CVD is a leading cause of morbidity, mortality and disability worldwide( Reference Roger, Go and Lloyd-Jones 1 ). Blood pressure (BP) has a strong and direct relationship with cardiovascular (CV) mortality( Reference Lewington, Clarke and Qizilbash 2 , Reference Miura, Daviglus and Dyer 3 ). More importantly, there is no evidence of a BP threshold. The risk of CV mortality increases progressively throughout the range of BP, including the range of pre-hypertensive BP (systolic BP 120–139 mmHg and diastolic BP 80–89 mmHg). Thus, small changes in BP due to dietary modification may have a significant impact on the prevalence of hypertension and the risk of CVD( Reference Lewington, Clarke and Qizilbash 2 ).
Tea, including black and green tea, is a popular beverage worldwide and is usually the major source of population flavonoid intake, often providing more than half of total intake( Reference Zamora-Ros and Luján-Barroso 4 ). Epidemiological studies have suggested that a high intake of both green and black tea is related to a reduction in the risk of CVD( Reference Kuriyama, Shimazu and Ohmori 5 , Reference de Koning Gans, Uiterwaal and van der Schouw 6 ). The reduction of CVD risk by tea intake may be largely due to the high levels of polyphenols, in particular flavonoids, present in both green and black tea. The beneficial effect of tea intake on endothelial function may suggest a mechanistic explanation for the reduced risk of CVD( Reference Grassi, Desideri and Di Giosia 7 ).
A substantial number of clinical trials have been performed to investigate the acute or chronic effects of tea beverages and extracts on the BP of subjects with CVD-related conditions as well as of healthy individuals( Reference Hodgson, Woodman and Puddey 8 – Reference Suliburska, Bogdanski and Szulinska 32 ). However, the results of these trials were inconsistent, sample sizes were relatively modest so studies were often underpowered to detect modest effects on BP, and most studies did not have BP as a primary outcome. Therefore, in the present study, we conducted a meta-analysis of all published randomised controlled trials to determine the acute and chronic effects of tea intake on systolic and diastolic BP.
Methods
Search strategy
According to the QUORUM (Quality of Reporting of Meta-analyses), we systematically searched PubMed (http://www.ncbi.nlm.nih.gov/pubmed; from 1967 to May 2014), EMBASE (http://www.embase.com; from December 1977 up to 2014), the Cochrane Library database (http://www.cochrane.org), and reviews and reference lists of relevant articles using the keywords ‘tea’, ‘green tea’, ‘black tea’, ‘tea polyphenols’, ‘blood pressure’, ‘hypertension’. The search was restricted to human research studies. No limit was placed on language. In addition, a manual search of references from the reports of clinical trials or review articles was performed to identify the relevant trials. Attempts were also made to contact investigators for unpublished results and full-text articles.
Study selection
Studies were included in the present meta-analysis if they met the following criteria: (1) studies evaluated the acute ( < 1 week) or chronic (>1 week) effects of tea on BP; (2) studies were randomised controlled trials with either a parallel or cross-over design; (3) studies reported net changes in BP or only follow-up BP measures, and the associated standard deviations (or data to calculate them); (4) food intake control regimen of the experimental group was consistent with that of the control group; (5) tea extract was not given as part of a multi-component supplement in either the experimental or control group. Studies were excluded from the analysis if only abstracts were published. Data of multiple published reports from the same study population were included only once.
Data extraction and quality assessment
Search, data extraction and quality assessment were completed independently by two reviewers (G. L. and X.-N. M.) according to the aforementioned inclusion criteria. Any discrepancies between the two reviewers were resolved by discussion until a consensus was reached. Study characteristics (including authors, year of publication, sample size, study design, study duration, dose and type of intervention) and population information (age, ethnicity, sex and initial healthy status) were extracted. For continuous outcomes in parallel studies, the means and standard deviations of changes from baseline to endpoint (for both intervention and control groups) were extracted. In cross-over studies, the means and standard deviations were used separately for interventions and controls. This step provided a conservative estimate of the effects and reduced the power of cross-over studies to show the real influences of interventions( Reference JPT and S 33 ).
Quality characteristics of the trials were assessed using the following criteria: (1) randomisation; (2) concealment of treatment allocation; (3) participant masking; (4) researcher masking; (5) reporting of withdrawals; (6) generation of random numbers. The Jadad score was also introduced in order to evaluate the quality of the included studies. Trials scored one point for each area addressed in the study design (randomisation, blinding, concealment of allocation, reporting of withdrawals and generation of random numbers), with a possible score ranging between 0 and 5 (highest level of quality)( Reference Moher, Pham and Jones 34 ). Higher numbers represented better quality (Jadad score ≥ 3).
Data synthesis and analysis
Net changes in each of the study variables, calculated from baseline and follow-up means and standard deviations (follow-up minus baseline), were used to estimate the principle effect. When the standard deviations were not available directly, they were calculated from standard errors or CI. If variances for net changes were not reported directly, they were calculated from CI, P values, or individual variances from the tea group and the control group. For trials in which variances for paired differences were reported separately for each group, we calculated a pooled variance for net changes using standard methods. Missing variances for paired differences were calculated from variances at baseline and at the end of the follow-up for each measure using correlation coefficient methods according to the Cochrane Handbook for Systematic Reviews of Interventions( Reference JPT and S 33 ). We assumed a correlation coefficient of 0·62( Reference JPT and S 33 ).
The present meta-analysis and statistical analyses were performed using STATA 12.0 (STATA Corporation LP). A P value < 0·05 was considered as statistically significant for all analyses. Weighted mean differences and 95 % CI were calculated for net changes in systolic and diastolic BP. The statistic heterogeneity of treatment effects between studies was formally tested with Cochran's test (P< 0·1). The I 2 statistic was also examined, and we considered I 2>50 % to indicate significant heterogeneity between trials( Reference Higgins, Thompson and Deeks 35 ). Results were obtained from a fixed-effects model if no significant heterogeneity was shown, and a random-effects model was selected for the analysis if significant heterogeneity was shown( Reference DerSimonian and Laird 36 ). Publication bias was assessed with funnel plots and Egger's regression test. Previously defined subgroup analyses were performed to examine the effects of factors (ethnicity, type of tea, polyphenol dose, health status, study duration and caffeine controlled) on the primary outcomes after chronic intake of tea, and to identify the possible source of heterogeneity within these studies. To test the robustness of the results, we performed a one-way sensitivity analysis. The scope of the present meta-analysis was to evaluate the influence of individual studies by estimating pooled changes in BP in the absence of each study.
Results
Results of the literature search
The method used for the selection of the studies is shown in Fig. 1. The initial search identified 714 reports, of which 682 were excluded because they were not clinical trials or because the interventions were not relevant to the purpose of the present meta-analysis. Through a manual reference search of primary and review articles, two additional articles were retrieved. Therefore, thirty-four potentially relevant articles were examined in more detail. Among them, nine were subsequently excluded. The reasons for the exclusion of the studies are presented in Fig. 1. Thus, a total of twenty-five articles were selected for the final analysis.
Fig. 1 Flow chart showing the number of citations retrieved by individual searches of trials included in the meta-analysis.
Study characteristics
A total of twenty-five eligible randomised controlled trials with 1476 subjects were included in the present meta-analysis( Reference Hodgson, Woodman and Puddey 8 – Reference Suliburska, Bogdanski and Szulinska 32 ). The characteristics of the included trials are shown in Table 1. The studies of Hodgson et al. ( Reference Hodgson, Puddey and Burke 13 ) and Duffy et al. ( Reference Duffy, Keaney and Holbrook 14 ) were both separated into two trials (acute and chronic effects of tea on BP). The trials varied in size from twelve to 240 subjects, and study duration varied from 1 h to 24 weeks. Of the twenty-five trials used in the meta-analysis, seven( Reference Belza, Toubro and Astrup 9 , Reference Quinlan, Lane and Moore 11 , Reference Bingham, Vorster and Jerling 12 , Reference Grassi, Mulder and Draijer 17 , Reference Frank, George and Lodge 26 , Reference Nantz, Rowe and Bukowski 27 , Reference Sone, Kuriyama and Nakaya 30 ) were conducted in healthy adults, and eighteen( Reference Hodgson, Woodman and Puddey 8 , Reference Hodgson, Burke and Puddey 10 , Reference Hodgson, Puddey and Burke 13 – Reference Mukamal, MacDermott and Vinson 16 , Reference Hodgson, Puddey and Woodman 18 – Reference Brown, Lane and Coverly 25 , Reference Nagao, Meguro and Hase 28 , Reference Brown, Lane and Holyoak 29 , Reference Bogdanski, Suliburska and Szulinska 31 , Reference Suliburska, Bogdanski and Szulinska 32 ) were conducted in patients with CV risk, among which, two( Reference Hodgson, Puddey and Burke 13 , Reference Bogdanski, Suliburska and Szulinska 31 ) enrolled hypertensive patients and eleven( Reference Hodgson, Woodman and Puddey 8 , Reference Duffy, Keaney and Holbrook 14 , Reference Mukamal, MacDermott and Vinson 16 , Reference Hodgson, Puddey and Woodman 18 , Reference Fukino, Shimbo and Aoki 19 , Reference Nagao, Hase and Tokimitsu 21 – Reference Hsu, Tsai and Kao 23 , Reference Brown, Lane and Coverly 25 , Reference Nagao, Meguro and Hase 28 , Reference Brown, Lane and Holyoak 29 ) included subjects with high-normal BP. Caffeine intake was controlled in fourteen trials( Reference Hodgson, Woodman and Puddey 8 , Reference Bingham, Vorster and Jerling 12 , Reference Hodgson, Puddey and Burke 13 , Reference Grassi, Mulder and Draijer 17 , Reference Hodgson, Puddey and Woodman 18 , Reference Nagao, Hase and Tokimitsu 21 , Reference Matsuyama, Tanaka and Kamimaki 24 – Reference Sone, Kuriyama and Nakaya 30 , Reference Suliburska, Bogdanski and Szulinska 32 ). Of the included studies, eighteen studies( Reference Hodgson, Woodman and Puddey 8 – Reference Hodgson, Puddey and Woodman 18 , Reference Diepvens, Kovacs and Vogels 20 , Reference Brown, Lane and Coverly 25 – Reference Nantz, Rowe and Bukowski 27 , Reference Brown, Lane and Holyoak 29 , Reference Bogdanski, Suliburska and Szulinska 31 , Reference Suliburska, Bogdanski and Szulinska 32 ) were performed in Whites, and the remaining seven( Reference Fukino, Shimbo and Aoki 19 , Reference Nagao, Hase and Tokimitsu 21 – Reference Matsuyama, Tanaka and Kamimaki 24 , Reference Nagao, Meguro and Hase 28 , Reference Sone, Kuriyama and Nakaya 30 ) were carried out in Asians. Most of the trials (sixteen trials)( Reference Hodgson, Woodman and Puddey 8 , Reference Mukamal, MacDermott and Vinson 16 , Reference Hodgson, Puddey and Woodman 18 – Reference Nagao, Meguro and Hase 28 , Reference Sone, Kuriyama and Nakaya 30 – Reference Suliburska, Bogdanski and Szulinska 32 ) adopted parallel study designs and seventeen( Reference Hodgson, Woodman and Puddey 8 , Reference Belza, Toubro and Astrup 9 , Reference Bingham, Vorster and Jerling 12 , Reference Grassi, Mulder and Draijer 17 , Reference Hodgson, Puddey and Woodman 18 , Reference Diepvens, Kovacs and Vogels 20 , Reference Nagao, Hase and Tokimitsu 21 , Reference Hsu, Tsai and Kao 23 – Reference Suliburska, Bogdanski and Szulinska 32 ) were double-blinded. A low-energy diet was administered in one trial( Reference Diepvens, Kovacs and Vogels 20 ), and in the remaining twenty-four trials, investigators attempted to maintain the usual lifestyles of participants.
Table 1 Characteristics of the twenty-five included randomised controlled trials
M, male; F, female; BP, blood pressure; RP, randomised parallel; DB, double-blinded; RC, randomised cross-over; OL, open-labelled; CAD, coronary artery disease; NR, not reported.
* Information on the number of males and females was unavailable.
The results of the validity of the included trials are presented in Table 2. Most of the trials (eighteen trials)( Reference Hodgson, Woodman and Puddey 8 , Reference Belza, Toubro and Astrup 9 , Reference Bingham, Vorster and Jerling 12 , Reference Duffy, Keaney and Holbrook 14 , Reference Mukamal, MacDermott and Vinson 16 – Reference Hodgson, Puddey and Woodman 18 , Reference Nagao, Hase and Tokimitsu 21 , Reference Hsu, Tsai and Kao 23 – Reference Suliburska, Bogdanski and Szulinska 32 ) were classified as high quality (Jadad score ≥ 3). Furthermore, twelve trials( Reference Hodgson, Woodman and Puddey 8 , Reference Bingham, Vorster and Jerling 12 , Reference Duffy, Keaney and Holbrook 14 , Reference Mukamal, MacDermott and Vinson 16 – Reference Hodgson, Puddey and Woodman 18 , Reference Hsu, Tsai and Kao 23 , Reference Brown, Lane and Coverly 25 , Reference Nantz, Rowe and Bukowski 27 , Reference Brown, Lane and Holyoak 29 , Reference Sone, Kuriyama and Nakaya 30 , Reference Suliburska, Bogdanski and Szulinska 32 ) reported the generation of random numbers, but only eight( Reference Hodgson, Woodman and Puddey 8 , Reference Mukamal, MacDermott and Vinson 16 , Reference Hodgson, Puddey and Woodman 18 , Reference Hsu, Tsai and Kao 23 , Reference Brown, Lane and Coverly 25 , Reference Brown, Lane and Holyoak 29 – Reference Bogdanski, Suliburska and Szulinska 31 ) reported details of allocation concealment. The details of dropouts were reported in twenty-four trials.
Table 2 Validity of the included trials
Main analysis
As shown in Fig. 2, the acute intake of tea had no effects on systolic and diastolic BP. The results of the long-term effects of tea intake on BP are shown in Fig. 3. Overall, compared with the tea-free control, the pooled mean decrease in systolic BP was − 1·8 (95 % CI − 2·4, − 1·1) mmHg (I 2= 17·4 %) and in diastolic BP was − 1·4 (95 % CI − 2·2, − 0·6) mmHg (I 2= 52·5 %) for tea intake. In addition, when stratified by type of tea, green tea exhibited a significant reduction in systolic BP of 2·1 (95 % CI − 2·9, − 1·2) mmHg (I 2= 21·8 %) and a decrease in diastolic BP of 1·7 (95 % CI − 2·9, − 0·5) mmHg (I 2= 59·9 %), and black tea showed a significant reduction in systolic BP of 1·4 (95 % CI − 2·4, − 0·4) mmHg (I 2= 9·7 %) and a decrease in diastolic BP of 1·1 (95 % CI − 1·9, − 0·2) mmHg (I 2= 22·9 %).
Fig. 2 Meta-analysis of the acute effects of tea intake on (a) systolic and (b) diastolic blood pressure compared with the control arms. WMD, weighted mean difference. (A colour version of this figure can be found online at http://www.journals.cambridge.org/bjn).
Fig. 3 Meta-analysis of the long-term effects of tea intake on (a) systolic and (b) diastolic blood pressure compared with the control arms. Subgroup analyses stratified by type of tea (black and green tea). WMD, weighted mean difference. (A colour version of this figure can be found online at http://www.journals.cambridge.org/bjn).
Subgroup and sensitivity analyses
The results of the subgroup analyses and sensitivity analyses on systolic and diastolic BP (long-term effects) are summarised in Table 3. The subgroup analyses by study duration suggested that tea intake over a median of 12 weeks had a pronounced reduction in systolic BP of − 2·6 (95 % CI − 3·5, − 1·7) mmHg and in diastolic BP of − 2·2 (95 % CI − 3·0, − 1·3) mmHg (Fig. 4), compared with the short-term subgroup ( < 12 weeks) (between groups P< 0·05). To explore the dose–effect relationship, polyphenol doses were divided into low ( ≤ 544 mg/d) and high (>544 mg/d) doses. The subgroup analyses found that the polyphenol doses were not an effect modifier. Meanwhile, we also stratified the subjects by health status into healthy and CV risk groups (overweight or obese, and diabetic), and found no significant difference between the two groups. In addition, the BP-lowering effects were not influenced by baseline BP status. To investigate whether the effects of tea intake were related to caffeine, the changes in BP were assessed separately between studies that controlled for caffeine intake and that did not. The pooled analysis indicated that tea ingestion with or without caffeine both significantly reduced systolic and diastolic BP, suggesting that caffeine intake could not modify the pooled BP-lowering effects of tea. The sensitivity analyses showed that the significance in the pooled changes in BP were not altered after the removal of the six trials( Reference Bingham, Vorster and Jerling 12 – Reference Hodgson, Croft and Mori 15 , Reference Grassi, Mulder and Draijer 17 , Reference Brown, Lane and Holyoak 29 ) with a cross-over design or the five trials( Reference Hodgson, Puddey and Burke 13 , Reference Hodgson, Croft and Mori 15 , Reference Fukino, Shimbo and Aoki 19 , Reference Diepvens, Kovacs and Vogels 20 , Reference Fukino, Ikeda and Maruyama 22 ) with low quality.
Table 3 Subgroup analyses of systolic and diastolic blood pressure (BP) stratified by previously defined study characteristics (Mean differences and 95 % confidence intervals)
CV, cardiovascular.
Fig. 4 Subgroup analyses of the effects of chronic intake of tea on (a) systolic and (b) diastolic blood pressure stratified by study duration ( ≥ 12 or < 12 weeks). WMD, weighted mean difference. (A colour version of this figure can be found online at http://www.journals.cambridge.org/bjn).
Publication bias
Publication bias of the trials was examined by analysing funnel plots and Egger's tests. As shown in Fig. 5, the funnel plots were symmetrical and Egger's tests indicated no significant publication bias (P= 0·947 for systolic BP and P= 0·653 for diastolic BP).
Fig. 5 Funnel plots of changes in (a) systolic (SBP) and (b) diastolic (DBP) blood pressure after chronic intake of tea. The vertical line represents the pooled mean effect size. (A colour version of this figure can be found online at http://www.journals.cambridge.org/bjn).
Discussion
The present meta-analysis showed that the acute intake of tea had no effects on BP. However, long-term consumption of black and green tea significantly reduced systolic and diastolic BP. Subgroup analyses indicated that the BP-lowering effects were apparent when the duration of the follow-up was over a median of 12 weeks. Differences in tea polyphenol doses, caffeine intake, study quality, ethnicity and health status of participants did not appear to significantly influence the pooled mean differences in BP.
A large population-based study that involved >40 000 middle-aged Japanese revealed that, compared with no tea drinking, habitual tea consumption (average of two cups (approximately 17 oz)/d for 10 years) was associated with a lower risk of death from CVD( Reference Kuriyama, Shimazu and Ohmori 5 ). However, reports on the effects of tea on CVD risk factors have been mixed. Some clinical studies have shown that green tea intake lowers total and LDL-cholesterol, and blood glucose levels( Reference Zheng, Xu and Li 37 , Reference Zheng, Xu and Li 38 ); however, some randomised trials have shown the lack of the effects of black tea intake on lipids( Reference Bingham, Vorster and Jerling 12 , Reference Duffy, Keaney and Holbrook 14 ). In addition, the BP-lowering effect of tea is also controversial. A previous meta-analysis has shown that tea intake had no significant effect on BP( Reference Taubert, Roesen and Schömig 39 ); however, the sample sizes of that study were relatively modest (343 subjects) and the duration of the study was short (mean 4 weeks). In the present meta-analysis involving a total of 1323 subjects at a mean follow-up of 12 weeks, we confirmed that tea ingestion resulted in a significant reduction of BP. More importantly, there was no indication of heterogeneity for systolic BP, and only modest heterogeneity was observed for diastolic BP. Therefore, it is reasonable to speculate that the BP-lowering effect of tea is also a contributor to the reduced risk of CVD mortality.
The BP-lowering effect of tea may be associated with its antioxidant properties and endothelial protection. Tea and their flavonoids could act as antioxidants by scavenging reactive oxygen species and nitrogen species, and chelating redox-active transition metal ions( Reference Brown, Khodr and Hider 40 , Reference Kerry and Rice-Evans 41 ). Studies on hypertensive animal models have shown that tea intake effectively attenuated increases in BP and, meanwhile, reduced the formation of vascular reactive oxygen species and improved endothelium-dependent relaxation in the aorta, which could account for the amelioration of hypertension( Reference Negishi, Xu and Ikeda 42 , Reference Ihm, Jang and Kim 43 ). In addition, there has been compelling evidence showing that ingestion of tea leads to increments in brachial artery flow-mediated dilation and improvement in endothelial function( Reference Grassi, Mulder and Draijer 17 , Reference Jochmann, Lorenz and Krosigk 44 ). However, the results from human intervention studies do not provide evidence that reduced reactive oxygen species formation contributes to the beneficial effects of tea intake on vascular health( Reference O'Reilly, Mallet and McAnlis 45 , Reference Widlansky, Duffy and Hamburg 46 ).
In the present meta-analysis, the beneficial effects of tea intake on BP were observed when the duration of consumption was slightly ≥ 12 weeks. We found that the acute intake of tea had no effects on BP. The results suggest that long-term benefits of tea intake on BP are unlikely to be due to acute changes. Because the improvement in endothelial function appears to be strongest in the hours after tea has been consumed( Reference Ras, Zock and Draijer 47 ), there may have been other mechanisms underlying the long-term benefits of tea ingestion in addition to the increase in the bioavailability of NO. Tea intake has been reported to have various beneficial effects on vascular function, such as anti-inflammatory effects, anti-platelet effects and anti-proliferative effects( Reference Deka and Vita 48 ). Thus, these effects may also be involved in potential mechanisms underlying the benefits of tea intake on BP. In the present subgroup analyses, the reduction in systolic BP by 2·6 mmHg after chronic intake of tea, as reported herein, would be expected to reduce stroke risk by 8 %, coronary artery disease mortality by 5 % and all-cause mortality by 4 % at a population level( Reference Whelton, He and Appel 49 ). These are profound effects and must be considered seriously in terms of the potential for dietary modification to modulate the risk of CVD. The beneficial effects of tea intake on endothelial function may more or less explain the reduced risk of CVD and stroke( Reference Grassi, Desideri and Di Giosia 7 ).
Tea intake has been shown to decrease BP in the present meta-analysis. However, the optimal dose that would best improve BP remains uncertain. The subgroup analyses found that the tea polyphenol dose was not an effect modifier. This finding should be interpreted with caution. Most of the polyphenols found in tea are flavonoids, and catechins constitute about 80 to 90 % of total flavonoids in green tea, whereas they only account for 20 to 30 % of total flavonoids in black tea because it can convert catechins into more complex condensed flavonoids, mainly thearubigins and theaflavins( Reference Balentine, Wiseman and Bouwens 50 ). It is difficult to conclude the active constituents of green and black tea that needs to be explored in further studies. Moreover, the differences in tea preparations and ethnicity might affect the effectiveness of tea. Therefore, variations in the study characteristics of the included trials made it difficult to assess the true dose–response relationship between tea intake and its BP-lowering effects.
Because tea also naturally contains caffeine in addition to flavonoids or other compounds, another potential issue is whether caffeine intake affects the BP-lowering effects of tea. Data from human and animal studies have reported that caffeine alone could increase BP by influencing arterial compliance and increasing arterial stiffness( Reference Giggey, Wendell and Zonderman 51 , Reference Potter, Haigh and Harper 52 ), and therefore it may have a potential to reverse the BP-lowering effect. However, the present meta-analysis showed that intake of tea with or without caffeine both resulted in a significant reduction of BP, indicating that caffeine did not alter the effectiveness of tea and their flavonoids. This could be explained by the fact that the dosage of caffeine contained in tea is relatively low when compared with that of flavonoids; therefore, the negative effect of caffeine on BP cannot overcome the positive effect of tea and their flavonoids.
Although we believe that the present meta-analysis provides useful information, there are some potential limitations that need to be addressed. First, as with any meta-analysis, internal validity relies on the quality of individual studies. Although all studies were randomised and most of the studies described withdrawals, the lack of blinding of participants or investigators to the intervention in a number of studies( Reference Hodgson, Woodman and Puddey 8 – Reference Bingham, Vorster and Jerling 12 , Reference Hodgson, Croft and Mori 15 , Reference Hodgson, Puddey and Woodman 18 ) increased the risk of expectation bias. In addition, the potential lack of blinding even in studies that were described as ‘double blind’ could also bias the results reported herein due to the nature of the use of the product.
Second, the present meta-analysis did not pool safety data. The dosage of tea polyphenols consumed daily ranged from low (116·1 mg/d) to high (1207 mg/d) in the present meta-analysis, and no subjects experienced serious adverse events. However, concern has been raised about the safety of supplementation with high doses of tea polyphenols, such as the possibility of hepatotoxicity( Reference Galati, Lin and Sultan 53 ). Therefore, safety issues need to be evaluated under conditions of long-term and high-dose exposure in the future.
An additional limitation was the size of these trials, which ranged between twelve and 240 participants. Therefore, the present meta-analysis may have been underpowered to detect a true effect.
In conclusion, BP is a consistent, strong and independent risk of CV mortality, and small changes in BP may have a significant impact on the risk of CV mortality. The findings of the present meta-analysis suggest that long-term ( ≥ 12 weeks) ingestion of tea (green and black tea) resulted in a significant reduction of systolic and diastolic BP, and the BP-lowering effects of tea were not influenced by ethnicity, caffeine intake, tea polyphenol doses, health status of participants and study quality.
Acknowledgements
The present study was supported by a grant from the Beijing Science and Technique Programs of China (Z131100006813039). The sponsor had no role in the design or conduct of the study, the collection, management, analysis or interpretation of the data, or the preparation, review and approval of the manuscript.
The authors' contributions are as follows: X.-H. H. and G. L. were responsible for the study concept and design; G. L. and X.-N. M. summarised the data and conducted the research; G. L. and X.-N. M. analysed and interpreted the data. All authors read and approved the final version of the manuscript
None of the authors has any conflicts of interest to declare.
