The NHANES database is a publicly available compendium, comprising participant data collected biennially through a multistage, stratified sampling method. This compendium encompasses a diverse range of data categories, including demographic profiles, questionnaire responses, dietary intakes, physical examinations, and laboratory evaluations. Using specific weighting protocols, researchers can adjust these data to make the resultant dataset representative of the broader US population (29).

This study used a cross-sectional design approach, incorporating data from three distinct phases of the ongoing NHANES, specifically the cycles spanning 2007–2008, 2009–2010, and 2017–2018. After performing the data weighting procedures, the subset extracted for further analytical examination included individuals aged 20 years or older, with available data on estimated glomerular filtration rate (eGFR) and proteinuria, as well as on the intake of flavonol and its subclasses.

The determination of daily nutrient intake from food sources was achieved through a complex methodology (16, 17, 30). Briefly, this segment of data collection was organized under the auspices of the US Department of Agriculture (USDA). To ensure precise acquisition of dietary intake data, trained interviewers used the USDA-developed Automated Multiple-Pass Method (AMPM). This methodology encompasses a comprehensive set of standardized queries tailored for specific food items, along with a wide array of potential response options. The process involved an initial 24 h dietary recall (Day 1), conducted face-to-face at the NHANES Mobile Examination Center, followed by a secondary recall (Day 2) completed by phone between 3 to 10 days later. The AMPM has been extensively validated in research studies and demonstrated to be an effective approach for measuring group energy intake and adult sodium intake (31, 32). The flavonol intake was calculated by multiplying the flavonol content in each food item by the consumption frequency obtained from the food frequency questionnaire. To accurately match foods containing flavonol and determine their flavonol content, food codes from the Food and Nutrient Database for Dietary Studies (FNDDS) were used. Specifically, the NHANES 2007–2008 dataset used FNDDS version 4.1 food codes, whereas the NHANES 2009–2010 and 2017–2018 datasets used FNDDS version 5.0 food codes (16, 17, 30). In this study, the aggregate flavonol intake during Day 1 and Day 2 was use as a metric to measure the flavonol consumption of each participant. Additionally, the flavonol compounds included: isorhamnetin, kaempferol, myricetin, and quercetin.

A solid-phase fluorescence immunoassay was performed to determine the levels of urinary albumin in human samples (33). The Jaffe rate method was used to measure the concentrations of creatinine in both serum and urine (34). The eGFR was calculated using the 2012 CKD-EPI equation with every serum creatinine measurement (35). The primary endpoint of this study was CKD, defined as an eGFR of less than 60 mL/min/1.73 m² and/or a urine albumin-to-creatinine ratio (UACR) exceeding 30 mg/g (3, 4). The secondary endpoints were separately identified as proteinuria (UACR exceeding 30 mg/g) and a decline in eGFR (eGFR less than 60 mL/min/1.73 m²).

This study considered multiple demographic covariates, including age, sex (male, female), race/ethnicity (non-Hispanic white, non-Hispanic black, Mexican American, and others), education level (high school or above, middle school or below), and economic status. Economic status was measured by the poverty income ratio (PIR), with a PIR less than 1 indicating poverty and PIR equal to or greater than 1 indicating non-poverty (36). Physical examinations included height, weight, body mass index (BMI), diastolic blood pressure (DBP), and systolic blood pressure (SBP). The BMI was calculated as weight in kilograms divided by height in meters squared, with a BMI of 25 kg/m² or greater defined as overweight or obese. Blood pressure was measured by trained staff, typically as an average of three readings. Personal habits included smoking history and physical activity. Smoking status includes never, former, and current smokers. “Never” is defined as having smoked fewer than 100 cigarettes in their lifetime. “Former” refers to individuals who have smoked at least 100 cigarettes in their lifetime but do not smoke at all currently. “Current” indicates individuals who have smoked at least 100 cigarettes in their lifetime and continue to smoke either on some days or every day. The physical activity was assessed in metabolic equivalent of task (MET) minutes per week and categorized into two groups based on whether it exceeded 600 MET minutes per week (37). Energy intake was the sum of calories consumed over the 2 days. Laboratory tests included total cholesterol, fasting lipids, fasting glucose, uric acid, and hemoglobin. The accuracy of the methods used to detect these substances was ensured by strict quality control procedures. Among the assessed complications were: hyperuricemia, defined as uric acid levels exceeding 360 μmol/L in women or 420 μmol/L in men (38); dyslipidemia, characterized by any of the following: triglyceride over 150 mg/dL, total cholesterol over 200 mg/dL, low-density lipoprotein above 130 mg/dL, high-density lipoprotein below 40 mg/dL for men or 50 mg/dL for women, or the use of lipid-lowering drugs (39); diabetes, diagnosed by one or more of the following criteria: glycated hemoglobin greater than 6.5%, fasting glucose above 7.0 mmol/L, random glucose over 11.1 mmol/L, postprandial glucose exceeding 11.1 mmol/L after 2 hours, use of hypoglycemic medications, or a clinical diagnosis (40); hypertension, defined by a systolic blood pressure higher than 140 mmHg or diastolic blood pressure over 90 mmHg, the use of antihypertensive drugs, or a clinical diagnosis (41).

The data from the NHANES database were collected using a complex, multilevel stratified sampling method. Thus, to ensure that the analysis reflects the broader demographics of the US, all collected data were subjected to a weighting procedure prior to analysis. Continuous variables are expressed as mean values with 95% confidence intervals (CIs), and categorical variables are expressed as percentages with their 95% CIs in Table 1. Statistical differences between CKD and non-CKD groups were evaluated using the rank-sum test for continuous variables and the chi-square test for categorical variables in Table 1. Based on the differences between the CKD and non-CKD populations, as well as clinical experience, we selected the appropriate variables to serve as calibration variables for subsequent analyses, which are specifically presented in the results section. The consumption of flavonol was divided into four distinct groups, each representing a quartile, namely the first quartile (Q1), the second quartile (Q2), the third quartile (Q3), and the fourth quartile (Q4). This study investigated the correlation between flavonol intake and CKD risk using a weighted multivariable logistic regression model, with an emphasis on trend analysis. The results were presented in Tables 2, 3. The dose–response relationship between flavonol consumption and CKD risk was analyzed using restricted cubic splines (RCS), as shown in Figure 1. In subgroup analyses, the association between flavonol intake and CKD risk was examined across various subgroups, complemented by interaction tests. The results were presented in Table 4. The relationships between specific flavonol compounds (isorhamnetin, kaempferol, myricetin, and quercetin) and CKD risk were analyzed using a weighted multivariable logistic regression method. The results were presented in Table 5. Statistical significance was assessed using a two-tailed p-value threshold of less than 0.05. All statistical evaluations were performed using the R software version 4.3.0 (R Development Core Team, University of Auckland, Auckland City, NZ).

Post-weighting, the data in this study represented 213,259,068 American adults aged 20 years and above, with 30,889,883 (14.5%) diagnosed with CKD. As shown in Table 1, there were significant statistical differences between the CKD and non-CKD populations in terms of age, sex, race, blood pressure (both systolic and diastolic), anthropometric measures (height, weight, BMI), lifestyle factors (physical activity status, smoking status, energy intake), renal function markers (eGFR, uACR), lipid profile (fasting triglycerides, total cholesterol), uric acid, fasting glucose, hemoglobin, and prevalence of hyperuricemia, dyslipidemia, hypertension, and diabetes. Additionally, dietary intake of isorhamnetin and kaempferol differed significantly between the groups. Compared to the non-CKD group, the CKD group was older, had a higher proportion of females, higher systolic blood pressure, lower diastolic blood pressure, shorter stature, higher weight, higher BMI, a higher proportion of overweight and obese individuals, a higher proportion of individuals engaging in less than 600 MET min/week of physical activity, lower energy intake, lower eGFR, higher uACR, elevated levels of fasting triglycerides, cholesterol, uric acid and blood glucose, lower hemoglobin, and increased prevalence of hypertension, diabetes, dyslipidemia, and hyperuricemia, along with lower daily intake of isorhamnetin and kaempferol.

As shown in Table 2, after multivariate adjustment for age, sex, BMI, ethnicity, smoking status, physical activity status, systolic blood pressure, diastolic blood pressure, uric acid, fasting glucose, cholesterol, fasting triglyceride, hyperuricemia, dyslipidemia, hypertension, and diabetes mellitus, the odds ratios (ORs) for CKD risk in the Q2, Q3, and Q4 quartiles of flavonol consumption were 0.598 (95% CI: 0.349, 1.023), 0.679 (95% CI: 0.404, 1.142), and 0.628 (95% CI: 0.395, 0.998), respectively, compared to the Q1 quartile. The p value for trend significance was 0.190. As shown in Table 3, after multivariate adjustment, compared to the Q1 quartile, the odds of a decline in eGFR in the Q2, Q3, and Q4 quartiles were 0.479 (95% CI: 0.179, 1.287), 0.372 (95% CI: 0.151, 0.915), and 0.665 (95% CI: 0.326, 1.358), respectively, with a p value for trend of 0.703. Furthermore, after multivariate adjustment, compared to quartile Q1, the odds of proteinuria risk in quartiles Q2, Q3, and Q4 were 0.589 (95% CI: 0.334, 1.041), 0.747 (95% CI: 0.417, 1.341), and 0.520 (95% CI: 0.301, 0.898), respectively, with a p value for trend significance of 0.060. As shown in Figure 1, RCS analysis revealed that flavonol consumption was non-linearly associated with proteinuria risk and eGFR decline, all with p values below 0.05.

The results of our investigation of the relationship between the intake of flavonol from dietary products and the risk of CKD are shown in Table 4. This analysis was divided across different groups, distinguished by factors such as sex, age, racial background, BMI, smoking status, and the presence of conditions like hypertension, diabetes, dyslipidemia, and hyperuricemia. After adjusting for multiple variables, the ORs for CKD incidence in the highest quartile (Q4) compared to the lowest (Q1) were as follows, with 95% confidence intervals (CIs): 0.313 (0.120, 0.815) for Black individuals, 0.338 (0.128, 0.894) for Mexican Americans, 0.562 (0.337,0.937) for dyslipidemia, 0.472 (0.252, 0.885) for hypertension, and 0.297 (0.115,0.771) for those with hyperuricemia. Trend tests for these subgroups mostly showed p-values above 0.05, except for Mexican Americans (p = 0.005) and those with hyperuricemia (p = 0.037). The interaction analysis indicated that for all subgroups, with the exception of the age subgroup, the p-values for the interaction tests exceeded 0.05.

The dose–response associations of consumption of isorhamnetin, kaempferol, myricetin, and quercetin with CKD are depicted individually in Figure 2, and the association between flavonol intake and CKD in US adults, adjusted for various factors are shown in Table 5. Notably, isorhamnetin intake showed a significant trend, with a reduction in CKD risk (OR: 0.860 [0.546, 1.343] in Q2, OR: 0.778 [0.515, 1.177] in Q3 and 0.637 [0.430, 0.943] in Q4; p = 0.013). Kaempferol, myricetin, and querectindid not show significant trends of CKD risk.