Lots of headlines last week warned young people (those under 55) that drinking more than 28 cups of coffee a week – four cups a day – may lead to an increased risk of all cause mortality. In other words, too much coffee might be a potential killer.

The headlines are based on a study published in the Mayo Clinic Proceedings, and have been reported in the Guardian, Daily Mail and Telegraph, to name but three.

But should we be worried?

The study is enormous (just under 50,000 people were included), and looks at a wide range of ages, which is a real strength. However, they only assess coffee drinking once, at the start of the study. The participants were asked to report number of cups drank per week. But questioning wasn't any more nuanced than that. It can't really be: in a study this big, collecting the data would take far too long.

There's a huge variation in caffeine from cup to cup. Compare a three-shot monster coffee from a chain to a half-a-teaspoon weak cup of instant you make yourself; all this kind of detail isn't picked up in this study. The authors of the paper point out that coffee drinking is relatively stable across people's adult life, but it may still be that people who drank many cups at the beginning of the study may have cut down by the end. For example, coffee use might be stable during your working life, but after retirement you might cut down. And this is important, since their headline finding is that lots of coffee is only bad for younger people.

This age-specific risk got me thinking, too. They found no association between coffee and death when they looked at the sample as a whole, it was only when splitting it by age that an association was seen, in the younger group. Now, it's perfectly possible that sub groups might be affected differently by various exposures. But generally, if you want to look at subgroups, it's good to have a theoretical reason as to why there might be differences.

What could be different between older and younger people which would mean they would be differently affected by coffee? The paper doesn't really have any suggestions as to why they performed this analysis, so it's hard to know.

One suggestion the authors make as to why the association is only seen in younger people is that the older people in the study have already made it this far in life, so in effect have been selected to be stronger. Perhaps, but this isn't the only possibility, and doesn't seem like the most likely one, to me at least.

The study takes account of some other potential differences between coffee drinkers and non coffee drinkers which could affect the relationship (what are known as confounders – people who drink coffee might also be more likely to smoke, or have higher blood pressure, for example). The differences this study controls for are: other hot drink use (caffeinated or not), physical activity, body mass index (a measure of weight taking height in to account), smoking, alcohol use, diabetes, hypertension, cholesterol level, and family history of cardiovascular disease.

This is a big list, but it doesn't consider the differences that might explain the age specific findings. For example, younger people might drink a lot of coffee because they have a stressful job. Stress is not good for your health, but this isn't taken in to account. Or perhaps use of drugs in general (caffeine is a stimulant) is related to leading a more risky lifestyle, and the dangers that come with that.

The authors don't present any descriptive data in their paper split by age. We therefore don't know how many people in the sample were over or under 55, and crucially how many people who died were over or under 55. It's probably safe to assume that more people died who were over 55.

The authors very kindly sent me the numbers of people who died, split by both gender and age. Although I still don't know about their coffee-drinking habits, there were more than three times as many deaths in the older group versus the younger group. Younger people and older people are likely to die for different reasons, too. Without a closer look at causes of death it's difficult to really make sense of these findings.

Looking at the graph in the paper, which is split by both age and gender, the confidence intervals (a measure of accuracy of the findings) are quite wide. This is particularly true for the women. In fact, from their graph it can be seen that, in the highest coffee consumers, there may actually be no difference between the age groups or genders, as the confidence intervals all cross each other.

Another slightly unusual aspect of the findings is that there's no 'dose-response' effect. If there was something in coffee that increased your risk of death, you would expect that a little bit of coffee would increase your risk a bit, and more would be more harmful still. However, their results and conclusions suggest that it's only for extreme levels of coffee consumption that a risk is seen.

This again makes me wonder whether there's something else at play, rather than coffee having a direct causal effect. The authors suggest genes may be playing a role in the interaction between age and risk, but this is a hypothesis; there may be other explanations.

All of this doesn't mean that coffee is good for you, and it doesn't mean that it doesn't increase the risk of death. Like all drugs, it's likely that moderation is a better bet than excess. But this paper isn't enough to put me off a morning cuppa, particularly when compared to another study with a huge sample size which found quite the opposite effect. This paper in the New England Journal of Medicine last year found those who drank coffee were actually less at risk of death (though it's worth noting their sample was older, with an average age in the 60s).