Everyone eats and drinks a variety of foods and beverages each day. The reasons for food choice are multiple, including culture, individual preference, cost, availability and the influence of advertising, as well as dietary advice from a variety of experts(Reference Caraher and Coveney1). Every day the public receives dietary advice from many sources, both reliable and unreliable, making it difficult to decide what is appropriate advice. Complete data on the intake of a particular food or dietary pattern in relation to health benefits or risks can take decades to emerge. Even then the data may not reach the level of evidence usually required for therapeutic interventions. Just how much evidence is sufficient to make public health guidelines? Tea and coffee present an interesting case study in regard to this issue.
Although tea and coffee are two of the beverages most commonly consumed, little dietary advice on their consumption exists(Reference Grigg2). Evidence has suggested that tea may play a role in the prevention of several cancers and CVD. As these are two of the largest causes of morbidity and mortality in the world(Reference Stewart and Kleihues3), waiting for complete data to become available before providing information to the public could result in unnecessary morbidity and mortality. The present paper examines the existing evidence, predominantly from epidemiological studies, on the relationship between coffee, tea and several chronic health conditions. It also considers whether evidence is currently sufficient to make any level of dietary recommendations.
Making dietary guidelines: how much evidence is required?
The objective of public health nutrition is to optimise longevity and health for the individual and diet is an extremely important determinant of health(Reference Binns and Lee4). Chronic diseases, by definition, have a long natural history, taking decades to develop and are likely to be multi-factorial in causation(Reference Binns and Lee4). This complicates nutritional epidemiology, and different approaches have been advocated for obtaining evidence on nutrition. The epidemiological or bottom-up approach, followed by most mainstream nutritional epidemiologists, involves studying particular diseases and then finding common elements within diets that increase or decrease their risk(Reference Trichopoulos, Lagiou and Trichopoulou5). The ecological or top-down approach instead uses the experience of different cultures whose diets appear to protect against morbidity and mortality(Reference Trichopoulos, Lagiou and Trichopoulou5). This latter approach has been advocated by experts(Reference Wahlqvist, Kouris-Blazos and Wattanapenpaiboon6) to illustrate the apparent advantages of the Mediterranean and Japanese diets with regard to CHD(Reference Keys7). Both approaches have strengths and weaknesses, so combining evidence may have considerable advantages.
Evidence on the optimal diet for longevity and health is most often communicated to the public through dietary guidelines(Reference Binns and Lee4). The process of development of these guidelines is sophisticated, with an increasing amount of scientific evidence required to justify decisions(Reference Binns and Lee4). Several classifications of the hierarchies of evidence have been developed; for example, by the Australian National Health and Medical Research Council (NHMRC)(8) (see Table 1). The gold standard for all systems is a meta-analysis or pooled analysis of data from a number of randomised controlled trials (RCT)(Reference Kroke, Boeing, Rossnagel and Willich9). However, epidemiological studies make up the bulk of evidence for nutritional guidelines and these are usually classified as Level III evidence, including cohort, case–control and comparative ecological studies with historical controls(8). Although these study designs have some inherent errors, they can provide helpful evidence of a likely role of diet in affecting health(Reference Margetts, Warm, Yngve and Sjostrom10).
Table 1 National Health and Medical Research Council levels of evidence for clinical practice guidelines(8)
The NHMRC states: ‘A decision should be made about what is feasible and appropriate in a given situation and the extent to which reasonable standards have been met by the available body of evidence’(8). There are important practical differences between the development of public health nutrition guidelines and guidelines for clinical practice(Reference Binns and Lee4). Since everyone eats every day, the option of no treatment (or no advice) is not available. Consideration must then be given to the level of evidence required to justify action in the context of dietary advice(Reference Wanless11). Achieving the NHMRC’s highest standard is typically unrealistic and inappropriate in the field of nutrition(Reference Kroke, Boeing, Rossnagel and Willich9). RCT frequently cannot be applied to evaluate public health interventions; it would either be unethical or impractical to randomise or use controls due to difficulty in blinding or the length of time required(Reference Binns and Lee4, (Reference Kroke, Boeing, Rossnagel and Willich9, (Reference Margetts, Warm, Yngve and Sjostrom10). Cost is usually substantial, and sometimes prohibitive, in public health RCT(Reference Wanless11). Consequently, observational studies or short-term interventions using biomarkers or interim end points, together with ecological studies, must be utilised(Reference Binns and Lee4). This situation is not perfect, yet waiting for complete data to become available could deprive the public of health benefits. Nutritionists are obliged to make dietary recommendations based on the best evidence available.
Case study: tea and coffee
There has been considerable interest in researching the health effects of tea and coffee. The majority of evidence comes from epidemiological studies, backed up by animal studies, in vitro studies and some short-term human interventions using biomarkers or other interim end points. The PubMed database was searched for systematic and other reviews of epidemiological studies, using the keywords ‘tea’+selected disease, such as ‘colorectal cancer’ and ‘cardiovascular disease’, and similarly for ‘coffee’+selected disease. Studies published in English since the year 2000 were included and full text copies of the papers were obtained. A total of 58 published reviews were obtained. Studies in the reviews were then tabulated, after which PubMed was searched again for additional epidemiological studies. Tables 2 and 3 summarise studies for coffee and tea, respectively. A full list of references can be provided upon request. A method employed by the World Cancer Research Fund(12) to rate the level of current evidence for foods and cancer prevention was adopted. The objective was to assess whether the evidence is convincing, probable, possible or insufficient that coffee or tea decreases risk, has no relationship to, or increases risk of a particular disease. Tables 4 and 5 list the level of evidence for coffee and tea, respectively, and particular diseases.
Table 2 Number of reviews and epidemiological studies on coffee and selected diseases
*For conditions with over 100 epidemiological studies, the number of studies was not counted.
Table 3 Number of reviews and epidemiological studies on tea and selected diseases
Table 4 Amount of evidence for coffee and selected conditions
Table 5 Amount of evidence for tea and selected conditions
Coffee: effects on health
Cancers
Coffee has been suspected of increasing the risk of several cancers but this has not been confirmed by further research. Two reviews on coffee consumption and breast cancer reported that studies consistently showed no relationship(12, (13). From the thirty-four studies identified, evidence that the consumption of coffee has no relationship with breast cancer risk is convincing.
While an early pooled analysis of eight studies on pancreatic cancer reported an OR of 1·4 (95 % CI 1·1, 1·8) for high v. low consumption(Reference La Vecchia, Liati, Decarli, Negri and Franceschi14), subsequent reviews(12, (13, (Reference Gordis15, (16) and a total of forty-five studies located indicated coffee could not be identified as a risk factor. The majority of evidence from two reviews of coffee and ovarian cancer(13, (Reference Larsson and Wolk17) and a total of seventeen studies also reported no association. In addition, two reviews(12, (Reference Dhote, Pellicer-Coeuret, Thiounn, Debre and Vidal-Trecan18) and a total of sixteen studies located on renal cancer, a pooled analysis of nine case–control studies on thyroid cancer(Reference Mack, Preston-Martin, Dal Maso, Galanti, Xiang and Franceschi19), and two reviews(12, (Reference Dagnelie, Schuurman, Goldbohm and van den Brandt20) and a total of sixteen studies on prostate cancer, all showed no substantial associations. On the basis of currently available evidence, coffee consumption probably has no relationship with the risk of pancreatic, ovarian, renal, thyroid or prostate cancer.
However, seven reviews located on urinary tract or bladder cancer reported possible slight increased risks with coffee consumption(12, (13, (Reference Viscoli, Lachs and Horwitz21–(Reference Villanueva, Cantor, King, Jaakkola, Cordier, Lynch, Porru and Kogevinas25). While two systematic reviews calculated increased risks of approximately 20 %(Reference Zeegers, Tan, Goldbohm and van den Brandt23) and 26 %(Reference Villanueva, Cantor, King, Jaakkola, Cordier, Lynch, Porru and Kogevinas25), a European-based review found an increased risk only for drinkers of 10 cups or more per d(Reference Sala, Cordier and Chang-Claude22). Despite the large number (eighty-eight) of studies located, evidence only suggested a possible slightly increased risk at high levels of consumption. Moreover, the majority of animal studies have shown no carcinogenic effects for coffee(Reference Zeegers, Kellen, Buntinx and van den Brandt26). Moderate consumption of coffee probably has no relationship with the risk of urinary tract or bladder cancer.
On the other hand, four reviews of coffee and colorectal cancer suggested a possible protective effect(12, (13, (Reference Giovannucci27, (Reference Tavani and La Vecchia28). A meta-analysis of twelve case–control studies reported a pooled relative risk (RR) of 0·72 (95 % CI 0·61, 0·84) for high v. low coffee consumption(Reference Giovannucci27). Among a total of forty-nine studies identified, the majority of case–control studies suggested an inverse association but the pattern of risk in cohort studies was unclear. Another review of five cohort studies did not support a protective effect(Reference Marques-Vidal, Ravasco and Ermelinda Camilo29). Several plausible biological mechanisms have been proposed, including reduction of cholesterol, bile acid and neutral sterol secretion into the colon, increased colonic motility, antioxidant properties of coffee(Reference Daglia, Papetti, Gregotti, Berte and Gazzani30, (Reference Anese and Nicoli31) and reduction of the genotoxicity of several carcinogens by coffee components(Reference Cavin, Holzhaeuser, Scharf, Constable, Huber and Schilter32). Animal studies also provide some support(Reference Cavin, Holzhaeuser, Scharf, Constable, Huber and Schilter32). The evidence suggests that coffee consumption is possibly related to a decreased risk of colorectal cancer.
CVD
Intake of coffee has been suspected as a risk factor for CVD. A recent review concluded that no clear association between coffee and the risk of hypertension, myocardial infarction or other CVD could be demonstrated in epidemiological studies(Reference Sudano, Binggeli, Spieker, Luscher, Ruschitzka, Noll and Corti33). Human intervention studies have explored only the short-term effects of coffee consumption on precursors of CVD. Two meta-analyses of RCT on blood pressure(Reference Jee, He, Whelton, Suh and Klag34, (Reference Noordzij, Uiterwaal, Arends, Kok, Grobbee and Geleijnse35) agreed that coffee consumption can increase blood pressure slightly. Another meta-analysis of fourteen intervention trials showed a significant dose–response relationship between coffee consumption and total or LDL cholesterol levels(Reference Jee, He, Appel, Whelton, Suh and Klag36). Although coffee consumption can have acute effects on precursors of CVD, overall, research so far has indicated that moderate coffee intake is probably not associated with increased risk of CVD(Reference Sudano, Binggeli, Spieker, Luscher, Ruschitzka, Noll and Corti33, (Reference Ranheim and Halvorsen37, (Reference Cornelis and El-Sohemy38).
Other conditions
The influence of coffee on risk of type 2 diabetes is a relatively new area of research, where five recent reviews have indicated a protective effect(Reference Ranheim and Halvorsen37, (Reference van Dam and Hu39–(Reference van Dam42). A systematic review of nine cohort and seven cross-sectional studies reported an RR of 0·65 (95 % CI 0·54, 0·78) for the highest v. the lowest category of coffee consumption(Reference van Dam and Hu39). One cohort study(Reference Greenberg, Axen, Schnoll and Boozer43) and subsequent review(Reference Greenberg, Boozer and Geliebter41) indicated that the protective effect may be linked to possible weight loss properties of coffee. While results from mechanistic studies in animals have been inconsistent, possible mechanisms for the effect of coffee components on glucose metabolism have been demonstrated. These include contributing to the in vitro antioxidant capacity of the diet(Reference Svilaas, Sakhi, Andersen, Svilaas, Strom, Jacobs, Ose and Blomhoff44), reduction in glucose concentrations(Reference Rodriguez de Sotillo and Hadley45) and increase in insulin sensitivity(Reference Shearer, Farah, de Paulis, Bracy, Pencek, Graham and Wasserman46), inhibition of glucose absorption in the intestine(Reference Johnston, Clifford and Morgan47) and benefits for lipid oxidation(Reference Yoshioka, Kogure, Yoshida and Yoshikawa48). Short-term human intervention studies have not shown positive effects on glucose metabolism(Reference Johnston, Clifford and Morgan47, (Reference van Dam, Pasman and Verhoef49–(Reference Battram, Arthur, Weekes and Graham52), but these results cannot be extrapolated to long-term effects. Coffee consumption may possibly decrease the risk of type 2 diabetes.
A role for coffee in the prevention of Parkinson’s disease is still being debated. A systematic review of thirteen studies(Reference Hernan, Takkouche, Caamano-Isorna and Gestal-Otero53) reported an overall RR of 0·69 (95 % CI 0·59, 0·80) for coffee v. non-coffee drinkers. Of seven additional studies, most demonstrated inverse associations in some groups. However, a biologically plausible mechanism for the role of coffee is currently unavailable(Reference Joghataie, Roghani, Negahdar and Hashemi54), so that the evidence for a preventive effect can only be rated as possible.
Epidemiological studies have reported a reduced incidence or mortality for liver cirrhosis with coffee consumption(Reference La Vecchia55). This has been linked to coffee’s effect on lowering γ-glutamyltransferase (GGT) levels, an indicator of cirrhosis risk. However, RCT have shown only slight or no depression of GGT levels by coffee(Reference La Vecchia55). Based on the limited data, it appears possible that coffee consumption is related to a reduced risk of liver cirrhosis.
About twenty-five reviews exist on coffee/caffeine and the risk of unfavourable reproductive outcomes. A review of sixty-eight studies(Reference Leviton and Cowan56) and a systematic review of congenital abnormalities(Reference Browne57) concluded that no convincing evidence has been presented which shows caffeine can increase the risk of any reproductive adversity. Nevertheless, an intake of less than 300 mg caffeine/d for pregnant women was recommended as a precaution(Reference Nawrot, Jordan, Eastwood, Rotstein, Hugenholtz and Feeley58). Another review of thirty-two studies on caffeine and bone health also concluded ‘no evidence that caffeine has any harmful effect on bone status or on the calcium economy in individuals who ingest the currently recommended daily allowances of calcium’(Reference Heaney59). Therefore, coffee consumption probably has no relationship with reproductive adversities or bone health.
Finally, coffee has been linked to increased risk of gallstones and rheumatoid arthritis. Data are sparse, but from twelve studies identified on gallstones and six on rheumatoid arthritis, there appears no association between coffee consumption and gallstones or rheumatoid arthritis.
Tea: effects on health
Of the total tea produced and consumed, 78 % is black, 20 % green and less than 2 % oolong(Reference Mukhtar and Ahmad60). Black tea is consumed primarily in Western countries while green tea is drunk mainly in China, Japan, India and a few countries in North Africa and the Middle East(Reference Mukhtar and Ahmad60). Epidemiological and laboratory evidence is emerging to support early ecological observations that tea drinking may be inversely associated with chronic diseases(Reference Cheng61–(Reference Cabrera, Artacho and Gimenez63). Tea contains various potentially protective compounds such as polyphenols, and much of the effect is attributed to catechins in green tea, especially epigallocatechin-3-gallate (EGCG), and theaflavins and thearubigins in black tea(Reference Graham64).
Cancers
Epidemiological evidence. Green and black teas have been studied separately for gastrointestinal cancers. A systematic review on gastric cancer suggested a protective effect of green tea on precursors of adenocarcinoma but no clear epidemiological evidence for a protective role in gastric cancer(Reference Borrelli, Capasso, Russo and Ernst65). Of twelve additional studies identified, nine showed some decrease in risk with consumption of green tea and almost all experimental studies have demonstrated inhibitory effects on gastrointestinal carcinogenesis(Reference Borrelli, Capasso, Russo and Ernst65). From twenty-seven studies, green tea appears to be related to a reduced risk of gastric cancer. A review of gastric cancer and black tea consumption(12) and subsequent studies have mostly reported no association. From nineteen studies, black tea possibly has no relationship with gastric cancer.
Four reviews have examined the relationship between green and/or black tea consumption and risk of colorectal cancer, and concluded there is inconsistent evidence for a protective effect from either tea type(Reference Tavani and La Vecchia28, (Reference Marques-Vidal, Ravasco and Ermelinda Camilo29, (Reference Arab and Il’yasova66, (Reference Suzuki, Tsubono, Nakaya, Koizumi, Shibuya and Tsuji67). However, a systematic review of eight studies on green tea reported an OR of 0·82 (95 % CI 0·69, 0·98)(Reference Sun, Yuan, Koh and Yu68) and other individual studies on green tea have indicated chemopreventive effects(Reference Nakachi, Matsuyama, Miyake, Suganuma and Imai69–(Reference Zhang, Binns and Lee73). From nine studies, green tea consumption is possibly related to a decreased risk of colorectal cancer, while black tea consumption possibly has no relationship with colorectal cancer based on twenty-one studies.
A systematic review of seven studies on green tea and breast cancer incidence reported a pooled RR of 0·89 (95 % CI 0·71, 1·10) for consumption of 5 cups or more per d(Reference Seely, Mills, Wu, Verma and Guyatt74). A meta-analysis of four studies also reported an approximate 20 % risk reduction with high consumption(Reference Sun, Yuan, Koh and Yu68). From nine studies, it is possible that intake of green tea is associated with a decreased risk of breast cancer. One meta-analysis(Reference Sun, Yuan, Koh and Yu75) and the majority of the seventeen studies concerning black tea and breast cancer reported no association, suggesting that black tea may have no relationship with breast cancer risk. Epidemiological studies on tea and ovarian cancer have generated inconsistent results(Reference Lee, Fraser and Binns76). However, a recent case–control and follow-up study reported a protective effect of green tea on both ovarian cancer risk and survival rates among Chinese women(Reference Zhang, Binns and Lee73, (Reference Zhang, Lee, Binns and Xie77). Most other studies conducted in populations drinking black tea showed no significant effect. From eleven studies, black tea possibly has no association with ovarian cancer risk whereas green tea consumption possibly decreases risk. Six reviews addressing tea and prostate cancer agreed that epidemiological evidence was not conclusive, but green tea could possibly afford chemopreventive effects(12, (Reference Lee, Fraser, Meng and Binns62, (Reference Saleem, Adhami, Siddiqui and Mukhtar78–(Reference Siddiqui, Adhami, Saleem and Mukhtar81). Of eleven studies identified, two case–control(Reference Jain, Hislop, Howe, Burch and Ghadirian82, (Reference Jian, Xie, Lee and Binns83) and a cohort study(Reference Heilbrun, Nomura and Stemmermann84) have shown a reduced risk. Therefore, tea consumption can possibly reduce the risk of prostate cancer.
There are several potential explanations for the more promising results for green tea than black tea. These include the higher catechin content of green tea and higher level of antioxidant activity(Reference Graham64), consumption of higher volumes of green tea, as well as confounding factors and the small number of published studies.
Other cancers investigated have shown inconsistent results. From fifteen studies on lung cancer, four reviews(12, (Reference Zeegers, Tan, Goldbohm and van den Brandt23, (Reference Zeegers, Kellen, Buntinx and van den Brandt85, (Reference Leppert, Shvarts, Kawaoka, Lieberman, Belldegrun and Pantuck86) and a total of thirty-nine studies on urinary tract/bladder cancer, and a pooled analysis of nine case–control studies on thyroid cancer(Reference Mack, Preston-Martin, Dal Maso, Galanti, Xiang and Franceschi19), the consumption of tea (black or green) does not appear to lower the risk of lung, urinary tract/bladder or thyroid cancer. Some studies on tea consumption and oesophageal cancer even reported a possible increased risk with the consumption of very hot beverages. From twenty-four studies, tea consumption at normal temperatures probably has no relationship with the risk of oesophageal cancer.
Animal and in vitro evidence. While epidemiological studies have yielded somewhat inconsistent results, quite strong evidence for cancer prevention has emerged from animal and in vitro cell culture studies, especially for green tea(Reference Cabrera, Artacho and Gimenez63, (Reference Chung, Schwartz, Herzog and Yang87, (Reference Yang, Lambert, Hou, Ju, Lu and Hao88). The majority have shown a protective effect at organ sites including the gastrointestinal tract, oral cavity, lung, oesophagus, skin, liver, pancreas, bladder, mammary gland and prostate(Reference Chung, Schwartz, Herzog and Yang87). The antioxidant property of tea polyphenols is the most researched cancer preventive mechanism(Reference Crespy and Williamson89). Antioxidants protect cells against the damaging effects of reactive oxygen species(Reference Crespy and Williamson89). Animal studies have shown that tea catechins increase total plasma antioxidant activity(Reference Yokozawa, Nakagawa and Kitani90, (Reference Skrzydlewska, Ostrowska, Farbiszewski and Michalak91). Tea catechins also increase the activity of several detoxifying and antioxidant enzymes that can metabolise carcinogens in the body into inactive products(Reference Skrzydlewska, Ostrowska, Farbiszewski and Michalak91, (Reference Donovan, Crespy, Manach, Morand, Besson, Scalbert and Rémésy92). Moreover, both EGCG and theaflavins in tea can induce apoptosis(Reference Lu, Lou, Li, Xie, Brash, Huang and Conney93, (Reference Gupta, Hastak, Ahmad, Lewin and Mukhtar94) and cell cycle arrest(Reference Chung, Schwartz, Herzog and Yang87). Other in vitro models have shown that tea catechins can inhibit signal transduction pathways mediated by epidermal growth factor and platelet-derived growth factor, favourably affecting angiogenesis(Reference Wiseman, Mulder and Rietveld95).
CVD
Epidemiological evidence. Two reviews on the relationship between tea consumption and CVD concluded that evidence was inconsistent(Reference Vita96, (Reference Kris-Etherton, Hecker, Bonanome, Coval, Binkoski, Hilpert, Griel and Etherton97). A meta-analysis of ten cohort and seven case–control studies examined stroke, myocardial infarction, CHD and tea(Reference Peters, Poole and Arab98). The results for stroke and CHD were heterogeneous, but the incidence rate of myocardial infarction could decrease by 11 % with an increase in tea consumption of 3 cups/d. Subsequent prospective cohort studies have suggested inverse associations with general CVD(Reference Kuriyama, Shimazu, Ohmori, Kikuchi, Nakaya, Nishino, Tsubono and Tsuji99), mortality(Reference Shimazu, Kuriyama, Hozawa, Ohmori, Sato, Nakaya, Nishino, Tsubono and Tsuji100), myocardial infarction(Reference Geleijnse, Launer, Van der Kuip, Hofman and Witteman101) and important vascular events(Reference Sesso, Gaziano, Liu and Buring102). Evidence has also emerged from a meta-analysis of seven cohort studies of flavonol intake (largely but not exclusively from tea), reporting a reduced risk of CHD mortality with higher consumption(Reference Huxley and Neil103). Although the majority of studies have been conducted in populations drinking black tea, positive findings are emerging for green tea. For example, the two existing studies on green tea and stroke have both shown protective effects(104, (Reference Chen, Li, Zhao, Zhou, Yang, Wang, Guo and Wu105). In addition, tea consumption has been inversely associated with precursors of CVD including atherosclerosis(Reference Geleijnse, Launer, Van der Kuip, Hofman and Witteman101, (Reference Sasazuki, Kodama and Yoshimasu106) and hypertension(Reference Hodgson, Devine, Puddey, Chan, Beilin and Prince107–(Reference Hodgson109).
Other evidence. Considerable evidence exists from short-term human intervention, animal and in vitro studies that tea consumption could reduce the risk of contributing factors for CVD including hypertension, atherosclerosis and thrombogenesis. Although biological mechanisms for these effects are not completely understood, tea has been shown to affect homocysteine, cholesterol, atherogenesis, inhibit LDL-cholesterol oxidation(Reference Trevisanato and Kim110, (Reference Stangl, Lorenz and Stangl111) and improve the vascular epithelium(Reference Cheng61, (Reference Vita96). A review of seventeen human clinical trials reported that tea flavonoids significantly increase the antioxidant capacity of the blood and protect DNA from oxidative damage(Reference Rietveld and Wiseman112). Furthermore, animal evidence for the positive effects of tea is more consistent than human evidence(Reference Zhao and Chen113).
Other conditions
Finally, it has been suggested that tea consumption protects against tooth cavities(Reference Wu and Wei114) and kidney stones(Reference Jeong, Kim, Kim and Kim115, (Reference Itoh, Yasui, Okada, Tozawa, Hayashi and Kohri116), improves immune function(Reference Trevisanato and Kim110), and may play a protective role in osteoarthritis(Reference Fajardo and Di Cesare117) and rheumatoid arthritis(Reference Mikuls, Cerhan, Criswell, Merlino, Mudano, Burma, Folsom and Saag118). However, only preliminary evidence is available so that further research is required.
Discussion
In summary, although coffee has been suspected of having a variety of adverse effects, research has confirmed that moderate consumption is safe. While potential positive effects of coffee on other conditions are being investigated, the evidence is currently weak. Tea, on the other hand, is a safe beverage that potentially plays a role in the prevention of several cancers and CVD. Evidence from epidemiological studies is not yet conclusive but when animal evidence is considered, especially for green tea, a role for tea in chronic disease prevention looks increasingly promising.
Coffee and tea are both consumed in most countries(Reference Grigg2). Worldwide, approximately three cups of tea are drunk for every cup of coffee. Between 1994 and 1996, 76·6 % of tea consumption took place in developing countries with tea drinking dominating throughout Asia, the former Soviet Union and Africa. In contrast, 71·5 % of coffee consumption took place in developed countries(Reference Grigg2). For example, the apparent per capita consumption of coffee by Australians increased from 0·3 kg in 1938–39 to 2·4 kg in the year ending June 1999. Meanwhile, tea consumption decreased from 3·1 kg to 0·9 kg over the same period(119).
World trends indicate that beverage consumption is moving away from the potentially healthier choice of tea towards coffee. Current dietary guidelines around the world make little mention of tea and coffee; consequently, the public is not receiving the best available advice on beverage consumption.
In light of the evidence, we are currently able to state that moderate consumption of both coffee and tea is safe. However, the current evidence on protective effects would not be regarded as strong enough to make recommendations for usage of either coffee or tea if the standards of a therapeutic substance were applied. But what if the question is posed in the way that members of the general public ask nutritionists? ‘Should I drink coffee or tea?’ ‘Which is better for me?’ If nutritionists are asked to recommend one or the other, we believe the evidence supports recommending tea as a healthy beverage and a better choice than coffee.
In considering whether it is appropriate to make guidelines on tea and coffee, principles underlying nutrition guidelines must be taken into account. All proposed health effects of tea are biologically plausible(Reference Binns and Lee4, (Reference McKay and Blumberg120). The recommendation is neither inconsistent with previous public health advice(Reference Binns and Lee4) nor requires any major deviation from the norm(Reference Binns and Lee4, (Reference McKay and Blumberg120). Recommending tea consumption is also conservative, with the near absence of risk, and is compatible with one of the traditional principles of medicine, often ascribed to Hippocrates: ‘First do no harm’.
The main concern about recommending tea and stating coffee as a safe beverage involves the effects of caffeine. In making guidelines, safe limits must be considered. Based on data reviewed by Nawrot et al.(Reference Nawrot, Jordan, Eastwood, Rotstein, Hugenholtz and Feeley58), an intake of caffeine at a dose level up to 400 mg/d (equivalent to 6 mg/kg body weight per d in a person weighing 65 kg) is not associated with any adverse effect in healthy adults. Using average values for caffeine content presented in Table 6(Reference Barone and Roberts121), this is equivalent to approximately 4–5 cups of coffee daily or 13 cups of tea (in the absence of other caffeine-containing items). However, such limits should be adjusted for at-risk groups. In the absence of conclusive data, pregnant women or those planning to become pregnant should limit caffeine intake to less than 300 mg/d (4·6 mg/kg body weight per d in a 65 kg person) and children to an upper intake of 2·5 mg/kg body weight per d(Reference Nawrot, Jordan, Eastwood, Rotstein, Hugenholtz and Feeley58).
Table 6 Standard values for caffeine content of tea and coffee(Reference Barone and Roberts121)
Dietary recommendations are most useful when quantified(12). Although we are able to quantify safe limits for intake of tea and coffee, current evidence is insufficient to quantify levels of tea intake necessary for potential health benefits. Epidemiological studies often do not collect data regarding tea preparation and type of tea consumed. Even though some evidence indicates green tea may have greater benefits than black tea, recommendations regarding type of tea cannot be made. The evidence for tea and coffee does not achieve the NHMRC’s standards for clinical practice guidelines. The impracticality and high cost of conducting long-term RCT means, realistically, that evidence will not reach this level any time in the near future. Evaluation of evidence for therapeutic and preventive measures should follow separate guidelines(Reference Kroke, Boeing, Rossnagel and Willich9). People worldwide will continue to drink tea and coffee, so that professional nutritionists must offer advice on the best available evidence rather than waiting until higher levels of evidence become available.
Conclusion
Tea and coffee are consumed by millions of people around the world every day. However, little dietary advice exists on the consumption of these beverages. Current evidence on the health effects of tea and coffee does not meet the NHMRC’s level for clinical practice guidelines(8). Nevertheless, evidence indicates that coffee is a safe beverage, while tea consumption is possibly protective against several cancers and CVD. Worldwide, consumption of tea is decreasing whereas coffee is increasing. In light of the evidence, nutritionists should advocate tea as part of a healthy diet and as a superior choice to coffee despite the fact that levels of intake cannot currently be recommended. The answer to our title question: we would like tea please.
Acknowledgements
There are no conflicts of interest to be declared by the authors. The funding for the project was provided by Curtin University, Perth, Western Australia.
Authorship responsibilities: All authors contributed significantly to the project. C.W.B. initiated the project, provided references and participated in rating the studies and editing of the manuscript. A.H.L. provided references and participated in rating the studies and editing of the manuscript. M.L.F. undertook literature searches and wrote the first draft of the manuscript.
