American Journal of Epidemiology

This Article Abstract FREE Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Archive Download to citation manager Search for citing articles in: ISI Web of Science (102) Request Permissions Disclaimer Google Scholar Articles by Peters, U. Articles by Arab, L. Search for Related Content PubMed PubMed Citation Articles by Peters, U. Articles by Arab, L. Social Bookmarking          What's this?

American Journal of Epidemiology Vol. 154, No. 6 : 495-503
Copyright © 2001 by The Johns Hopkins University School of Hygiene and Public Health

ORIGINAL CONTRIBUTIONS

Does Tea Affect Cardiovascular Disease? A Meta-Analysis

Ulrike Peters^1,2, Charles Poole¹ and Lenore Arab¹

¹ Department of Epidemiology, University of North Carolina School of Public Health, Chapel Hill, NC.
² Division of Cancer Epidemiology and Genetics, National Cancer Institute, Bethesda, MD.

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX REFERENCES

 
This meta-analysis of tea consumption in relation to stroke, myocardial infarction, and all coronary heart disease is based on 10 cohort studies and seven case-control studies. The study-specific effect estimates for stroke and coronary heart disease were too heterogeneous to be summarized (homogeneity p < 0.02 for stroke, p < 0.001 for coronary heart disease). Only the relative risk estimates for myocardial infarction (seven studies) appeared reasonably homogeneous (homogeneity p = 0.20). The incidence rate of myocardial infarction is estimated to decrease by 11% with an increase in tea consumption of 3 cups per day (fixed-effects relative risk estimate = 0.89, 95% confidence interval: 0.79, 1.01) (1 cup = 237 ml). However, evidence of bias toward preferential publication of smaller studies that suggest protective effects urges caution in interpreting this result. The geographic region where the studies were conducted appeared to explain much of the heterogeneity among coronary heart disease, myocardial infarction, and probably stroke results. With increasing tea consumption, the risk increased for coronary heart disease in the United Kingdom and for stroke in Australia, whereas the risk decreased in other regions, particularly in continental Europe.

cardiovascular diseases; cerebrovascular accident; coronary disease; meta-analysis; myocardial infarction; tea

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX REFERENCES

 
Tea is the second most common drink in the world. Because of the high consumption of tea, even small effects in persons could have a large impact on public health. The polyphenolic flavonoids in tea are thought to have a protective effect on cardiovascular disease. Oxidized low density lipoproteins occur in atherosclerotic plaques in the vascular system and heart (1). Flavonoids have antioxidative properties that prevent oxidation of low density lipoproteins in vitro and, as recently shown, in vivo as well (2–6). High concentrations of autoantibodies against oxidized low density lipoproteins have been found in patients with atherosclerosis (7, 8). Nevertheless, the relevance of these laboratory findings for a hypothetically protective effect of tea on cardiovascular disease remains unknown. Therefore, the goal of the present study was to analyze the estimated effect of tea consumption on cardiovascular disease in all published epidemiologic studies.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX REFERENCES

 
To search for observational studies of tea consumption in relation to cardiovascular disease, we conducted a literature search in MEDLINE for papers published from January 1, 1966, to October 14, 2000. The keywords for the search were tea, coffee, caffeine, or flavonoids together with one of the following outcomes: cardiovascular disease, coronary heart disease, stroke, myocardial infarction, ischemic heart disease, cerebrovascular disease, or cardiovascular risk. We included references listed in the recovered articles as well as in review articles. We examined all observational studies directly, because some studies did not report results for tea and cardiovascular disease as the main result and did not mention these results in the abstracts of the articles.

Four reports on three early case-control studies (9–11) were excluded because they did not include sufficient information to compute an estimate of relative risk or its standard error. (Throughout this paper, the term "relative risk" is used for incidence rate ratios estimated directly in cohort studies and by the exposure odds ratio in case-control studies.) In addition, two cross-sectional studies that measured serum levels and included undiagnosed cases (12, 13) were excluded. However, data from the latter cohort (13) were included as drawn from a separate publication (14). A report by Klatsky et al. (15) was replaced by a more recent analysis of the same cohort (16). In the report by Sato et al. (17), results from the prospective part of the study were used. We contacted the authors of 11 studies (14, 16–25) for information not included in their papers, such as variance estimates, numbers of subjects in categories of tea consumption, and typical tea cup sizes in specific locales. Nine authors responded to our request (see Acknowledgments).

The studies addressed a diversity of cardiovascular diseases (International Classification of Diseases, Ninth Revision, codes 390–459), which were combined in different ways (appendix table 1). We included the three outcome categories (myocardial infarction, stroke, and the broader category of coronary heart disease), which were examined in at least three studies. If a study provided more than one effect estimate for a single outcome category (14, 16), only one estimate was used.

The following paragraphs in this section pertain to statistical methods. The statistical analyses included 1) extracting or computing a common, comparable relative risk estimate from each study; 2) searching for evidence of publication bias; 3) analyzing interstudy variation; and 4) computing summary effect estimates as indicated. Evidence of publication bias and heterogeneity was taken to contraindicate reliance on overall summaries (23). In the presence of heterogeneity of effect, the main purpose of the meta-analysis became the identification of sources of interstudy variation.

We attempted to place the studies on a common footing by estimating the relative risk of a 3 cups per day increase in tea consumption (e.g., from no tea consumption to 3 cups per day) (1 cup = 237 ml). For studies reporting relative risk estimates in categories of tea consumption, we used inverse variance-weighted categorical regression to estimate linear exposure-response curves. These analyses were conducted on the incidence rate scale for cohort studies and on the logit scale for case-control studies. We used the covariance-corrected method of Greenland and Longnecker (26) for studies that provided the relative risk of tea consumption for more than two categories of tea consumption and provided the person-time per number of all subjects and cases per category of tea consumption (18, 20, 21, 25, 27–30). For other studies (14, 17, 19, 22, 23, 33, 34), we used the method described by Berlin et al. (31) and Greeland (32) or the relative risk for tea consumption as continuous (16, 24).

We assigned exposure values, in cups per day, to the tea consumption categories in the original studies as follows. For case-control studies, we used the tea consumption of the control groups. The various measures of tea consumption (cups, grams, milliliters) were transformed to a common measure of cups per day (1 cup = 8 ounces = 237 ml). When category medians (14, 20) or means (29, 33) were available, they were used. Category midranges were applied for the remaining closed-ended categories if they were no broader than 2 cups per day. If the highest, open-ended category included no more than 20 percent of the study subjects, we assigned that category a value equal to 1.2 times its lower boundary (18, 21–23, 25, 27). For the study by Stensvold et al. (19), more detailed data from an earlier report (35) on the same cohort were used to assign values to the tea consumption categories. Some of the closed-ended categories were wider than 2 cups per day in the studies by Rosenberg et al. (30), Jick et al. (34), and the Boston Collaborative Drug Surveillance Program (28). For these studies, we assigned category medians from the National Health and Nutrition Examination Survey I Epidemiologic Followup Study (36). In the Japanese study by Sato et al. (17), 41 percent of the subjects were in the highest, open-ended category. For this study, we used category midranges for the closed-ended categories. We used the factor 1.4 times the lower boundary of 7 cups per day for the highest, open-ended category instead of 1.2, because we expected the distribution to be skewed to the right, given the large proportion of people in the highest category.

The log-rank test of Begg and Mazumdar (37) was used for evidence of publication bias, with low p values suggesting the presence of bias. In addition, funnel graphs in which the study-specific effect estimates are displayed in relation to the reciprocals of their estimated variances were created (38). In the absence of publication bias, these graphs resemble a funnel with the estimates from the larger studies in the center, flanked symmetrically on either side by the less precise estimates.

To explore sources of heterogeneity among the studies, we performed stratified and meta-regression analyses (32, 39). The meta-regressions were fit using inverse variance-weighted, linear regression. The dependent variable was the log relative risk, and the independent variables were the study characteristics suspected of being sources of heterogeneity. After transformation back to the original ratio scale, the meta-regressions estimate the ratio of the average relative risk estimates reported by studies with one characteristic to the average estimates reported by studies with another characteristic. In this way, these models quantify the degree to which characteristics of the studies are associated with their results. The fit of the meta-regression models was checked by calculating the residual sum of squares (32).

We examined the following study characteristics in the stratified and meta-regression analyses: study design (cohort, case-control); mortality or morbidity data; geographic region (United States, United Kingdom, continental Europe, Asia, Australia); gender; adjustment for potential confounders by modeling, restriction, or stratification (sex and/or age, socio-economic status, smoking, other nondietary risk factors for cardiovascular disease including alcohol, dietary factors); publication year (before or during 1980, 1981–1990, during or after 1991); age of subjects (<50 years, >=50 years, all ages); participation rate (among the controls in case-control studies if separately reported); frequency of tea drinking in the study population (<50 percent, 50 percent drinking 1cup per day); and years of follow-up (cohort studies only). We were not able to investigate whether differences in dietary assessment methods accounted for heterogeneity because of the lack of description of the dietary assessment methods applied. Study characteristics were initially examined one by one, because of the lack of power to run them jointly due to the small number of studies. An attempt was made to examine jointly those characteristics that appeared in these analyses to be appreciably associated with the study-specific effect estimates.

To examine the effect of differences in the strength of tea in different regions quantitatively, we multiplied the assigned categorical dose by 0.5 and recalculated the summarized risk estimate for studies conducted in the United States on coronary heat disease or myocardial infarction. According to this calculation, we assume that the tea in the United States is half as strong as that in Europe. All analyses were conducted with Statistical Analysis System (SAS) software (40).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX REFERENCES

 
We identified and included 10 cohort studies and seven case-control studies (table 1, Appendix). Most studies suggested a decrease in the rate of cardiovascular disease outcomes with increasing tea consumption. Results from coronary heart disease or myocardial infarction are shown in figure 1. Two studies from the United Kingdom (14, 27) and two studies from the United States (21, 23) indicated an increased risk with increasing tea consumption, whereas the other studies indicated a decrease in risk. In table 1, the risk estimates were standardized for measuring the effects per 3 cups of tea per day. In the four studies showing an increased risk of coronary heart disease or myocardial infarction, the risk increased by 4–126 percent with each 3 cups/day (14, 21, 23, 27). In the other studies on coronary heart disease or myocardial infarction, the risk decreased between 1 and 75 percent per each 3 cups/day. For stroke, one of six studies showed an increased risk of 51 percent with each 3 cups/day (25), whereas the other studies indicated a decrease in risk of 26–66 percent with each increment of 3 cups per day.

View this table: [in this window] [in a new window]   TABLE 1. Overview and reanalysis of 17 observational epidemiologic studies of the effect of tea consumption on cardiovascular diseases

 

View larger version (20K): [in this window] [in a new window]   FIGURE 1. Relative risks of myocardial infarction or coronary heart disease with tea consumption as a categorical variable in observational epidemiologic studies (categorical values are shown as provided in the studies). One cup = 237 ml. *, the categorical risk estimates calculated from the continuous variable provided. , BCDSP, Boston Collaborative Drug Surveillance Program, which also included subjects from Canada, New Zealand, and Israel.

 
The analyses of publication bias are summarized in figure 2. The small numbers of studies limit the interpretation, but the funnel graphs and the Begg-Mazumdar test both suggest the presence of publication bias for myocardial infarction and particularly for stroke. Specifically, it appears that smaller studies producing results inconsistent with the hypothesis of a preventive effect for myocardial infarction may have a very low probability of becoming published.

View larger version (22K): [in this window] [in a new window]   FIGURE 2. Funnel graphs of the relative risk for each cup of tea per day (beta coefficients) by the fixed-effects weights of a total of 17 observational epidemiologic studies of different cardiovascular diseases including myocardial infarction, coronary heart disease, or stroke. SE, standard error; 1 cup = 237 ml.

 
The results for cardiovascular disease categories were very heterogeneous (p < 0.02). The characteristic most strongly associated with the study-specific effect estimates was geographic region (table 2). Studies on coronary heart disease and myocardial infarction conducted in continental Europe reported much stronger inverse associations than did studies conducted elsewhere. On average, the relative risk estimates from continental Europe were about one third to one fourth of the magnitude of the estimates from the United States, with the exception of the two studies (14, 27) in the United Kingdom that reported a moderately strong positive association between tea intake and coronary heart disease. Neither the basic study design nor any other characteristic of the studies mentioned appeared to explain the heterogeneity of their results for coronary heart disease or myocardial infarction to any appreciable degree.

View this table: [in this window] [in a new window]   TABLE 2. Estimated effect of the covariates' region and study design on the risk estimates of tea consumption on cardiovascular diseases in observational epidemiologic studies (results of a multiple linear meta-regression)

 
For studies on stroke, the follow-up time in cohort studies appeared to explain the heterogeneity among the studies. With each year of follow-up time the risk estimates decreased by 5 percent. In addition, the geographic region or study design explained heterogeneity. Because the one study from Australia showing a very different effect (increasing risk with increasing tea consumption) from the other studies from other geographic regions is also the only case-control study, it is not possible to determine if the geographic region or the study design explained heterogeneity.

The test for heterogeneity of the effect of tea on myocardial infarction showed no strong evidence of heterogeneity (p = 0.20) (table 3). The fixed-effects summary suggests a decrease in incidence of myocardial infarction of 11 percent associated with an increment of 3 cups of tea per day (summary relative risk = 0.89, 95 percent confidence interval: 0.79, 1.01) (table 3). The evidence of publication bias, however, urges caution in interpreting this result.

View this table: [in this window] [in a new window]   TABLE 3. Stratified effect estimates of tea consumption on cardiovascular diseases in observational epidemiologic studies

 
Results stratified by region and study design are shown in table 3. For the broader outcome in which myocardial infarction is considered together with coronary heart disease, the results from the three studies in continental Europe are homogeneous (p = 0.95) and suggest that drinking an additional 3 cups of tea per day can reduce incidence by two thirds, an immense preventive effect. The eight studies conducted in the United States were less homo-geneous (p = 0. 30) with a beneficial effect, if any, being a reduction in incidence of only about 5 percent for each 3 cups/day. An increasing risk with increasing tea consumption was indicated for the two studies from the United Kingdom on coronary heart disease. These estimates also did not indicate strong evidence of heterogeneity (p = 0.30) with an increase in risk of 62 percent with each increment of 3 cups/day.

Additionally, we calculated the summarized risk estimate for studies conducted in the United States on coronary heart disease or myocardial infarction, assuming that tea in the United States is half as strong as that in Europe. In this case, the summarized risk estimate for studies conducted in the United States for coronary heart disease or myocardial infarction is 0.90 (95 percent confidence interval: 0.81, 1.01) with each increment of 3 cups/day rather than 0.95 (table 3).

The geographic stratified risk estimates of coronary heart disease alone were fairly similar to the risk estimates of coronary heart disease or myocardial infarction combined, although the results of the three studies from the United States were too heterogeneous (p = 0.16) to be summarized.

For myocardial infarction, the effect estimates from studies in the United States were reasonably homogeneous and suggested little or no effect. The fixed-effects summary from these six studies from the United States was almost identical to the estimate from the one study from Italy that was included (relative risk = 0.89 vs. relative risk = 0.91).

For stroke, the only case-control study conducted in Australia indicated a nonsignificant increased risk of 51 percent with each 3 cups/day (25). Cohort studies conducted in the United States, continental Europe, or Asia did not indicate strong evidence of heterogeneity (p = 0.21) with a significant reduction in stroke incidence of 12 percent per 3 cups of tea/day. The protective effect of tea on stroke increased in cohort studies by 5 percent with each year of follow-up (4–15 years).

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX REFERENCES

 
The main purpose of this study was to examine the published literature on tea consumption and cardiovascular disease for evidence of publication bias and heterogeneity of effect and to summarize the effect estimates as indicated. Evidence of publication bias in the literature was apparent for myocardial infarction and particularly for stroke. The relative risk estimates for studies of myocardial infarction appeared homogeneous overall, although publication bias urges caution in interpreting this result. For coronary heart disease or myocardial infarction other than geographic region, no characteristic of the studies appeared to explain the heterogeneity or to be strongly associated with the study-specific results. Three studies from continental Europe were consistent in suggesting a strong preventive effect, whereas eight studies from the United States were consistent in suggesting little or no effect for coronary heart disease or myocardial infarction. Heterogeneity of risk estimates for stroke could be explained by length of follow-up together with study design and/or geographic region. Cohort studies not conducted in Australia indicated a decreased risk of stroke with increasing tea consumption. However, evidence of publication bias suggests extreme caution in interpretation of these data.

The small number of published studies for any specific cardiovascular disease outcome severely limited the ability to detect publication bias or heterogeneity. Tests of homogeneity are well known to have low power. Begg and Mazumdar (37) stated that their test for publication bias has moderate power with 25 studies and high power with 75 studies. It is impressive, therefore, that the p values from the tests for heterogeneity and publication bias were so low for stroke, despite the small number of studies in the published literature on the topic at hand. It would be of considerable interest to learn if epidemiologic researchers have examined results for tea and myocardial infarction and stroke and refrained from publishing them because they were not indicative of pronounced preventive effects. Only one of many possibilities is that publication bias on this question has been stronger in Europe, where more people drink tea, than in the United States.

If studies with positive associations have not been published, as suggested by the tests of publication bias and funnel graphs, the central tendency would be closer to the null value. One cannot predict what the stratified and meta-regression analysis of study characteristics would show if a sizable number of unpublished studies were brought to light. The evidence of publication bias did not disprove the hypothesis of a protective effect, but it tempered the strengths of the conclusion regarding preventive potential.

The outcomes investigated in the studies had different definitions and were combined in more or less broad categories. If tea has different effects on different aspects of cardiovascular diseases, a combined estimate from various outcomes may minimize a tea effect. The results of the pres-ent meta-analysis suggest this possibility because the study results were less heterogeneous for myocardial infarction alone than for any less specific outcome. These results suggest that a more precise definition of the outcome might improve the homogeneity among studies and the precision in the effect estimate.

Case-control studies tend to have a higher potential for recall and selection bias. Nevertheless, we saw little or no difference in results between cohort studies and case-control studies for coronary heart disease or myocardial infarction. For stroke, it remains unclear if the differences were due to regional differences or differences in study design. The low number of studies limited the power to detect differences among the studies. If we expected the differences to be small, as possibly for differences in adjustment for confounding, we have limited ability to detect small differences in the meta-analysis.

For cohort studies on stroke, we found an increased protective effect with increasing length of follow-up between 4 and 15 years, which might indicate the importance of early prevention.

The findings from the United Kingdom and Australia of a positive association were unique. One possible explanation involves the polyphenolic antioxidant flavonoids hypothesized to be one mechanism by which tea might reduce cardiovascular disease incidence (1–7). In the United Kingdom, milk commonly is added to tea. Hertog et al. (27) reported that more than 99 percent of tea drinkers added milk and argued that this difference might explain why they did not find evidence of a protective effect of tea. Hasalam (41) showed that flavonoids are bound to protein. Further, indirect evidence suggested by Serafini et al. (42) showed that adding milk to tea abolished its in vivo plasma antioxidant potential. In contrast to these findings, however, Hollman et al. (43) and van het Hof et al. (44) did not find different flavonoid plasma concentrations in subjects given tea with or without milk. The hypothesis has been stated for the findings from the United Kingdom studies on coronary heart disease. This is also a possible explanation for the findings for stroke in Australia, a population that is strongly influenced by immigrants from the United Kingdom. In any event, the hypothesis of Hertog et al. (27) and Hasalam (41) might explain why an inverse association would not be seen in the United Kingdom and Australia, but that hypothesis would not explain why a positive association was seen. The amount of fat in the milk that is added to the tea would seem woefully insufficient to increase cardiovascular disease risk.

Sesso et al. (20) suggested that higher tea consumption might be a surrogate for a healthier lifestyle. Weak inverse associations of tea consumption with smoking, body mass index, and dietary risk factors have been reported (16, 19, 29). Residual confounding and lack of control for lifestyle factors might explain why some studies appear to suggest a protective effect of tea on cardiovascular disease. These problems may also explain regional differences. In the United Kingdom and in continental Europe, tea consumption is very common and therefore may not be restricted to people with healthier behaviors. Perhaps residual confounding and lack of control for lifestyle factors are especially pronounced in the United States, where fewer people drink tea and where weaker associations between tea and cardiovascular disease have been reported.

An important limitation of the studies was imprecision of the exposure measurement. Only the study from Japan investigated green tea. All other studies referred simply to tea. Some studies mention that the subjects were asked only about the frequency of tea consumption without any more detailed questions about the kind or preparation of tea. Tea comprises a heterogeneous group of beverages, including fermented black tea, half fermented oolongs, unfermented green tea, and sweetened or unsweetened ice tea, and it might even be understood to include fruit tea or herbal teas. It is to be expected that study subjects give a summary answer for any kind of tea if they are asked only about their tea consumption without more detailed questions. Different kinds of tea differ in the kind and quantity of substances and, even within the same kind of tea, differences exist. According to the analysis of Prior and Cao (45), the phenol content and antioxidant capacity of black, green, and herbal or berry teas can vary more than twofold. The mean total phenol content of black tea is 129.3 mg/g, of green tea, 71.7 mg/g, and of herbal/berry tea, 51.7 mg/g. In addition, the method of preparation affects the content of tea.

The information available was insufficient for us to address the kinds of tea, the methods of preparation, or the differences in tea strength. These factors might help to explain the regional differences we found, however. For instance, it might be expected that the kind of tea, method of preparation, and preference of tea strength differ among the regions and that there might be more similarity within a region than between regions. It is likely that the varying characteristics of tea have different effects on cardiovascular disease. If, for instance, Europeans tend to drink stronger tea than North Americans do, the effect per cup of tea could be higher in the European studies. We examined this hypothesis by recalculating the summarized risk estimate for coronary heart disease or myocardial infarction in the way that the tea in the United States is assumed to be only half as strong as that in Europe. In this case, the risk estimate for studies conducted in the United States decreased only from 0.95 to 0.90 with each 3 cups/day and is still very different from the summarized risk estimate in continental Europe of 0.27. It appears, therefore, that differences in tea strength may explain only a small fraction of the regional differences.

Because of the high consumption and distribution of tea worldwide, hypothetical health effects of tea are important public health issues. It appears worthwhile to address the regional differences in future research, while improving control for potential confounders and measurement of the many characteristics of tea and its preparation and consumption. Of greatest and most immediate importance would be for all investigators who have unpublished results on tea and cardiovascular disease to bring those results forward.

	APPENDIX

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX REFERENCES

 

APPENDIX TABLE 1. Summary of adjustment of potential confounder and description of the outcome as used in 17 observational epidemiologic studies of the effect of tea consumption on cardiovascular diseases

Study Outcome in meta-analysis Disease outcome Description of outcome (ICD-9* code) Adjustment for potential confounder Hirvonen et al., 2000 (22) Stroke Fatal and nonfatal stroke incidence 430–431, 433–434 1, 2, 3, 11, 12, 15, 17, 21, 22, 23, 24 Yochum et al., 1999 (24) CHD* Mortality from CHD 410–414, 429.2 1, 2, 3, 6, 11, 12, 13, 15, 17, 23, 24, 25, 27, 28, 29, 31 Stroke Mortality from stroke 430–438 Woodward and Tunstall-Pedoe, 1999 (14) CHD Mortality from CHD ICD code not specified 1, 2, 5, 8, 11, 12, 13, 21, 22, 23, 24, 30, 37 Hertog et al., 1997 (27) CHD Mortality IHD* ICD code not specified 1, 5, 11, 12, 15, 21, 22, 24, 25, 26, 30, 31, 32 Rimm et al., 1996 (23) CHD Mortality from CHD ICD code not specified 1, 2, 3, 11, 12, 16, 17, 18, 22, 24, 27, 29, 31 Keli et al., 1996 (33) Stroke Fatal and nonfatal stroke incidence 430–438 1, 11, 21, 22, 24, 25, 33, 36 Hertog et al., 1993 (29) CHD Mortality from CHD 410–414 1, 11, 12, 13, 15, 21, 22, 23, 24, 25, 27, 28, 29, 30, 31, 32, 37 Klatsky et al., 1993 (16) CHD Mortality from CHD ICD code not specified, chronic coronary 1, 2, 3, 4, 6, 11, 12, 24 MI* Mortality from acute MI ICD code not specified Stroke Mortality from stroke 430–434 Stensvold et al., 1992 (19) CHD Mortality from CHD 410–413, 414.0–414.1, 414.9, 798.1–798.2 1, 11, 21, 22 Sato et al., 1989 (17) Stroke Mortality from stroke 430–438 1, 2, 10, 11, 24, 34 Sesso et al., 1999 (20) MI Nonfatal MI ICD code not specified, confirmed by evidence of creatine kinase 1, 2, 8, 11, 12, 13, 14, 15, 18, 19, 24, 25, 27 Thrift et al., 1996 (25) Stroke Nonfatal cerebral hemorrhage ICD code not specified, based on discharge diagnosis 1, 2, 5, 7, 11, 12, 15, 22, 24 Gramenzi et al., 1990 (18) MI Nonfatal MI ICD code not specified 1 Rosenberg et al., 1988 (21) MI Nonfatal MI ICD code not specified, based on discharge diagnosis 1, 2, 3, 7, 8, 9, 10, 11, 12, 13, 15, 17, 18, 20, 22, 24, 37 Rosenberg et al., 1980 (30) MI Nonfatal MI ICD code not specified, based on discharge diagnosis Jick et al., 1973 (34) MI Nonfatal MI ICD code not specified, based on discharge diagnosis BCDSP,* 1972 (28) MI Nonfatal MI ICD code not specified, based on discharge diagnosis 1, 2

^* ICD-9, International Classification of Diseases, Ninth Revision; CHD, coronary heart disease; IHD, ischemic heart disease; MI, myocardial infarction; BCDSP, Boston Collaborative Drug Surveillance Program.

1, age; 2, gender; 3, education/profession; 4, race; 5, poverty index/social class; 6, marital status; 7, religion; 8, Framingham type A personality score/Bortner personality score; 9, year of interview; 10, geographic area; 11, smoking; 12, body mass index; 13, physical activity; 14, aspirin use; 15, history of myocardial infarction, coronary heart disease, hypertension; 16, other baseline disease; 17, diabetes; 18, family history of myocardial infarction or coronary heart disease; 19, family history of diabetes; 20, number of visits of physician in the previous year; 21, systolic and/or diastolic blood pressure; 22, serum cholesterol; 23, serum high density lipoprotein; 24, alcohol; 25, calories/energy; 26, fat; 27, saturated fat; 28, cholesterol; 29, dietary fiber/whole grain; 30, vitamin C; 31, vitamin E; 32, beta-carotene; 33, antioxidant vitamins; 34, salt intake; 35, calcium; 36, fish; 37, coffee.

Numbers in parentheses, reference citation.

	ACKNOWLEDGMENTS

 
The authors thank Dr. Arthur L. Klatsky, Mary A. Armstrong, Dr. Howard D. Sesso, Dr. C. La Vecchia, Dr. I. Stensvold, Dr. Eric Rimm, Dr. Aaron Folsom, Dr. Amanda Thrift, and Dr. Mark Woodward for giving us information about their studies in addition to their published papers. Further, they want to thank Dr. Cande Ananth from Robert Wood Johnson Medical School, New Brunswick, New Jersey, as well as Dr. Pietro Ferrari from the International Agency for Research on Cancer, Lyon, France, for offering them a Statistical Analysis System macro to accomplish the covariance-corrected categorical regression method of Greenland and Longnecker (26).

	NOTES

 
Correspondence to Dr. Lenore Arab, Department of Epidemiology (CB 7400), University of North Carolina School of Public Health, Chapel Hill, NC 27599-7400 (e-mail: Lenore{at}unc.edu).

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX REFERENCES

 

1. Shaikh M, Martini S, Quiney JR, et al. Modified plasma-derived lipoproteins in human atherosclerotic plaques. Atherosclerosis 1988;69:165–72.[Web of Science][Medline]
2. de Whalley CV, Rankin SM, Hoult JR, et al. Flavonoids inhibit the oxidative modification of low density lipoproteins by macrophages. Biochem Pharmacol 1990;39:1743–50.[Web of Science][Medline]
3. Negre-Salvayre A, Salvayre R. Quercetin prevents the cytotoxicity of oxidized LDL on lymphoid cell lines. Free Radic Biol Med 1992;12:101–6.[Web of Science][Medline]
4. Rice-Evans CA, Miller NJ, Paganga G. Structure-antioxidant activity relationships of flavonoids and phenolic acids. Free Radic Biol Med 1996;20:933–56.[Web of Science][Medline]
5. Wu TW, Fung KP, Wu J, et al. Morin hydrate inhibits azo-initiator induced oxidation of human low density lipoprotein. Life Sci 1996;58:PL 17–22.[Web of Science]
6. Yoshida H, Ishikawa T, Hosoai H, et al. Inhibitory effect of tea flavonoids on the ability of cells to oxidize low density lipo-protein. Biochem Pharmacol 1999;58:1695–703.[Web of Science][Medline]
7. Salonen JT, Yla-Herttuala S, Yamamoto R, et al. Autoantibody against oxidised LDL and progression of carotid atherosclerosis. Lancet 1992;339:883–7.[Web of Science][Medline]
8. Bergmark C, Wu R, de Faire U, et al. Patients with early-onset peripheral vascular disease have increased levels of autoantibodies against oxidized LDL. Arterioscler Thromb Vasc Biol 1995;15:441–5.[Abstract/Free Full Text]
9. Yudin J, Morland J. Sugar intake and myocardial infarction. Am J Clin Nutr 1967;20:503–6.[Abstract]
10. Little J, Shanoff H, Csima A, et al. Diet and serum lipids in male survivors of myocardial infarction. Lancet 1965;1:933–5.[Medline]
11. Howell RW, Wilson DG. Dietary sugar and ischaemic heart disease. Br Med J 1969;3:145–8.
12. Imai K, Nakachi K. Cross sectional study of effects of drinking green tea on cardiovascular and liver diseases. BMJ 1995;310:693–6.[Abstract/Free Full Text]
13. Brown CA, Bolton-Smith C, Woodward M, et al. Coffee and tea consumption and the prevalence of coronary heart disease in men and women: results from the Scottish Heart Health Study. J Epidemiol Community Health 1993;47:171–5.[Abstract/Free Full Text]
14. Woodward M, Tunstall-Pedoe H. Coffee and tea consumption in the Scottish Heart Health Study follow up: conflicting relations with coronary risk factors, coronary disease, and all cause mortality. J Epidemiol Community Health 1999;53:481–7.[Abstract]
15. Klatsky AL, Friedman GD, Armstrong MA. Coffee use prior to myocardial infarction restudied: heavier intake may increase the risk. Am J Epidemiol 1990;132:479–88.[Abstract/Free Full Text]
16. Klatsky AL, Armstrong MA, Friedman GD. Coffee, tea, and mortality. Ann Epidemiol 1993;3:375–81.[Medline]
17. Sato Y, Nakatsuka H, Watanabe T, et al. Possible contribution of green tea drinking habits to the prevention of stroke. Tohoku J Exp Med 1989;157:337–43.[Web of Science][Medline]
18. Gramenzi A, Gentile A, Fasoli M, et al. Association between certain foods and risk of acute myocardial infarction in women. BMJ 1990;300:771–3.
19. Stensvold I, Tverdal A, Solvoll K, et al. Tea consumption: relationship to cholesterol, blood pressure, and coronary and total mortality. Prev Med 1992;21:546–53.[Web of Science][Medline]
20. Sesso HD, Gaziano JM, Buring JE, et al. Coffee and tea intake and the risk of myocardial infarction. Am J Epidemiol 1999;149:162–7.[Abstract/Free Full Text]
21. Rosenberg L, Palmer JR, Kelly JP, et al. Coffee drinking and nonfatal myocardial infarction in men under 55 years of age. Am J Epidemiol 1988;128:570–8.[Abstract/Free Full Text]
22. Hirvonen T, Virtamo J, Korhonen P, et al. Intake of flavonoids, carotenoids, vitamins C and E, and risk of stroke in male smokers. Stroke 2000;31:2301–6.[Abstract/Free Full Text]
23. Rimm EB, Katan MB, Ascherio A, et al. Relation between intake of flavonoids and risk for coronary heart disease in male health professionals. Ann Intern Med 1996;125:384–9.[Abstract/Free Full Text]
24. Yochum L, Kushi LH, Meyer K, et al. Dietary flavonoid intake and risk of cardiovascular disease in postmenopausal women. Am J Epidemiol 1999;149:943–9.[Abstract/Free Full Text]
25. Thrift AG, McNeil JJ, Forbes A, et al. Risk factors for cerebral hemorrhage in the era of well-controlled hypertension. Melbourne Risk Factor Study (MERFS) Group. Stroke 1996;27:2020–5.[Abstract/Free Full Text]
26. Greenland S, Longnecker MP. Methods for trend estimation from summarized dose-response data, with applications to meta-analysis. Am J Epidemiol 1992;135:1301–9.[Abstract/Free Full Text]
27. Hertog MG, Sweetnam PM, Fehily AM, et al. Antioxidant flavonols and ischemic heart disease in a Welsh population of men: the Caerphilly Study. Am J Clin Nutr 1997;65:1489–94.[Abstract/Free Full Text]
28. Coffee drinking and acute myocardial infarction. Report from the Boston Collaborative Drug Surveillance Program. Lancet 1972;2:1278–81.[Medline]
29. Hertog MG, Feskens EJ, Hollman PC, et al. Dietary antioxidant flavonoids and risk of coronary heart disease: the Zutphen Elderly Study. Lancet 1993;342:1007–11.[Web of Science][Medline]
30. Rosenberg L, Slone D, Shapiro S, et al. Coffee drinking and myocardial infarction in young women. Am J Epidemiol 1980;111:675–81.[Abstract/Free Full Text]
31. Berlin JA, Longnecker MP, Greenland S. Meta-analysis of epidemiologic dose-response data. Epidemiology 1993;4:218–28.[Web of Science][Medline]
32. Greenland S. Meta-analysis. 1998;2:643–74.
33. Keli SO, Hertog MG, Feskens EJ, et al. Dietary flavonoids, antioxidant vitamins, and incidence of stroke: the Zutphen Study. Arch Intern Med 1996;156:637–42.[Abstract/Free Full Text]
34. Jick H, Miettinen OS, Neff RK, et al. Coffee and myocardial infarction. N Engl J Med 1973;289:63–7.
35. Solvoll K, Selmer R, Loken EB, et al. Coffee, dietary habits, and serum cholesterol among men and women 35–49 years of age. Am J Epidemiol 1989;129:1277–88.[Abstract/Free Full Text]
36. US Department of Health and Nutrition (DHHS). Second National Health and Nutrition Examination Survey I Epidemiologic Followup Study, 1982–1984. Hyattsville, MD: National Center for Health Statistics, 1992.
37. Begg CB, Mazumdar M. Operating characteristics of a rank correlation test for publication bias. Biometrics 1994;50:1088–101.[Web of Science][Medline]
38. Light R, Mazumdar M. Summing up: the science of reviewing research. Cambridge, MA: Harvard University Press, 1984:63–72.
39. Wallenstein S, Bodian C. Epidemiologic programs for computers and calculators. Inferences on odds ratios, relative risks, and risk differences based on standard regression programs. Am J Epidemiol 1987;126:346–55.[Abstract/Free Full Text]
40. SAS Institute, Inc. SAS v8.0. Cary, NC: SAS Institute, Inc, 1999.
41. Hasalam E. Plant phenols: vegetable tannins revisited. Cambridge, United Kingdom: Cambrige University Press, 1989:154–219.
42. Serafini M, Ghiselli A, Ferro-Luzzi A. In vivo antioxidant effect of green and black tea in man. Eur J Clin Nutr 1996;50:28–32.[Web of Science][Medline]
43. Hollman PC, Feskens EJ, Katan MB. Tea flavonols in cardiovascular disease and cancer epidemiology. Proc Soc Exp Biol Med 1999;220:198–202.[Medline]
44. van het Hof KH, Wiseman SA, Yang CS, et al. Plasma and lipoprotein levels of tea catechins following repeated tea consumption. Proc Soc Exp Biol Med 1999;220:203–9.[Medline]
45. Prior RL, Cao G. Antioxidant capacity and polyphenolic components of teas: implications for altering in vivo antioxidant status. Proc Soc Exp Biol Med 1999;220:255–61.[Medline]

Received for publication March 27, 2000. Accepted for publication February 14, 2001.

CiteULike    Connotea    Del.icio.us    What's this?

This article has been cited by other articles:

				L. Arab, W. Liu, and D. Elashoff Green and Black Tea Consumption and Risk of Stroke: A Meta-Analysis Stroke, May 1, 2009; 40(5): 1786 - 1792. [Abstract] [Full Text] [PDF]

				L. Arab and J. B. Blumberg Proceedings of the Fourth International Scientific Symposium on Tea and Human Health. September 18, 2007. Washington, DC, USA. J. Nutr., August 1, 2008; 138(8): 1526S - 1588S. [Full Text] [PDF]

				D. Grassi, A. Aggio, L. Onori, G. Croce, S. Tiberti, C. Ferri, L. Ferri, and G. Desideri Tea, Flavonoids, and Nitric Oxide-Mediated Vascular Reactivity J. Nutr., August 1, 2008; 138(8): 1554S - 1560S. [Abstract] [Full Text] [PDF]

				I. C. W. Arts A Review of the Epidemiological Evidence on Tea, Flavonoids, and Lung Cancer J. Nutr., August 1, 2008; 138(8): 1561S - 1566S. [Abstract] [Full Text] [PDF]

				C. Angeloni, E. Leoncini, M. Malaguti, S. Angelini, P. Hrelia, and S. Hrelia Role of quercetin in modulating rat cardiomyocyte gene expression profile Am J Physiol Heart Circ Physiol, March 1, 2008; 294(3): H1233 - H1243. [Abstract] [Full Text] [PDF]

				J. O. Clarke and G. E. Mullin A Review of Complementary and Alternative Approaches to Immunomodulation Nutr Clin Pract, February 1, 2008; 23(1): 49 - 62. [Abstract] [Full Text] [PDF]

				S. Debette, D. Courbon, N. Leone, J. Gariepy, C. Tzourio, J.-F. Dartigues, P. Barberger-Gateau, K. Ritchie, A. Alperovitch, P. Amouyel, et al. Tea Consumption Is Inversely Associated With Carotid Plaques in Women Arterioscler Thromb Vasc Biol, February 1, 2008; 28(2): 353 - 359. [Abstract] [Full Text] [PDF]

				S. M. Shenouda and J. A. Vita Effects of Flavonoid-Containing Beverages and EGCG on Endothelial Function J. Am. Coll. Nutr., August 1, 2007; 26(4): 366S - 372S. [Abstract] [Full Text] [PDF]

				J. Lin, K. M. Rexrode, F. Hu, C. M. Albert, C. U. Chae, E. B. Rimm, M. J. Stampfer, and J. E. Manson Dietary Intakes of Flavonols and Flavones and Coronary Heart Disease in US Women Am. J. Epidemiol., June 1, 2007; 165(11): 1305 - 1313. [Abstract] [Full Text] [PDF]

				D. Taubert, R. Roesen, and E. Schomig Effect of Cocoa and Tea Intake on Blood Pressure: A Meta-analysis Arch Intern Med, April 9, 2007; 167(7): 626 - 634. [Abstract] [Full Text] [PDF]

				M. E. Widlansky, N. M. Hamburg, E. Anter, M. Holbrook, D. F. Kahn, J. G. Elliott, J. F. Keaney Jr., and J. A. Vita Acute EGCG Supplementation Reverses Endothelial Dysfunction in Patients with Coronary Artery Disease J. Am. Coll. Nutr., April 1, 2007; 26(2): 95 - 102. [Abstract] [Full Text] [PDF]

				J. W. Erdman Jr., D. Balentine, L. Arab, G. Beecher, J. T. Dwyer, J. Folts, J. Harnly, P. Hollman, C. L. Keen, G. Mazza, et al. Flavonoids and Heart Health: Proceedings of the ILSI North America Flavonoids Workshop, May 31-June 1, 2005, Washington, DC J. Nutr., March 1, 2007; 137(3): 718S - 737S. [Abstract] [Full Text] [PDF]

				H. Yamada and H. Watanabe Tea polyphenols in preventing cardiovascular diseases Cardiovasc Res, January 15, 2007; 73(2): 439 - 440. [Full Text] [PDF]

				K. J. Mukamal, M. Alert, M. Maclure, J. E. Muller, and M. A. Mittleman Tea Consumption and Infarct-Related Ventricular Arrhythmias: The Determinants of Myocardial Infarction Onset Study J. Am. Coll. Nutr., December 1, 2006; 25(6): 472 - 479. [Abstract] [Full Text] [PDF]

				C. Cabrera, R. Artacho, and R. Gimenez Beneficial effects of green tea--a review. J. Am. Coll. Nutr., April 1, 2006; 25(2): 79 - 99. [Abstract] [Full Text] [PDF]

				B. M Popkin, L. E Armstrong, G. M Bray, B. Caballero, B. Frei, and W. C Willett A new proposed guidance system for beverage consumption in the United States Am. J. Clinical Nutrition, March 1, 2006; 83(3): 529 - 542. [Abstract] [Full Text] [PDF]

				S. Tokudome, I. Nahomi, C. Goto, Y. Tokudome, and M. A Moore Black tea and cardiovascular disease Int. J. Epidemiol., April 1, 2005; 34(2): 482 - 483. [Full Text] [PDF]

				C. Poole, U. Peters, D. Il'yasova, and L. Arab Authors' response Int. J. Epidemiol., April 1, 2005; 34(2): 483 - 483. [Full Text] [PDF]

				J. A Vita Polyphenols and cardiovascular disease: effects on endothelial and platelet function Am. J. Clinical Nutrition, January 1, 2005; 81(1): 292S - 297S. [Abstract] [Full Text] [PDF]

				I. C. Arts and P. C. Hollman Polyphenols and disease risk in epidemiologic studies Am. J. Clinical Nutrition, January 1, 2005; 81(1): 317S - 325S. [Abstract] [Full Text] [PDF]

				Y.-C. Yang, F.-H. Lu, J.-S. Wu, C.-H. Wu, and C.-J. Chang The Protective Effect of Habitual Tea Consumption on Hypertension Arch Intern Med, July 26, 2004; 164(14): 1534 - 1540. [Abstract] [Full Text] [PDF]

				K. W. Lee and G. Y. H. Lip Effects of Lifestyle on Hemostasis, Fibrinolysis, and Platelet Reactivity: A Systematic Review Arch Intern Med, October 27, 2003; 163(19): 2368 - 2392. [Abstract] [Full Text] [PDF]

				A. Rietveld and S. Wiseman Antioxidant Effects of Tea: Evidence from Human Clinical Trials J. Nutr., October 1, 2003; 133(10): 3285S - 3292. [Abstract] [Full Text] [PDF]

				J. M Hodgson, A. Devine, I. B. Puddey, S. Y. Chan, L. J. Beilin, and R. L. Prince Tea Intake Is Inversely Related to Blood Pressure in Older Women J. Nutr., September 1, 2003; 133(9): 2883 - 2886. [Abstract] [Full Text] [PDF]

				H. D Sesso, R. S Paffenbarger Jr, Y. Oguma, and I-M. Lee Lack of association between tea and cardiovascular disease in college alumni Int. J. Epidemiol., August 1, 2003; 32(4): 527 - 533. [Abstract] [Full Text] [PDF]

				C. Poole, U. Peters, D. Il'yasova, and L. Arab Commentary: This study failed? Int. J. Epidemiol., August 1, 2003; 32(4): 534 - 535. [Full Text] [PDF]

				D. J. Maron, G. P. Lu, N. S. Cai, Z. G. Wu, Y. H. Li, H. Chen, J. Q. Zhu, X. J. Jin, B. C. Wouters, and J. Zhao Cholesterol-Lowering Effect of a Theaflavin-Enriched Green Tea Extract: A Randomized Controlled Trial Arch Intern Med, June 23, 2003; 163(12): 1448 - 1453. [Abstract] [Full Text] [PDF]

				J. M Hodgson, V. Burke, L. J Beilin, K. D Croft, and I. B Puddey Can black tea influence plasma total homocysteine concentrations? Am. J. Clinical Nutrition, April 1, 2003; 77(4): 907 - 911. [Abstract] [Full Text] [PDF]

				A. G. Thrift and G. A. Donnan RE: "DOES TEA AFFECT CARDIOVASCULAR DISEASE? A META-ANALYSIS" Am. J. Epidemiol., September 1, 2002; 156(5): 490 - 490. [Full Text] [PDF]

				U. Peters, C. Poole, and L. Arab THE AUTHORS REPLY Am. J. Epidemiol., September 1, 2002; 156(5): 490 - 491. [Full Text] [PDF]

				RE: "DOES TEA AFFECT CARDIOVASCULAR DISEASE? A META-ANALYSIS" Am. J. Epidemiol., April 15, 2002; 155(8): 781 - 781. [Full Text] [PDF]

This Article Abstract FREE Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Archive Download to citation manager Search for citing articles in: ISI Web of Science (102) Request Permissions Disclaimer Google Scholar Articles by Peters, U. Articles by Arab, L. Search for Related Content PubMed PubMed Citation Articles by Peters, U. Articles by Arab, L. Social Bookmarking          What's this?

Online ISSN 1476-6256 - Print ISSN 0002-9262

Copyright © 2009 Johns Hopkins Bloomberg School of Public Health

Oxford Journals Oxford University Press

• Site Map
• Privacy Policy
• Frequently Asked Questions

Other Oxford University Press sites: Oxford University Press Oxford Journals China Oxford Journals Japan American National Biography Booksellers' Information Service Children's Fiction and Poetry Children's Reference Corporate & Special Sales Dictionaries Dictionary of National Biography Digital Reference English Language Teaching Higher Education Textbooks Humanities International Education Unit Law Medicine Music Online Products Oxford English Dictionary Reference Rights and Permissions Science School Books Social Sciences Very Short Introductions World's Classics