and statistical significance was defined as P < 0.05.

We analyzed data from 13,862 respondents in the NHANES database, spanning the years 1999 to 2010. Within this cohort, 2,449 individuals were identified with CKD, presenting a mean age of 59.63 ± 18.25 years, while 11,413 individuals were identified without CKD, with a mean age of 44.10 ± 15.49 years. Age, sex, race, education, PIR, marital status, smoking status, drinking status, BMI, hypertension, cancer, diabetes mellitus, hyperuricemia, anemia, phosphorus, calcium, eGFR, UACR, and CLR were significantly different between the CKD and non-CKD groups (p<0.005). Detailed weighted and unweighted data for participants with and without CKD are provided in Table 1 and Supplementary Table S1 , respectively.

The weighted logistic regression analysis revealed a significant association between CKD and CLR. In the unadjusted model, elevated CLR levels were linked to an increased risk of CKD (Odds Ratio [OR] = 1.54, 95% Confidence Interval [CI] = 1.30–1.83, P < 0.001). After adjusting for age, sex, race, education, PIR, marital status, smoking status, drinking status, and BMI in Model 2, CLR remained significantly associated with CKD (OR = 1.35, 95% CI = 1.17–1.56, P < 0.001). In Model 3, which accounted for common CKD complications and various biochemical indicators, the positive correlation between CLR and CKD persisted ( Figure 2 ). The RCS model demonstrated a non-linear relationship between CLR and CKD prevalence; however, the graphical representation suggested a more pronounced linear correlation ( Figure 3 ).

Figure 4 illustrates the correlation between CLR levels and markers of kidney injury (eGFR and UACR) using survey-weighted multiple linear regression analysis. In Model 3, which adjusted for demographic information, lifestyle habits, chronic conditions associated with kidney disease, and biochemical indicators, CLR was found to be negatively associated with eGFR (β = -1.181, 95% CI = -2.071 to -0.291, P = 0.010). Conversely, the fully adjusted model demonstrated a positive correlation between CLR levels and UACR levels (β = 12.392, 95% CI = 1.272 to 23.513, P = 0.030). The RCS model showed no significant non-linear correlation between CLR and markers of kidney injury (eGFR and UACR) ( Figure 5 ).

To evaluate the predictive ability of CLR for CKD, we utilized the ROC curve and calculated the AUC. The results indicated that the AUC of CLR in predicting CKD was 0.653 (95% CI, 0.641–0.665), suggesting that CLR has a moderate predictive ability in distinguishing between CKD patients and non-CKD individuals ( Figure 6 ). The OCV, determined using the maximum Youden’s index method, was 0.856, at which point the sensitivity was 0.703 and the specificity was 0.526. Furthermore, the positive predictive value at the OCV was 0.874, while the negative predictive value was 0.275.

Figure 7 presents a subgroup analysis of the relationship between CLR and CKD, demonstrating a significant association across all strata, except for other Hispanics and other Race-Including Multi-Racial (P < 0.05). Interaction tests revealed that age, race, PIR, and anemia modified the association between CLR and CKD (P < 0.05). Supplementary Figure S1 details a subgroup analysis of the relationship between CLR and eGFR. After excluding Age 20-44, other Hispanic, other Race-Including Multi-Racial, Married/living with a partner, PIR ≥ 4, Obese (BMI ≥ 30), Nondrinker, Patients with Cancer, Patients with Diabetes, and Patients without Hypertension, the association between CLR and eGFR remained significant (P < 0.05) across all strata. Interaction tests indicated that gender, age, race, education, PIR, marital status, smoking status, BMI, hypertension, and hyperuricemia influenced the association. Supplementary Figure S2 performs subgroup analyses of the relationship between CLR and UACR. After removing Age ≥ 65, Mexican American, other Hispanic, other Race-Including Multi-Racial, Never married, college or above, PIR 2-3.99, Current smoker, Low to moderate drinker, and Nondrinker, the relationship between CLR and UACR was significant (P < 0.05) in all strata. Interaction tests showed that race and cancer affected the association.

Supplementary Tables S8 - S19 present the results of the sensitivity analysis, where we employed four methods to assess the robustness of the findings. First, after adjusting for all covariates (including age, sex, BMI, hypertension, diabetes, etc.), and further including stroke and heart disease variables, the results indicated that the associations between CLR and CKD, eGFR, and UACR remained significant (OR = 0.78, 95% CI, 0.70–0.88, P < 0.001; β = -0.978, 95% CI, -1.848 to -0.108, P = 0.028; β = 11.768, 95% CI, 0.830–22.706, P = 0.035). Second, unweighted model analysis was performed to assess whether the results were influenced by the weighted data processing, and it showed that the association between CLR and CKD, as well as CLR and eGFR, remained significant (OR = 0.81, 95% CI, 0.75–0.87, P < 0.001; β = -1.144, 95% CI, -1.666 to -0.623, P < 0.001). However, the association between CLR and UACR approached the significance level (β = 11.744, 95% CI, -0.019 to 23.506, P = 0.050). Third, CKD was redefined using stricter diagnostic criteria (eGFR < 45 ml/min/1.73 m² or ACR ≥ 30 mg/g), and the results of these analyses were largely consistent with the original findings. Finally, after adjusting for all covariates and incorporating the use of prescribed medications (including anti-inflammatory drugs, lipid-lowering drugs, glucose-lowering drugs, and antihypertensive medications), the association between CLR and CKD, as well as CLR and eGFR, remained significant (OR = 0.78, 95% CI, 0.69–0.88, P < 0.001; β = -1.101, 95% CI, -1.943 to -0.259, P = 0.011). However, the association between CLR and UACR was no longer significant (P > 0.05). These sensitivity analyses suggest that CLR has strong robustness in its correlation with CKD and eGFR, but its predictive ability for UACR may be influenced by factors such as pharmacological interventions.