Flavan-3-ols or flavanols, terms used interchangeably, are compounds that belong to a polyphenol subclass called flavonoids, which share a common C6–C3–C6 skeleton. Flavan-3-ols are perhaps the most structurally complex in the flavonoid subclass ranging from simple monomers (such as catechin and its isomer epicatechin) to oligomers (from dimers to decamers), polymers (>10mers) and other derived compounds (e.g. theaflavins and thearubigins). The oligo and polymers of flavan-3-ols are also referred to as condensed tannins or proanthocyanidins (PA), named for their ability to yield anthocyanidins when heated in acidic media(Reference Santos-Buelga and Scalbert1). Enzymatic and non-enzymatic oxidation of (gallo)catechins, reactions characteristic of green tea fermentation, results in flavanol-derived compounds: theaflavins and high-molecular-weight thearubigins(Reference Santos-Buelga and Scalbert1, Reference de la Rosa, Alvarez-Parrilla and González-Aguilar2). Being much larger molecules, PA and flavanol-derived compounds tend to be less bioavailable and have different functional properties; therefore they are often considered as separate groups of flavonoids(Reference Santos-Buelga and Scalbert1, Reference Manach, Williamson and Morand3). Although their average degree of polymerisation can be estimated, the structures of some high-molecular weight polymers of PA and of most of the thearubigins have not been well defined due to inadequate analytical methods(Reference Santos-Buelga and Scalbert1).
Flavan-3-ol monomers are found ubiquitously in plants as secondary metabolites(Reference Aron and Kennedy4). Flavan-3-ol monomers are most abundant in fruits (e.g. berries, apples/pears, stone fruits), barley, cocoa beans, nuts(Reference Hellstrom, Torronen and Mattila5) and their derived products(Reference Auger, Al-Awwadi and Bornet6). Gallocatechins are found, almost exclusively, in green tea infusions(Reference de la Rosa, Alvarez-Parrilla and González-Aguilar2, Reference Williamson and Manach7) while flavanol-derived compounds, theaflavins and thearubigins, are abundant in fermented black and oolong teas(Reference de la Rosa, Alvarez-Parrilla and González-Aguilar2). Common PA-rich foods are cocoa, berries, nuts and some raw beans(Reference Hellstrom, Torronen and Mattila5, Reference Gu, Kelm and Hammerstone8). Transformations and losses of some flavonoid compounds during processing and cooking are common and vary for different subclasses and even for the individual compounds(Reference Crozier, Lean and McDonald9–Reference Beecher11).
Total and individual compounds of flavan-3-ols have been studied extensively in vitro for their antioxidant, anti-inflammatory, immunomodulator and anti-carcinogenic effects(Reference Ramiro-Puig and Castell12–Reference Yang, Wang and Lu14). A plethora of human intervention studies currently available strongly suggests beneficial effects on human health, particularly the effects of flavan-3-ol-rich foods, such as tea, cocoa and chocolate(Reference Williamson and Manach7, Reference Hooper, Kroon and Rimm15, Reference Williamson, Sies and Heber16). Intervention studies involving various PA-rich sources such as red wine, grape, pomegranate, chocolate and cranberry juice showed numerous positive effects on antioxidant, CVD and endothelial maintenance biomarkers(Reference Manach, Williamson and Morand3). Given the limited reported bioavailability of PA, particularly those having a high degree of polymerisation (>3)(Reference Cos, de Bruyne and Hermans17), the observed action of PA-rich foods in the body, with the exception of perhaps the intestinal lumen, may be attributed to flavan-3-ol monomers, which are systematically associated to PA and comprise 5–25 % in these foods, or to other yet unidentified PA(Reference Manach, Williamson and Morand3, Reference Schroeter, Heiss and Balzer18). Furthermore, PA may also exert their action after their degradation by the colonic microbiota and subsequent absorption(Reference Monagas, Urpi-Sarda and Sánchez-Patán19). Indeed, two Italian case–control studies suggested inverse associations between PA, but not flavan-3-ol monomer, intake and gastric and colorectal cancers(Reference Rossi, Negri and Parpinel20, Reference Rossi, Rosato and Bosetti21). Furthermore, the inverse association augmented with increasing degrees of polymerisation of PA in colorectal cancer cases(Reference Rossi, Negri and Parpinel20) despite their observed decrease in absorption. Further research, particularly in prospective studies, on individual flavan-3-ol and PA compounds is clearly needed to clarify and confirm these potential effects.
Total intake of any nutrients is usually related to sex, cultural, lifestyle and socio-economic factors that may affect accessibility to and habitual consumption of health-promoting foods. Studies in European Prospective Investigation into Cancer and Nutrition (EPIC) Spain and the USA found significant differences in total flavonoid intake between sexes and among different age groups, socio-economic levels and ethnic groups(Reference Chun, Chung and Song22, Reference Zamora-Ros, Andres-Lacueva and Lamuela-Raventós23). Therefore, these factors need to be taken into consideration when looking into associations of these compounds and their dietary sources in disease prevention. To our knowledge, there are few data on individual flavan-3-ol intakes in the European population. The present study aims to evaluate total, subclasses and individual dietary intake of flavan-3-ols and their main food sources by EPIC centre and geographical region, while taking into account lifestyle, anthropometric and socio-demographic factors.
Materials and methods
EPIC is an ongoing prospective cohort study designed to investigate the associations between diet, lifestyle and cancer throughout ten Western European countries: Denmark, France, Germany, Greece, Italy, Norway, Spain, Sweden, The Netherlands and UK(Reference Slimani, Kaaks and Ferrari24). The cohort includes approximately 366 000 women and 153 000 men, most of them aged 35–74 years, who were enrolled between 1992 and 2000 by twenty-three centres. Some differences in methods of recruitment exist between centres. Part of the Oxford (UK) cohort was recruited from subjects who consumed a vegetarian-type diet. This was designated a ‘Health-conscious’ group and shall be distinguished from the UK General population cohort which is a combined group of the UK Cambridge and UK Oxford general population. The female part of the cohort in Florence (Italy) and Utrecht (The Netherlands) is composed of women who underwent breast cancer screening. The French cohorts recruited women only, but from the members of the health insurance scheme for the state-school employees. The centres in Italy and Spain recruited mostly blood donors. For the purpose of dietary consumption patterns analysis, the initial twenty-three centres were later redefined by geographical areas into twenty-seven centres(Reference Slimani, Ferrari and Ocke25). The calibration subsample of the EPIC cohort study composed of 36 994 subjects (8 % of the whole EPIC cohort), who were recruited to be a random sample stratified by age, sex and centre, and weighted for expected cancer cases in each stratum of the main EPIC cohort study, was considered herein. After exclusion of 945 subjects under 35 or over 74 years of age because of low participation in these age categories, and sixteen subjects without baseline dietary data, a total of 36 037 subjects were included. Approval for the study was obtained from the ethical review boards of all local recruiting research institutions. All participants provided written informed consent.
Measurements of diet and other lifestyle factors
Dietary intake was measured with a standardised 24 h dietary recall (24-HDR) administered via a computerised interview programme (EPIC-SOFT) developed specifically for the EPIC calibration study(Reference Slimani, Ferrari and Ocke25, Reference Slimani, Deharveng and Charrondiere26). The 24-HDR was administered in a face-to-face interview, except in Norway where it was obtained by telephone(Reference Brustad, Skeie and Braaten27). A detailed description of the rationale and methodology of the 24-HDR calibration study in the EPIC cohort has been described elsewhere(Reference Slimani, Kaaks and Ferrari24, Reference Kaaks, Plummer and Riboli28–Reference Kaaks and Riboli30). Data on socio-demographic and lifestyle factors, including educational level, physical activity and smoking history were collected at baseline through standardised questionnaires and clinical examinations for the calibration sample(Reference Haftenberger, Schuit and Tormo31–Reference Slimani, Fahey and Welch34). Age as well as body weight and height were self-reported by the participants during the 24-HDR interview. The mean time interval between these baseline questionnaire measures and the 24-HDR interview varied by country, from 1 d to 3 years(Reference Slimani, Kaaks and Ferrari24).
Flavonoid Food Composition Database
The US Department of Agriculture (USDA) released a PA database in 2004 and an updated flavonoid database in 2007, with more analytical values for raw, cooked, canned and commercially processed foods(35, 36). In the process of combining the two USDA databases, we observed data duplicity of the monomers. Since flavan-3-ols monomers (USDA database on flavonoids)(35) and PA monomers (USDA database on PA)(36) are the same molecules, the PA monomer data was removed. We expanded these databases with analytical values from the Phenol-Explorer database released in 2009(Reference Neveu, Perez-Jimenez and Vos37). Approximately, 6·5 and 0·6 % of our database came from USDA and Phenol-Explorer, respectively. Thus far, these databases are the most complete and updated databases on flavonoids/polyphenols and they evaluate and compile the most worldwide food composition data published. We further expanded our EPIC-specific food composition database (FCDB) by estimating values for foods not present in either of the two databases, but that had occurred in the 24-HDR. Therefore, for our FCDB, we calculated estimated values (92·9 %) including logical zeros (25·3 %), estimations based on similar food items (22·5 %), application of retention factors (27·7 %) and recipes (17·3 %). When there were no analytical data provided for cooked foods by either USDA or Phenol-Explorer, retention factors were applied. The retention factors reported in various foods were between 42 and 74 % for catechins and 0 and 95 % for tannins(Reference Arts, van de Putte and Hollman38). Therefore, to simplify and homogenise the calculations, we used the same retention factors for all flavonoids, as in our previous studies(Reference Zamora-Ros, Andres-Lacueva and Lamuela-Raventós23, Reference Zamora-Ros, Knaze and Lujan-Barroso39, Reference Zamora-Ros, Knaze and Lujan-Barroso40). They were 70, 35 and 25 % after frying, cooking in a microwave oven and boiling, respectively(Reference Rechner, Wagner and van Buren41). The final FCDB created contained a total of 1877 food items. The unknown composition values, without any analytical or estimated data, were calculated as a zero by default and ranged from 2 % (theaflavin gallates) to 16 % (epicatechin-3-gallates). Finally, the 24-HDR food items were linked with the expanded flavonoid FCDB using an ad hoc SQL (Structured Query Language) application.
General linear modelling was used for the calculation of the adjusted daily mean (least squared) intake and standard error using SPSS (version 17.0.0, SPSS, Inc.) for total flavan-3-ols, their individual compounds and subgroups. The mean intake was adjusted for age, weighted by season and day of 24-HDR and stratified by EPIC centre and age. Flavan-3-ol monomers as aglycones included: catechin, epigallocatechin, epicatechin, epicatechin-3-gallate, epigallocatechin-3-gallate, gallocatechin and catechin-3-gallate. PA were divided into the following subgroups: dimers, trimers, 4–6mers, 7–10mers and >10mers (polymers). Theaflavins included compounds: theaflavin, theaflavin-3,3′-digallate, theaflavin-3′-gallate and theaflavin-3-gallate. Although present in the USDA database, due to the extensive limitations in analytical methods currently employed to identify and quantify thearubigins, we did not include them in our analysis of total flavan-3-ols(Reference Santos-Buelga and Scalbert1, Reference Rechner, Wagner and van Buren41). Epigallocatechin, epicatechin-3-gallate, epigallocatechin-3-gallate, gallocatechin and catechin-3-gallate were later combined into a single group called ‘(epi)gallocatechins’ due to the resemblance among the chemical structures. Flavan-3-ol monomer, PA and theaflavin intakes are calculated as the sum of the individual compounds or subgroups and expressed in mg/100 g of fresh weight. During the analysis of the related factors and of the main food sources, EPIC centres were combined by geographical regions into a Mediterranean (MED) region (Greece, Italy, Spain and South of France) and non-MED (non-MED) region (France other than the South centre, Germany, The Netherlands, Norway, Denmark and Sweden). The UK General population cohort and the Health-conscious cohort presented similar intakes for flavan-3-ols and their food sources but markedly different from all others; therefore, in the socio-demographic analysis they were kept as a separate UK region. The contribution of each food and food group to the total and individual intake of flavan-3-ols was calculated as a percentage. The general linear modelling was also used in the comparison of the mean intakes by socio-demographic, anthropometric and lifestyle factors, adjusting for age, region, energy intake and BMI, and weighted for season and day of 24-HDR. P values < 0·05 indicated significance.
A south-to-north gradient in the daily mean intake of total and monomers of flavan-3-ols and of theaflavins was observed among EPIC centres in both men and women (Table 1). The highest total flavan-3-ol intake was observed in the UK Health-conscious men (453·6 mg/d) and in women of the UK General population cohort (377·6 mg/d). The lowest total intake was observed in Greek men (160·5 mg/d) and women (124·8 mg/d). Flavan-3-ol monomer intake was the highest in the UK General population (213·5 mg/d in men, 178·6 mg/d in women) and the lowest in Greece (26·6 mg/d in men, 20·7 mg/d in women). Theaflavin intake was the highest in the UK General population for both men (29·3 mg/d) and women (25·3 mg/d). Daily theaflavin intake was close to 0 mg in Greece and in Spanish and southern Italian centres (Ragusa, Naples and Florence). In contrast, daily intake of total PA was the highest in Spanish centres (455·2 mg in men from Asturias and 237·9 mg in women from San Sebastian), followed by men in Turin (Italy) and women in Asturias (Spain), respectively. However, PA intake was the lowest in Greece (134·6 mg/d in men and 101·0 mg/d in women). Intake amounts of the individual flavan-3-ols, theaflavins and of PA subgroups are presented in Annexes 1 and 2. PA subclass, particularly the group of polymers (>10mers), was the highest contributor to the total flavan-3-ol intake (Table 2). Flavan-3-ol monomers were the second highest contributors to the total intake, providing contribution of between 18·6 % in the MED region and 44·9 % in the UK. Catechins and epicatechins, equally, were the main single-compound contributors in the MED region, while in the non-MED and UK regions it was the epigallocatechin-3-gallate monomers. Theaflavins were the lowest contributors to the total flavan-3-ol intake. The four theaflavin compounds contributed almost equally to the total theaflavins in all three regions.
* Adjusted for age and weighted by season and day of recall.
† Sum of catechin, epicatechin and epicatechin 3-gallate, epigallocatechin 3-gallate, gallocatechin and catechin-3-gallate.
‡ Sum of theaflavin, theaflavin-3′-gallate, theaflavin-3 gallate and theaflavin-3,3′-digallate.
MED, Mediterranean; Non-MED, non-Mediterranean; PA, proanthocyanidins.
* Values are percentages derived from models adjusted for sex, age and weighted by season and day of recall.
Non-citrus fruit, particularly apples/pears, was the most important food source of total flavan-3-ols in the MED (56·2 %) and non-MED (34·1 %) regions (Table 3). Wine and then tea were the other two major sources of flavan-3-ols in these two regions. On the other hand, tea (51·3 %) was the most prominent source of total flavan-3-ols in the UK, followed by non-citrus fruit (19·9 %) and wine (6·1 %). The major food sources of catechins and epicatechins in all three regions were tea, non-citrus fruits and wine; however, chocolate candy/bars were also noteworthy dietary contributors. Tea was the lone source of theaflavins and a major source of (epi)gallocatechins in all three regions (77·7 % in the MED, 90·5 % in the non-MED and 95·1 % in the UK region). The principal dietary source of total PA in the MED region was non-citrus fruit (62·3 %) followed by wine (17·3 %) and chocolate candy/bars (4·6 %). Non-citrus fruits were also the main source of PA in the non-MED and UK regions, but their contributions were smaller (48·0 and 37·2 %, respectively). In the non-MED region, wine was the second most important source (12·6 %), followed by chocolate/candy (6·6 %) and tea (5·0 %). Whereas in the UK region, the secondary sources of PA were tea (15·0 %), wine (10·0 %), cakes/pies/pastries/puddings (7·6 %) and pulses (7·1 %).
* Values are percentages derived from models adjusted for centre, age and sex, and weighted by season and day of recall.
† Sum of epigallocatechin, epicatechin 3-gallate, epigallocatechin 3-gallate, gallocatechin and catechin-3-gallate.
‡ Sum of theaflavin, theaflavin-3′-gallate, theaflavin-3 gallate and theaflavin-3,3′-digallate.
Sex-stratified analysis of the related factors showed similar results; therefore the data are presented for men and women combined (Table 4). Total flavan-3-ol intake and also the intake stratified by monomers, PA and theaflavins were shown to significantly vary between the geographical regions. The intake of flavan-3-ol monomers and theaflavins in the UK region was almost 4-fold and over 16-fold that of the MED region, respectively. Conversely, PA intake was significantly higher in the MED region (217·2 mg/d) compared to the non-MED (177·9 mg/d) and the UK (198·4 mg/d) regions. After adjusting for BMI and energy, women had significantly higher intakes of total flavan-3-ols and their subclasses. The intake of total flavan-3-ols and their subclasses was significantly different between the age groups, being the highest in the 55- to 64-year-olds. It also increased with the level of education completed and the level of physical activity. On the other hand, current smokers and obese participants (BMI ≥ 30 kg/m2) had the lowest intakes of total flavan-3-ols and their subclasses.
* General linear model adjusted for age, sex, region, BMI (whenever not stratified for the respective variables) and energy intake, and weighted by season and day of recall.
† Sum of catechins, epicatechins and (epi)gallocatechins.
‡ Sum of theaflavin, theaflavin-3′-gallate, theaflavin-3 gallate and theaflavin-3,3′-digallate.
§ P values are for overall differences in mean consumption by general linear model among the socio-demographic, lifestyle and anthropometric subgroups.
To our knowledge, this is the only study thus far assessing the intake of total and flavan-3-ol monomers, PA and flavan-3-ol-derived compounds as well as their food sources and associated factors in all twenty-seven EPIC centres of ten European countries using a common expanded flavonoid FCDB and dietary assessment method (24-HDR). Our results show a wide range of total flavan-3-ol intakes following a south-to-north geographical gradient. When stratified by regions, total flavan-3-ol intake in the UK was about 2-fold that of the MED region. This relatively steep gradient in flavan-3-ol intake was mainly due to higher intakes of theaflavins and epigallocatechins in northern EPIC cohorts; indeed the main source of these subclasses of flavan-3-ols was found to be tea. On the other hand, PA intake was found to be statistically higher in the MED region, although large differences were also noted among centres within the same region. The main source of PA in the MED region was non-citrus fruit, chiefly apples and pears, followed by wine, similar to what was previously reported for the EPIC Spanish cohort(Reference Zamora-Ros, Andres-Lacueva and Lamuela-Raventós23). Furthermore, the almost-nil intake of theaflavins in Greece, Spain and southern Italy indicates minimal consumption of tea in these countries. Even so, the major sources identified for the total and individual flavanols, PA and theaflavins were quite similar except in the UK where pulses also formed a considerable food source of PA.
A well-established inverse geographical gradient of CVD mortality exists(Reference Levi, Chatenoud and Bertuccio42), which may seem paradoxical with the north-to-south gradient for flavan-3-ol intake and the observed beneficial effects of these compounds and flavonoid-rich foods against CVD(Reference Suzuki, Isobe and Morishita13, Reference Hooper, Kroon and Rimm15, Reference Williamson, Sies and Heber16). Though far-fetched at this point to imply that flavan-3-ols have a significant role in CVD, a few factors could be considered to help elucidate this. Despite their higher observed antioxidant activity in vitro (Reference Henning, Niu and Lee43, Reference Auger, Mullen and Hara44), galloylated flavan-3-ol monomers (mainly found in fermented/black teas) have lower bioavailability than non-galloylated monomers(Reference Manach, Williamson and Morand3, Reference Auger, Mullen and Hara44) (found more commonly in non-citrus fruit, green tea and cocoa). However, it is more likely that other risk factors of CVD may be more prevalent in the northern countries, such as high intake of SFA(Reference Linseisen, Welch and Ocke45), low intake of MUFA(Reference Lopez-Miranda, Perez-Jimenez and Ros46), low intake of fruits and vegetables(Reference Agudo, Slimani and Ocke47), low wine consumption(Reference Vidavalur, Otani and Singal48), sedentary lifestyle(Reference Thorp, Healy and Owen49, Reference Jakes, Day and Khaw50) and social class influences(Reference McFadden, Luben and Wareham51).
The present study also demonstrated that statistical differences exist in flavan-3-ol intakes among groups with different socio-demographic, anthropometric and lifestyle characteristics. Consumption of total flavan-3-ols, monomers and PA increased with age up to about 64 years of age and then it fell slightly. Similar results were seen in Spanish-EPIC(Reference Zamora-Ros, Andres-Lacueva and Lamuela-Raventós23), US(Reference Gu, Kelm and Hammerstone8, Reference Chun, Chung and Song22) and Australian(Reference Johannot and Somerset52) studies in adults. The intakes were significantly higher in former and never smokers. Since the major sources were tea and fruits, respectively, this suggests possible interaction between the consumption of these food sources and smoking habits(Reference Whichelow, Erzinclioglu and Cox53, Reference Dauchet, Montaye and Ruidavets54). Additionally, two case–control studies suggested that a flavonoid-rich diet may protect against pancreatic and lung cancer in smokers only(Reference Cui, Morgenstern and Greenland55, Reference Bobe, Weinstein and Albanes56). Total flavan-3-ol intakes have been shown to be significantly associated with a slower increase of BMI in women in The Netherlands Cohort Study after adjusting for confounders including dieting, and some healthy habits such as fruit and vegetable intake(Reference Hughes, Arts and Ambergen57). Our study showed that the intakes of total and subclasses of flavan-3-ols were lower in obese subjects and, in addition, higher in physically active subjects. Further investigation explaining the association between all these factors and their role in obesity is needed. In our study, intakes of PA increased with higher level of education, and were found to be considerably low in the group without formal schooling. In line with our findings, consumption of fruits, fruit juice, wine and tea has been previously associated with higher socio-economic status(Reference Hulshof, Brussaard and Kruizinga58). These factors should be taken into consideration when looking at disease prevention, planning of healthy diets within a specific population and may also be instrumental when establishing ‘lifespan essential’ dietary reference intakes for flavan-3-ols, PA and theaflavins in different populations, and in subgroups with unfavourable lifestyle factors(Reference Williamson and Holst59).
Major differences in total flavan-3-ol intake can be observed between our study and other descriptive studies (Table 5). Our estimates of total flavan-3-ol intake were much higher than those reported in other countries and the main reason being that some individual compounds or subclasses of flavan-3-ols were not included in the total estimation of flavan-3-ols in those studies. Comparison of the subclasses (monomers, PA and theaflavins) may therefore be preferable. Even then, the intakes may vary because of other important factors involved in the estimation, such as different methodologies used to assess dietary intake of cohorts (FFQ, 24-HDR, diet history, etc); compositional data of the compounds may come from different sources or from older versions of the USDA database(Reference Zamora-Ros, Andres-Lacueva and Lamuela-Raventós23, Reference Otaki, Kimira and Katsumata60), or the median intake reported rather than the mean. The latter is particularly the case in the Greek EPIC study(Reference Dilis and Trichopoulou61). Finally, cohort characteristics, especially those that are associated with flavan-3-ol consumption (e.g. age and sex) may vary considerably. All these are to be kept in mind when making comparisons. Flavan-3-ol monomer intake reported recently for the UK and Ireland by Beking & Vieira(Reference Beking and Vieira62) and based on food balance sheets was about one-third that of our reported intake for the UK region. The monomer intake reported in an elderly Dutch cohort(Reference Arts, Hollman and Feskens63) as closer but still slightly lower than our estimates for the Dutch cohorts of Bilthoven and Utrecht even though, in our study, intakes were found to be higher in older age groups. Monomer intakes reported recently in Italy(Reference Rossi, Rosato and Bosetti21) and for Spanish(Reference Zamora-Ros, Andres-Lacueva and Lamuela-Raventós23) and Greek(Reference Dilis and Trichopoulou61) EPIC cohorts were within our ranges for those countries. As for other countries known to have a tea culture, Otaki et al. (Reference Otaki, Kimira and Katsumata60) have estimated monomer intake in Japanese women to be around 380 mg/d, more than double the value we reported for the UK region. This is probably because of higher consumption of non-fermented tea, such as green tea, in Japan. Green tea is a rich source of flavan-3-ol monomers but not a source of theaflavins, which was found exclusively in black tea, the tea more commonly consumed in the UK. In contrast, monomer intake (188 mg/d) reported in Australia was comparable to our value in the UK region. PA intake for Spain in our study is slightly higher than previously reported for the EPIC Spain cohort; however, the previous study used dietary history questionnaires and only the USDA food composition values in their estimation of flavan-3-ols(Reference Zamora-Ros, Andres-Lacueva and Lamuela-Raventós23). A recent Italian case–control study assessed the mean PA intake to be around 290 mg/d which is within the range of the Italian values in our study(Reference Rossi, Rosato and Bosetti21). However, most PA estimations have been done in case–control studies, which assess small groups of controls and not always all the subgroups of PA were included(Reference Rossi, Negri and Parpinel20, Reference Rossi, Rosato and Bosetti21, Reference Kyle, Sharp and Little64). Surprisingly, Greece being a MED country had the lowest intake of PA of all EPIC centres. Among the other previously mentioned factors for this difference, this finding is also supported by the lower consumption of fruit in Greece compared to Italy and Spain reported previously in EPIC studies(Reference Agudo, Slimani and Ocke47). Perez-Jimenez et al. (Reference Perez-Jimenez, Fezeu and Touvier65) reported intakes of flavan-3-ol monomers (114 mg/d), PA (191 mg/d) and theaflavins (16 mg/d) in French women that are within the range of our values for the French EPIC centres. A limited number of descriptive studies is available on PA to facilitate a comparison with non-EPIC countries. Using their own composition database, Ovaskainen et al. (Reference Ovaskainen, Torronen and Koponen66) reported lower PA intakes for a Finnish population compared to the northern EPIC countries such as Sweden and Norway. The sources of PA in northern EPIC countries were found to be similar to those in Finland, with the exception of berries, which were not singled out in our study but were an important source of PA in Finland. These differences in intake and food sources compared to our study are most probably due to the varying study variables already exposed earlier. Finally, Wang et al. (Reference Wang, Chung and Song67) recently estimated PA intake for the US population to be about 95 mg/d. This is still slightly lower than the lowest intakes found in our study (Greek cohorts). Clearly, more consistent methods of intake estimation between and within countries are needed. Parallel to that, improved methods for identification and quantification of some flavan-3-ol compounds, such as thearubigins, are needed to allow for more exhaustive flavonoid composition databases.
C, catechins; EC, epicatechins; EGC, (epi)gallocatechins; PA, proanthocyanidins; ns, not specified; –, value not provided by the original study; DR, dietary recall; USDA, US Department of Agriculture.
* Where applicable and when not provided by the study, total flavan-3-ols were calculated as the sum of the subgroups.
† Median values given instead of the mean.
‡ Total flavan-3-ols were calculated as the sum of the values in Rossi et al. (Reference Rossi, Rosato and Bosetti21) for C, EC, theaflavins and thearubigins combined and in Wang et al. (67) for PA.
The use of a common expanded flavonoid database provided us with greater coverage of foods representative of the EPIC countries while allowing for comparisons of results across the countries. Despite the fact that we applied retention factors to foods prepared by cooking, we estimated higher intakes than in the previous studies. Moreover, our values are likely to be underreported due to spices and herbs often not accounted for during diet assessment and because a small proportion (2–16 %) of flavan-3-ol analytical values in our study was still missing. The underestimation of intakes is also probably due to the omission of dietetic supplements in this analysis. However, few consumers of herb/plant supplements participated in this study (the highest was 5 % reported in Denmark)(Reference Skeie, Braaten and Hjartaker68).
To our knowledge, this is the largest study to date describing flavan-3-ol and PA intake across several European countries. Since not all the EPIC cohorts are representative of the population, the observed level of intake cannot be extrapolated to the general population of each region.
In summary, this study provides total and individual flavan-3-ol, PA and theaflavin intakes for ten EPIC countries by sex and EPIC centre. The major dietary contributors of these flavonoid subclasses are described by the MED, non-MED and UK regions. In addition, we show that socio-demographic, anthropometric and lifestyle factors associated with differential consumption of flavan-3-ols, PA and theaflavins exist. Combined with more elucidated information on the bioavailability of these compounds, these descriptive data will be valuable in future evaluations of total and individual flavan-3-ols and their role in health and disease in the European population.
The present work was carried out with the financial support of the European Commission: Public Health and Consumer Protection Directorate 1993–2004; Research Directorate-General 2005; Ligue contre le Cancer, Institut Gustave Roussy, Mutuelle Générale de l'Education Nationale, Institut National de la Santé et de la Recherche Médicale (INSERM, France); German Cancer Aid; German Cancer Research Centre; German Federal Ministry of Education and Research; Danish Cancer Society: Health Research Fund (FIS) of the Spanish Ministry of Health (RTICC (DR06/0020); the participating regional governments and institutions of Spain; Cancer Research UK; Medical Research Council, UK; the Stroke Association, UK; British Heart Foundation; Department of Health, UK; Food Standards Agency, UK; the Wellcome Trust, UK; Hellenic Ministry of Health; the Stavros Niarchos Foundation and the Hellenic Health Foundation; Italian Association for Research on Cancer; Compagnia San Paolo, Italy; Dutch Ministry of Public Health, Welfare and Sports; Dutch Ministry of Health; Dutch Prevention Funds; LK Research Funds; Dutch ZON (Zorg Onderzoek Nederland); World Cancer Research Fund (WCRF); Swedish Cancer Society; Swedish Scientific Council; Regional Government of Skane, Sweden; Nordforsk – Centre of Excellence programme. Some authors are partners of ECNIS, a network of excellence of the 6FP of the EC. R Z.-R. is thankful for a postdoctoral programme Fondo de Investigación Sanitaria (FIS; no. CD09/00133) from the Spanish Ministry of Science and Innovation. The authors thank Raul M. García for developing an application to link the FCDB and the 24-HDR. The authors declare that there are no conflicts of interest. The authors' contributions are as follows: R. Z.-R. and C. A. G. designed the research; V. K. and R. Z.-R. conducted the research; V. K. and L. L.-B. performed the statistical analysis; V. K. and R. Z.-R. wrote the manuscript. V. K., R. Z.-R., L. L.-B., I. R., A. S., N. S., E. R., C. T. M. v. R., H. B. B.-d. M., A. T., V. D., K. T., G. S., D. E., J. R. Q., E. M., J. M. H., F. C., E. W., U. E., P. H. M. P, R. K., B. T., G. J., I. J., R. T., H. B., D. D., P. A., A. M., K.-T. K., R. L., V. Kr., E. A., C. S., S. S., K. O., A. T., A. O., M.-C. B.-R., G. F., F. P. and C. A. G. read, critically reviewed and approved the final manuscript.
* Adjusted for sex and age, and weighted by season and day of recall.
* Adjusted for sex and age, and weighted by season and day of recall.