Objective: To give an overview of the association between tea consumption and iron status.
Methods: A PUBMED search was performed (up to June 2001) for all publications containing the words: tea and ferritin, h(a)emoglobin, iron status or an(a)emia. Sixteen studies were evaluated in groups with high (infants, children and premenopausal women) or low prevalence of iron deficiency (men and the elderly).
Results and Discussion: Of the 16 studies reviewed, six included infants and children, six premenopausal women, two men and two the elderly. In study groups with high prevalence of iron deficiency, tea consumption was inversely associated with serum ferritin and/or haemoglobin. The association disappeared when adjusting for confounding (dietary) factors, except for one study including 40% of iron deficient women. In groups with low prevalence of iron deficiency, tea consumption was not inversely associated with serum ferritin and/or haemoglobin. In those at risk for iron overload, such as middle-aged men, tea consumption may lower serum ferritin concentrations as reported in one study. This finding awaits further confirmation.
Conclusion: This overview shows that tea consumption does not influence iron status in Western populations in which most people have adequate iron stores as determined by serum ferritin concentrations. Only in populations of individuals with marginal iron status does there seem to be a negative association between tea consumption and iron status.
Iron deficiency affects in particular infants, children, teenagers, pregnant and lactating women and women of child-bearing age (Beaton, 1974; Murray & Lopez, 1996) and iron overload mainly middle-aged men (Milman et al, 1999). Physiological iron demands are high in periods of tissue growth during early childhood and adolescence, because of losses of blood and surface cells of the gut, and in women during the reproductive years (Cook, 1990). Serum ferritin, a sensitive marker of the fullness of iron stores, is considered to be the best available index for iron status. In the earliest phase of iron deficiency this iron storage is gradually depleted. A serum ferritin concentration lower than 12 and 10 µg/l for children younger than 6 y indicates exhausted iron stores or iron deficiency (Bender & Bender, 1997; MacPhail, 1998) and concentrations higher than 200 µg/l (women) or 300 µg/l (men) elevated iron stores (Fleming et al, 2001). Once iron stores have become depleted, ferritin concentrations no longer reflect the severity of the iron lack and other measures such as haemoglobin concentrations must be added to diagnose iron deficiency anaemia. Measures of haemoglobin alone are relatively insensitive to iron depletion as concentrations are also reduced in the presence of chronic infection or inflammation regardless of iron status. High serum ferritin alone might indicate underlying disease processes, as ferritin is a positive acute phase protein (Gabay & Kushner, 1999). This may be especially important in the elderly.
In Western populations 10–20% of the infants and children up to 2–4 y (Gibson, 1999; Looker et al, 1997; Thane et al, 2000) and of premenopausal women (Galan et al, 1985; Looker et al, 1997; Soustre et al, 1986) are iron deficient and about 3% (Galan et al, 1985; Gibson, 1999; Looker et al, 1997; Soustre et al, 1986; Thane et al, 2000) iron deficient anaemic, with higher risk in minority and poverty groups (Looker et al, 1997). In men and in the elderly, the prevalence of iron deficiency and iron deficiency anaemia is low (below 3%; Looker et al, 1997). Thirteen percent of the elderly (Fleming et al, 2001) and 18% (Milman et al, 1999) of men can be classified as iron overloaded. The body has adaptation mechanisms to prevent deficiency or excess of iron stores by regulating mucosal iron absorption according to one's iron stores, increasing absorption when iron stores are depleted and reducing absorption as iron stores are repleted (Cook, 1990). The body is more capable in regulating the absorption of non-haem than haem iron. Polyphenol-containing beverages, such as tea, reduce non-haem iron bioavailability by the formation of insoluble complexes (Brune et al, 1989; Disler et al, 1975; Reddy et al, 2000). However, this does not necessarily mean that high tea consumption is associated with an unfavourable iron status at the population level. This review describes the association between tea consumption and iron status and is presented in groups in which high or low prevalence of iron deficiency was expected because of higher iron requirements in infants, children and women than in men and the elderly.
Human studies were evaluated for the association between tea consumption and iron status. We did not perform a meta-analysis, because not enough studies on the relation between tea consumption and iron status have been performed. Instead, we presented an overview of studies published, retrieved from PUBMED up to June 2001. The search was restricted to papers published in English. Full publications as well as abstracts are included in the overview; no attempt was made to search for unpublished results. Some publications or abstracts were retrieved through scanning relevant reference lists of articles. The keyword ‘tea’ was always included in the search. In addition, we considered as outcome measure for iron status the keywords ‘(serum) ferritin’ and/or ‘h(a)emoglobin’ or more generally described as ‘iron status’ or ‘an(a)emia’. The association of tea consumption with serum ferritin and/or haemoglobin had to be reported in the result section of the publication to be included in this overview.
The review is organised according to groups with expected high (infants, children and premenopausal women) and low (men and the elderly) prevalence of iron deficiency and within those groups according to iron status of the study sample. Two studies were not included in the overview because analyses were not stratified according to these groups (Mehta et al, 1992; Yen & Su, 1999). Iron status of each study population was characterised by the percentage of iron deficient (ID) and/or iron deficient anaemic (IDA) subjects. When this figure was absent in the publication, the percentage of anaemic (A) subjects was given. The dietary assessment method was retrieved as well as the main dietary factors important for iron intake/bioavailability such as dietary iron, percentage of haem iron and vitamin C intake. If these data were not given, the intake of meat and vegetables and fruit was reported instead. Tea consumption was calculated as ml/day when possible, but in some references tea intake was given as cups/day, times/week, g/day. When tea consumption was analysed together with coffee consumption it was indicated in the table and when possible the type of tea was given. The association of tea consumption with iron status (concentrations of serum ferritin or haemoglobin) is presented in the tables in the way it was described in the publication. It was stated which factors the association between tea consumption and iron status was controlled for. We evaluated the studies taking into account the study design, sample size, type of population, quality of dietary and iron status assessment and quality of data analysis.
Infants and children
Studies addressing the relationship between tea consumption and iron status in infants and children are summarised in Table 1. Of the six studies found, three studies were carried out within the UK, a high tea consuming country. Two of them (Gibson, 1999; Thane et al, 2000) used data of the UK National Diet and Nutrition Survey (NDNS) to evaluate the association between tea drinking and iron status in children aged 1.5 to 4.5-y-old. Twenty percent of the children were iron deficient, 8% anaemic and 3% iron deficient anaemic. Gibson's cross-sectional study (Gibson, 1999) focused on the relationship between consumption of breakfast cereals and iron intake/status and observed a small inverse association (r=−0.09) between tea consumption and serum ferritin, but did not adjust for possible confounding factors. In a later study with NDNS data (Thane et al, 2000) tea consumption, included in the model as a binary variable, was not significantly associated with serum ferritin and/or haemoglobin concentrations after adjusting for age and gender. Another study (Cowin et al, 2001) compared haemoglobin and ferritin concentrations of children who did or did not consume various food groups. Tea consumption was not associated with low haemoglobin or ferritin concentrations. The amount of cow's milk and calcium was negatively associated with serum ferritin and low energy-adjusted vitamin C and/or low consumption of meat and poultry with low haemoglobin concentrations, after adjustment for confounding factors. This study (Cowin et al, 2001) is well performed including a large number of children (n=701) and applying a 3 day record for food consumption assessment. The factors that remained in the model to predict high serum ferritin concentrations were high mothers' parity, absence of recent infections, high birth weight, high energy adjusted intakes of non-haem iron, high vitamin C and low calcium intakes (Cowin et al, 2001). Two case–control studies (Kuvibidila et al, 1992; Merhav et al, 1985) focused on the relationship between tea consumption and the prevalence of anaemia without determining the iron deficiency status specifically with serum ferritin measurements. Both studies observed (Kuvibidila et al, 1992; Merhav et al, 1985) that anaemia, as determined by haemoglobin concentrations only, is more prevalent among tea drinking infants than among non-tea drinkers, even with a longer period of beef and poultry feeding in the tea group (Merhav et al, 1985) or adjusting for episodes of sickness (Kuvibidila et al, 1992). Differences in other dietary factors than tea were, however, not reported (Kuvibidila et al, 1992; Merhav et al, 1985) and the low weight suggested a lower total energy intake in tea drinkers in the Zairian study (Kuvibidila et al, 1992). In a study from New Zealand (Wilson et al, 1999) in 206 hospitalised children of whom 29% were iron deficient, the diets of 69% of the anaemic children included factors which may have contributed to their iron deficiency such as early introduction of cow's milk, late introduction of meat or regular consumption of tea (eight children drank tea). The investigators, however, did not study the presence of dietary factors in a non-anaemic (control) group of children. From this study it is, therefore, not possible to clarify whether and how the dietary factors really correlated with iron status.
Table 2 summarises studies that investigated the association between tea consumption and iron status in women. No significant associations between high tea consumption and low iron status parameters were found in a European study (Van de Vijver et al, 1999) and in a Chinese study (Root et al, 1999) with a low percentage of iron deficient women. The cross-sectional study (Van de Vijver et al, 1999) among European girls (n=1080; mean 13.5 y) and young women (n=524; mean 22.0 y) from six countries focused on the association between calcium intake and iron status. Tea and coffee consumption was not significantly associated with serum ferritin. The Chinese study (Root et al, 1999) showed that women can adapt successfully to a wide range of iron intakes and bioavailability. Root et al, (1999) examined the iron status of middle-aged Chinese women (n=400; 32–66 y old) randomly selected from five counties in rural China. Twenty percent of these women were postmenopausal. The women consumed diets rich in non-haem iron and low in vitamin C. The level of intake of black tea, even in very high amounts (9–38 g dry tea per day) and dietary fibre were not associated with any measure of iron status.
A negative association between tea (and coffee) consumption and iron status (mainly measured by serum ferritin) was observed in studies including a large percentage of iron deficient women (Galan et al, 1985; Pate et al, 1993; Razagui et al, 1991). The cross-sectional studies (Galan et al, 1985; Soustre et al, 1986) included women in the same age range (16–53 y) and with similar percentages of iron deficient subjects (16 and 21%, respectively). In both studies, the duration of menses inversely correlated with serum ferritin concentrations, while dietary iron intake did not significantly correlate with serum ferritin. Both tea and dairy products inversely correlated with serum ferritin concentrations (r=−0.18 and r=−0.20 respectively, Galan et al, 1985). This result was not adjusted for intakes of other foods that could have affected iron bioavailability. The significant association might have disappeared since intakes of coffee and dairy products, but not tea consumption, inversely associated with serum ferritin, after controlling for all other foods and beverages (Soustre et al, 1986).
Two studies included more than 40% iron deficient women (Pate et al, 1993; Razagui et al, 1991). Only 15 mentally handicapped women aged 28 y (range 19–43 y old) with mean daily iron intake of 9.5±1.5 mg (range from 6 to 11.7 mg/day) were investigated (Razagui et al, 1991). Tea intake with meals was higher in iron deficient women (563 ml/meal/day), compared to women with sufficient iron stores (184 ml/meal/day), whereas vitamin C intake with meals was lower. Tea consumption at meal times was significantly negatively correlated with serum ferritin concentrations (r=−0.67; all subjects combined) and positively with vitamin C intake (r=0.71; all subjects combined). However, these correlations were not adjusted for differences in other dietary variables. In female long-distance runners (n=111) and inactive women (n=65) with a mean age of 29 y, 50% of the runners and 22% of the inactive group were iron deficient (Pate et al, 1993). No significant differences were found between the two groups in mean dietary intake of iron, vitamin C, coffee and tea or plant food products. The runners consumed significantly fewer meat products per week (P<0.01), more fibre and a higher percentage of energy from carbohydrates. Coffee and tea intake and minutes run per week were significantly, independently and negatively associated with serum ferritin concentrations (total group).
The association between tea intake and iron status was addressed in two studies (Hunt & Roughead, 2000; Imai & Nakachi, 1995), of which one was an experiment (Hunt & Roughead, 2000; Table 3). The parallel designed experiment (Hunt & Roughead, 2000) in 31 healthy men (age≥32 y) showed successful adaptation of iron absorption to diets with low iron bioavailability containing tea (from 1 g dry, black instant) with each meal. Serum ferritin and the other blood indexes of iron status were insensitive to the 12 week diets with low or high iron bioavailability. Faecal ferritin excretion changed within a few days in response to differences in dietary iron bioavailability. The cross-sectional Japanese study (Imai & Nakachi, 1995) including 1371 Japanese men aged 40 y and above suggested positive effects of green tea on iron status as well as lipid peroxidation. Lower serum ferritin, haemoglobin and lipid peroxide concentrations were observed among men consuming more than 10 cups of green tea compared with less than three cups a day after controlling for age, cigarette smoking and alcohol consumption. We assume that the unit given in Table 4 of this publication is misprinted (µg/l instead of mg/l). Concentrations of lipid peroxides in smokers reduced to levels of non-smokers when they consumed more than 10 cups of green tea a day. Serum ferritin and lipid peroxide concentrations were significantly associated (r=0.25, P<0.001). Dietary iron intake and enhancers or inhibitors of iron absorption were not included in the study.
Elderly women and men
The relationship between food consumption, inclusive of tea, and biochemical markers of iron status was investigated in elderly persons in two studies (Table 4; Doyle et al, 1999; Roebothan & Chandra, 1996). In the National Diet and Nutrition Survey (n=1268; Doyle et al, 1999) 10% of the free-living and 45% of the institutionalised seniors were found to be anaemic. Ninety-five percent of the participants consumed tea. Tea consumption was negatively associated with haemoglobin concentrations in men. Tea consumption was positively associated with energy intake and may be associated with a higher level of social activity of the seniors. For high serum ferritin, high intake of vegetables and low dairy foods were the best predictors. Roebothan (Roebothan & Chandra, 1996) studied 127 elderly (33 men and 94 women) above the age of 60 y. Dietary intakes of haem iron, ascorbic acid, calcium, dietary fibre and also of tea and coffee were not significantly different in seniors with adequate (n=108) compared with inadequate (n=19) iron stores. The sample size of this study was relatively small. The two studies do not provide evidence for an inverse association between tea consumption and iron status in the elderly. Tea drinking in the two studies is associated with a good social life and a higher energy intake.
A wide variety of studies with different designs, from different countries, and carried out in different age and gender groups, addressed the association between tea consumption and iron status. All studies, except for one in China and one in Zaire, were carried out in Western populations. The conclusions are mainly based on cross-sectional studies determining tea consumption on the one and iron status on the other hand. Most studies addressed the effects of a variety of dietary factors, among them tea, that could affect iron status and calculated correlation factors between tea consumed and serum ferritin (Doyle et al, 1999; Galan et al, 1985; Gibson, 1999; Pate et al, 1993; Root et al, 1999; Soustre et al, 1986; Van de Vijver et al, 1999). These analyses were carried out with (Doyle et al, 1999; Pate et al, 1993; Root et al, 1999; Soustre et al, 1986) or without (Galan et al, 1985; Gibson, 1999; Van de Vijver et al, 1999) adjusting for other iron bioavailability factors. Other case–control studies (Kuvibidila et al, 1992; Mehta et al, 1992; Merhav et al, 1985; Razagui et al, 1991; Roebothan & Chandra, 1996) compared iron deficient (or anaemic) cases with non-iron deficient (and/or non-anaemic) controls with respect to tea consumption and other dietary habits. Only one experimental study in men was available for this overview. The sample size in general was acceptable except for the study of Razagui and co-workers (Razagui et al, 1991).
Main factors determining the strength of the association between tea consumption and iron status, are the mean iron status of the population under study and the adjustment for confounding factors. Iron absorption rates are higher in iron deficient subjects (with low serum ferritin concentrations) compared with subjects with sufficient iron stores. In populations including a high proportion of these subjects, therefore, it is more likely to observe significant associations of iron status parameters with enhancers and inhibitors of iron bioavailability. For calculating iron absorption from the diet, taking into account serum ferritin levels, newly developed algorithms can be applied (Hallberg & Hulthen, 2000). Besides factors such as social class or recent illnesses that may confound the association between tea consumption and iron status, dietary factors other than tea consumption alone influence iron intake and bioavailability and must be taken into account. In the children from the UK studies (Cowin et al, 2001; Gibson, 1999; Thane et al, 2000) and all Western women study populations (Galan et al, 1985; Pate et al, 1993; Razagui et al, 1991; Soustre et al, 1986; Vijver et al, 1999) the average iron intake was below the recommended daily allowances. Comparing anaemic with non-anaemic subjects, diets were, in addition, lower in vitamin C (Razagui et al, 1991; Roebothan & Chandra, 1996). The studies that found a negative association between tea consumption and iron status (Galan et al, 1985; Gibson, 1999; Razagui et al, 1991) or anaemia (Kuvibidila et al, 1992; Merhav et al, 1985), except for the study of Pate et al (1993) did not adjust for differences in iron intake and other iron bioavailability factors. Recent experimental work suggests that not polyphenols, but animal tissue (beef, poultry and seafood), phytic acid and vitamin C might be the most important factors determining iron bioavailability (Reddy et al, 2000). Those factors should be taken into account when addressing the association between tea consumption and iron status.
In conclusion, this overview shows that tea consumption does not influence iron status in Western populations in which most people have adequate iron stores as determined by serum ferritin concentrations. Only in populations of individuals with marginal iron status, there seems to be a negative association between tea consumption and iron status.
The authors thank Dr Jianjun Zhang, Dr Evert Schouten and Dr Hugo Kesteloot for critically evaluating the manuscript. This study was supported by a grant from the Unilever Chair in Nutritional Epidemiology.
About this article
European Journal of Clinical Nutrition (2007)