Clinical Nutrition xxx (2014) 1e8
Contents lists available at ScienceDirect
Clinical Nutrition journal homepage: http://www.elsevier.com/locate/clnu
Meta-analyses
Black tea consumption and serum cholesterol concentration: Systematic review and meta-analysis of randomized controlled trials Yimin Zhao, Sailimuhan Asimi, Kejian Wu, Jusheng Zheng, Duo Li* Department of Food Science and Nutrition, Zhejiang University, Hangzhou, People's Republic of China
a r t i c l e i n f o
s u m m a r y
Article history: Received 25 November 2013 Accepted 9 June 2014
Background & aims: The results of randomized controlled trials in relation to the effect of regular black tea consumption on serum cholesterol concentration were inconsistent. We aimed to investigate and quantify the effect of black tea consumption on serum concentrations of total, LDL and HDL cholesterol. Methods: We systematically searched and identified relevant literatures in PubMed, Scopus and the Cochrane Library. Inclusion and exclusion of studies, data extraction, quality assessment and metaanalysis were conducted according to the PRISMA statement. Results: Ten eligible studies with 411 participants were identified in the present meta-analysis. No significant heterogeneity was found between studies. Consumption of black tea significantly reduced LDL cholesterol concentration (4.64 mg/dL; 95% CI: 8.99, 0.30 mg/dL; P ¼ 0.036). No remarkable change was detected in total cholesterol (2.04 mg/dL; 95% CI: 6.43, 2.35 mg/dL; P ¼ 0.363) or HDL cholesterol (1.15 mg/dL; 95% CI: 3.04, 0.75 mg/dL; P ¼ 0.236). Subgroup analysis showed that the lowering effect on LDL cholesterol was more effective in subjects with higher cardiovascular risk. Conclusions: Black tea consumption significantly lowered serum concentration of LDL cholesterol, especially in subjects with higher cardiovascular risk. Black tea intake did not impose obvious effect on serum concentrations of total and HDL cholesterol. © 2014 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.
Keywords: Black tea Systematic review Meta-analysis Cholesterol
1. Introduction Tea is one of the most popular drink in the world with about 80% being consumed as black tea [1], thus tiny effect on individual health caused by tea arouses wide concern and could have an enormous influence on public health [2]. Tea is produced from Camellia sinensis, it can be classified into three major types according to the level of the fermentation process: unfermented green tea, partially fermented Oolong tea and fully fermented black tea [3]. As a rich dietary source of flavonoids, black tea has potential beneficial health effects on human body. Previous meta-analyses of observational studies suggested that high black tea intake was associated with reduced risk of type 2 diabetes and stroke [4,5]. Randomized controlled trial showed that black tea intervention can improve flow-mediated dilation which indicated a cardiovascular protective effect of black tea [6]. Hypercholesterolemia is closely correlated with risk of cardiovascular diseases, such as atherosclerosis and coronary artery * Corresponding author. Department of Food Science and Nutrition, Zhejiang University, 866 Yuhangtang Road, Hangzhou 310058, People's Republic of China. Tel./fax: þ86 571 88982024. E-mail address: [email protected] (D. Li).
disease [7]. Modification of serum cholesterol profile remains an important method for cardiovascular disease prevention and it was estimated that for every 1% reduction in total cholesterol concentration, the risk of cardiovascular diseases decreased by an average of 2% and 1 mg/dL reduction of LDL cholesterol concentration can reduce coronary artery disease risk by 1% [8,9]. To date, the results from randomized clinical trials concerning the effect of black tea on serum cholesterol concentration remained inconsistent. Therefore, the objective of this study was to systematically review and quantify the randomized controlled trials regarding the effect of black tea consumption on serum cholesterol concentration. 2. Methods This systematic review and meta-analysis has followed the recommendations of the PRISMA statement [10]. 2.1. Study selection and eligibility criteria We conducted a systematic search for publications before November 2013 using the databases of PubMed (http://www.ncbi. nlm.nih.gov/pubmed), Scopus (http://www.scopus.com) and the
http://dx.doi.org/10.1016/j.clnu.2014.06.003 0261-5614/© 2014 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.
Please cite this article in press as: Zhao Y, et al., Black tea consumption and serum cholesterol concentration: Systematic review and metaanalysis of randomized controlled trials, Clinical Nutrition (2014), http://dx.doi.org/10.1016/j.clnu.2014.06.003
2
Y. Zhao et al. / Clinical Nutrition xxx (2014) 1e8
Cochrane Library (http://www.thecochranelibrary.com). We used the text words: tea, black tea and Camellia sinensis, paired with the following words: blood lipid, blood cholesterol, low density lipoprotein cholesterol and high density lipoprotein cholesterol. All searches were restricted in English-language publications. An additional manual search was conducted using reference lists from original research papers, previous meta-analysis and review articles. To be included, a study must meet all the following criteria: (1) conducted in adults; (2) randomized controlled trials of either parallel or crossover design; (3) used black tea as the only active treatment intervention; (4) provided available data to calculate the difference between baseline and endpoint for cholesterol measures including total cholesterol, LDL cholesterol and HDL cholesterol; (5) with an intervention duration 2 weeks and a sample size 10. Besides, a study will be excluded if it employed pregnant or severely ill participants.
there was no obvious heterogeneity among trials, otherwise the random-effects model developed by DerSimonian and Laird was adopted to calculate the pooled effect [13]. Possible publication bias was examined by Begg's test [14], a P < 0.1 indicated potential publication bias may exist. Trim and fill method was used to correct potential publication bias and corresponding filled funnel plot was also presented for visual inspection of publication bias [15]. To explore the source of heterogeneity, we carried out a series of predesigned subgroup analyses. Subgroup analyses were implemented by comparing trials stratified by study and population characteristics including healthy status of subjects, duration, study design, daily flavonoids intake from black tea, type of intervention and control. Furthermore, we also conducted additional sensitivity analyses to test if the results were steady [10].
2.2. Data extraction and quality assessment
3.1. Results of the study selection
Data were extracted independently by two researchers and any discrepancy was resolved via discussion. Study characteristics were extracted including author, country, study design, intervention duration, sample size, population information, type of intervention and control, diet change. In each trial, the means and SDs of cholesterol measures at baseline and endpoint in both intervention and control groups were extracted. For studies that had multiple time points for the same participants, only the last endpoint was used for analysis. If a study adopted different intervention protocols, intervention group which met the inclusion criteria was used. Extracted data were converted to conventional units and for cholesterol 1 mmol/L was converted to 38.6 mg/dL. When SD was not reported directly, we calculated them from SEM and 95% CI using the following equations according to the Cochrane handbook for systematic reviews of interventions [11]:
The PRISMA flow diagram is shown in Fig. 1. Finally we identified ten studies that met the inclusion criteria [16e25]. Total of 597 potentially relevant titles and abstracts were identified from searches on PubMed, Scopus and the Cochrane Library together with other reference sources. After ruling out animal studies and those irrelevant to the aim of this meta-analysis, 27 articles were retrieved as full text and assessed for inclusion. We excluded studies with a duration shorter than two weeks [6] or with a small sample size less than 10 [26]. Furthermore two studies were excluded because black tea was used as a part of multicomponent supplement in intervention group [27,28]. Three articles were ruled out because they used Pu'er tea as intervention treatment [29e31]. Pu'er tea is a sort of wet-fermented tea despite it was called Chinese black tea in these studies. Another ten studies were excluded due to a lack of sufficient detail for inclusion or a cross-sectional design.
SD ¼ SEM
pffiffiffi n
(1)
3. Results
3.2. Study characteristics
And
pffiffiffi n ðupperlimit lowerlimitÞ÷3:92
The R is the correlation coefficient. To be conservative, a minimum correlation coefficient of 0.5 was used. Quality assessment was conducted in duplicate independently and quality characteristics included the following items: randomization, random sequence generation, allocation concealment, blinding of both participants and researchers, description of dropouts. The Jadad score was adopted to assess the quality of each study and trials scored one point for each area addressed [12]. Trials with Jadad score 4 were classified as high quality. Discrepancy was resolved through discussion until a consensus was reached.
The characteristics of all the included studies are listed in Table 1. The research of Bahorun et al. [16] was separated into 2 trials (effects on males and females respectively). Therefore, 11 trials involved 411 adult participants were included in this metaanalysis. Six trials from 5 studies were conducted in healthy people [16,17,21,23,25] while the other five trials were performed in subjects with higher cardiovascular risk such as prediabetes or hypercholesterolemia [18e20,22,24]. The sample size ranged from 15 to 77 with a median of 31. Six trials lasted for 4 weeks [17,19e21,23,25], four trials lasted longer than 10 weeks [16,22,24] and the average duration was up to 8 weeks. Only two trials were conducted in East Asian countries [21,24] and the rest were mostly performed in America or Europe. Black tea extract capsule was used in two trials [22,24] and the remaining adopted black tea beverages as the intervention treatment. Controlled background diet was adopted only in one trial [18]. Eight trials used parallel design [16,20e25] and the rest 3 trials adopted crossover design [17e19].
2.3. Data synthesis and analysis
3.3. Quality assessment
Statistical analysis was performed using Stata/SE 12.0 for Windows (StataCorp, College Station, TX, USA). Heterogeneity among trials was assessed by I2 statistic and Cochrane Q test. We suggested there was no significant heterogeneity among trials when I2 4 weeks). Because four of the ten studies did not report the dose of daily flavonoids intake from black tea [17,20,22,24], we did not perform subgroup analysis stratified by daily flavonoids intake from black tea. For trials which employed subjects with mild hypercholesterolemia [18,20,24], coronary disease [19] or prediabetes [22], we treated them together as ‘trials performed in subjects with higher cardiovascular risk’. The results of subgroup analyses showed that the pooled effect of total cholesterol or HDL cholesterol was not influenced by type of intervention, type of control, study design, duration or healthy status of participants (Table 2). However, the subgroup analysis stratified by healthy status manifested that regular intake of black tea significantly reduced LDL cholesterol concentration in subjects with higher cardiovascular risk (8.26 mg/dL; 95% CI: 15.13, 1.40 mg/dL; P ¼ 0.018) rather than in healthy subjects (2.22 mg/dL; 95% CI: 7.83, 3.39 mg/dL; P ¼ 0.438). Besides, the pooled effect of trials with intervention duration longer than 4 weeks showed a marginally significant reduction in LDL cholesterol concentration (8.49 mg/dL; 95% CI: 17.42, 0.44 mg/dL; P ¼ 0.062) but not in those trials with shorter duration (3.45 mg/ dL; 95% CI: 8.42, 1.53 mg/dL; P ¼ 0.174). A marginally significant reduction also was detected in the trials which used black tea extract as intervention treatment (9.87 mg/dL; 95% CI: 19.92, 0.19 mg/dL; P ¼ 0.054) other than those trials using black tea beverage (3.44 mg/dL; 95% CI: 8.26, 1.38 mg/dL; P ¼ 0.162). Sensitivity analyses showed that the pooled effects of total cholesterol, LDL cholesterol and HDL cholesterol together with the corresponding CIs were not obviously changed when using a coefficient R ¼ 0.68 which was used in a previous meta-analysis [32]. Random-effects model was suggested to be more conservative and could provide a broader CI. The significance in the pooled effects in each cholesterol measure was not altered after we used randomeffects model to calculate the pooled effects. Besides the inclusion
of three studies which used Pu'er tea as intervention treatment did not significantly influence the final results. 3.5. Publication bias Results from Begg's rank correlation test suggested that no obvious publication bias was detected in the meta-analysis of LDL cholesterol (P ¼ 0.754) or HDL cholesterol (P ¼ 0.858), but not in total cholesterol (P ¼ 0.074). After trim and fill method was adopted, the pooled effect and corresponding CI of total cholesterol remained stable (Table 3). Filled funnel plot of each cholesterol measure was presented for visual inspection (Fig. 5). 4. Discussion The aim of the present study was to examine the relationship between black tea consumption and serum cholesterol concentration. This meta-analysis of 10 randomized controlled trials showed a significant reduction in serum concentration of LDL cholesterol after black tea administration, however no remarkable effect was detected on total cholesterol or HDL cholesterol. No obvious heterogeneity among studies was observed. Subgroup analyses suggested that the mean differences in total cholesterol and HDL cholesterol concentration were not affected by type of intervention or control, study design, healthy status of participant or study duration. However the effect of black tea consumption on LDL cholesterol concentration was more effective in subjects with higher cardiovascular risk like hypercholesterolemia. The results of sensitivity analyses indicated that the conclusions of the present meta-analysis were steady. Flavonoids, especially for theaflavins and catechins, are the major bioactive components in black tea. Black tea makes an important contribution to dietary intake of flavonoids. Epidemiological evidence recommended that serum cholesterol decreased with increasing tea intake and the daily consumption of >3 cups of black tea may protect against stroke and reduce all-cause mortality
Please cite this article in press as: Zhao Y, et al., Black tea consumption and serum cholesterol concentration: Systematic review and metaanalysis of randomized controlled trials, Clinical Nutrition (2014), http://dx.doi.org/10.1016/j.clnu.2014.06.003
6
Y. Zhao et al. / Clinical Nutrition xxx (2014) 1e8
Table 2 Results of subgroup and sensitivity analyses. Variables
Total cholesterol
LDL cholesterol
No. of Pooled effect (95% CI) P for No. of Pooled effect (95% CI) trials mg/dL heterogeneity trials mg/dL Subgroup analysis Type of intervention Black tea beverage 8 Black tea extract 2 Type of control Placebo 4 Water 6 Healthy status Healthy 6 With higher 4 cardiovascular risk Duration >4 weeks 3 4 weeks 7 Design Parallel 7 Crossover 3 Sensitivity Included Pu'er tea studiesa 13 Coefficient R ¼ 0.68 10 Using random-effects model 10 a
HDL cholesterol P for No. of Pooled effect (95% CI) P for heterogeneity trials mg/dL heterogeneity
3.16 (8.26, 1.94) 1.19 (7.46, 9.85)
0.492 0.222
6 3
5.19 (10.44, 0.07) 3.47 (11.20, 4.26)
0.536 0.632
7 3
2.43 (5.05, 1.19) 0.06 (2.58, 2.70)
0.477 0.994
1.93 (8.38, 4.53) 2.14 (8.14, 3.86)
0.398 0.339
4 5
4.71 (10.74, 1.31) 4.56 (10.84, 1.71)
0.214 0.720
4 6
0.20 (2.56, 2.16) 2.88 (6.07, 0.31)
0.983 0.383
0.62 (6.30, 5.06) 4.16 (11.09, 2.77)
0.267 0.657
5 4
2.22 (7.83, 3.39) 0.566 8.26 (15.13, 1.40) 0.609
6 4
1.50 (3.64, 0.63) 0.17 (3.93, 4.26)
0.249 0.996
0.24 (9.74, 10.23) 2.59 (7.48, 2.30)
0.272 0.448
4 5
8.49 (17.42, 0.44) 3.45 (8.42, 1.53)
0.740 0.358
4 6
0.89 (5.39, 3.61) 1.20 (3.29, 0.89)
0.227 0.724
0.51 (5.90, 4.88) 5.07 (12.65, 2.52)
0.293 0.817
7 2
3.73 (8.68, 1.23) 7.69 (16.75, 1.36)
0.493 0.440
8 2
1.18 (3.26, 0.89) 0.97 (5.65, 3.71)
0.410 0.914
4.68 (12.32, 2.96) 2.45 (6.09, 1.18) 2.04 (6.43, 2.35)
Contents lists available at ScienceDirect
Clinical Nutrition journal homepage: http://www.elsevier.com/locate/clnu
Meta-analyses
Black tea consumption and serum cholesterol concentration: Systematic review and meta-analysis of randomized controlled trials Yimin Zhao, Sailimuhan Asimi, Kejian Wu, Jusheng Zheng, Duo Li* Department of Food Science and Nutrition, Zhejiang University, Hangzhou, People's Republic of China
a r t i c l e i n f o
s u m m a r y
Article history: Received 25 November 2013 Accepted 9 June 2014
Background & aims: The results of randomized controlled trials in relation to the effect of regular black tea consumption on serum cholesterol concentration were inconsistent. We aimed to investigate and quantify the effect of black tea consumption on serum concentrations of total, LDL and HDL cholesterol. Methods: We systematically searched and identified relevant literatures in PubMed, Scopus and the Cochrane Library. Inclusion and exclusion of studies, data extraction, quality assessment and metaanalysis were conducted according to the PRISMA statement. Results: Ten eligible studies with 411 participants were identified in the present meta-analysis. No significant heterogeneity was found between studies. Consumption of black tea significantly reduced LDL cholesterol concentration (4.64 mg/dL; 95% CI: 8.99, 0.30 mg/dL; P ¼ 0.036). No remarkable change was detected in total cholesterol (2.04 mg/dL; 95% CI: 6.43, 2.35 mg/dL; P ¼ 0.363) or HDL cholesterol (1.15 mg/dL; 95% CI: 3.04, 0.75 mg/dL; P ¼ 0.236). Subgroup analysis showed that the lowering effect on LDL cholesterol was more effective in subjects with higher cardiovascular risk. Conclusions: Black tea consumption significantly lowered serum concentration of LDL cholesterol, especially in subjects with higher cardiovascular risk. Black tea intake did not impose obvious effect on serum concentrations of total and HDL cholesterol. © 2014 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.
Keywords: Black tea Systematic review Meta-analysis Cholesterol
1. Introduction Tea is one of the most popular drink in the world with about 80% being consumed as black tea [1], thus tiny effect on individual health caused by tea arouses wide concern and could have an enormous influence on public health [2]. Tea is produced from Camellia sinensis, it can be classified into three major types according to the level of the fermentation process: unfermented green tea, partially fermented Oolong tea and fully fermented black tea [3]. As a rich dietary source of flavonoids, black tea has potential beneficial health effects on human body. Previous meta-analyses of observational studies suggested that high black tea intake was associated with reduced risk of type 2 diabetes and stroke [4,5]. Randomized controlled trial showed that black tea intervention can improve flow-mediated dilation which indicated a cardiovascular protective effect of black tea [6]. Hypercholesterolemia is closely correlated with risk of cardiovascular diseases, such as atherosclerosis and coronary artery * Corresponding author. Department of Food Science and Nutrition, Zhejiang University, 866 Yuhangtang Road, Hangzhou 310058, People's Republic of China. Tel./fax: þ86 571 88982024. E-mail address: [email protected] (D. Li).
disease [7]. Modification of serum cholesterol profile remains an important method for cardiovascular disease prevention and it was estimated that for every 1% reduction in total cholesterol concentration, the risk of cardiovascular diseases decreased by an average of 2% and 1 mg/dL reduction of LDL cholesterol concentration can reduce coronary artery disease risk by 1% [8,9]. To date, the results from randomized clinical trials concerning the effect of black tea on serum cholesterol concentration remained inconsistent. Therefore, the objective of this study was to systematically review and quantify the randomized controlled trials regarding the effect of black tea consumption on serum cholesterol concentration. 2. Methods This systematic review and meta-analysis has followed the recommendations of the PRISMA statement [10]. 2.1. Study selection and eligibility criteria We conducted a systematic search for publications before November 2013 using the databases of PubMed (http://www.ncbi. nlm.nih.gov/pubmed), Scopus (http://www.scopus.com) and the
http://dx.doi.org/10.1016/j.clnu.2014.06.003 0261-5614/© 2014 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.
Please cite this article in press as: Zhao Y, et al., Black tea consumption and serum cholesterol concentration: Systematic review and metaanalysis of randomized controlled trials, Clinical Nutrition (2014), http://dx.doi.org/10.1016/j.clnu.2014.06.003
2
Y. Zhao et al. / Clinical Nutrition xxx (2014) 1e8
Cochrane Library (http://www.thecochranelibrary.com). We used the text words: tea, black tea and Camellia sinensis, paired with the following words: blood lipid, blood cholesterol, low density lipoprotein cholesterol and high density lipoprotein cholesterol. All searches were restricted in English-language publications. An additional manual search was conducted using reference lists from original research papers, previous meta-analysis and review articles. To be included, a study must meet all the following criteria: (1) conducted in adults; (2) randomized controlled trials of either parallel or crossover design; (3) used black tea as the only active treatment intervention; (4) provided available data to calculate the difference between baseline and endpoint for cholesterol measures including total cholesterol, LDL cholesterol and HDL cholesterol; (5) with an intervention duration 2 weeks and a sample size 10. Besides, a study will be excluded if it employed pregnant or severely ill participants.
there was no obvious heterogeneity among trials, otherwise the random-effects model developed by DerSimonian and Laird was adopted to calculate the pooled effect [13]. Possible publication bias was examined by Begg's test [14], a P < 0.1 indicated potential publication bias may exist. Trim and fill method was used to correct potential publication bias and corresponding filled funnel plot was also presented for visual inspection of publication bias [15]. To explore the source of heterogeneity, we carried out a series of predesigned subgroup analyses. Subgroup analyses were implemented by comparing trials stratified by study and population characteristics including healthy status of subjects, duration, study design, daily flavonoids intake from black tea, type of intervention and control. Furthermore, we also conducted additional sensitivity analyses to test if the results were steady [10].
2.2. Data extraction and quality assessment
3.1. Results of the study selection
Data were extracted independently by two researchers and any discrepancy was resolved via discussion. Study characteristics were extracted including author, country, study design, intervention duration, sample size, population information, type of intervention and control, diet change. In each trial, the means and SDs of cholesterol measures at baseline and endpoint in both intervention and control groups were extracted. For studies that had multiple time points for the same participants, only the last endpoint was used for analysis. If a study adopted different intervention protocols, intervention group which met the inclusion criteria was used. Extracted data were converted to conventional units and for cholesterol 1 mmol/L was converted to 38.6 mg/dL. When SD was not reported directly, we calculated them from SEM and 95% CI using the following equations according to the Cochrane handbook for systematic reviews of interventions [11]:
The PRISMA flow diagram is shown in Fig. 1. Finally we identified ten studies that met the inclusion criteria [16e25]. Total of 597 potentially relevant titles and abstracts were identified from searches on PubMed, Scopus and the Cochrane Library together with other reference sources. After ruling out animal studies and those irrelevant to the aim of this meta-analysis, 27 articles were retrieved as full text and assessed for inclusion. We excluded studies with a duration shorter than two weeks [6] or with a small sample size less than 10 [26]. Furthermore two studies were excluded because black tea was used as a part of multicomponent supplement in intervention group [27,28]. Three articles were ruled out because they used Pu'er tea as intervention treatment [29e31]. Pu'er tea is a sort of wet-fermented tea despite it was called Chinese black tea in these studies. Another ten studies were excluded due to a lack of sufficient detail for inclusion or a cross-sectional design.
SD ¼ SEM
pffiffiffi n
(1)
3. Results
3.2. Study characteristics
And
pffiffiffi n ðupperlimit lowerlimitÞ÷3:92
The R is the correlation coefficient. To be conservative, a minimum correlation coefficient of 0.5 was used. Quality assessment was conducted in duplicate independently and quality characteristics included the following items: randomization, random sequence generation, allocation concealment, blinding of both participants and researchers, description of dropouts. The Jadad score was adopted to assess the quality of each study and trials scored one point for each area addressed [12]. Trials with Jadad score 4 were classified as high quality. Discrepancy was resolved through discussion until a consensus was reached.
The characteristics of all the included studies are listed in Table 1. The research of Bahorun et al. [16] was separated into 2 trials (effects on males and females respectively). Therefore, 11 trials involved 411 adult participants were included in this metaanalysis. Six trials from 5 studies were conducted in healthy people [16,17,21,23,25] while the other five trials were performed in subjects with higher cardiovascular risk such as prediabetes or hypercholesterolemia [18e20,22,24]. The sample size ranged from 15 to 77 with a median of 31. Six trials lasted for 4 weeks [17,19e21,23,25], four trials lasted longer than 10 weeks [16,22,24] and the average duration was up to 8 weeks. Only two trials were conducted in East Asian countries [21,24] and the rest were mostly performed in America or Europe. Black tea extract capsule was used in two trials [22,24] and the remaining adopted black tea beverages as the intervention treatment. Controlled background diet was adopted only in one trial [18]. Eight trials used parallel design [16,20e25] and the rest 3 trials adopted crossover design [17e19].
2.3. Data synthesis and analysis
3.3. Quality assessment
Statistical analysis was performed using Stata/SE 12.0 for Windows (StataCorp, College Station, TX, USA). Heterogeneity among trials was assessed by I2 statistic and Cochrane Q test. We suggested there was no significant heterogeneity among trials when I2 4 weeks). Because four of the ten studies did not report the dose of daily flavonoids intake from black tea [17,20,22,24], we did not perform subgroup analysis stratified by daily flavonoids intake from black tea. For trials which employed subjects with mild hypercholesterolemia [18,20,24], coronary disease [19] or prediabetes [22], we treated them together as ‘trials performed in subjects with higher cardiovascular risk’. The results of subgroup analyses showed that the pooled effect of total cholesterol or HDL cholesterol was not influenced by type of intervention, type of control, study design, duration or healthy status of participants (Table 2). However, the subgroup analysis stratified by healthy status manifested that regular intake of black tea significantly reduced LDL cholesterol concentration in subjects with higher cardiovascular risk (8.26 mg/dL; 95% CI: 15.13, 1.40 mg/dL; P ¼ 0.018) rather than in healthy subjects (2.22 mg/dL; 95% CI: 7.83, 3.39 mg/dL; P ¼ 0.438). Besides, the pooled effect of trials with intervention duration longer than 4 weeks showed a marginally significant reduction in LDL cholesterol concentration (8.49 mg/dL; 95% CI: 17.42, 0.44 mg/dL; P ¼ 0.062) but not in those trials with shorter duration (3.45 mg/ dL; 95% CI: 8.42, 1.53 mg/dL; P ¼ 0.174). A marginally significant reduction also was detected in the trials which used black tea extract as intervention treatment (9.87 mg/dL; 95% CI: 19.92, 0.19 mg/dL; P ¼ 0.054) other than those trials using black tea beverage (3.44 mg/dL; 95% CI: 8.26, 1.38 mg/dL; P ¼ 0.162). Sensitivity analyses showed that the pooled effects of total cholesterol, LDL cholesterol and HDL cholesterol together with the corresponding CIs were not obviously changed when using a coefficient R ¼ 0.68 which was used in a previous meta-analysis [32]. Random-effects model was suggested to be more conservative and could provide a broader CI. The significance in the pooled effects in each cholesterol measure was not altered after we used randomeffects model to calculate the pooled effects. Besides the inclusion
of three studies which used Pu'er tea as intervention treatment did not significantly influence the final results. 3.5. Publication bias Results from Begg's rank correlation test suggested that no obvious publication bias was detected in the meta-analysis of LDL cholesterol (P ¼ 0.754) or HDL cholesterol (P ¼ 0.858), but not in total cholesterol (P ¼ 0.074). After trim and fill method was adopted, the pooled effect and corresponding CI of total cholesterol remained stable (Table 3). Filled funnel plot of each cholesterol measure was presented for visual inspection (Fig. 5). 4. Discussion The aim of the present study was to examine the relationship between black tea consumption and serum cholesterol concentration. This meta-analysis of 10 randomized controlled trials showed a significant reduction in serum concentration of LDL cholesterol after black tea administration, however no remarkable effect was detected on total cholesterol or HDL cholesterol. No obvious heterogeneity among studies was observed. Subgroup analyses suggested that the mean differences in total cholesterol and HDL cholesterol concentration were not affected by type of intervention or control, study design, healthy status of participant or study duration. However the effect of black tea consumption on LDL cholesterol concentration was more effective in subjects with higher cardiovascular risk like hypercholesterolemia. The results of sensitivity analyses indicated that the conclusions of the present meta-analysis were steady. Flavonoids, especially for theaflavins and catechins, are the major bioactive components in black tea. Black tea makes an important contribution to dietary intake of flavonoids. Epidemiological evidence recommended that serum cholesterol decreased with increasing tea intake and the daily consumption of >3 cups of black tea may protect against stroke and reduce all-cause mortality
Please cite this article in press as: Zhao Y, et al., Black tea consumption and serum cholesterol concentration: Systematic review and metaanalysis of randomized controlled trials, Clinical Nutrition (2014), http://dx.doi.org/10.1016/j.clnu.2014.06.003
6
Y. Zhao et al. / Clinical Nutrition xxx (2014) 1e8
Table 2 Results of subgroup and sensitivity analyses. Variables
Total cholesterol
LDL cholesterol
No. of Pooled effect (95% CI) P for No. of Pooled effect (95% CI) trials mg/dL heterogeneity trials mg/dL Subgroup analysis Type of intervention Black tea beverage 8 Black tea extract 2 Type of control Placebo 4 Water 6 Healthy status Healthy 6 With higher 4 cardiovascular risk Duration >4 weeks 3 4 weeks 7 Design Parallel 7 Crossover 3 Sensitivity Included Pu'er tea studiesa 13 Coefficient R ¼ 0.68 10 Using random-effects model 10 a
HDL cholesterol P for No. of Pooled effect (95% CI) P for heterogeneity trials mg/dL heterogeneity
3.16 (8.26, 1.94) 1.19 (7.46, 9.85)
0.492 0.222
6 3
5.19 (10.44, 0.07) 3.47 (11.20, 4.26)
0.536 0.632
7 3
2.43 (5.05, 1.19) 0.06 (2.58, 2.70)
0.477 0.994
1.93 (8.38, 4.53) 2.14 (8.14, 3.86)
0.398 0.339
4 5
4.71 (10.74, 1.31) 4.56 (10.84, 1.71)
0.214 0.720
4 6
0.20 (2.56, 2.16) 2.88 (6.07, 0.31)
0.983 0.383
0.62 (6.30, 5.06) 4.16 (11.09, 2.77)
0.267 0.657
5 4
2.22 (7.83, 3.39) 0.566 8.26 (15.13, 1.40) 0.609
6 4
1.50 (3.64, 0.63) 0.17 (3.93, 4.26)
0.249 0.996
0.24 (9.74, 10.23) 2.59 (7.48, 2.30)
0.272 0.448
4 5
8.49 (17.42, 0.44) 3.45 (8.42, 1.53)
0.740 0.358
4 6
0.89 (5.39, 3.61) 1.20 (3.29, 0.89)
0.227 0.724
0.51 (5.90, 4.88) 5.07 (12.65, 2.52)
0.293 0.817
7 2
3.73 (8.68, 1.23) 7.69 (16.75, 1.36)
0.493 0.440
8 2
1.18 (3.26, 0.89) 0.97 (5.65, 3.71)
0.410 0.914
4.68 (12.32, 2.96) 2.45 (6.09, 1.18) 2.04 (6.43, 2.35)
