Posted on March 18, 2015
Last week we saw glorious news for coffee addicts: drinking coffee prevents myocardial infarction, as well as famine, poverty and global warming (okay, maybe not). “Coffee prevents heart attacks”, read the Sky News headline. The media adored this story. “Three coffees a day cuts the risk of heart disease and strokes,” proclaimed the Mirror. The Telegraph lead with the slightly more accurate: “Three to five cups of coffee a day could reduce the risk of heart attack”, while the BBC try to avoid hyperbole by saying: “Regular coffee drinkers have ‘cleaner’ arteries”.  But even this is at best a simplistic summary of the conclusions reached by Yuni Choi et al in their paper published in the journal Heart.
Rather than investigating the risk of heart attack as the papers suggested, the researchers were examining the effects of coffee consumption on coronary artery calcium in asymptomatic Korean adults, the majority of whom were male (83.7%). Coffee consumption was measured through a self-administered questionnaire, and coronary calcium was measured by CT scan – incidentally that’s an awful lot of unnecessary radiation. It is quite a large study, with over 25,000 participants, but due to its cross-sectional nature, does not provide data concerning the long term effects of coffee on coronary calcium.
Coffee is unlikely to be a cardiovascular risk-free drink. Caffeine is a stimulant, and excessive consumption is thought to be proarrhythmic. The Heart paper says: “coffee consumption has also been associated with increased cholesterol concentrations,” and acknowledges, “concerns about a potential increase in cardiovascular disease risk associated with coffee consumption”. However, Choi and colleagues point to a recent meta-analysis that found moderate coffee consumption to be associated with a lower risk of heart disease. If one thing is clear, it is that this topic is not.
The Mirror begins: “Drinking three cups of coffee a day can slash your risk of a heart attack, new research reveals. There has previously been much debate on the effect of the beverage on cardiovascular health.” And these findings end that debate? This suggests that the Korean study is a panacea answering an age old question; it certainly is not.
Both the Guardian and the Mirror attempt to add credibility to their story by quoting statistics from the paper. Starting a paragraph with: “The ratio was 0.59 for those consuming three to five cups per day, compared with non-coffee drinkers,” is most unhelpful. What is this ratio of which they speak? Well, it’s a value comparing mean coronary calcium score between groups. Is this relevant? Does a higher score predict mortality? Who knows? The papers didn’t seem to think it was important to consider those points.
The news outlets also fail to mention that only the group who drank between three and five cups of coffee per day had a statistically significant difference in coronary artery calcium score when compared with those who drank no coffee. So when the Mirror says: “Their findings showed the calcium ratios were…0.81 for people having at least five cups or more every day compared with non-coffee drinkers,” they perhaps should have said: “we can be 95% sure that the true value for the ratio of coronary artery calcium scores for people having at least five cups of coffee per day compared with those who have no coffee lies somewhere between 0.46 to 1.43.”
The 95% confidence interval here is wide, and includes and exceeds 1. In other words, drinking lots of coffee might decrease the level of calcium in coronaries (as reported), but on the other hand it might increase considerably. Perhaps this is pedantic, but it is an important distinction. They don’t know what effect drinking five cups of coffee has.
Furthermore, the score ratio looks at the amount of calcium in arteries. The coronary calcium score ranges between 0 and >100, based on the CT scan findings. But coronary artery calcium was detected in only 13.4% of participants (the vast majority of participants had a score of 0). So if instead of the score ratio we look at the odds ratio of presence of coronary artery calcium (the chance of having coronary calcium score >0 compared to the control group), the results are less impressive. Again, only the three to five cups a day group reaches significance (OR 0.80, 95% CI 0.66 – 0.97); this is the only group in which the 95% confidence intervals for the odds ratio does not include 1. And, even in this group, the difference might be very small.
Because studies cannot include the whole population, there is some uncertainty around the true value of an outcome. Confidence intervals are calculated in order to express the range of this uncertainty. Crucially, if the CI crosses 1 there is very little evidence of a difference between the outcomes for the control group and the group that received the intervention.
No, not really.
The study doesn’t look into an association between coffee drinking and the risk of cardiovascular disease as the headlines suggest. Rather it uses a surrogate marker of coronary artery atherosclerosis. The research focuses on young healthy Koreans, not older Western adults at risk of heart disease and stroke – who the published articles were directed at. The statistics, where reported, are incomplete and therefore muddling. For the sake of a few column inches, you could argue that the public has been misled. Clearly the papers’ readers aren’t looking for a scientific commentary on the study, but they would presumably appreciate simple facts rather than what are at best tenuous interpretations of the evidence.
This is just the latest in a continual stream of research papers that are highlighted in the media. Reporting findings from medical literature is of course nothing new, particularly when they relate to diets, fat, smoking, drugs or alcohol – consider the recent news that the 30 year old guidance which says that fat is bad for us is wrong. Breakthroughs in health research should be publicised in the national media, but sensational headlines, even if they refer to valuable studies, may well harm the health of those who read them. It seems that any scientific paper presenting “new research”, especially if it involves tens of thousands of participants, must be impressive.
Perhaps this particular story won’t make a huge difference to the public’s health beliefs, but you could argue that in this case the evidence has been misrepresented in order to sell newspapers. That would be irresponsible.