Abstract
Objective
To examine the association between dietary flavonoid intake and gallstone prevalence in a representative United States sample using National Health and Nutrition Examination Survey 2017–2018 data.
Methods
This cross-sectional study included 4087 adults aged ≥20 years. Flavonoid intake was assessed using the United States Department of Agriculture Food and Nutrient Database for Dietary Studies and covered 29 flavonoids across six subclasses. Gallstones were defined by self-reported physician diagnosis. Covariates included demographics, lifestyle, and health status. Multivariate logistic regression, weighted quantile sum, quantile g-computation, and Bayesian kernel machine regression models were used to evaluate the individual and combined effects of flavonoids.
Results
Gallstone prevalence was 11.00% (n = 475). A U-shaped relationship was observed for total flavonoid intake, with lower intakes negatively and higher intakes positively associated with gallstone prevalence. Among females, higher total flavonoid and flavonol intakes were significantly associated with increased gallstone prevalence. Specific flavonoids, including theaflavins and thearubigins, were associated with gallstone prevalence in females. Both the weighted quantile sum and quantile g-computation models showed a significant positive correlation in the overall population and among females; however, this association was not observed in males. Bayesian kernel machine regression identified flavan-3-ols as a key contributor in females.
Conclusions
A U-shaped relationship was observed between total flavonoid intake and gallstone prevalence. Dietary flavonoids, particularly in females, are associated with the prevalence of gallstone, highlighting the need for further research into potential mechanism and consideration of sex in dietary guidelines.
Keywords
Introduction
Gallstones are a common digestive disorder affecting approximately 11% of US adults and are associated with complications including cholecystitis and pancreatitis.1,2 Current effective treatment options, including laparoscopic cholecystectomy, do not address the underlying risk factors associated with gallstone formation, highlighting the need for further research into preventive strategies. 3 Understanding the nature of gallstone disease is crucial, as it encompasses genetic predispositions, dietary habits, and lifestyle factors that may contribute to its pathogenesis. 4
The development of gallstones is influenced by multiple factors, among which diet is a key modifiable risk factor. 5 Dietary flavonoids, which are a category of polyphenols found in plant-derived foods, have undergone extensive research due to antioxidant and anti-inflammatory properties. 6 These properties suggest that flavonoids may be associated with gallstone prevalence, either by modulating cholesterol metabolism or by reducing inflammation in the gallbladder. 7 The link to gallstone prevalence is still not definitive. Certain research indicates that flavonoids may protect against gallstone formation because of their ability to reduce inflammation and oxidative stress. 8 However, other studies have reported no significant associations, highlighting the need for further investigation.9,10 The complexity of this relationship is further compounded by the diverse nature of flavonoids. Moreover, previous research has not adequately explored the potential nonlinear association between flavonoid intake and gallstone prevalence or the combined effects of multiple flavonoids. 11
To address the existing gaps, this study used data from the 2017–2018 National Health and Nutrition Examination Survey (NHANES). Advanced statistical methods, including multivariate logistic regression, weighted quantile sum (WQS) regression, quantile g-computation (QGC), and Bayesian kernel machine regression (BKMR), were applied to assess the individual and combined effects of 29 flavonoids across 6 subclasses. This comprehensive analysis of the association between flavonoid intake and gallstone prevalence may improve our understanding of dietary influences on gallstone disease and provide guidance for potential prevention strategies.
Methods
Research cohort
This cross-sectional study drew upon data from the NHANES, which took place in the US from 2017 to 2018 (https://www.cdc.gov/nchs/nhanes/). NHANES is a cross-sectional survey that is representative of the nation and aims to assess the health and nutritional status of both adults and children in the US. The reporting of this retrospective study conforms to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines for cross-sectional studies. 12 The dataset used in this study is publicly available and was fully deidentified by the National Center for Health Statistics (NCHS) before release. No further deidentification was performed by the authors. All personally identifiable information, including names, addresses, telephone numbers, and social security numbers, was removed from the public-use data files to ensure participant confidentiality. The study included individuals aged ≥20 years. In adherence to the ethical protocols of NHANES, every participant provided written informed consent prior to their involvement in the survey. Initially, 9254 participants were eligible. After excluding individuals younger than 20 years, those with incomplete gallstone questionnaire data, those with incomplete 2-day flavonoid intake data, and pregnant women, 4087 participants were included in the final analysis (Figure 1).
Flowchart of the study NHANES, US, 2017–2018.
Measurement of flavonoids
Data on dietary flavonoid intake were obtained from the United States Department of Agriculture (USDA) Food and Nutrient Database for Dietary Studies (FNDDS), which incorporates information from two 24-h dietary recall interviews that were part of the NHANES. 13 This database contains values for 29 individual flavonoids across six subclasses: anthocyanidins, flavan-3-ols, flavanones, flavones, flavonols, and isoflavones. Each subclass comprises specific flavonoids, such as cyanidin and delphinidin in anthocyanidins and quercetin and kaempferol in flavonols. Total dietary flavonoid intake was calculated as the sum of these six subclasses, which include the 29 individual flavonoids. The final dietary flavonoid intake was calculated as the average of the 2-day dietary recalls.
Assessment of gallstones
The presence of gallstones was determined using responses to a standardized questionnaire. Participants were inquired whether they had ever been informed by a physician that they had gallstones. Those who responded affirmatively were categorized as having gallstones.
Covariates
Covariates were selected based on previous research and clinical judgment. In this study, the healthy eating index-2020 (HEI-2020) 14 was used to reflect participants’ overall dietary intake. This index helped account for the influence of other dietary factors by providing a comprehensive assessment of diet quality. Demographic characteristics, including age, sex, race, education level, and poverty income ratio (PIR), were collected through standardized household interviews. Age was categorized into four groups: 20–39, 40–59, 60–79, and ≥80 years. Race was classified as Mexican American, non-Hispanic White, non-Hispanic Black, and other races. Educational level was categorized as below high school, high school, and above high school. Poverty status was classified into three categories according to PIR: <1.3, 1.3–3.5, and ≥3.5. Alcohol consumption was classified into three categories: heavy drinkers (≥15 g/d for women and ≥30 g/d for men), moderate drinkers (0–15 g/d for women and 0–30 g/d for men), and nondrinkers. Total physical activity was measured in metabolic equivalent of task (MET) and further categorized as inactive (0–600 MET-minutes/week), moderate (600–1500), and active (>1500) groups. 15 Body mass index (BMI) was classified as obesity (≥30 kg/m2), overweight (25–30 kg/m2), and nonobese (<25 kg/m2). Smoking status was categorized as current, former, or never. Anemia, liver disease, chronic kidney disease (CKD), cardiovascular disease (CVD), diabetes, hypertension, hyperlipidemia, hyperuricemia, depression, and cancer were identified using questionnaire responses or laboratory criteria (Table S1).
Statistical analysis
Statistical analyses were performed in accordance with NHANES guidelines, taking into consideration the complex sampling design and applying appropriate sample weights.
Data preprocessing
Missing data were evaluated using the ‘dlookr’ package, revealing PIR and hyperuricemia as the variables with the most missing data at 11.01% and 5.55% (Table S2), respectively. The Recursive Partitioning and Regression Trees (RPART) method was used for imputation because of its effectiveness with both categorical and continuous data. Data distributions were examined before and after imputation to ensure consistency (Figure S1). Flavonoid intake data, which were skewed, were survey-weighted into quartiles and ln-transformed to stabilize variance before further analysis.
Descriptive statistics and statistical testing
Continuous variables were summarized as means and SDs for normally distributed data and as medians with interquartile ranges (IQRs) for skewed data. Categorical variables were presented as weighted percentages. The Wilcoxon rank-sum test was used for skewed continuous variables, whereas the chi-square test was used for categorical variables. Pearson’s correlation analysis was performed to evaluate associations among ln-transformed flavonoid values (Figure S2).
Multivariate logistic regression
Weighted multivariate logistic regression analyses, accounting for the complex sampling design, were conducted to examine the association between flavonoid intake and gallstones. Restricted cubic spline (RCS) analysis was performed to explore potential nonlinear associations. For the RCS analysis, we used three knots placed at the 10th, 50th, and 90th percentiles of the flavonoid intake distribution, following Harrell's recommended default. The nonlinearity was tested using the Wald test for the combined null hypothesis that all spline coefficients were zero. The fully adjusted model incorporated covariates like age, HEI-2020, sex, race, education level, PIR, BMI, physical activity level, smoking status, drinking status, depression, CKD, diabetes, hypertension, hyperlipidemia, liver disease, anemia, CVD, hyperuricemia, and cancer as covariates.
WQS regression and QGC model
The WQS Regression 16 and QGC Model 17 were used to evaluate the impact of combined flavonoid on gallstone prevalence. WQS assigns weights ranging from 0 to 1 to flavonoids, assuming a uniform, linear impact, but this assumption can limit its accuracy. The QGC model, however, did not assume uniform direction or linearity, allowing for a more detailed analysis. Both methods used quintiles and 5000 bootstrap iterations to improve robustness. In the WQS analysis, 40% of the dataset was used as the validation set to evaluate model reliability and accuracy.
BKMR
The BKMR model 18 was used to assess the combined effects of 29 flavonoids across six subclasses. It uses Markov chain Monte Carlo (MCMC) to estimate the relationship between chemical mixtures and outcomes through Gaussian kernel regression. A hierarchical variable selection method was applied to estimate the posterior inclusion probabilities (PIPs), including group and conditional PIPs, which indicate the probability of including groups or specific chemicals in the model. This approach helps identify the most influential flavonoids within each group and the most influential group overall. The exposure–response relationship was estimated by fixing the remaining flavonoids at their median levels, and the effect of changing the exposure of one flavonoid was assessed at different percentiles. The model was run for 10,000 iterations separately for males and females. Note that BKMR results were not interpreted based on conventional statistical significance.
Sensitivity analysis
We utilized multiple mixture analysis models to cross-validate our findings, ensuring the robustness of our results. The E-value 19 is a metric used in observational studies to assess the robustness of an association against potential unmeasured confounding. Specifically, it represents the minimum strength of association, on the risk ratio scale, that an unmeasured confounder would need to have with both the exposure and the outcome to fully explain the observed association.
Statistical analyses were carried out utilizing R (version 4.4.1) along with packages such as ‘dietaryindex,’ 20 ‘bkmrhat,’ ‘qWQS,’ ‘qgcomp,’ ‘rms,’ ‘survey,’ and ‘ggplot2.’ Regression results are presented as odds ratios (OR) accompanied by 95% confidence interval (CI), with significance established at p <0.05.
Ethics statement
This study was conducted in accordance with the Declaration of Helsinki of 1975, as revised in 2024.
Results
Participant characteristics
A total of 4087 individuals, representing 235,764,780 people after complex sample weighting, were included in the present study. The overall prevalence of gallstones was 11.00% (n = 475). The average age of participants was 49 years, with those having gallstones being significantly older than those without gallstones (p < 0.001). Females accounted for 51.80% of the total participants. Notably, among participants with gallstones, females constituted a significantly higher proportion (72.58%) than males (27.42%). However, in the group without gallstones, males accounted for a slightly higher proportion (50.87%) than females (49.13%). Participants with gallstones tended to be more overweight and engaged in lower levels of physical activity. Additionally, they had a higher prevalence of several comorbidities, including depression, CKD, and diabetes. Other significant differences were observed in the prevalence of hyperlipidemia, liver disease, anemia, CVD, and cancer. Furthermore, participants with gallstones had a lower median intake of total flavones (p = 0.044) (Table 1). The distribution of 29 individual flavonoids was also examined among participants with and without gallstones (Table S3). Specifically, participants with gallstones had significantly higher median intakes of theaflavin, thearubigins, luteolin, theaflavin-3,3′-digallate, theaflavin-3′-gallate, and theaflavin-3-gallate than those without gallstones.
Population characteristics by gallstones status in adults, National Health and Nutrition Examination Survey (NHANES 2017–2018).
Median (IQR) for skewed distribution; mean and SD for normal distribution; n (unweighted) (%); N (weighted).
Wilcoxon rank-sum test for complex survey samples; chi-squared test with Rao–Scott's second-order correction.
p < 0.05.
BMI: body mass index; CKDs: chronic kidney diseases; CVDs: cardiovascular disease; HEI-2020: healthy eating index-2020; IQR: interquartile range; PIR: poverty income ratio.
In this study, females had a significantly higher prevalence of gallstones than males (15.96% vs. 6.48%, p < 0.001) (Table S4). Additionally, females were tending to be older and had a higher healthy eating index score. They were less likely to smoke but had lower levels of physical activity. However, there were no significant sex differences in the prevalence of most comorbidities. Notably, males had a higher prevalence of CVD, whereas females had a higher prevalence of anemia. Notable disparities were found in the consumption of several individual flavonoids among males and females, with females generally having higher intakes except for kaempferol. Among the flavonoid subclasses, females had higher intakes of anthocyanidins and flavanones but a lower intake of flavonols than males.
Associations of total and subclassed of flavonoids with gallstones
In the analysis of the overall population, the findings did not indicate a significant correlation between total or flavonoid subclass intake and gallstone prevalence (Table 2). Nevertheless, a U-shaped nonlinear relationship was detected for total flavonoid intake (p < 0.001) (Figure 2), suggesting that at lower levels of intake, there is a negative association with gallstone prevalence, which transitions to a positive association at higher intake levels. Similarly, subclasses such as isoflavones, flavan-3-ols, and flavonols also exhibited U-shaped relationships.
Association between ln-transformed subclassed and total sum favonoids intake and gallstones prevalence. (a) Results in the total population; (b) results in sex subgroups are shown in blue (males) and pink (females). The horizontal dashed line represents the reference odds ratio of 1.0. All the models were adjusted for age, sex (except for sexsubgroup analysis), HEI-2020, race, education level, PIR, BMI, physical activity level, smoking status, drinking status, depression, chronic kidney disease (CKD), diabetes, hypertension, hyperlipidemia, liver conditions, anemia, cardiovascular diseases (CVDs), hyperuricemia, and cancer.
Association between ln-transformed sum and subclass flavonoids concentration and gallstones risk by weighted logistic regression and RCS model.
p < 0.05.
The model was adjusted for age, HEI-2020, sex, race, education level, poverty income ratio (PIR), body mass index (BMI), physical activity level, smoking status, drinking status, depression, chronic kidney disease (CKD), diabetes, hypertension, hyperlipidemia, liver conditions, anemia, cardiovascular diseases (CVDs), hyperuricemia, and cancer.
CI: confidence interval; HEI-2020: healthy eating index-2020; OR: odds ratio; RCS: restrict cubic spine.
When stratified by sex, a significant positive association was observed among females. The aggregate of all flavonoids and the flavonol subclass intake were significantly associated with an increased prevalence of gallstones, with ORs of 1.166 (95% CI: 1.053–1.290, p = 0.003) and 1.253 (95% CI: 1.107–1.417, p < 0.001), respectively. Moreover, both isoflavones and flavan-3-ols exhibited U-shaped relationships among females. In males, no significant association was observed between total flavonoid intake and gallstone prevalence. However, flavanone intake was significantly associated with a lower prevalence of gallstones (OR = 0.912; 95% CI: 0.833–0.998; p = 0.046), whereas the other flavonoid subclasses showed no significant associations. For total flavonoid intake, a J-shaped relationship was observed in males, suggesting a negative association at lower intake levels followed by a plateau at higher levels. Among the subclasses, only flavanones exhibited a similar J-shaped pattern. Similar findings were observed in the multivariable logistic regression analyses using categorized original flavonoid values (Figure 3), where only the intake of the sum of all flavonoids and flavonols showed a positive and increasing trend with gallstone prevalence in females, whereas no such trends were observed in the overall population or among males nor for the other flavonoid subclasses.
Associations of ln-transformed subclassed and total sum flavonoids intake with gallstones prevalence in all participants. (a) Overall population; (b) females; (c) males. All the models were adjusted for age, sex (except for sex subgroup analysis), HEI-2020, race, education level, PIR, BMI, physical activity level, smoking status, drinking status, depression, chronic kidney disease (CKD), diabetes, hypertension, hyperlipidemia, liver conditions, anemia, cardiovascular diseases (CVDs), hyperuricemia, and cancer.
We further analyzed the associations between each of the 29 flavonoids intake with gallstones. In the overall population, the flavan-3-ols subclass was where most significant associations were detected (Table S5). For instance, specific flavan-3-ols like theaflavin-3-gallate, theaflavin-3,3′-digallate, theaflavin-3′-gallate, theaflavin, and thearubigins were all significantly positively associated with gallstone prevalence (p < 0.05). These relationships were consistently positive in females but not in males. Additionally, epicatechin and epigallocatechin within the flavan-3-ol subclass, as well as myricetin and quercetin within the flavonol subclass, were positively associated with gallstones in females. In contrast, naringenin of flavones, luteolin and kaempferol of flavonols were negatively associated with gallstones in males.
Regarding nonlinear relationships (Figure S3 and S4), in males, only genistein within the isoflavone subclass showed an increasing trend, whereas the remaining significant associations were either decreasing or exhibited nearly flat fluctuations. In females, however, nearly all nonlinear relationships demonstrated a trend of initially decreasing and then increasing, with some showing a globally upward trend. Those results indicated single flavonoid presents diverse effect in different sexs, and the flavan-3-ols acts as a key subclass.
Associations of flavonoids mixed exposure with gallstones using WQS regression and QGC model
Considering the inconsistencies in the associations between the subclasses of flavonoids and gallstone prevalence, as well as those of individual flavonoids, we employed mixture models to provide a more comprehensive assessment. Both directions of the WQS index (Table 3) were modeled, yielding consistent results. In the overall population, there was a positive link between the WQS index and gallstone prevalence (OR = 1.774; 95% CI: 1.218–2.586; p = 0.003). This association was also significant for females (OR = 1.945; 95% CI: 1.202–3.149; p = 0.007) but not for males (p = 0.876). The E-value for the overall population was 2.946 (CI = 1.733), indicating that an unmeasured confounder would need to be associated with both flavonoid intake and gallstone prevalence by an OR greater than 2.946 to fully explain the observed association. For females, the E-value was 2.137 (CI = 1.421), indicating a moderate degree of resistance to unmeasured confounding.
The outcomes of the QGC model (Table 3) were aligned with the WQS regression results, revealing a positive association between the combined intake of 29 flavonoids and gallstone prevalence in the overall population (OR = 1.945; 95% CI: 1.180–3.204; p = 0.009). This correlation was also significant among females (OR = 2.056; 95% CI: 1.090–3.877; p = 0.026) but not among males (p = 0.191). In the QGC model, the E-value for the overall population was 3.301, and for females, it was 2.227, both of which suggest robust findings.
Association between mixture of 29 flavonoids and gallstones prevalence by WQS regression and QGC model.
The E-value represents the minimum effect size, indicating that it has a lower bound but no upper limit.
p < 0.05.
The models were adjusted for age, HEI-2020, sex, race, education level, poverty income ratio (PIR), body mass index (BMI), physical activity level, smoking status, drinking status, depression, chronic kidney disease (CKD), diabetes, hypertension, hyperlipidemia, liver conditions, anemia, cardiovascular diseases (CVDs), hyperuricemia, and cancer.
CI: confidence interval; HEI-2020: healthy eating index-2020; OR: odds ratio; QGC: quantile G-computation; WQS: weighted quantile sum.
Weights of flavonoids intake
In the positive WQS regression analysis, the weights of individual flavonoids were calculated to determine their relative contributions to gallstone prevalence (Figure 4; Table S6). In the overall population, thearubigins and theaflavin-3-gallate, both belonging to the flavan-3-ol subclass, had the highest weights at 0.466 and 0.158, respectively. Among females, major contributors included thearubigins, theaflavin-3,3′-digallate, and theaflavin, all of which also belong to the flavan-3-ol subclass. Additionally, the isoflavones glycitein and genistein had weights above the threshold in both the overall population and among females.
Estimated weights of 29 flavonoids intake in both positive and negative directions using WQS regression in different populations. (a) Overall population; (b) females, and (c) males. The vertical red dot lines indicate the threshold of 1/29. All the models were adjusted for age, sex (except for sex subgroup analysis), HEI-2020, race, education level, PIR, BMI, physical activity level, smoking status, drinking status, depression, chronic kidney disease (CKD), diabetes, hypertension, hyperlipidemia, liver conditions, anemia, cardiovascular diseases (CVDs), hyperuricemia, and cancer.
In the QGC model (Figure S5 and Table S7), theaflavin-3,3′-digallate, theaflavin, and thearubigins were identified as the three largest contributors in both the overall population and among females. However, it is noteworthy that theaflavin exhibited a negative weight in the overall population, and theaflavin-3,3′-digallate showed a negative weight among females, which was inconsistent with the direction observed in the logistic regression analyses. This discrepancy may be attributed to the complex interactions among the various components of the flavonoid mixture. Additionally, glycitein emerged as a major contributor, consistent with the findings from both the WQS model and multivariable logistic regression. The results suggest that flavan-3-ols and isoflavones, including specific constituents such as thearubigins, theaflavin-3,3′-digallate, theaflavin, and glycitein, are associated with gallstone prevalence, particularly among females.
BKMR model
The BKMR model was used to assess the combined effects of flavonoids on gallstone prevalence after stratification by sex. Among females, a positive association was observed as the percentile exposure levels of all flavonoids increased simultaneously (Figure 5(a)). In contrast, males exhibited an inverted U-shaped relationship, with effect estimates consistently below zero (Figure 5(b)). Among females, the posterior probability that the overall mixture effect was positive was 0.98, indicating strong directional evidence of a positive association. In males, this probability was 0.21, suggesting no evidence of a positive association. To assess individual effects, all other flavonoid exposures were fixed at their 50th percentile levels. Univariate exposure–response analyses demonstrated upward trends among females for theaflavin-3,3′-digallate, theaflavin-3-gallate, daidzein, genistein, and glycitein, whereas isorhamnetin showed a downward trend (Figure 5(c) and Figure S6(a)). Other flavonoids did not exhibit clear trends. In males, none of the individual flavonoids showed significant trends (Figure 5(d) and Figure S6(b)). The BKMR model identified the flavan-3-ol subclass as a key contributor to the overall association among females (group PIP = 0.989, Table S8). In addition, bivariate exposure–response analyses suggested that apigenin, cyanidin, epicatechin, and eriodictyol may interact extensively with other flavonoids (Figures S7 and S8).
Overall association estimates with 95% CI for the overall 29 flavonoids mixture intake and gallstones prevalence in the Bayesian kernel machine regression (BKMR) models stratified by sexs. All the flavonoids at particular percentiles (from 0.25 to 0.75 increment by 0.05) were compared to all the flavonoids at their 50th percentile in (a) females and (b) males. In females, the posterior probability of a positive overall association was 0.98, indicating strong directional evidence; in males, this probability was 0.21. These Bayesian posterior probabilities provide evidence for a positive association in females, even though the 95% credible interval includes the null value. Association of 29 flavonoids intake individually with gallstones risk using the BKMR model in (c) females and (d) males. Overall association estimates were adjusted for adjusted for age, sex (except for sex subgroup analysis), HEI-2020, race, education level, PIR, BMI, physical activity level, smoking status, drinking status, depression, chronic kidney disease (CKD), diabetes, hypertension, hyperlipidemia, liver conditions, anemia, cardiovascular diseases (CVDs), hyperuricemia, and cancer.
Discussion
This study identified a U-shaped relationship between total dietary flavonoid intake and gallstone prevalence among US adults, with the novel finding of marked sex specificity: significant positive associations were observed among females but not among males. These findings highlight the complex, nonlinear nature of flavonoid–gallstone relationships and underscore the importance of considering sex as a critical effect modifier in dietary epidemiology. In the overall population, no significant linear association was detected between total or flavonoid subclass intake and gallstone prevalence. However, a U-shaped nonlinear relationship was observed (p < 0.001), suggesting a negative association at lower intake levels and a positive association at higher intake levels. Similar U-shaped relationships were observed for subclasses such as isoflavones, flavan-3-ols, and flavonols. This biphasic pattern suggests a complex dose–response association. Lower to moderate flavonoid intake levels have been associated with favorable antioxidant and anti-inflammatory biomarkers, which in turn have been linked to a lower prevalence of gallstones in observational studies.21,22 Conversely, at higher intake levels, excessive flavonoid intake may interfere with enterohepatic bile acid circulation, increase biliary cholesterol saturation, or alter gallbladder motility, thereby increasing gallstone prevalence.23,24 Similar biphasic dose–response relationships have been reported for other dietary components (e.g. alcohol and coffee) in gallstone epidemiology. Similarly, another study of 4113 participants demonstrated significant inverse associations between total flavonoid intake and hepatic steatosis, further supporting the association of flavonoids with favorable metabolic health indicators. 25 However, contrasting findings have also been reported. A study involving 6629 NHANES participants found no significant association between saturated fatty acid intake and gallstone prevalence, suggesting that dietary components may interact differently with gallstone development across populations and dietary patterns. 26 The discrepancies in findings might stem from variations in study design, sample characteristics, and the specific dietary components examined.
When stratified by sex, significant positive associations were observed among females. The aggregate of all flavonoids and the flavonol subclass intake were significantly associated with increased gallstone prevalence, with ORs of 1.166 (95% CI: 1.053–1.290, p = 0.003) and 1.253 (95% CI: 1.107–1.417, p < 0.001), respectively. Moreover, both isoflavones and flavan-3-ols exhibited U-shaped relationships among females, with lower intakes negatively and higher intakes positively associated with gallstone prevalence. In males, however, no significant association was observed between total flavonoid intake and gallstone prevalence. Only flavanone intake was significantly associated with a lower prevalence of gallstones (OR = 0.912; 95% CI: 0.833–0.998; p = 0.046), whereas the remaining flavonoid subclasses showed no significant associations. This sex-specific response underscores the complexity of the association between diet and gallstone prevalence. Females typically have higher estrogen levels, which are known to influence bile acid synthesis and cholesterol metabolism. 27 Estrogen upregulates the expression of the cholesterol transporter ABCG5/G8, promoting biliary cholesterol secretion. 28 Because many flavonoids are phytoestrogens or can modulate estrogen receptor activity, they may influence bile acid homeostasis and cholesterol transport through estrogen-related pathways.29,30 This provides a biologically plausible explanation for why the positive association between flavonoid intake and gallstone prevalence was observed in females but not in males.
These sex-specific findings were further supported by the mixture analyses. Both the WQS and QGC models demonstrated significant positive associations between combined flavonoid exposure and gallstone prevalence in the overall population (WQS: OR = 1.774, 95% CI: 1.218–2.586, p = 0.003; QGC: OR = 1.945, 95% CI: 1.180–3.204, p = 0.009) and among females (WQS: OR = 1.945, 95% CI: 1.202–3.149, p = 0.007; QGC: OR = 2.056, 95% CI: 1.090–3.877, p = 0.026), whereas no significant association was observed among males. The corresponding E-values (2.946 for the overall population in the WQS model and 3.301 in the QGC model) indicate moderate to strong robustness to unmeasured confounding. In the positive WQS regression, thearubigins and theaflavin-3-gallate, both belonging to the flavan-3-ol subclass, had the highest weights at 0.466 and 0.158, respectively. Among females, significant contributors included thearubigins, theaflavin-3,3′-digallate, and theaflavin, all of which are also part of the flavan-3-ol subclass. Additionally, isoflavones such as glycitein and genistein had weights above the threshold. The QGC model identified theaflavin-3,3′-digallate, theaflavin, and thearubigins as the three largest contributors in both the overall population and among females. The BKMR model further identified flavan-3-ols as the predominant contributor among females (group PIP = 0.989). In females, a positive association was observed as the percentile exposure levels of all flavonoids increased simultaneously, whereas males exhibited an inverted U-shaped relationship with effect estimates consistently below zero. Univariate exposure–response analyses demonstrated increasing trends among females for theaflavin-3,3′-digallate, theaflavin-3-gallate, daidzein, genistein, and glycitein. Collectively, these findings from multiple mixture models consistently identify flavan-3-ols and isoflavones as key contributors to the observed associations among females. An important consideration regarding the positive associations observed among females, particularly for theaflavins and thearubigins, is the possibility of protopathic bias. These compounds are derived almost exclusively from black tea. It is plausible that individuals, particularly women experiencing undiagnosed or early symptoms of gallstone disease, such as postprandial bloating or fatty food intolerance, altered their dietary habits before participating in the NHANES survey. For example, patients with biliary symptoms may have reduced or eliminated black tea consumption if they perceived it as a symptom trigger, which would be expected to produce a spurious inverse association that was not observed in this study. Conversely, if participants increased tea consumption because of its antioxidant effects before diagnosis, this could exaggerate a positive association. Although the cross-sectional design cannot establish temporal relationships, protopathic bias represents a plausible noncausal explanation for the association between theaflavins/thearubigins and gallstone prevalence among females. Prospective cohort studies with repeated dietary assessments before and after gallstone diagnosis are needed to further evaluate this possibility.
The predominance of flavan-3-ols, particularly theaflavins and thearubigins from black tea, is biologically plausible. Flavan-3-ols are the most abundant flavonoids in black tea, which is a major dietary source in the US population. 31 These compounds have higher bioavailability than many other flavonoid subclasses and are extensively metabolized by the gut microbiota into active conjugates that can enter the enterohepatic circulation. 32 Mechanistically, theaflavins and thearubigins have been shown to inhibit cholesterol-7α-hydroxylase (CYP7A1), the rate-limiting enzyme in bile acid synthesis, thereby increasing cholesterol saturation in bile. 33 The gallate moiety in theaflavin-3-gallate and theaflavin-3,3′-digallate may enhance binding to nuclear receptors such as farnesoid X receptor (FXR), which regulates bile acid homeostasis. 34 Additionally, flavonoid metabolites can alter bile acid pool composition through interactions with the gut microbiota, potentially promoting lithogenic bile, as shown in Figure 6. 35 The interaction between flavonoid signaling and estrogen-mediated pathways may further amplify these effects among females. Estrogen receptors are known to interact with flavonoid signaling pathways, 36 and the higher baseline estrogen levels among females may create a metabolic environment in which flavonoid-induced alterations in bile acid synthesis and cholesterol transport are more pronounced. Future research should focus on elucidating the key transcription factors that regulate these specific cellular subtypes, as understanding their roles could provide insights into targeted interventions for gallstone prevention.
Conceptual diagram of flavonoids and gallstone mechanisms.
In contrast, males showed no significant positive associations in the mixture analyses. In the multivariable logistic regression analysis, higher flavanone intake was associated with a lower prevalence of gallstones (OR = 0.912; 95% CI: 0.833–0.998; p = 0.046), and specific flavonoids, including naringenin, luteolin, and kaempferol, also showed inverse associations. However, these associations were not corroborated by the WQS or QGC models, and the effect size was small. Given the inconsistency across analytical approaches, these findings should be interpreted cautiously as exploratory. The mechanisms underlying this sex disparity may involve differences in estrogen-mediated bile acid metabolism and baseline flavonoid intake, although the lack of statistical robustness precludes definitive conclusions. Additionally, males generally have lower baseline flavonoid intake and different dietary patterns, which may alter the dose–response relationship. Beyond sex, other demographic factors may also modify the association between flavonoid intake and gallstone prevalence. Ethnic differences in flavonoid metabolism and gallstone prevalence may exist because of genetic polymorphisms in genes involved in flavonoid absorption (e.g. SGLT1 and BCRP) and phase II metabolism (e.g. UGT1A1 and COMT). 35 These protective associations among males warrant further investigation and should not be overlooked when formulating population-wide dietary recommendations.
The consistency of findings across the multivariable logistic regression, WQS, QGC, and BKMR models strengthens the robustness of our conclusions. Each analytical approach addresses different methodological challenges: logistic regression evaluates individual exposure–outcome associations while controlling for confounders; WQS and QGC assess mixture effects under different directional assumptions; and BKMR accounts for complex nonlinear interactions and identifies key contributors within the mixture. The positive association between flavonoid intake and gallstone prevalence among females was consistently observed across all four approaches, whereas the null or inverse findings among males were similarly consistent. This multifaceted analytical strategy provides a more nuanced understanding of how dietary factors may be associated with gallstone prevalence and underscores the importance of considering both individual and combined effects of dietary flavonoids.
This study provides important insights into the association between dietary flavonoid intake and gallstone prevalence, especially emphasizing a positive correlation in females with no significant association observed in males. Nevertheless, several limitations should be acknowledged. First of all, the NHANES data are cross-sectional, precluding the establishment of causal relationships. Subsequent studies could employ a prospective cohort design to better investigate the causal association between flavonoid intake and gallstone prevalence. Then, dietary intake data are based on participants’ recollection, potentially introducing recall bias. Future studies could use more precise dietary assessment methods to improve data accuracy. The presence of gallstones is based on the self-reported physician diagnosis by the participants, lacking confirmation through imaging or medical records. This may involve recall bias or classification bias. Although previous studies have supported the validity of self-reported diagnoses in epidemiologic surveys, future research still recommends using imaging studies to verification. Additionally, reverse causality and protopathic bias cannot be excluded. Individuals with symptomatic gallstone disease or a recent diagnosis may have already modified their dietary habits, including reducing or avoiding black tea consumption, before participating in NHANES. Such behavioral changes could have influenced the observed associations, and the cross-sectional design does not allow us to distinguish whether flavonoid intake precedes or follows gallstone formation. Finally, flavonoid intake was assessed using two 24-h dietary recalls, which reflected short-term intake not long-term consumption. Future studies using repeated dietary assessments or food frequency questionnaires designed for flavonoid intake would help confirm the observed relationships. Although we accounted for numerous potential confounding factors, unmeasured confounders may still exist. Further studies could delve into these confounding factors such as ethnic differences to strengthen the reliability of these findings.
Conclusions
In summary, this study provides cross-sectional insights into factors associated with gallstone prevalence. The findings suggest that future dietary guidance may consider both flavonoid intake and sex. Among females, higher flavonoid intake was associated with increased gallstone prevalence, warranting further attention in observational studies. Among males, certain flavonoids were inversely associated with gallstone prevalence, although these findings require confirmation in prospective studies. Sex-specific dietary recommendations regarding flavonoid intake may therefore be warranted. Future research should further investigate the specific mechanisms underlying the effects of flavonoids to inform the development of new preventive and therapeutic strategies.
Supplemental Material
sj-docx-1-imr-10.1177_03000605261466549 - Supplemental material for U-shaped association between dietary flavonoid and gallstone prevalence with sex-specific effects: Insights from a cross-sectional study
Supplemental material, sj-docx-1-imr-10.1177_03000605261466549 for U-shaped association between dietary flavonoid and gallstone prevalence with sex-specific effects: Insights from a cross-sectional study by Jiaying Lin, Xin Lin, Yuquan Chen, Lihang Yang and Xiaoyan Xiu in Journal of International Medical Research
Supplemental Material
sj-docx-2-imr-10.1177_03000605261466549 - Supplemental material for U-shaped association between dietary flavonoid and gallstone prevalence with sex-specific effects: Insights from a cross-sectional study
Supplemental material, sj-docx-2-imr-10.1177_03000605261466549 for U-shaped association between dietary flavonoid and gallstone prevalence with sex-specific effects: Insights from a cross-sectional study by Jiaying Lin, Xin Lin, Yuquan Chen, Lihang Yang and Xiaoyan Xiu in Journal of International Medical Research
Footnotes
Acknowledgments
We acknowledge the NHANES participants and staff.
Ethics statement
This study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving research study participants were approved by the Institutional Review Board of the NCHS. Written informed consent was obtained from all participants. No additional Institutional Review Board approval is required.
Author contributions
Jiaying Lin, Xin Lin, and Yuquan Chen: Conceptualization, Methodology, Investigation, and Writing–original draft. Lihang Yang: Validation and Data curation. Xiaoyan Xiu: Supervision and Writing–review & editing. All authors have read and agreed to the current version of the manuscript.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Declaration of conflicting interests
The authors declare no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data availability statement
Supplemental material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
