A number of potential health effects have lately been accorded to tea consumption. It is, however, not clear whether an increase in tea consumption increases the risk of iron depletion in a normal apparently healthy adult population. We have therefore evaluated this.
A total of 954 men (aged 52–68 years) and 1639 women (aged 42–68 years), who were participants of SU.VI.MAX Study, completed a detailed questionnaire on tea consumption. To determine the iron status of the participants, a venous blood sample was drawn and serum-ferritin was measured. Iron depletion was defined as a serum ferritin concentration <16 μg/l. Three 1-day food records were used to estimate the intake of other dietary enhancing or inhibiting factors of iron absorption, which were included in the logistic regression models.
The mean serum-ferritin concentration was not related to black, green and herbal tea consumption in men, pre- or postmenopausal women. Also the risk of iron depletion was in the multivariate model not related to any kind of tea drinking or to the strength of tea, the infusion time or the time of tea drinking.
The data suggest that normal apparently healthy adults are not at risk of iron depletion owing to any kind of tea drinking.
The SU.VI.M.AX project received financial support from public and private sectors. Special support was received from Unilever Bestfoods France (UBF) and to the ‘Centre d’Information Scientifique Thé & Santé de Lipton’, UBF.
Over 2000 million people in the world, most of them living in developing countries suffer from iron deficiency. It has been estimated that 99% of Southeast Asian population are iron-deficient, and even in Europe 17% of the population are iron-deficient (Ramakrishnan, 2002).
The inhibition of non-heme iron absorption owing to simultaneous tea consumption is well known; a high consumption of tea may increase the risk of iron depletion in populations of individuals with marginal iron status (Temme and van Hoydonck, 2002). Especially children, pregnant and lactating women and women of childbearing age are at risk (Beaton, 1974). The inhibition of non-heme iron absorption is owing to the polyphenols in tea, which form insoluble complexes with iron (Disler et al., 1975; Brune et al., 1989). It is not known exactly which polyphenols are responsible, but it has been suggested that the content of iron-binding galloyl groups might be a major determinant of the inhibitory effect of phenolic compounds on iron absorption from the diet (Brune et al., 1989). Although herbal teas contain polyphenols, they are mainly monomeric flavonoids and not cathechins (Lallement and Bezanger, 1970). It is, however known that they are also capable of inhibiting iron absorption Hurrell et al. (1999) have shown that several polyphenolic-containing beverages (coffee, (herbal) tea and cocoa) can be potent inhibitors of non-heme iron absorption. The inhibition of non-heme iron absorption is strongly dose-related, the polyphenol content of tea is important, which in turn depends largely on the tea consumption behaviour: the type of tea, the time of the day at which tea is consumed (with or between meals) the mode of tea preparation (infusion time and the quantity of tea leaves); the longer the infusion time and the higher the amount, the higher the polyphenol content in tea (Astill et al., 2001). These factors may, therefore, also have an impact on iron absorption.
In a recent meta-analysis tea consumption was observed to be associated with a lower risk of cardiovascular disease, probably owing to high-antioxidant capacity of polyphenols in tea (Peters et al., 2001). Before establishing recommendations to increase tea consumption in view of the prevention of cardiovascular diseases, it is essential to know if regular tea drinking in an adult population with a normal iron status is related to an increase risk of iron deficiency. In epidemiologic studies carried out so far, no clear association between tea consumption and iron deficiency has been observed. This could be owing to the fact that great deal of dietary iron is in the heme form, which is readily absorbed without interference of polyphenols or other inhibitors/enhancers (Hallberg and Hulthén, 2000). However, the fact that no association has generally been observed may also be owing to insufficient adjustment for confounding factors and factors affecting iron absorption (consumption of coffee, vitamin C, animal protein and calcium) or the inclusion of a small number of subjects (Temme and van Hoydonck, 2002). Therefore, we performed a study in an apparently healthy adult population on (herbal) tea consumption and iron status taking into account other variables affecting iron absorption.
Subjects and methods
Subjects were participants of the SU.VI.MAX (SUpplémentation en VItamines et Minéraux AntioXydants) Study, which is a double–blind, placebo–controlled, primary prevention trial undertaken to determine whether supplementation with antioxidant vitamins and minerals at nutritional doses can reduce the incidence of cancers and ischemic heart disease. The rationale, design and methods of the study as well as the major characteristics of the participants have been described in detail elsewhere (Hercberg et al., 1998, 2004). In brief, 13 017 eligible subjects (women aged 35–60 years and men aged 45–60 years) were included in 1994 to be followed for 8 years. Participants underwent a yearly visit consisting every alternate year of biological sampling or a clinical examination. Participants were free of chronic diseases and apparently healthy at baseline. A random selection of 2600 subject among the 5287 subjects for whom dietary intake data were available was made for the analysis of serum-ferritin. Subjects who had hemochromatosis or a high serum-ferritin level owing to a known infection were excluded. The present analysis includes 2593 subjects for whom all necessary data were available. These subjects do not differ from the total population with respect to the major characteristics.
The SU.VI.MAX Study has been approved by the ethical committee for studies with human subjects (CCPPRB no. 706) of Paris–Cochin and the ‘Commision National Informatique et Liberté’ (CNIL no. 334641) which advocates that all medical information is confidential and anonymous.
Assessment of tea consumption and nutrient intake
Tea consumption was estimated using a specially developed tea questionnaire, which included questions on consumption of black, green and herbal tea, the strength of the tea, the infusion time, the tea recipe as well as the moment of tea drinking (with or between meals). The questionnaire was based on the validated Arizona tea questionnaire (Hakim et al., 2001), which was adapted to fit the French lifestyle. It was completed by over 7000 subjects in 2001/2002.
Data on intakes of coffee, iron, vitamin C, calcium and animal protein were obtained through 1-day food records that participants were asked to complete every 2 months (altogether six times per year) through the Minitel Telematic Network. The Minitel is a small terminal used in France as an adjunct to the telephone. At the beginning of the study, participants received free of charge a tiny central processing unit specifically developed for the study and loaded with specialized software that allows subjects to fill out the computerized food record offline and to transmit data during brief telephone connections. An instruction manual for codification of foods guided the participants during the completion of the records. The manual contains photographs showing portions in three sizes and along with the two intermediate portions between the three shown sizes and the two extreme portion sizes (less than the smallest size or more than the largest size), seven choices are available to indicate the consumed portion. Photos of portion sizes were previously validated using 780 subjects in a pilot study (Le Moullec et al., 1996). For the present study the mean of three food records collected in 2001/2002 were used.
Assessment of iron status and other data
Serum-ferritin concentration was used to determine the iron status of the participants. At the sixth annual visit (2001/2002), a 35 ml venous blood sample was drawn in participants who had been fasting for 12 h at the time of their visit. After collection (Becton Dickinson tubes, Pont de Chaix, France) blood was kept at +4°C in the dark until centrifugation and preparation of aliquots. Serum-ferritin concentrations were measured using a nephelometric assay (BNII Behring, Paris La Défense, France). The laboratory quality assurance included analysis of serum from standard pools with each run and international standards. Complete data on biochemical markers of iron status were obtained for 1548 men and 2502 women. Iron depletion was defined as a serum-ferritin level smaller than 16 μg/ml (Hallberg et al., 1993).
Data on menopausal status and smoking were obtained through an interview by a physician during the sixth annual visit in 2001/2002. Menopausal status was based on self-declaration and age.
The amount and frequency of tea consumption was estimated from the questionnaire and the mean (s.d.) for descriptive variables were calculated. The mean (s.d.) serum-ferritin concentration was calculated by group of tea drinking and differences between the groups were evaluated using analysis of variance, adjusting for age. To evaluate the relation between tea drinking and iron status, odds ratios (95% confidence interval) were calculated adjusting for age, smoking and intakes of coffee, iron, calcium, animal protein and vitamin C. Being in the intervention or control group for the original study was no confounder for these analyses. Subjects were divided in three groups for the analyses: men, pre- and postmenopausal women.
The general characteristics of the study population are presented in Table 1. As expected the concentration of serum-ferritin was lowest in premenopausal women and highest in men. Consumption of black tea was most common, but herbal tea was also consumed by a large part of the postmenopausal women.
Most of the subjects declared to drink tea of normal strength, with about a fifth-to-a-quarter declaring to drink strong tea for all types (Table 2). The strength of tea was positively correlated to the amount of tea used for making tea (data not shown). The most common infusion time is 1–5 min and most subjects declared to drink black and green tea always with meals, whereas they drank herbal tea mostly between meals.
Serum-ferritin concentration was not related to black, green and herbal tea consumption in men, pre- and postmenopausal women (Table 3). Neither was there any difference in serum-ferritin concentration between the tea-drinking groups with respect to strength of tea, infusion time and time of tea drinking.
When the population was divided in depleted and nondepleted subjects, no relation was observed between any kind of tea drinking and iron status (Table 4). Furthermore, neither the strength of tea, the infusion time and the time of tea drinking were related to being iron-depleted or not (data not shown).
We found no relation between black, green or herbal tea and iron depletion in a general apparently healthy adult population.
We have to remind the reader that the population consisted of subjects that initially took part in a long-term intervention trial. Therefore, they may have been health conscious and thus in a better health than the general French population. This means that the range of serum-ferritin concentration may have been too small to show a relation. However, compared to previous studies, our analyses involved a high number of subjects from all over France, which would increase the range of intake and serum concentrations.
We have chosen to take serum-ferritin concentrations as a biomarker of iron status as it is directly related to the level of storage iron in normal subjects. It has been suggested that a difference of 1 μg/l in serum-ferritin is equivalent to about 8 mg of storage iron in normal adults (Walters et al., 1973). Hallberg et al. (1993) have shown that the best discrimination for iron deficiency is found at a concentration of less than 16 μg/l. However, this marker also increases in response to infection and even though we eliminated subjects with known infection, it is possible that some iron-depleted persons having an unknown infection are misclassified as healthy. However, we believe this number of such subjects is small and that bias owing to this is limited.
Since 1985 several studies have been published evaluating iron status in relation to tea consumption. The first study, among 157 female students showed an inverse relation between iron status and tea consumption, but no correction was made for other enhancing and inhibiting factors (Galan et al., 1985). Shortly after, the same group published data of 127 French women (16–53 years of age) in which no relation was observed after multiple adjustments were made (Soustre et al., 1986). A small study among mentally handicapped women in Britain showed that iron-depleted women, based on serum-ferritin concentrations, had a higher tea intake than nondepleted women (Razagui et al., 1991). It has to be noted, however, that also their vitamin C consumption was significantly lower. Data from the NHANES II study (Mehta et al., 1992) showed similar results, but tea and coffee intake were not separated and dietary intake was assessed using only one 24-h recall, which is probably not enough to estimate intake when relating these data to measures of health and disease (Mennen et al., 2002). Another study from the US among female runners also showed an inverse association between serum-ferritin and tea/coffee consumption, but no correction was made for other factors (Pate et al., 1993). A Japanese study provided data in men over 40 years of age, showing a lower serum-ferritin concentration in men consuming over 10 cups of green tea per day, without taking into account other dietary factors (Imai and Nakchi, 1995). This seems contrary to the data from the British National Diet and Nutrition Survey where no relation between serum-ferritin and tea consumption was observed in elderly subjects (Doyle et al., 1999). The latter finding is in line with results from a Canadian study where no difference was observed in terms of tea consumption between elderly subjects with an inadequate and those with an adequate iron status, based on serum-ferritin concentrations (Roebothan and Chandra, 1996). Four Danish studies have been published with different results. The first is the only longitudinal study in this field, evaluating dietary intake obtained through a food frequency questionnaire and serum-ferritin during a 6-year follow-up in 238 middle-aged Danes. The authors concluded that dietary factors are poor predictors of serum-ferritin (Osler et al., 1998). This tallies with the results from a later study in 1319 middle-aged men in which no relation between tea consumption and serum-ferritin was observed (Milman et al., 2000). The other two studies in 1332 middle-aged Danish women and in 232 Danish eldery showed an inverse relation between tea and coffee consumption and serum-ferritin, but again other dietary factors were not included in the analyses (Milman et al., 1999, 2004). Finally, a study among 400 Chinese women (aged 32–66 years) did not show a relation between tea consumption (even in very high amounts) and serum-ferritin (Root et al., 1999).
Overall out of 13 studies six did not observe a relation, and the studies which showed the contrary, either did not correct for the other enhancing and inhibiting factors (five studies) or showed also lower vitamin C intakes when looking at anemic subjects (two studies). It seems that although drinking a cup of black tea with (or within an hour of) a meal may reduce absorption of non-heme iron with 60–80% as suggested by the algorithms of Hallberg and Hulthén (2000) and by Zijp et al. (2000), within a complete diet this has no or limited effect on iron status in healthy adults. This is well in line with the results from our study.
The fact that tea consumption seems to have a limited effect on iron status, may have a mechanistic explanation. Hunt and Roughead (2000) and Hunt (2003) showed that men and premenopausal women are capable to adapt iron absorption according to its bioavailability; that is, on a high-bioavailability diet the absorption of non-heme iron decreased, whereas it increased on a low-bioavailability diet. So when bioavailability of non-heme iron is low owing to a high-tea consumption and a low-vitamin C consumption for example, these adaptations prevent a deterioration of iron status. Another explanation may be that dietary inhibiting and promoting factors, such as phytic acid and animal protein are more important for iron absorption than polyphenols (Reddy et al., 2000).
To our knowledge there are no other studies published that include herbal tea consumption in their evaluation of iron status. Inhibitory effects of herbal tea have been established, although they are much lower than those from black tea (Hurrell et al., 1999). Despite the effects shown on iron absorption, herbal tea does not seem to influence iron status, based on serum-ferritin concentrations.
In conclusion, the results from our study suggest that in an apparently healthy adult population tea consumption as a part of a total diet does not seem to lead to an increased risk of iron depletion, irrespective of the type and time of tea consumption and the strength of tea.