Introduction

Plants produce a great variety of secondary metabolites. Phenolic compounds, usually called (poly)phenols (PPs), are the most widely distributed in dietary sources. Although they are not essential for humans, PPs may have a potential positive impact on human health. Due to their abundance in certain foods, they are important candidates responsible for the beneficial effects, associated with vegetable and fruit-rich diets. Consumption of PPs in high amounts may present some adverse effects, mainly pharmacokinetic interactions and inhibition of non-heme iron absorption. However, this could be more relevant when PPs are consumed as supplements [1]. Moreover, there is growing evidence that a continuous and prolonged intake of PPs may be associated with a lower risk of several chronic non-communicable diseases, such as diabetes, cardiovascular diseases, neurodegenerative diseases and cancer, as well as low-grade inflammation and symptoms associated with ageing and menopause [2–6]. However, several limitations exist in epidemiological studies, and in most of the clinical trials, and no conclusive answers linking dietary PP exposure, PP metabolites and disease biomarkers have yet been reached [7–9].

PPs are characterized by the presence of one or more hydroxyl groups attached to an aromatic ring [10]. They can be classified into different classes according to the number of phenol rings that they contain and on the basis of structural elements that bind these rings to one another [11, 12] (Fig. 1). PPs may occur in a non-conjugated form (aglycone) or conjugated with sugars, carboxylic and organic acids, amines, lipids and other phenols. In addition, some PPs may also be linked to plant cell wall macromolecules. Larger PPs include hydrolysable phenols such as ellagitannins (ET) and condensed tannins such as proanthocyanidins (PA). Ellagitannins are oligomers of ellagic and gallic acid, which are only hydrolysed by various treatments such as hot water, acids and bases or enzymatically by tannases [13]. PA are dimers, oligomers and polymers of flavan-3-ols, mostly the aglycones catechin and epicatechin [14]. Main classes, subclasses and some common PPs are shown in Fig. 1. PPs that are not so common and are not represented in Fig. 1 include other subclasses of flavonoids such as chalcones, hydroxychalcones and coumarins, and other phenols such as alkylphenols, cucurminoids, hydroxybenzaldehydes, hydrocinnamaldehydes, hydroxyphenylpropenes, methoxyphenols, naphtoquinones, phenolic terpenes and tyrosols [15].

Fig. 1
figure 1

Main (poly)phenol classes according with Phenol-Explorer database (http://phenol-explorer.eu/, accessed on June 2015). Classes are in bold, subclasses are in italic bold and families within subclasses are in italic. A common aglycone is represented with the respective name in brackets (structures were drawn with ChemBioDraw Ultra 14.0)

Full size image

Epidemiological studies can be very helpful in studying the association of PP intake with real disease endpoints. However, prediction of health effects of PPs in these studies is dependent on calculating the amount of ingested PPs by the studied population (PP exposure) and the bioavailable fraction of PPs. Exposure, on the one hand, relies on the correct estimation of food intake and, on another hand, on the correct estimation of PP content of foods. Moreover, estimation of amounts of circulating PP and/or its metabolites depend on reliable biomarkers. These should be compounds measured in a biological sample that reflect habitual intake of some PP rich food, or the individual PP, and should be subjected to little intraindividual variability [16]. This review aims to gather information from several studies focused on PP intake in different countries, compare methods used for both assessment of food intake and PP quantification, discuss its nutritional relevance and point existing gaps and future directions.

Overview of (poly)phenol content in foods and its nutritional relevance

There are two main approaches to determine PP content in foods: (1) Folin–Ciocalteu colorimetric assay, based on the reducing properties of the phenolic groups, used to estimate total amount of PPs [17] and (2) chromatographic techniques, most commonly, HPLC linked with several possible detection systems, used to quantify individual compounds and also total PP content [18]. Two major public databases gather the (poly)phenol content of foods and beverages: the United States Department of Agriculture (USDA) databases for flavonoids [19], proanthocyanidins [14] and isoflavones [20], and the Phenol-Explorer database [15]. USDA databases only present values for PPs of the flavonoid class, expressed as aglycones, leaving out other classes that are abundant in many foods. On the other hand, Phenol-Explorer gathers information on all classes and types of PPs, including aglycones, conjugates and esters, being therefore more comprehensive. This database provides content values for individual PPs in three categories:

  1. 1.

    Chromatography, which presents data for extractable PPs (EPP), PPs extractable with aqueous/organic solvents (methanol/ethanol/acetone and water), determined normally by reverse-phase high-performance liquid chromatography (HPLC) and gas chromatography. EPPs are low molecular weight PPs from several classes and subclasses, including PA (dimers and trimers) and hydrolysable tannins of low molecular weight [21].

  2. 2.

    Normal-phase HPLC for PA from dimers to polymers.

  3. 3.

    Chromatography after hydrolysis, which contains data for those PPs that are attached to the food matrix and can only be solubilized and quantified after basic or acid hydrolysis, mostly lignans in many foods and phenolic acids in cereals [22]. PPs under this category are part of what is defined by Saura-Calixto as non-extractable PPs (NEPP), the residue of aqueous/organic extracts. As originally defined, NEPP includes high molecular weight proanthocyanidins and hydrolysable PPs (HPP) which are PPs linked to cell wall macromolecules or trapped within the food matrix and can only be released by hydrolysis treatments [21].

To highlight the nutritional relevance of PP in the human diet, we looked for the fifty richest foods in PPs and the distribution of the different subclasses of PPs among the selected foods (Tables 1, 2). Phenol-Explorer database was used to estimate the total PP content per common serving sizes for each food. The 50 richest foods were selected and ranked within each food group according to total PP content per portion normally consumed, to get a clear picture of the nutritional relevance of the PP intake per food. Some spices and herbs, particularly cloves, are foods with the highest contents in PPs per 100 g of fresh weight; however, serving sizes are very small, and thus, the actual PP intake is lower or equivalent to other foods less rich in PPs. Considering serving sizes, fruits are clearly the richest in PPs among food groups, accounting for 44% of the fifty selected foods. Within the selected fruits, 50% are berries, emphasizing the importance of these fruits to PP intake. Particularly, high values are observed for black chokeberry and elderberry, followed by blackcurrant and blueberry. Although apples and oranges do not have such a high PP content as berries, they are very important contributors to total PP intake, having a huge dietary relevance due to its high consumptions among the population [23, 24]. According to FAOSTAT food balance data of 2011, oranges account for 13.33 kg of the food supply per capita, followed by banana with 12.05 kg and apple 9.4 kg [25]. Despite being one of the most consumed fruits, banana is not included in the 50 richest foods in PPs presented in Table 1, because in Phenol-Explorer, its content in PPs is very low (5.4 mg/100 g fw of flavanols, including PAs up to 6 mers, and 1 mg/100 g fw of hydroxybenzoic acids). However, when considering NEPP, bananas scored values as high as 980 mg/100 g fw for PA [26]. Thus, this is a good example of a food that is highly consumed by the population and is an important contributor to overall PP intake, but may be underestimated due to incomplete information of the composition database on NEPP. Underestimation in total PP intake may therefore also happen for other foods with nutritional significance. Furthermore, PP contents may vary depending on species, seasonal variation, maturity, and various agricultural, food-processing and storage practices [27, 28].

Table 1 Fifty most rich foods in PPs, ranked according to content per portion in the different food groups
Full size table

Table 2 ascertains the relevance of the different classes and subclasses of PPs for the selected foods, according to the data extracted from Phenol-Explorer. Considering this classification, there is a clear differential distribution among foods, with some subclasses being generally more abundant than others. For example, within flavonoids, flavanones, flavones and isoflavones are the less abundant subclasses in the 50 richest foods. This, however, may not reflect their dietary relevance. They may be highly consumed, such as flavanones, which are present in oranges, one of the most consumed fruits. Other less common phenols are included in the general class “other” (Table 2). Some spices, for example cloves, stand out with enormous amounts of hydroxyphenylpropenes, and even though they are used in low quantities, they still contribute to a high phenol intake when consumed. In cereals, most phenols are alkylphenols, which occur mainly in whole grain products and are therefore lost during refining. Tyrosols are present in high amounts exclusively in olives, and stilbenes are found in low amounts in wine and lingonberry. Lignans, a class of NEPP, are the main PP found in fruits such as pear and apricot, in vegetables like broccoli and cauliflower and in sesame seeds. Other fruits and vegetables such as peach, orange, nectarine and carrot also include considerable percentages of lignans.

Common subclasses of flavonoids in the selected foods, include: anthocyanins, mostly present in red and black fruits, as well as in red wine; flavanols, abundant in fruits, tea and some seeds, particularly cocoa products; proanthocyanidins, which constitute more than 90% of the flavanols in apples, berries and seeds; flavonols, mostly abundant in vegetables and some herbs. Phenolic acids are also very abundant PPs, being present in all food groups. Foods with high percentages of hydroxycinnamic acids include coffee, the main contributor for the intake of this subclass of phenolic acids, some vegetables, particularly, potato and some fruits such as loquat and pear and rye bread. Hydroxybenzoic acids, on the other hand, are mainly present in chestnut, red raspberry and cloudberry.

Overview of (Poly)phenol intake in the world

PP intake was assessed in many epidemiological studies and clinical trials studying the effect of PP on human health. However, in many of these studies, the study population was small, and the studies were difficult to compare due to different methods of assessment of food intake and calculation of PP intake. In this review, we gathered information on PP intake among different countries all over the world, to compare methodologies and values of PP intake. A systematic search was performed (Supplemental Fig. 1), and the following criteria were used to select studies for analysis:

  1. 1.

    Assessment of intake of all plant based food groups contributing to PP intake: cereals, vegetables, legumes, fruits, wine, cocoa products, tea and coffee;

  2. 2.

    Large study population (n ≥ 1000);

  3. 3.

    Both sexes included in the study;

  4. 4.

    Total PP content or total content of major classes (flavonoids or phenolic acids); studies assessing only a few subclasses were not considered;

  5. 5.

    In epidemiological studies assessing the association between PP intake and a disease, only healthy controls were considered.

Based on these criteria, 49 records were eligible for full text analysis, and 24 were selected for qualitative analysis (flow chart of study selection and search terms are presented on Supplemental Fig. 1). The selected studies (Table 3) were performed in thirteen different countries, from Europe (Spain [24, 29, 30], Finland [31], Poland [32, 33], France [34], Italy [35, 36], UK and Ireland [37]), Australia [38], USA [39–44], Brazil [45, 46], China [47], Korea [48] and Fiji [49]. One study was a large European cohort [50] involving 10 countries from Europe (Denmark, France, Germany, Greece, Italy, Norway, Spain, Sweden, Netherlands and UK). Another study also analysed the polyphenol intake in Europe [51], based on a European food consumption database that included 14 countries from Europe (Belgium, Czech Republic, Denmark, Finland, France, Germany, Hungary, Ireland, Italy, Latvia, Spain, Sweden and UK). Comparative analysis of the selected studies highlights some important differences on the methods used for assessing dietary intake and conversion to PP intake.

Table 2 Percentage of the different classes and subclasses of (poly)phenols in the 50 most rich foods in PP
Full size table

Assessment of food intake

Selection of the dietary data collection method determines the accuracy of the result in terms of absolute estimates. The main methods of individual diet assessment were twenty-four hour diet recalls (24hR, 11 studies) and semiquantitative food frequency questionnaires (FFQ, 8 studies). Other methods used to assess individual consumption were: a forty-eight hour recall (48hR, 1 study), a Diet History Questionnaire (DHQ, 1 study) and a National Survey Questionnaire (NSQ, 1 study) that assessed the foods present at home. One study relied on values from Food Balance Sheets (FBS) to estimate the national food consumption (Table 3). FBS give information on a country’s food supply during a reference period. Although they show the amount of foods potentially available for consumption by the population, they may not cover all the foods and do not represent the actual consumption, providing only a gross estimate [52]. As such, they should not be considered an adequate instrument to base assessment of PP intake.

The food frequency questionnaire is normally used to obtain information on how often a food is consumed. Quantitative information is obtained if the questionnaire includes questions on portion sizes, but requires validation before use [53]. This is the most used method to collect long-term dietary intake data from large numbers of individuals, since it can be self-administered and is cost-effective. However, it is based on an individual’s perceptions of usual intake over a large period, normally the previous year, and although shorter periods may be selected (such as previous month), it may lack detail about the consumed foods and is prone to large random errors [54]. Furthermore, the type of information provided to participants on portion sizes may also influence individual perception, changing the report of consumed foods. Portion sizes are normally reported as natural units, household measures or grams and may be accompanied by photographs of different serving portions. In fact, portion sizes may be considered country-specific, and thus, questionnaires need to be adapted to culture habits, including the foods consumed in the country and the respective portion sizes. From the eight studies that used FFQ as the diet assessment method, seven were country-specific [24, 33, 35, 40, 41, 44, 47] and five were clearly validated (Table 3).

The 24-hour diet recall provides detailed information about all foods and beverages consumed over the previous day, recorded by a trained interviewer, face-to-face, by telephone or computer assisted. Although more accurate than the FFQ method, data from one single 24hR are not representative of habitual intake, and information from multiple recalls repeated in non-consecutive occasions should be collected [54]. Only two of the analysed studies provided multiple 24hR in non-consecutive days [34, 45]. This is particularly important when estimating PP intake since some of the foods providing PPs are consumed seasonally.

The DHQ is carried out in three steps to collect detailed information on both frequency and quantity of consumed foods over a determined period [55]. It is the most time consuming and expensive method, not being feasible for large populations.

Most of the analysed studies were based on diet assessment data from large epidemiological studies [24, 30, 33–36, 40, 41, 44, 47, 50] or national dietary surveys [29, 31, 32, 38, 39, 42, 43, 45, 46, 48, 49]. However, dietary data from many of these studies report to consumption 10 years ago or more [30, 34, 36, 38, 41–43, 49, 50]. Dietary habits of the population may have changed meanwhile, and these data may thus not represent actual food intake.

Since all methods have some limitations, the use of a combination of different types of dietary assessment methods has been suggested to improve accuracy and facilitate interpretation of the dietary data. However, for estimation of PP intake most, studies rely only on just one method.

Calculation of (poly)phenol intake

Converting food consumption data to PP intake requires a food composition database with specific characteristics. A realistic calculation of PP intake depends on a complete and accurate database, with all possible foods contributing to PP intake, as well as accurate values for all existing classes and subclasses of PPs. This goal could be hampered not only by assessment of food intake, but also by an incorrect estimation of PP content in foods. The most representative databases of PP content in foods used worldwide were already discussed in the previous section. Although Phenol-Explorer is more inclusive, it is not complete and any missing value for a food consumed by the population limits the estimation. Another neglected aspect is values for foods eaten after cooking. In particular, PP bioaccessibility and bioavailability are seriously affected by food processing, failing to predict raw-to-cook differences in PP yield as well as PP losses [56].

Six of the selected studies relied merely on USDA database, only estimating values for flavonoid intake (Table 3) [30, 37–39, 42, 47]. Furthermore, three of the studies did not include proanthocyanidins [37–39], thus not having a realistic estimation of flavonoid intake. In fact, as shown in Table 3, within flavonoids, the main subclasses contributing to PP intake were flavanols, mainly proanthocyanidins. A study in the USA that estimated PA intake has reported that PA oligomers and polymers may be responsible for half or more of flavanol intake [57]. Some common foods, such as fruits like strawberry and apple, beans and cocoa products, are rich in flavanols (Table 2), with low contents of flavanol monomers and high contents of PA (90% or more of flavanols being dimers, oligomers and polymers [15]). Therefore, in diets rich in these types of foods, PA may well be the main contributors to flavonoid intake, as well as to total PP intake (Table 3).

Table 3 Polyphenol intake in several European and non-European countries: comparison of methodologies for diet assessment and PP quantification
Full size table

Whenever other classes of PPs, in addition to flavonoids, are to be considered, the Phenol-Explorer database should be used or food should be analysed in the laboratory. Nine of the twenty-four analysed studies presented EPP intake; six of them used the Phenol-Explorer database and included values for both flavonoids and phenolic acids [24, 33, 34, 45, 46, 50]; and three presented only total content in EPP, quantified by the Folin–Ciocalteu method [29, 32, 49]. Only two studies have determined total PP intake, including EPP and NEPP—Spain [29] and Finland [31]. EPP values for Europe [50, 51], Spain [24], France [34] and Poland [33] were not very different, ranging from 820 to 1741 mg of PP/day. Brazil shows lower values (377 and 460 mg of PP/day), probably due to the low consumption of fruits and vegetables by the population [45, 46]. The lowest values were reported for Fiji Islands (267 mg/day); as refereed by the authors, this was due to a dietary shift away from fruits and vegetables, as well as limitations in reporting techniques [49]. The studies that consider EPP+NEPP present values ranging from 2591 to 3016 mg/day [29] and 863 mg/day [31]. The much higher values found in one of the studies from Spain may reflect an overestimation of food intake by the assessment method (questionnaire assessing the quantity of foods present at homes or purchased by hotels and institutions, instead of the actual consumed foods), and also the utilization of a less accurate method such as Folin–Ciocalteu for the quantification of total PP [29]. On the other hand, the process followed for extraction of NEPP includes both enzymatic and chemical hydrolysis of the residue after aqueous-organic extraction of EPP, allowing a higher release of PPs from the food matrix. Furthermore, the contribution of NEPP to total PP content in food may be between 60 and 90% in some fruits and nuts [26], emphasizing the importance of including NEPP analysis to estimate total PP intake. In the study from Finland, a more realistic information on food assessment was performed with two 24hR in consecutive days, specific methods were used to extract and quantify different PPs and total PP was calculated as the sum of individual PPs. However, enzymatic hydrolysis was not performed, which may explain a lower content of PP in foods.

Another important issue that should be addressed is the energy adjustment of intake values. PP intake may be associated with total energy intake, both by amount of consumed food and by type of consumed food, as shown in the cohort studies of Poland [33] and Italy [35]. Adjustment of nutrient and PP intake for total energy intake is particularly important in epidemiologic studies, because diseases may be influenced by caloric intake, and thus, the effects of the studied PPs may be distorted or confounded by energy intake [58, 59]. Although in some studies energy adjustment does not seem to have an influence on PP intake [24, 60], in others, values may be about 50% lower after energy adjustment [33]. Most of the recent studies shown in Table 3 already present energy adjusted PP intake [33, 42–44, 47, 50].

When considering the estimation by PP classes, only seven of the analysed studies estimated the intake of two main PP classes: flavonoids and phenolic acids [24, 31, 33, 34, 45, 46, 50]. High variations were observed in flavonoid intake (including PA), the lowest was 55 mg/day, in Brazil, and the highest was 898 mg/day, in Poland (Table 3). Flavonoid values for Spain, France and Europe were similar (443, 506 and 519 or 428 mg/day, respectively, Table 3). Regarding phenolic acids intake in Europe, Spain showed the lowest intake (304 mg/day), Finland, France and Europe had similar values (641, 639 and 609 mg/day, respectively) and Poland had the highest values (800 mg/day) (Table 3). The high heterogeneity clearly reflects different diet patterns that ultimately are the reflection of cultural differences associated with traditional food production and availability. Diets richer in coffee will tend to have higher phenolic acid intake, mainly hydroxycinnamic acids (Tables 2 and 3). Coffee consumption may contribute to as high as 52% of total EPP intake [50]. On the other hand, diets richer in fruits and tea will tend to have higher flavonoid intake, with tea contributing to intake of flavanol monomers and fruits to intake of PA (Tables 2 and 3). If NEPP are considered, cereals become important contributors to total PP intake due to its high content in hydrolysable PP, which comprise hydrolysable tannins, phenolic acids and hydroxycinnamic acids that are released from the food matrix by strong acidic hydrolysis [29]. Fruits and legumes also raise its importance as contributors due to its content in condensed tannins [29].

Biomarkers for (poly)phenol intake

Until now, most of the studies on PP intake reflect data based on dietary questionnaires. However, this approach is subjected, as discussed, to important biases and errors associated with dietary assessment and inaccuracies in food composition databases. Therefore, biomarkers may be chosen as objective measures not depending on dietary assessment methods. Due to large interindividual variability in absorption and metabolism, biomarkers should also reflect tissue exposure to PP more precisely [3]. Moreover, PP exposure also depends on factors such as microbiota, oxidative stress, presence of other xenobiotics, inflammation and disease.

Consequently, developing biomarkers of PP intake, measured in blood and urine, is essential to correctly identify associations between PP intake and health-related outcomes, in epidemiologic and clinical studies. However, the relationship between PP intake and specific biomarkers is often highly complex. The PPs most commonly ingested in the human diet are not necessarily the most active within the body, either because of their low intrinsic activity or due to one or several of the following factors [61]:

  1. 1.

    Poor digestion, which may result from difficult release of PPs from the food matrix, interactions with other food compounds present in the meal, or simply because the individual PPs are not digested by enzymes present in stomach and small intestine;

  2. 2.

    Poor absorption in small intestine;

  3. 3.

    Metabolizing by colon microbiota, with the production of metabolites that are absorbed and may be more active than parent compounds;

  4. 4.

    Extensive biotransformation by phase I and II reactions (mainly phase II) in enterocytes upon uptake, or in liver after bloodstream transport;

  5. 5.

    Rapid excretion.

It would thus be very valuable to have biomarkers for PP intake, especially if we could distinguish between metabolites arising from different classes or subclasses of PPs. This specificity is very important since some PPs present in urine may arise from the ingestion of their main dietary source or may be formed when other consumed PPs are metabolized. For example, the presence of ferulic acid in urine may derive from both consumption of foods rich in ferulic acid, such as cereals, and when caffeic acid is metabolized to ferulic acid [62]. Moreover, other metabolites, such as hippuric acid, may arise from different PPs and also from endogenous metabolism of other substances [63].

It has been suggested that urine biomarkers are more suitable to assess PP intake, since absorbed metabolites may be rapidly removed from circulation, either by entering tissues, or by excretion in urine, in amounts that are higher than the metabolites found in plasma [64]. Another advantage is that urine is a noninvasive method. On the other hand, time of appearance of metabolites in urine varies widely; for example, after ingestion of a berry-fruit pure rich in PPs, different metabolites could be identified in urine collected at different interval times [65]. Thus, determination of urinary metabolites in pooled 24-h urine collection may provide a more realistic assessment of total PP intake. A meta-analysis performed with forty PPs has shown that daidzein, glycitein, enterolactone and hydroxytyrosol, had both a high recovery yield in urine and a high correlation with the dose of consumed PP or food (isoflavones from supplements or soy products, lignans from flaxseed and olive oil, respectively), suggesting good sensitivity and robustness as biomarkers for intake [63].

Results obtained from other studies may suggest biomarkers for dietary exposures of other PP classes or subclasses, or for intake of PP rich foods: Dihydrocaffeic acid-3′-O-sulphate and feruloylglycine were pointed as suitable urine biomarkers for coffee ingestion [64]; conjugates of the flavanones hesperidin and naringenin for orange juice [66]; (epi)gallocatechin, (epi)catechin and methyl-(epi)catechin metabolites for flavan-3-ols [64], and free and conjugated forms of urolithins may be used as markers of ellagitannins [13, 67]. However, some flavonoid subclasses give rise to a broad range of urine metabolites, such as anthocyanins [68] and flavonols [64], thus making it difficult to ascertain the determination of a compound or a few urine compounds to this subclass of ingested PPs. This may be further complicated by sample processing before analysis, leading to different yields in recoveries of metabolites and, therefore, differences in their relative abundances. In some cases, sample preparation methods may lead to losses of unstable metabolites. Moreover, discrepancies could also be associated with the enzymatic treatment of samples; for example, the extent of hydrolysis of sulphates metabolites may be different, depending on the origin of the enzymes utilized [69]. The method of analysis, sensibility of instrumentation and availability of standards are other factors that may influence the accuracy of metabolite analysis and identification. An easier and faster method may be used, based on Folin–Ciocalteu that determines total PPs in urine collected after overnight fast [70]. Although a high variability was observed, total PP excreted has shown a positive association with total PP food intake [4, 70].

Recently, a metabolomics approach applied to a European cohort has identified specific sets of metabolites correlated with consumption of coffee, red wine, tea, citrus fruit, apples and chocolate, suggesting that measurement of food metabolome in future epidemiologic studies should be used as a complement or substitute to the traditional methods based on questionnaires, thus improving PP dietary exposure assessment [71]. This, however, may be difficult in studies with large populations; therefore, the use of questionnaires for the entire population, together with food metabolome analysis for a subset of the population as a validation procedure, could be a feasible alternative.

Conclusions

Nowadays, a broad perception of PP intake in the world and a comparison of intakes among different countries are not possible with the available data. Few countries have conducted reliable national surveys with the specific aim of estimating PP intake. We could identify some major limitations related to the accuracy of available data on PP intake: (1) Different methods are used for dietary assessment, mostly FFQs and 24-h recalls. Although easier to use, FFQs lack detail about consumed foods, must be country-specific and are prone to large random errors. On the other hand, the 24-h recall is considered as a reference method, has detailed and specific information on food consumption, but information on habitual intake requires multiple recalls; (2) Many studies are based on data collected a long time ago, not reflecting actual diet; (3) Different databases are used for calculating PP intake from consumed food, and information on some foods or PPs, particularly NEPP, is missing; (4) Lacking of reliable tools for food metabolome analysis, as a complement to the traditional questionnaires, either by lack of robust biomarkers or analytic limitations. This is a very important gap due to the large interindividual variability in PP absorption and metabolism.

Furthermore, different methods for collecting data on food consumption and calculation of PP intake make it very difficult to compare intakes among countries. Additionally, PP contents are not available for some foods or specific varieties characteristically consumed in some countries, decreasing the reliability of PP intake estimation. This, however, is important if we want to fully understand the relevance of the effect of PPs in human health and design new foods with bioactive compounds targeted to the needs of specific populations in different countries.

To minimize all the factors that may influence the study of the effect of PPs on health, it is essential to have some guidelines harmonizing data collection for such studies. National surveys should be conducted in each country, with the following criteria:

  1. 1.

    Data on food intake should be collected by multiple 24-h recalls distributed along the year (at least one per trimester). Whenever a cohort is covering several years, initial and concluding year should be covered;

  2. 2.

    Total PP intake should include extractable and non-extractable PPs. A harmonized protocol for extraction and determination of PPs must be developed and used for construction of national databases regarding the foods actually consumed by the population. These data should then be included on a wider database, such as Phenol-Explorer.

  3. 3.

    PP content in food should be calculated as the sum of individual components for the main different PP classes: flavonoids (extractable and non-extractable); phenolic acids (extractable and non-extractable); lignans; stilbenes; and tyrosols. Values should be adjusted for energy intake.

  4. 4.

    If possible, PP metabolome in human urine should be determined, at least for a subset of the population, to validate overall PP intake. A less accurate alternative could be total urine PP, which showed a positive association with PP intake.

Many small population studies are also conducted to address health effects of individual or specific mixtures of foods or (poly)phenols. In such studies, the same criteria should be applied concerning the methods to assess food intake and PP content in the consumed food. In addition, when available, they should include validated biomarkers, either individual compounds or panel of compounds, for the food or PP under study.