“More than four cups of coffee a day increases the risk of an early death,” The Daily Telegraph warns, on the basis of a large, long-term – but rather flawed – study.
The result that captured the media’s attention was men aged under the age of 55 who drank 28 cups or more of coffee a week had a 56% increased risk of premature death, and women in the same group had double the risk, compared to non-coffee drinkers of a similar age.
No increase in death risk was found in those over 55 or those who consumed lower amounts of coffee.
However, the study also found that confounders – that is, other influences such as smoking – were important in muddying the link between coffee and death rates. This pointed to the possibility that drinking lots of coffee might be an indication of a generally less healthy lifestyle in these people, which may cause higher death rates in the long term.
Taken on its own, this study does not prove that drinking less coffee makes you live longer, or that drinking three “grandé” americanos each day does any direct harm.
However, drinking such a large amount of coffee is not generally recommended, as while it may not shorten your lifespan, it could lead to symptoms of anxiety, irritability and insomnia.
The study was carried out by researchers from US and UK universities and was funded by US National Institutes of Health grants.
The study was published in the peer-reviewed medical journal Mayo Clinic Proceedings.
The media reporting was generally accurate, but didn’t fully discuss the limitations of the research. The Daily Mail deserves some praise for pointing out that greater coffee consumption may be a sign of a generally less healthy lifestyle, which could explain the higher death rates. Other explanations (or speculations) offered by the media included that coffee could increase blood pressure.
This was a cross-sectional study looking at whether coffee consumption was linked to death from any cause, or death from cardiovascular disease – one of the most common causes of death in the UK.
There have been many studies investigating coffee and health, some have found coffee consumption to be beneficial and some detrimental, so the situation is not clear.
Unfortunately, a lot of the research uses a cross-sectional design. The problem with this is that it cannot prove cause and effect, only that some link exists. This means that a lack of clarity remains despite continued study.
Furthermore, there are many confounders that could affect death rates, such as smoking, age, and alcohol consumption, so it is tricky to isolate the influence of coffee within this mix of other large and complex factors.
The researchers used lifestyle and medical information collected from the Aerobics Center Longitudinal Study, which had 43,727 adult participants who were monitored from 1970 to 2002. Data on lifestyle, including coffee consumption, was collected by face-to-face interview and a medical examination gathered information on blood pressure, blood chemistry and cardiovascular fitness levels, among other things.
Information on deaths was taken from electronic death registers and death certificates.
The researchers compared different levels of reported coffee consumption to see if any were linked to an increased risk of death from any cause, or from any specific cause such as cardiovascular disease. They analysed the potential link in men and women separately.
Some of the analysis adjusted for other factors (confounders) known to be linked to both coffee consumption and risk of death, such as smoking and exercise levels, which would muddy the results.
The average time the participants were followed up in the study was 17 years. Over this time there were 2,512 deaths from any cause, with about a third (804, 32%) due to cardiovascular disease.
Men and women who reported consuming higher amounts of coffee were more likely to smoke and had lower levels of cardiovascular fitness.
A crude statistical model of coffee consumption showed higher levels of coffee consumption was linked to higher rates of death from any cause, and death from cardiovascular disease for both men and women. However, once confounders such as physical activity, body mass index, smoking, alcohol and tea consumption were also factored into the analysis, the link either disappeared or was reduced to a level where it failed to reach a statistically significant level. This means that the link could have been the result of chance.
Further sub-group analysis reported that men and women under the age of 55 who drank more than 28 cups of coffee per week (the highest category recorded in the study) had a higher risk of death from any cause than those who drank no coffee.
Men’s risk of death increased 56% if they were in this high coffee drinking category (hazard ratio (HR) 1.56, 95% confidence interval (CI) 1.04 to 1.40), while the risk for women more than doubled (hazard ratio 2.13, 95% CI 1.26 to 3.59). Coffee consumption was not associated with death from any cause in men and women over the age of 55.
Analyses of the death rates from cardiovascular disease for older and younger groups were not performed, because no link was found between coffee consumption and death from cardiovascular disease for all ages.
No statistically significant links were found between coffee consumption and death from cardiovascular disease once confounders were included in the analysis.
The researchers reported a “positive association between coffee consumption and all-cause mortality [death from any cause] was observed in men and women younger than 55 years. On the basis of these findings, it seems appropriate to suggest that younger people avoid heavy coffee consumption (averaging >4 cups per day). However, this finding should be assessed in future studies of other populations”.
This study suggests men and women under the age of 55 who drink large amounts of coffee (more than 28 cups a week) may be at a higher risk of death from any cause than those who report drinking no coffee.
The study was large and covered a wide age range (20 to 90 years), but it also had many drawbacks, meaning the results may not be accurate or reliable overall.
One of the main limitations was the issue of “confounding”. The results clearly showed that smoking, alcohol, exercise levels and cardiovascular fitness (confounders) were in part responsible for the reported relationship between coffee consumption and death rates. For instance, not taking account of these (and other) confounders led to a statistically significant link reported between higher coffee intake and death.
But this link disappeared when the confounders were accounted for in the analysis. This means that some of the link between coffee and death rates was because of these other factors, rather than coffee itself. So it is possible that there are other confounders, not measured in the study or not present in the analysis, that may be causing all or part of the residual link between coffee intake and death reported in under 55 year olds.
The study also only measured coffee consumption at one time point. It did not assess any changes in drinking habits over a person’s lifetime. This means the categories are only a snapshot of coffee intake and may not be an accurate picture of average coffee consumption over a lifetime.
The mechanism by which coffee might cause death or disease is not clear, but there are some theories. One of the study’s co-authors explained in The Guardian that “the exact mechanism between coffee and mortality still needs clarification. Coffee is high in caffeine, which has the potential to stimulate the release of epinephrine, inhibit insulin activity, and increase blood pressure”. This could lead to a higher risk of certain diseases and possibly death.
The reason why the under-55s were the only group to show a significant link was also not clear. One explanation would be that coffee may be linked to deaths that occur more frequently in under-55s. This would mean the potential risk of coffee-related deaths would peak somewhere under the age of 55, and reduce after 55 to a point where it no longer is a risk. This is highly speculative and is not proven. The authors did not put forward a clear explanation for this age-specific effect. Instead a co-author reinforced the possibility that it may have been an unreliable result possibly due to confounding by suggesting "heavy coffee consumption behaviour might cluster with other unhealthy behaviours such as sleeping late, and eating a poor diet" and this could be responsible for the higher death rates.
Taken on its own this study does not prove that drinking less coffee makes you live longer, or that drinking even very high levels of coffee (greater than 28 cups a day) does any harm. However, drinking such a large amount of coffee may be a sign of a generally unhealthy lifestyle, which may go some way to explain the link that was observed.
It is important to remember that in its caffeinated form, coffee is a stimulant, so drinking large amounts of it can lead to unpleasant side effects such as irritability, problems sleeping, restlessness and even in some cases, nausea and vomiting.