The effect of coffee consumption on serum lipids: a meta-analysis of randomized controlled trials

Article metrics

Abstract

Background/objectives:

Numbers of epidemiological studies assessing coffee consumption and serum lipids have yielded inconsistent results. We aimed to evaluate the effects of coffee intake on serum lipids.

Subjects/methods:

We searched several English and Chinese electronic databases up to September 2011 for randomized controlled trials of coffee on serum lipids. Weighted mean effect size was calculated for net changes in serum lipids by using random-effect models or fixed-effect models. Subgroup and meta-regression analyses were conducted to explore possible explanations for heterogeneity among trials.

Results:

Twelve studies conducted in Western countries with a total of 1017 subjects were identified. Meta-analyses showed, on average, drinking coffee for 45 days was associated with an increase of 8.1 mg/dl (95% confidence interval (CI): 4.5, 11.6; P<0.001) for total cholesterol (TC), 5.4 mg/dl (95% CI: 1.4, 9.5; P=0.009) for low-density lipoprotein cholesterol (LDL-C) and 12.6 mg/dl (95% CI: 3.5, 12.6; P=0.007) for triglyceride (TG). The increase in TC were greater in trials using unfiltered coffee and caffeinated coffee as the treatment group. Those who had hyperlipidemia were more sensitive to the cholesterol-raising effect of coffee. Meta-regression analysis revealed a positive dose-response relation between coffee intake and TC, LDL-C and TG.

Conclusion:

The intake of coffee especially unfiltered coffee is contributed significantly to the increase in TC, LDL-C and TG, and the changes were related to the level of intake. Studies of coffee intake on serum lipids in Asian populations should be performed.

Introduction

Coffee is one of the most commonly consumed beverages and even small health effects of coffee could have considerable public health consequences. Controversies regarding its risks and benefits still exist as substances in coffee may have either unfavorable or beneficial effects on the cardiovascular system.1 A large body of epidemiology studies, including recent studies,2, 3, 4, 5 examining the association between coffee consumption and serum lipids have yielded inconsistent results.

A meta-analysis of nine randomized controlled trials (RCTs) published before December 1998 concluded that consumption of unfiltered, but not filtered, coffee increased serum levels of total cholesterol (TC) and low-density lipoprotein cholesterol (LDL-C).6 However, this meta-analysis was only based on English-language literature and the authors failed to include two important RCTs,7, 8 which included 345 participants. In addition, the active factors such as caffeine and coffee oils, which were responsible for the plasma lipoprotein-increasing effect of coffee, were not determined.

To determine the effect of caffeinated coffee and decaffeinated coffee consumption on plasma lipoprotein cholesterol levels in healthy men, Fried et al.9 performed a RCT by three intervention coffee including 720 ml/day of caffeinated coffee, 360 ml/day of caffeinated coffee and 720 ml/day of decaffeinated coffee. They found that only men in the third group had mean increases in plasma levels of TC (0.24 mmol/l), LDL-C (0.17 mmol/l) and high-density lipoprotein cholesterol (HDL-C) (0.08 mmol/l). In another RCT, 181 men consumed a standard caffeinated coffee for 2 months followed by randomization to continue caffeinated coffee (control), change to decaffeinated coffee or no coffee for 2 months.10 They reported that change from caffeinated to decaffeinated coffee increased plasma LDL-C, which suggests that a coffee component other than caffeine is responsible for the LDL-C change. Therefore, whether the changes were attributable to the caffeine or not remained unclear.

Much of the currently available information on the effects of coffee consumption is derived from epidemiological research. Clinical trials used various intake levels of coffee consumption with varying type of coffee and different protocols.7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 Meta-analysis is a statistical technique in which results of separate studies are combined to increase statistical power, clarify results and more accurately estimate the size of treatment effects. To more precisely evaluate the effects of coffee intake on serum lipids and to clarify the active factors of coffee, we performed a meta-analysis of 12-RCTs.

Materials and methods

Literature search

PubMed (updated till September 2011), the Cochrane Controlled Trials Register, EMBASE (1985–September 2011), Science Citation Index Expanded, China National Knowledge Infrastructure (1979–September 2011), VIP database for Chinese Technical Periodicals (1989–September 2011) and Wanfang database (1982–September 2011) were used to search articles (in English and Chinese) that described RCTs investigating the effect of coffee consumption on serum lipids.

Selection of studies

Titles, abstracts and subject headings in the databases were searched with the use of the following Boolean phrases: (‘coffee’ or ‘caffeine’ or ‘coffea’ or ‘caffeic acid’ or ‘cafestol’ or ‘kahweol’) and (‘cholesterol’ or ‘lipoproteins’ or ‘apolipoproteins’ or ‘triglyceride’ (TG) or ‘lipids’). We also examined all references of related reviews and papers identified by the search. We did not try to contact the experts for the obtaining of unpublished data.

Studies were selected for analysis if they met all of the following criteria: (1) subjects ingested coffee for at least 7 days; (2) the RCTs included a parallel control group; and (3) TC, LDL-C, HDL-C or TG was used as an index of lipid profile. Reasons for the exclusion of studies were: (1) nonrandomized treatment allocation; (2) a lack of concurrent control group; (3) insufficient data to calculate the net change in serum lipids and their variances. If the study sample was found to overlap with that in another article or if two articles described aspects of the same study, only the publication with the largest sample was used. If the study reported some comparisons, we included all comparisons in the meta-analysis.

Data extraction

Data from original articles were abstracted independently by two reviewers (Li Cai and Zhaoyan Liu) using a standardized data collection form. Any disagreements between the two reviewers were resolved through discussion with a third reviewer (Defu Ma). The following data were collected: (1) author and publication year; (2) population information including country, sample size, age, sex, healthy status and baseline coffee consumption; (3) study characteristics, including study design, type of intervention and control treatment, cups of coffee/day during the trial and intervention duration; and (4) lipid profile, including baseline and final concentrations or net changes of TC, LDL-C, HDL-C and TG.

Statistical analysis

The mean lipid profiles at baseline and at the end of intervention period were used to calculate mean net change for each study. For parallel trials, net change in serum lipids was calculated as the mean difference (coffee minus control) of the change (follow-up minus baseline) in serum lipids. For crossover trials, net change was calculated as the mean difference in values between the end of the coffee supplementation and control periods. The variances for net changes in serum lipids were not reported directly in some studies. Therefore, they were calculated from confidence intervals (CIs) and P values for intervention and control groups (parallel trials) or intervention and control periods (crossover trials) by using standard methods.21 The variance of paired differences was not reported in one parallel trial,7 we calculated it from the variance at baseline and at the end of follow-up by using correlation coefficient methods.21

To calculate the pooled mean net changes, each comparison was assigned a weight consisting of the reciprocal of its variance. Weighted mean effect of coffee consumption on lipid profile and the corresponding 95% CIs were calculated. Heterogeneity of effect size across trials was tested by Q statistics in which statistical significance was established at P<0.10. We also calculated the I2 statistic, which describes the proportion of total variation in study estimates that is caused by heterogeneity. Random-effects model was used in the meta-analysis of TC, LDL-C and TG.

To explore the potential influences of study designs and participant characteristics on net change in lipid profile and identify the possible source of heterogeneity within studies, we further conducted prespecified subgroup analyses stratified by study design, type of coffee, brewing methods, intervention duration, intervention dose, age, sex and hyperlipidemic status. When information on mean age was missing in one trial, an average of the minimum and maximum values for that trial was used.17 Sensitivity analysis was conducted to investigate the influence of a single trial on the overall effect estimated by omitting one study in each turn. Univariate and multivariate meta-regression models were employed to explore the influence of coffee dose or study duration on net change in lipid profile. Covariates included in the multivariate model were coffee dose, coffee brewing method (filtered or unfiltered) and study duration.

Publication bias was assessed with funnel plots, which plotted the s.e. of the studies against their corresponding effect sizes. In addition, Egger's linear regression test and Begg's rank correlation test were conducted to detect publication bias. When publication bias was identified, a nonparametric trim and fill method was performed to adjust the publication bias. All analyses were conducted in STATA (version 9.2; Stata Corp., College Station, TX, USA). P<0.05 was considered statistically significant, except where otherwise specified.

Results

The trial flow chart was illustrated in Figure 1. Our literature search identified 36 trials including 2 of them obtained from the reference lists. Twenty four studies were excluded because of non-randomization (2 studies), lack of a control group (20 studies, including one study in Chinese) and insufficient original data (2 studies). Thus, 12 studies with a total of 1017 subjects were included in this meta-analysis.

Figure 1
figure1

Flow chart of articles identified and evaluated in the review.

The characteristics of the trials included were shown in Table 1. The trials varied in sample size from 12 to 261. A crossover design was used in six trials and others used parallel design. Caffeine coffee was used in the total 12 studies, and decaffeinated coffee was used in 3 studies. Unfiltered coffee was used as intervention in 12 comparisons and filtered coffee was used in other comparisons. The control group received no coffee (17 comparisons) or tea (5 comparisons). The duration of treatment varied widely, ranging from 14 to 79 days (average, 45 days). Coffee intake varied from 2.4 to 8.0 cups/day in the various treatment groups. Nine studies were conducted in Europe, including three in Netherland, two in the United Kingdom, two in Finland, one each in Italy and Norway and three were performed in the United States. Six of these 12 studies were focused on men and the remanding studies included both sexes. In five comparisons, the participants had hyperlipidemia at baseline according to the definition of the original study. All studies reported no significant differences regarding baseline characteristics such as age, body mass index, TC, LDL-C, HDL-C and TG between groups. There were also no significant weight changes or negative side effects reported.

Table 1 Characteristics of study design, type of intervention and study populations in the included studies

In this meta-analysis, the pooled estimate of the effect of coffee consumption on TC in 12 studies was an increase of 8.1 mg/dl (95% CI: 4.5, 11.6; P<0.001) (Figure 2). Eight studies with 15 comparisons reported values of LDL-C before and after coffee or control treatments (Figure 2), and the pooled estimate was 5.4 mg/dl (95% CI: 1.4, 9.5; P=0.009). When the 9 studies with 17 comparisons reported values of HDL-C were combined, the mean difference was −0.1 mg/dl (95% CI: −0.6, 0.4; P=0.644) in HDL-C (Figure 2). The pooled mean net change for TG in 7 studies (11 comparisons) was 12.6 mg/dl (95% CI: 3.5, 12.6; P=0.007) (Figure 2). In the sensitivity analyses, none of the trials omitted in each turn seemed to substantially influence the treatment effect (data not shown).

Figure 2
figure2

Meta-analysis of the effect of coffee consumption on TC (a), LDL-C (b), HDL-C (c) and TG (d) as compared with control. WMD, weighted mean difference.

Subgroup analyses of the effects of coffee intake on TC, LDL-C and TG was shown in Table 2. Interestingly, we found that caffeine coffee had significant effects on LDL-C, TC and TG but decaffeinated coffee had not. Moreover, unfiltered coffee exerted significant increasing effects on TC, LDL-C and TG comparing with that of filtered coffee. The increase in TC, LDL-C and TG were more pronounced in studies with coffee intake 6 cups/day or studies performed in both male and female. There was higher increasing effect on TC in hyperlipidemic than that in normal subjects (28.9 mg/dl vs 5.6 mg/dl). However, after excluding the hyperlipidemic participants, coffee intake still had a significant effect on LDL-C, TC and TG in normal subjects.

Table 2 Subgroup analyses of TC, LDL-C and TG in stratified by study design, type of intervention and study populations

Univariate and multiple meta-regression analyses for cups of coffee administered, and duration of intervention were conducted (Table 3). In the multiple model, a significant dose-response relation between coffee consumption and TC, LDL-C and TG was found. Results from the multiple model confirmed that coffee dose was independently associated with net change in TC (P=0.006), LDL-C (P=0.006) and TG (P=0.003), even after adjusted for types of coffee and study duration. Coffee dose was therefore proved to be a significant and independent predictor for heterogeneity, strengthening the results of the subgroup analyses. There was also a tendency for a positive association between duration of intervention and serum TC levels after adjusted for cups and types of coffee (P=0.051).

Table 3 Meta-regression models investigating effect of amount of coffee and study duration on TC, LDL-C and TG

The potential publication bias was detected by funnel plots, Egger's test and Begg's test. Funnel plots suggested asymmetry in the meta-analysis of TC, but not in the mea-analysis of LDL-C, HDL-C and TG (data not shown). Egger's test and Begg's test yielded similar results with funnel plots: TC (Egger P=0.01; Begg P=0.01), LDL-C (Egger P=0.40; Begg P=0.73), HDL-C (Egger P=0.62; Begg P=0.43) and TG (Egger P=0.23; Begg P=0.1).

To correct the publication bias identified in the meta-analysis of TC, the trim and fill method was employed. After filling in the missing study identified by the trim and fill method and recalculating the pooled estimate of TC, we found that coffee drinking increased TC by 7.8 mg/dl (95% CI: 4.1, 11.5; P<0.001), which changed little compared with 8.1 (95% CI: 4.5, 11.6).

Discussion

This meta-analysis showed a significant increase in TC, LDL-C and TG, but not in HDL-C, during coffee interventions. The effect of coffee drinking on lipid profile was more pronounced in trials in which the participants drank boiled coffee, caffeinated coffee, had hyperlipidemia, drank more coffee and were older. Meta-regression analyses further revealed significant positive associations between coffee consumption and the net change in TC, LDL-C and TG.

These dose-response relationships we found are consistent with observational studies. One prospective study reported that drinking one cup of regular coffee a day was associated with about a 2-mg/dl increase in TC over 16.7 months of follow-up after adjustment for age and changes in other potential confounders.22 Nystad et al.5 found that total coffee consumption was positively associated with TC levels. Yamashita et al.2 also reported that coffee consumption showed positive association with TC and LDL-C, even though consumption of unfiltered coffee would be very little in the study sample. In this meta-analysis, the summary effect for every cup/day increment in coffee intake was 3.74 mg/dl in TC, 3.38 mg/dl in LDL-C and 6.56 mg/dl in TG. In observational studies, the effect of coffee additives, such as cream and milk, might be responsible for the association between coffee consumption and higher serum cholesterol levels. However, this is not the case in the RCTs used in this meta-analysis because use of additives was controlled in almost all trials.

In the subgroup analysis, the effect of coffee drinking on lipids was more prominent in trials with a shorter intervention duration (<8 weeks). However, a tendency that change in TC was associated with longer intervention duration was shown; and there was no association between coffee intervention duration and change in LDL-C or TG, after adjusted for cups and types of coffee. The lack of statistical significance in the present study may be due to the various choice of comparison group, different intervention populations or confounding. It should be noted that all the trials included in this meta-analysis focused on Western populations. Only one trial23 investigating the effect of coffee consumption on serum lipids in Chinese was found and excluded owing to a lack of control group. This Chinese study reported that consumption of 5 g caffeinated coffee/day for 1 month exerted significant favorable effects on serum lipids in young women, which is inconsistent with our study. More trials are needed to investigate the health effect of coffee intake in Asian population before definitive conclusions can be made.

This study suggests that sex may be a modifier of the effect of coffee on the lipid profiles. The effect of coffee on lipids was significantly larger in trials that included both men and women, compared with trials that included only men. Sex differences in lipid profiles in response to coffee have not been particularly investigated in previous RCTs. Further studies should be conducted which specifically enroll women.

Whether the cholesterol-raising effect of coffee is partly due to caffeine remained unclear. In our meta-analysis, only three randomized control trials involving decaffeinated coffee have been identified.9, 10, 13 Results indicated that caffeinated coffee had significant effects on TC, LDL-C and TG but decaffeinated coffee had not. It was reported that a significant dose-response relationship was found between caffeine consumption and TC in women but not in men, regardless of source of caffeine.24 In another cross-sectional study, after adjusting for age and adiposity, the mean serum cholesterol level was 11 mg/dl higher for women consuming 200 mg or more of caffeine per day compared with those consuming less.25 Nevertheless, a recent study found that serum caffeine concentrations were closely positively related to TG in caffeine-drug users, but no associations were found between serum caffeine concentrations with TC or LDL-C levels either in caffeine-drug users or nonusers.26 In addition, Du et al.26 concluded that chronic intake of caffeine (exclusively from diet source) might slightly increase HDL-C concentrations in women. Caffeine has been shown to exhibit several biological effects, such as increased fat oxidation and mobilization of glycogen in muscle, increased lipolysis and decreased body fat.27 Further confirmation is required in large prospective studies, especially RCTs.

It is nowadays a general viewpoint that coffee oils, such as cafestol and kahweol, are responsible for the lipid-raising effects of coffee consumption.27, 28 Consistent with this viewpoint, we found that trials using boiled or unfiltered coffee had a stronger cholesterol-raising effect than did those using filtered coffee. Boiled coffee has a higher concentration of coffee oils because of the higher temperatures used during its preparation and the longer contact time between the coffee grounds and water.15 Filtration of coffee through a paper filter can remove the most of coffee oils from the coffee extract.29 Coffee oils, including cafestol and kahweol, increase the synthesis of cholesterol by decreasing excretion of bile acids and neutral sterols.27

The lipid-raising effects of coffee drinking pose a possible threat to coronary health. In general, case–control studies have found high-coffee intakes to be associated with significantly increased risk of CHD.30 However, cohort studies generally have not supported a significant association.31, 32, 33 An up-to-date meta-analysis of prospective studies indicated that moderate coffee consumption may be weakly inversely associated with risk of stroke.34 Coffee contains several biologically active substances other than diterpenes and caffeine, such as chlorogenic acid, flavonoids, melanoidins and various lipid-soluble compounds.35 The antioxidants present in coffee are useful in lowering the risk of coronary heart disease.36 For example, chlorogenic acid improves the antioxidative status of the body and reduces LDL oxidation;35 chlorogenic acid and caffeic acid slow down the process of inflammation, hence providing protection from the hazardous effect of free radicals and against endothelial damage.37

Our study has several strengths. In the previous meta-analysis,6 the study search was not comprehensive. Two primary studies7, 8 published before 1998 addressing the research question were not identified. In addition, one trial20 that met the inclusion criteria was published in 2000. These three trials7, 8, 20 involved 409 participants were included in our study. Second, the trials included in our meta-analysis were all well-conducted RCTs, which provide much stronger support for a causal association than observational studies. Furthermore, improved statistical methods, such as meta-regression analysis and Egger's test, were employed in our analysis, making the results more convincible.

It is noteworthy that the changes of TC and TG were smaller and greater, respectively, compared with the previous meta-analysis.6 With regard to TC, one possible explanation for this difference is that we further included two trials7, 8 (345 participants) with a coffee dose <6 cups/day. On the other hand, results from Grubben et al.20 showed that consumption of 1l unfiltered coffee/day for 2 weeks significantly raised TG concentrations by 35.4 mg/dl, which had a great impact on the pooled mean net change for TG.

The limitations of this study should be considered. Although we searched the articles in English and Chinese, we did not find any Chinese paper meeting our inclusion criteria. In addition, we found some evidence for publication bias in the meta-analysis of TC. However, after filling in the missing study identified by the trim and fill method and recalculating the combined value of TC, we found that coffee drinking increased TC by 7.8 mg/dl, which means that the publication bias had little effect on the results. Furthermore, the sensitivity analysis showed minimal influence on the combined results for any single trial.

In conclusion, our results suggested that the intake of coffee especially unfiltered coffee is contributed significantly to the increase in TC, LDL-C and TG. Coffee dose is a potential modifier to the effect of coffee consumption on serum lipids. Coffee oils as the active factors are a major contributor to the cholesterol-raising effect of coffee. More studies need to be conducted on the effect of coffee on serum lipids particularly in Asian populations.

References

  1. 1

    Butt MS . Sultan MT. Coffee and its consumption: benefits and risks. Crit Rev Food Sci Nutr 2011; 51: 363–373.

  2. 2

    Yamashita K, Yatsuya H, Muramatsu T, Toyoshima H, Murohara T, Tamakoshi K . Association of coffee consumption with serum adiponectin, leptin, inflammation and metabolic markers in Japanese workers: a cross-sectional study. Nutrition and Diabetes 2012; 2: e33.

  3. 3

    Balk L, Hoekstra T, Twisk J . Relationship between long-term coffee consumption and components of the metabolic syndrome: the Amsterdam Growth and Health Longitudinal Study. Eur J Epidemiol 2009; 24: 203–209.

  4. 4

    Naidoo N, Chen C, Rebello SA, Speer K, Tai ES, Lee J et al. Cholesterol-raising diterpenes in types of coffee commonly consumed in Singapore, Indonesia and India and associations with blood lipids: a survey and cross sectional study. Nutr J 2011; 10: 48.

  5. 5

    Nystad T, Melhus M, Brustad M, Lund E . The effect of coffee consumption on serum total cholesterol in the Sami and Norwegian populations. Public Health Nutr 2010; 13: 1818–1825.

  6. 6

    Jee SH, He J, Appel LJ, Whelton PK, Suh I, Klag MJ . Coffee consumption and serum lipids: a meta-analysis of randomized controlled clinical trials. Am J Epidemiol 2001; 153: 353–362.

  7. 7

    D'Amicis A, Scaccini C, Tomassi G, Anaclerio M, Stornelli R, Bernini A . Italian style brewed coffee: effect on serum cholesterol in young men. Int J Epidemiol 1996; 25: 513–520.

  8. 8

    Caan B, Limb ES, Sweetnam MP, Fehily AM, Amarah L, Hutchings A . Instant coffee and cholesterol: A randomised controlled trial. Eur J Clin Nutr 1995; 49: 779–784.

  9. 9

    Fried RE, Levine DM, Kwiterovich PO, Diamond EL, Wilder LB, Moy TF et al. The effect of filtered-coffee consumption on plasma lipid levels. Results of a randomized clinical trial. JAMA 1992; 267: 811–815.

  10. 10

    Superko HR, Bortz WJ, Williams PT, Albers JJ, Wood PD . Caffeinated and decaffeinated coffee effects on plasma lipoprotein cholesterol, apolipoproteins, and lipase activity: a controlled, randomized trial. Am J Clin Nutr 1991; 54: 599–605.

  11. 11

    Van Dusseldorp M, Katan MB, Van Vliet T, Demacker PNM, Stalenhoef AFH . Cholesterol-raising factor from boiled coffee does not pass a paper filter. Arterioscler Thromb 1991; 11: 586–593.

  12. 12

    Rosmarin PC, Applegate WB, Somes GW . Coffee consumption and serum lipids: a randomized, crossover clinical trial. Am J Med 1990; 88: 349–356.

  13. 13

    Burr ML, Gallacher JE, Butland BK, Bolton CH, Downs LG . Coffee blood pressure and plasma lipids: a randomized controlled trial. Eur J Clin Nutr 1989; 43: 477–483.

  14. 14

    Bak AA, Grobbee DE . The effect on serum cholesterol levels of coffee brewed by filtering or boiling. N Engl J Med 1989; 321: 1432–1437.

  15. 15

    Aro A, Tuomilehto J, Kostiainen E, Uusitalo U, Pietinen P . Boiled coffee increases serum low density lipoprotein concentration. Metabolism 1987; 36: 1027–1030.

  16. 16

    Forde OH, Knutsen SF, Arnesen E, Thelle DS . The Tromso heart study: coffee consumption and serum lipid concentrations in men with hypercholesterolaemia: an randomised intervention study. Br Med J (Clin Res Ed) 1985; 290: 893–895.

  17. 17

    Aro A, Kostiainen E, Huttunen JK, Seppala E, Vapaatalo H . Effects of coffee and tea on lipoproteins and prostanoids. Atherosclerosis 1985; 57: 123–128.

  18. 18

    Christensen B, Mosdol A, Retterstol L, Landaas S, Thelle DS . Abstention from filtered coffee reduces the concentrations of plasma homocysteine and serum cholesterol--a randomized controlled trial. Am J Clin Nutr 2001; 74: 302–307.

  19. 19

    McAnlis GT, McEneny J, Pearce J, Young IS . Black tea consumption does not protect low density lipoprotein from oxidative modification. Eur J Clin Nutr 1998; 52: 202–206.

  20. 20

    Grubben MJ, Boers GH, Blom HJ, Broekhuizen R, de Jong R, van Rijt L et al. Unfiltered coffee increases plasma homocysteine concentrations in healthy volunteers: a randomized trial. Am J Clin Nutr 2000; 71: 480–484.

  21. 21

    Higgins J, Green S Cochrane Handbook for Systematic Reviews of Interventions Version 5.0.0 [updated February 2008]. The Cochrane Collaboration, 2008 Available from www.cochrane-handbook.org.

  22. 22

    Wei M, Macera CA, Hornung CA, Blair SN . The impact of changes in coffee consumption on serum cholesterol. J Clin Epidemiol 1995; 48: 1189–1196.

  23. 23

    Bai Z, Gao X, Han H, Bian Y . Effects of noise and caffeine on serum lipids. Occup Med 1992; 6: 326–327.

  24. 24

    Mathias S, Garland C, Barrett-Connor E, Wingard DL . Coffee plasma cholesterol, and lipoproteins. A population study in an adult community. Am J Epidemiol 1985; 121: 896–905.

  25. 25

    Shirlow MJ, Mathers CD . Caffeine consumption and serum cholesterol levels. Int J Epidemiol 1984; 13: 422–427.

  26. 26

    Du Y, Melchert HU, Knopf H, Braemer-Hauth M, Gerding B, Pabel E . Association of serum caffeine concentrations with blood lipids in caffeine-drug users and nonusers—results of German National Health Surveys from 1984 to 1999. Eur J Epidemiol 2005; 20: 311–316.

  27. 27

    Ranheim T, Halvorsen B . Coffee consumption and human health--beneficial or detrimental? --Mechanisms for effects of coffee consumption on different risk factors for cardiovascular disease and type 2 diabetes mellitus. Mol Nutr Food Res 2005; 49: 274–284.

  28. 28

    Urgert R, Katan MB . The cholesterol-raising factor from coffee beans. Annu Rev Nutr 1997; 17: 305–324.

  29. 29

    Cornelis MC, El-Sohemy A . Coffee caffeine, and coronary heart disease. Curr Opin Lipidol 2007; 18: 13–19.

  30. 30

    Higdon JV, Frei B . Coffee and health: a review of recent human research. Crit Rev Food Sci Nutr 2006; 46: 101–123.

  31. 31

    Lopez-Garcia E, Van Dam RM, Willett WC, Rimm EB, Manson JE, Stampfer MJ et al. Coffee consumption and coronary heart disease in men and women: a prospective cohort study. Circulation 2006; 113: 2045–2053.

  32. 32

    Kawachi I, Colditz GA, Stone CB . Does coffee drinking increase the risk of coronary heart disease? Results from a meta-analysis. Br Heart J 1994; 72: 269–275.

  33. 33

    Reis JP, Loria CM, Steffen LM, Zhou X, van Horn L, Siscovick DS et al. Coffee, decaffeinated coffee, caffeine, and tea consumption in young adulthood and atherosclerosis later in life: the CARDIA study. Arterioscler Thromb Vasc Biol 2010; 30: 2059–2066.

  34. 34

    Larsson SC, Orsini N . Coffee consumption and risk of stroke: a dose-response meta-analysis of prospective studies. Am J Epidemiol 2011; 174: 993–1001.

  35. 35

    Cornelis MC, El-Sohemy A . Coffee caffeine, and coronary heart disease. Curr Opin Clin Nutr Metab Care 2007; 10: 745–751.

  36. 36

    Butt MS . Sultan MT. Coffee and its consumption: benefits and risks. Crit Rev Food Sci Nutr 2011; 51: 363–373.

  37. 37

    Corti R, Sudano I, Spieker L, Binggeli C, Hermann F, Toenz D et al. Coffee--poison or medicine? Ther Umsch 2005; 62: 629–633.

Download references

Author information

Correspondence to D Ma.

Ethics declarations

Competing interests

The authors declare no conflict of interest.

Rights and permissions

Reprints and Permissions

About this article

Keywords

  • coffee
  • caffeine
  • lipids
  • meta-analysis
  • randomized controlled trials

Further reading