Higher habitual intakes of flavonoids and flavonoid-rich foods are associated with a lower incidence of type 2 diabetes in the UK Biobank cohort

Aim To examine the associations of a diet high in flavonoid-rich foods, as reflected by a “Flavodiet Score” (FDS), the major individual food contributors to flavonoid intake, and flavonoid subclasses with type 2 diabetes (T2D) risk in the UK Biobank cohort. Materials and methods Flavonoid intakes were estimated from ≥2 dietary assessments among 113,097 study participants [age at enrolment: 56 ± 8 years; 57% female] using the U.S Department of Agriculture (USDA) databases. Multivariable Cox proportional hazards models were used to investigate associations between dietary exposures and T2D. Results During 12 years of follow-up, 2628 incident cases of T2D were identified. A higher FDS (compared to lower [Q4 vs. Q1]), characterised by an average of 6 servings of flavonoid-rich foods per day, was associated with a 26% lower T2D risk [HR: 0.74 (95% CI: 0.66–0.84), ptrend = <0.001]. Mediation analyses showed that lower body fatness and basal inflammation, as well as better kidney and liver function partially explain this association. In food-based analyses, higher intakes of black or green tea, berries, and apples were significantly associated with 21%, 15%, and 12% lower T2D risk. Among individual flavonoid subclasses, 19–28% lower risks of T2D were observed among those with the highest, compared to lowest intakes. Conclusions A higher consumption of flavonoid-rich foods was associated with lower T2D risk, potentially mediated by benefits to obesity/sugar metabolism, inflammation, kidney and liver function. Achievable increases in intakes of specific flavonoid-rich foods have the potential to reduce T2D risk.


INTRODUCTION
Type 2 diabetes (T2D) remains one of the top four major noncommunicable diseases (NCDs), contributing to >4 million diabetes-related deaths annually [1].Across the globe, 415 million people live with diabetes [1], with an estimated 90% of these people having T2D.Widely preventable, T2D is associated with higher risk of cardiovascular disease, dementia, and cancer, causing extensive global economic burden in healthcare costs [2].Modifiable risk factors such as overweight and obesity are considered as major contributors to T2D, and thus diet is crucial for primary prevention [3,4].Previous studies suggest a preventive association of healthy plant-based dietary patterns with T2D risk beyond their effects on energy balance [5].
Favourable health effects of plant-food consumption have been attributed to flavonoids, a class of polyphenolic compounds that are commonly found in fruits, vegetables, dark chocolate, tea, and red wine [6].Flavonoids have been categorised into six main subclasses (flavanones, flavones, flavan-3-ols (including proanthocyanidins), flavonols, anthocyanins, and isoflavones), with different bioavailability and bioactivity [7,8].Several prospective studies have reported inverse associations between numerous flavonoid subclasses and incident T2D risk [9].There is growing evidence from short-term randomised controlled trials (RCTs) to suggest that higher intakes of various flavonoid subclasses, namely anthocyanins and flavan-3-ols consumed through flavonoid-rich food sources, such as cocoa, tea and blueberries improves insulin sensitivity, insulin resistance, and lipid profiles [10][11][12].
To better understand the role of dietary flavonoid intake in T2D prevention, and given that flavonoids are typically consumed as part of a dietary pattern through flavonoid-rich foods, we investigated whether a flavonoid-rich diet represented by a novel Flavodiet Score (FDS), which is associated with lower mortality risk [13], is associated with incident T2D in a large-scale populationbased study, the UK Biobank, for the first time.We further carried out mediation analyses to evaluate which potential biological mechanisms may mediate associations between a flavonoid-rich diet and T2D risk.We also aimed to evaluate associations between the major food contributors to flavonoid intake and habitual intakes of flavonoid subclasses with incident T2D.

MATERIALS AND METHODS Study population
The UK Biobank is a large-scale population-based prospective cohort study of >500,000 participants aged between 40 and 69 years and recruited between 2006 and 2010 from the UK.Participants attended one of 22 assessment centres located across England, Scotland, and Wales, where they completed a baseline assessment comprising a touchscreen questionnaire and a series of physical assessments.Further information on the study protocol has been previously reported [14].All UK Biobank participants provided written informed consent at recruitment and the study received ethical approval from the NHS North West Multicentre Research Ethics Committee (Ref.11/NW/0382).
Participants who completed fewer than two 24-h dietary assessments (n = 372,173), had implausible energy intakes (>17,573 KJ or <3347 KJ for men and >14,644 KJ or < 2092 KJ for women [15]) (n = 3953), or withdrew their consent during the study follow-up were excluded from the analysis.Participants were also excluded if they had diabetes (n = 5009), CVD (n = 1868), cancer (n = 5920), were diagnosed with T2D before the last 24h dietary assessment (n = 301), or if they were pregnant at baseline assessment (n = 46) (Supplemental Fig. S1).

Flavodiet Score (FDS)
Further details on the Oxford WebQ dietary assessment tool, including the method used to estimate flavonoid intakes are provided in the Supplementary Appendix.The top three foods that contributed the highest percent to total flavonoid and flavonoid subclass intakes were considered eligible for inclusion in the FDS (excluding fruit juices).In addition, dark chocolate was also considered eligible for inclusion within the score as it typically contains high concentrations of flavonoids.However, in this UK population, intakes of dark chocolate were low, with a median (range) of 0 (0-2.5)servings/day, and thus, did not contribute highly to flavonoid intakes.Based on mean intakes (servings/day) from a minimum of two 24-h dietary assessments, the score was made up of ten Oxford WebQ food items (tea (black and green), red wine, apples, berries, grapes, oranges (including satsumas), grapefruit, sweet peppers, onions, and dark chocolate).A final FDS was calculated by summing the total number of servings consumed across the ten selected food items.Tea (black and green) intakes were capped at a maximum of 4 servings/day due to the high intakes reported in this British population (median (range) 2.3 (0-11.5)).From this, the FDS was categorised into sex-specific quartiles.As we were interested in flavonoids rather than alcohol consumption, (a) we adjusted for total alcohol consumption in our statistical models on the FDS and T2D risk, and (b) we re-ran our statistical analyses on the FDS excluding red wine consumption.
As previously conducted for food groups in the UK Biobank [16], intraclass coefficients (ICCs) were calculated to test the reproducibility of the FDS over time and Spearman's rank correlation coefficients were calculated to test the reliability of flavonoid intakes over time, using average flavonoid-rich food intakes from the Oxford WebQ dietary assessments at the second and third assessment (collected in February-April 2011 and June-August 2011) versus the fourth and fifth (October-December 2011 and April-June 2012).A subgroup of 21,543 participants who completed each of these dietary assessments were included in this analysis.Further, Spearman's rank correlation coefficients were calculated to test the agreement between flavonoid intakes estimated from U.S Department of Agriculture (USDA) and Phenol-Explorer databases.

Covariates
Sociodemographic, dietary, and lifestyle factors were self-reported at the baseline assessment between 2006 and 2010.Covariates considered in this study included: sex, age at recruitment, ethnicity, geographical region of recruitment, education, physical activity, alcohol intake, smoking status, energy intake, BMI, waist circumference, polypharmacy index, multimorbidity index, genetic risk of T2D (polygenic risk score (PRS) generated and issued by the UK Biobank for use upon request [17,18]), Townsend deprivation index, family history of diabetes, hypercholesterolemia, hypertension, menopausal status, number of completed dietary assessments, wholegrain intake, sugar-sweetened beverage intake, red and processed meat intake, coffee intake and the healthful plant-based diet index [19].Further details on how the covariates were classified for this study can be found in Table S1.

Case ascertainment
Incident T2D cases were defined as primary type 2 diabetes mellitus according to the International Classification of Diseases 10th edition (ICD-10) (E11), using UK Biobank linked hospital inpatient data on admissions and diagnoses available until the 30th of September 2021 from the Hospital Episode Statistics (HES) for England, 31st of July 2021 for Scottish Morbidity Records (SMR), and 31st of March 2016 for the Patient Episode Database for Wales (PEDW).Follow-up for incident T2D analyses was censored at date of hospitalisation, death, or end of follow-up, whichever occurred first.

Statistical analyses
Cox proportional hazards regression models were used to assess the relationship between the FDS, flavonoid subclass intakes, and intakes of individual flavonoid-rich foods, and incident T2D, producing hazard ratios (HR) and 95% confidence intervals (CI).The FDS was divided into sexspecific quartiles according to summed scores.Flavonoid intakes and major flavonoid-food contributors were also grouped into sex-specific quartiles, with the lowest quartile serving as the referent group across all analyses.The FDS, flavonoid subclass intakes, and flavonoid-rich foods were also modelled as continuous exposure variables when carrying out linear trend tests (P-trend).
To investigate effect modification, stratified analyses were carried out across potential risk modifiers smoking status (never, ever), sex (male, female), BMI (<25, ≥25 kg/m 2 ), education (low: GSEs/O-Levels/GCSEs or equivalent, NVQ/HND/HNC/A-Levels/AS-Levels or equivalent; high: Other professional qualifications, College/university degree), ethnicity (white, non-white), and alcohol intake (<1 g/day, >1 g/day).Further, heterogeneity was assessed across strata of polygenic risk (T2D) (PRS tertiles; low, intermediate, high).Likelihood ratio tests (LRT) were used to test for potential effect modification of associations between the FDS and T2D risk by covariates.Multiplicative terms between the FDS (continuous) and covariates along with the main effect term were added to fully adjusted Cox regression models.LRTs were used to compare models with and without multiplicative terms.
A mediation analysis was also carried out to assess the association between potential mediators (biomarkers of obesity/sugar metabolism, inflammation, kidney and liver function, and lipid metabolism) on the FDS-T2D pathway.Details on the rationale for the choice of potential mediators and on the statistical methods are provided in the Supplementary Appendix.
Stata version 17.0 (Stata Corp LP, College Station, TX) was used to conduct all analyses.All reported p values were two-sided, with statistical significance set at p < 0.05 for main analyses.To account for multiple testing in flavonoid-rich food and flavonoid subclass analyses, Bonferroni correction was used [20].

Characteristics of the study population
Baseline characteristics and key nutrient intakes among 113,097 study participants with data available from a minimum of 2 or more dietary assessments at baseline (mean (SD), 3.0 (0.9)) are presented across quartiles of the FDS in Table 1 and Table S2.During a median of 12 years of follow-up (1,358,384 person-years), 2628 cases of incident T2D cases occurred.Of the participants included, the median age (IQR) was 57 [12], 63,855 (56.5%) were female, 7864 (7.0%) were smokers, and 66,708 (59.0%) had a BMI over 25 kg/m 2 .The FDS ranged between 0 and 20.8 points, with higher scoring participants being more likely to be female, be of older age, be more physically active, and have a higher degree of education compared to participants with lower scores.The reproducibility ICC (range) of the FDS over time was 0.68 (0.1-30.5).For flavonoid intakes, Spearman's coefficients for correlations over time ranged between 0.36 and 0.68 (Table S3).When comparing flavonoid intakes estimated through USDA and Phenol-Explorer databases, Spearman's rank correlation coefficients were high for all comparisons (r > 0.79) except for the flavone subclass (Spearman's r = 0.10) (Table S4).

Flavonoid intake and incident type 2 diabetes risks
Associations between flavonoid subclass intakes and risk of T2D across quartiles are shown in Tables S7 and S8.Comparing high (Q4) to low (Q1) flavonoid subclass intakes, a higher intake of anthocyanins, flavan-3-ols, flavonols, flavones, polymers, and proanthocyanidins was associated with a 19%, 26%, 28%, 19%, 26%, and 27% lower risk of T2D after multivariable adjustments for demographic and lifestyle factors respectively.These associations remained statistically significant after Bonferroni correction.When assessing flavonoid subclass intake derived from Phenol-Explorer and T2D risk, results remained unchanged for all subclasses apart from flavones, where significant associations were lost upon multivariable adjustment (Table S9).

Subgroup and sensitivity analyses
Stratified associations between the FDS and T2D are shown in Table S10, with no indication of heterogeneity across smoking status, sex, BMI, education, ethnicity or alcohol intake (P interaction > 0.05).In the multivariable model adjusted for demographic and lifestyle characteristics, a higher FDS (Q4) compared to a lower FDS (Q1) across strata of genetic T2D risk only showed statistically significant inverse associations with incident T2D in intermediate and high-risk groups (HR: 0.71 [95%CI: 0.57-0.90],p trend = 0.02; HR: 0.71 [95%CI: 0.60-0.84],p trend = <0.001)(Table S11).However, there was no statistical heterogeneity in associations across strata of genetic risk.Further, when adjusting for the healthful plantbased diet index as a potential confounder, significant inverse associations remained for a higher FDS (Q4) compared to a lower FDS (Q1) (Table S12).Excluding study participants with ≤2 years of follow-up did not notably alter results (Table S13).

Mediation analysis
Mediation analyses based on biomarker measurements were carried out to explore potential biological mechanisms (Table 4).Results showed BMI, IGF-1, C-reactive protein, cystatin C, urate, gamma glutamyl transferase (GGT) and alanine aminotransferase (ALT) to potentially mediate the FDS-T2D associations by 2-5%.Cumulative mediation analyses of significant mediators showed the observed associations between the FDS and T2D to be mediated by a proportion of 28% overall.Other factors (including creatinine, aspartate aminotransferase (AST), LDL-direct cholesterol or lipoprotein(a)) were not identified as potential mediators between the FDS and T2D risk.

DISCUSSION
For the first time in a population-based study, we examined the relationship between consumption of a flavonoid-rich diet as reflected by a higher FDS with risk of T2D.Results showed that a greater FDS, characterised by an average of 6 servings of flavonoid-rich foods per day (from several sources) was associated with a 26% lower risk of T2D.This association did not differ across groups of high genetic T2D risk or other covariates.Mediation analyses indicated that the protective association of flavonoid intake with T2D risk may be mediated, in part, by benefits of a flavonoid-rich diet on obesity and sugar metabolism (BMI and IGF-1), basal inflammation (C-reactive protein levels), kidney function (cystatin C and urate), and liver status (GGT and ALT).Of the major food contributors to flavonoid intake, red wine, tea (black and green), berry and apple intake were inversely associated with T2D risk.While our study was the first to examine associations between the FDS, a novel tool that allows a real-world compatible operationalization of a flavonoid-rich diet, and T2D risk in a large prospective cohort, associations observed for flavonoid subclass intakes are in agreement with several previous studies.In a metaanalysis of 9 prospective cohort studies, higher dietary intakes of flavanols, flavonols, flavan-3-ols, and isoflavones were associated with up to a 14% lower risk of T2D compared to lower intakes [9].Discrepancies in findings on flavanone intakes and T2D risk currently exist; an observational study in nondiabetic participants in the PREDIMED trial has previously shown inverse associations between flavanone intakes and T2D risk [21], while a large US prospective study showed positive associations [22].In line with meta-analyses on flavonoid subclass intake and T2D risk [9,23], our study found no associations between flavanone intakes and T2D.This lack of association may be attributed to fruit juice consumption, which Sugar-sweetened beverage intake, mean (SD) c 0.7 (0.9) 0.5 (0.8) 0.4 (0.7) 0.4 (0.7) 0.5 (0.8) Red and processed meat intake, mean (SD) c 0.9 (0.8) 0.9 (0.8) 0.9 (0.8) 0.9 (0.8) 0.9 (0.   largely contributes to flavanone intakes [24] (38% orange juice in this study), and has previously been associated with increasing the risk of T2D due to its high glycaemic load [25].Tea, berries, apples, grapes, oranges, grapefruit, sweet peppers, onions, and dark chocolate were considered as major contributors to flavonoid intake and were therefore considered in the FDS.In a metaanalysis of 16 cohort studies, linear inverse associations were reported between tea consumption and T2D risk [26].Likewise, meta-analyses on berry fruits, apples, grapes, and grapefruit and risk of T2D have demonstrated similar associations [27][28][29].Similar to the meta-analysis carried out by Halvorsen et al. [29] on fruit and vegetable consumption and risk of T2D, we also reported no significant associations between orange and onion intake and T2D.As one of the few meta-analyses to have published on fruit and vegetable subtypes, Halvorsen et al. [29] did not report on associations between 'other' vegetable intake, including peppers.Of the previous studies published on fruit and vegetable intake and risk of T2D, only one study considered pepper intake within 'other vegetables', reporting inverse associations [30].In contrast to results from our analyses where no significant associations were found between pepper intake and T2D risk, inverse associations reported by Villegas et al. [30] may partly be explained by peppers being grouped with 'other' vegetables, exerting a greater combined benefit.Therefore, direct comparisons between our findings cannot be made.Contrary to findings from a metaanalysis of randomised controlled trials reporting on the beneficial effects of cocoa and dark chocolate consumption on cardiovascular health (including T2D) [10], our results showed no significant associations between dark chocolate intake and risk of T2D, potentially due to the high number of participants with zero intakes (85.6%).
In this study we investigated the potential mechanisms by which flavonoids might impact T2D risk, using mediation analysis.In agreement with previous studies on flavonoid intakes and weight maintenance [31], our analysis suggests that lower T2D risk among those with a flavonoid-rich diet may be largely mediated by a lower BMI.However, in multivariable-adjusted models, associations were only marginally attenuated when BMI and waist circumference were included as potential confounders.In turn, our finding underlines that flavonoid intake may reduce weight gain through numerous mechanisms, including increased satiety [32] and energy metabolism [33].Chronic inflammation has also been shown to play a major role in the development of metabolic dysfunction and particularly hyperglycaemia [34].In line with our previous work where we showed that higher anthocyanin and flavonol intake was associated with anti-inflammatory effects [35], mediation analyses from this study also suggest that basal inflammation plays a considerable role in mediating the protective effects of a flavonoid-rich diet.These results are further supported by mechanistic insights from in vitro and animal studies suggesting higher flavonoid intakes to decrease T2D risk through reduced inflammation and oxidative stress [36,37].In addition to lower inflammation, better liver function may explain lower T2D among persons with higher FDS in our study.Again, these results  are in agreement with previous studies to suggest that habitual flavonoid intake may protect against liver dysfunction [38], while non-alcoholic fatty liver disease is considered an independent diabetes risk factor [39].Similarly, it has been proposed that flavonoids may enhance kidney function [40], and our findings suggest that habitual flavonoid intake may indeed protect against early kidney dysfunction before diabetes is diagnosed.Several other mechanisms may explain the potential beneficial effects of a diet rich in flavonoids on T2D.Antidiabetic effects of several subclasses are suggested to arise through improved insulin signalling and secretion; firstly, through regulation of insulin-responsive glucose transporters e.g., GLUT4, whereby translocation facilitates glucose transportation and uptake in the body, and secondly, through the promotion of pancreatic β-cell proliferation, attenuating oxidative stress and apoptosis [37].The anthocyanins, flavan-3-ols, and flavonols have raised particular interest due to their beneficial antidiabetic effects surrounding improved insulin resistance and glycemia metabolism [41].The strong mediation by body fatness we observed may suggest that flavonoids exert beneficial effects in the early phase of diabetes development before manifesting hyperglycaemia.
In this study, a higher intake of red wine, tea, berries, and apples was associated with lower risks of T2D, even after adjusting for several demographic and lifestyle characteristics as potential confounders.This indicates that the observed beneficial effects on health may not be solely explained through overall improved diet quality, but rather something specifically present in flavonoid-rich plant foods.Due to its significant contribution to total anthocyanin intake (41% in this study), red wine was added as a FDS component.However, it is notable that on exclusion of red wine from the score, inverse associations were only marginally attenuated between a higher FDS and incidence of T2D.
Although the USDA database was used to estimate flavonoid intakes in this UK cohort, upon assessing the agreement between flavonoid intakes estimated from the USDA and Phenol-Explorer databases, all flavonoid subclasses showed excellent agreement (r > 0.79), except for the flavone subclass (Spearman's r = 0.10).This is consistent with a previous study which compared flavonoid intakes derived from USDA and Phenol-explorer databases [42].The association between flavone intake based on Phenol-Explorer data and diabetes risk was not statistically significant, possibly because flavone intake levels according to Phenol-Explorer were mostly driven by orange juice consumption, unlike USDA-derived flavone intakes [43].
The strengths of this study include the prospective design, large sample size, and long follow-up time of 12 years.Upon completion of subgroup analyses, this study showed no heterogeneity across strata of potential confounders.Sensitivity analysis excluding cases that occurred within the first 2 years of follow-up showed almost no change in results, supporting the assumption that associations are not a result of reverse causality.However, there are also some limitations to consider.Firstly, the UK Biobank cohort is made up of British middle-aged adults, >90% of which are of White-European ancestry which limits the generalisability of results to other non-European and ethnic groups.To better represent habitual dietary preferences, we used dietary data from participants who completed at least two 24-h dietary assessments, supported by our finding of a good reproducibility of the FDS over time based on these assessments.Although the Oxford WebQ dietary questionnaire has shown reasonable validity against objective biomarkers [44], we cannot rule out over-or underreporting and recall bias.Further, it can be assumed that in some cases, flavonoid sub-class intakes have been misclassified due to specific food items not being represented in the Oxford WebQ dietary questionnaire e.g., types of berries were not accounted for individually, but were grouped into one overarching category, 'berries'.Additionally, flavonoid intake values may have also been misclassified through being unable to directly match an Oxford WebQ food item with a USDA flavonoid code, potentially attenuating risk estimates.Statistical mediation effects in our study were modest, with 28% of the association between the FDS and T2D attributable to mediators.This modest proportion may be due to potential additional, unmeasured mediators and regression dilution related to single biomarker measurements.It must also be considered that although multivariable-adjusted models accounted for several confounding factors, findings from this study were ascertained through observational data, possibly exposing associations to residual confounding.

CONCLUSION
In summary, in this large prospective cohort of 113,097 middleaged adults, a higher Flavodiet Score, achieved via a diet abundant in specific flavonoid-rich foods was associated with lower risks of incident T2D, irrespective of genetic predisposition and other established risk factors for T2D.Further, the observed inverse associations may be mediated by the beneficial effects of flavonoids on obesity/sugar metabolism, inflammation, and kidney and liver function.This study supports the current advice on increasing fruit consumption to reduce T2D risk, but points to a specific role for berries and apples.Encouraging an achievable increase in habitual intake of specific flavonoid-rich foods and beverages, namely tea, berries, and apples may lower T2D risk.

Table 1 .
Baseline characteristics across quartiles (Q) of the Flavodiet Score in the UK Biobank (n = 113,097).

Table 1 .
continued FDS flavodiet score, BMI body mass index, MET metabolic equivalent task, PRS polygenic risk score, T2D type 2 diabetes mellitus, SD standard deviation.
a Relative frequencies (%) include missing values which may not equate to 100%.b Other includes any race or ethnic group not otherwise specified.c Portion sizes were specified as a "serving" in the Oxford WebQ tool.

Table 2 .
Hazard ratios (95% confidence intervals) of type 2 diabetes across sex-specific quartiles (Q) of the Flavodiet Score (n = 113,097).Model 1 was adjusted for sex and education; stratified by age (5-year categories) and region.Model 2 was adjusted for sex, BMI, waist circumference, ethnicity, physical activity, smoking status, alcohol intake, education, energy intake, polypharmacy index, multimorbidity index, Townsend deprivation index, family history of diabetes, hypercholesterolemia, hypertension, menopausal status, PRS (T2D), number of completed dietary assessments, and intake of wholegrains, red and processed meat, sugar-sweetened beverages, and coffee; stratified by age (5-year categories) and region.Q quartile, FDS flavodiet score, BMI body mass index, PRS polygenic risk score, T2D type 2 diabetes mellitus, HR hazard ratios, CI confidence intervals.
a Reference categories.P-trend is for linear trend.

Table 3 .
continued Model 1 was adjusted for sex and education; stratified by age (5-year categories) and region.Model 2 was adjusted for sex, BMI, waist circumference, ethnicity, physical activity, smoking status, alcohol intake (excluding red wine analysis), education, energy intake, polypharmacy index, multimorbidity index, Townsend deprivation index, family history of diabetes, hypercholesterolemia, hypertension, menopausal status, PRS (T2D), number of completed dietary assessments, and intake of wholegrains, red and processed meat, sugar-sweetened beverages, and coffee; stratified by age (5-year categories) and region.Q quartile, BMI body mass index, T2D type 2 diabetes mellitus, PRS polygenic risk score, HR hazard ratios, CI confidence intervals.Intake values are median (range) and portion sizes were specified as a "serving" in the Oxford WebQ tool. a

Table 4 .
Mediation analysis between Flavodiet Score and type 2 diabetes.BMI (excluding when considered as a potential mediator), waist circumference, ethnicity, physical activity, smoking status, alcohol intake, education, energy intake, polypharmacy index, multimorbidity index, Townsend deprivation index, family history of diabetes, hypercholesterolemia, hypertension, menopausal status, PRS (T2D), number of completed dietary assessments, and intake of wholegrains, red and processed meat, sugar-sweetened beverages, and coffee; stratified by age (5-year categories) and region.
a Hazard Ratios with 95% Confidence Intervals (CI) for Flavodiet Score (1-point increments), adjusted for sex, b Potential mediators are modelled on the continuous scale.