Does coffee cut stroke risk?

Monday September 19 2011

“Two cups of coffee every day ‘may cut stroke risk’,” reported the Daily Mail . It said that “a comprehensive analysis of the health benefits of coffee has confirmed it can have a powerful preventative effect against one of Britain’s biggest killers”.

As reported, this was an analysis of the health benefits of coffee, incorporating a systematic review and meta-analysis of 11 previous studies of whether coffee consumption is linked to stroke risk. It found that, compared with people who drank none or minimal coffee, those who drank moderate amounts (between two and six cups a day) had a lower risk of having a stroke.

This review was well-conducted, but is limited by the fact that there were several major differences between the individual studies. Also, the participants were only asked about their coffee consumption once at the start of the study and then followed for up to 20 years. Many studies have investigated whether or not coffee has health effects, some finding it to be beneficial, and others harmful. Although this research found an association between coffee and stroke, it does not confirm that drinking coffee reduces the risk of stroke.

Where did the story come from?

The study was carried out by researchers from the Karolinska Institute in Sweden. Funding was provided by the Swedish Council for Working Life and Social Research and the Karolinska Institute.

The study was published in the peer-reviewed American Journal of Epidemiology .

The media reports of how the study was carried out were generally accurate. More emphasis could have been placed on the limitations of this review, which prevent any firm conclusions from being made.

What kind of research was this?

This was a meta-analysis of the findings from several previous studies of whether coffee consumption is associated with the risk of stroke. This analysis pooled the data from these 11 studies, which in total included more than 10,000 cases of stroke in 479,689 participants.

A meta-analysis is a type of research method that pools the results of multiple studies. Such pooling can increase the ‘power’ (or ability) to detect an association, and decrease the likelihood that any found associations were due to chance. As the number of subjects included in a study increases, the power of the study increases as well. However, systematic reviews are often limited by the methodological quality of the individual studies.

What did the research involve?

The researchers searched two databases for prospective cohort studies examining the association between coffee and stroke that had been published between 1966 and 2011. To be included in the analysis, studies had to meet the criteria of having measured at least three categories of coffee consumption (for instance, 0 to 1 cups, 2 to 3 cups, and 4 or more cups a day), and having calculated the relative risk of stroke for each of these categories. The three levels were required in order to detect whether the association between drinking coffee and risk of stroke changed depending on how much coffee was consumed. The researchers also collected data on the age and sex of the participants, and the location and year of the studies.

The researchers extracted data from each of the studies, including the average amount of coffee consumed (median and mean) and the relative risk of stroke. These data were pooled and used to estimate the relative risks for different levels of coffee consumption. The pooled data were then separated into five groups:

  • a reference group that was equal to the lowest category of consumption in each study (for example, some studies classed this as none or never, some as less than one cup a day and some as less than one cup a month)
  • fewer than three cups a day
  • three to four cups a day
  • five to six cups a day
  • seven or more cups a day

The relative risks for each of these groups were calculated and compared to the lowest group to estimate the link between different levels of coffee consumption and the risk of stroke.

The researchers then carried out statistical analyses of their results by subgroups, including study location, sex, years of follow-up and stroke subtype, to detect whether or not any of these factors confounded the relationship between coffee consumption and stroke risk.

What were the basic results?

The researchers identified 138 articles in their literature search. They excluded 127 articles because they did not meet the inclusion criteria, which left 11 studies to be included in the meta-analysis. In total, the 11 studies reported 10,003 cases of stroke among 467,689 participants. Seven studies were carried out in Europe, two in the US and two in Japan. Individual studies accounted for various risk factors for stroke, such as age, smoking status, level of alcohol consumption, history of diabetes, history of high blood pressure, level of physical activity and diet.

The researchers found that compared to drinking no coffee:

  • Those who drank one cup of coffee a day had an 8% reduced risk of stroke (Relative Risk [RR] = 0.92, 95% CI 0.89 to 0.96).
  • Those who drank two cups of coffee a day had a 14% reduced risk of stroke (RR = 0.86, 95% CI 0.78 to 0.94).
  • Those who drank three to four cups of coffee a day had a 17% reduced risk of stroke (RR = 0.83, 95% CI 0.74 to 0.92).
  • Those who drank six cups of coffee a day had a 13% reduced risk of stroke (RR = 0.83, 95% CI 0.74 to 0.92).
  • There was no significant reduction in risk of stroke with drinking eight cups of coffee a day (RR = 0.93, 95% CI 0.79 to 1.08).

When the researchers removed three studies that had included patients with a history of heart attack and diabetes, the results did not change substantially. However, when they pooled the data into four categories (fewer than three cups a day, three to four cups a day, five to six cups a day and seven or more cups a day), only the lowest category was statistically significant (RR = 0.88, 95% CI 0.86 to 0.90).

Subgroup analysis revealed that the relative risks were similar in different geographic locations and throughout the follow up period. The results did not change substantially between men and women either. When the researchers analysed the effect of coffee on different types of stroke, coffee had a similar effect for both ischaemic (due to a clot) and haemorrhagic (due to a bleed) strokes. However, this association was only statistically significant in the ischaemic group.

How did the researchers interpret the results?

The researchers concluded that moderate coffee consumption was weakly associated with a reduced risk of stroke. That is, the more coffee consumed the lower the risk of stroke, up to a point. They say that the strongest association occurred at three to four cups of coffee a day, which equated to a 17% lower risk of stroke.


This was a large meta-analysis of prospective cohort studies that had examined the association between drinking coffee and the risk of stroke.

The meta-analysis was well devised and carefully carried out. However, as well as having strengths it is subject to several weaknesses.
Cohort studies are well suited for examining associations between different factors. As these studies were also prospective (following people over time) the studies could also collect information on potential confounders (which might confuse the association) and take them into account. This increases the confidence that this relationship is not due to other factors.

Meta-analyses have the advantage of a larger sample size than any single study, which improves the power to detect a difference. However, they rely heavily on the quality of the individual studies. The results of a meta-analysis are only as good as the design of its component studies.

The researchers say that using prospective cohort studies should eliminate some of the bias that can influence meta-analyses. They also say that many of the included studies had a large number (from around 1,600 to more than 120,000) of participants, and followed them up for a long time (2 to 24 years), which improves the credibility of the individual data.

However, the researchers also point out that the individual studies had a large limitation, in that all but one collected information of coffee consumption only once, at the start of the study. As the studies had a large follow-up period, there is no way to confirm that the amount of coffee consumed did not change over 2 to 25 years.

The manner in which meta-analyses are reported often makes it difficult to assess the quality of the underlying studies. The studies were in varied populations. They looked at different age groups, some looked at mixed populations and some looked at only men or women. However, other details of these populations are not given. Importantly, it is not possible to tell whether all participants were free from history of stroke, mini-stroke (TIA) or other cardiovascular disease at the study start. If the person already had cardiovascular disease at the time that they were asked about their coffee consumption it would not be possible to assess the link between the two. Also, the individual studies appear to have varied widely regarding the potential confounders that they adjusted for in their analyses.

The researchers do say that a meta-analysis of randomised controlled trials would have been preferable to the observational studies that were used. However, they say that such trials are expensive and difficult to implement due to the lifestyle nature of the exposure (coffee consumption) and the long period of follow-up that would be needed in order to allow for a reasonable number of the outcomes (in this case strokes) to be observed.

The researchers point out that coffee is a complicated mixture of substances, and as such may affect health in both positive and negative ways. They say that some of the substances may benefit one’s health through their action on low-density lipoprotein cholesterol (LDL, or ‘bad’ cholesterol) and sensitivity to insulin. On the other hand, some research has suggested that caffeine consumption is associated with increased hypertension. These are theories that this research is not able to assess.

Overall, this study suggests that consuming coffee in moderation is unlikely to increase your risk of stroke, but it can’t tell us anything else relating to the positive or negative health effects of coffee. As it is not possible to say that consuming coffee directly causes reduction in stroke risk, if you don’t already drink coffee, this study does not provide a reason to start.

Analysis by Bazian
Edited by NHS Choices