Breaking down the daily grind

17:00, May 25 2012
NOT ALL THAT: Coffee probably isn't keeping you alive.

You probably saw the headlines last week: "Coffee is good for you", "Coffee and good health go together", and perhaps a little optimistically "Coffee Drinkers Have Lower Risk of Death".

They all describe a study published last week in the New England Journal of Medicine.

The study certainly provides good news for coffee drinkers, but it's not quite time to count your early morning flat white as part of your health regime.

The good news is that coffee is not killing you. There were some good reasons to think a coffee habit might have a negative effect on health. Doctors know that caffeine raises your blood pressure and drinking coffee can increase production of the "bad" cholesterol known as LDL.

Evidently these effects are not harmful enough to dent the life-expectancy of coffee drinkers.

The study that elicited all those headlines looked at information from about 400,000 members of a US National Institute of Health study on diet and found coffee drinkers were, all else being equal, slightly /less likely to die over the 14-year period of the study.

So, that cup of coffee isn't just helping you make it through another day, it's actually prolonging your life at the same time? Well, maybe, but there are few reason to be a little sceptical about that result.

First, the life-prolonging effect detected by the study is pretty small. Almost every article on the story quoted the fact that a woman who drinks 4 or 5 cups of coffee a day was 16 per cent less likely to die over the course of the study.

That, of course, is the largest of 8 effects reported in the study, and it's not really all that big.

Journalist love statistics of the form "x  per cent more or less likely to die" (called relative risks), but these statistics often inflate the importance of the result they represent.

In order to know what a 16 per cent decrease in risk means we need to know what the base level of risk is.

In this case, around 10 per cent of the women included in the study died over the  14 years it ran, so a 16 per cent decrease in risk moves the background risk from 10 per cent to 8.4 per cent.

For this result a 16 per cent decrease in relative risk is a 1.6 per cent decrease in risk of dying during the study (the "absolute risk").

A 1.6 per cent decrease in risk is not nothing, but even that effect might disappear on closer inspection.

This study used an "observational" design, meaning participants weren't randomly assigned to a particular doses of coffee, rather, people simply reported the amount of coffee they drank.

That's really the only way you can do large-scale studies of diet, but it makes it hard to tie any effects these studies find to a particular cause.

The coffee study is a case in point. If the researchers had looked only at the raw numbers of their study they would have concluded that coffee drinking was a road to an early death.

This result arises from the fact people who drink a lot of coffee are more likely to be smokers. It was only after the effect of smoking was taken out of the picture that the protective effects of coffee emerged.

It is possible that the life-extending properties of coffee are the result of another spurious correlation.

So, by all means enjoy your coffee, but it might be a bit early to conclude that its keeping you alive.

David Winter is an Evolutionary Biologist at the University of Otago who blogs about bugs, biology and leading a sceptical life at The Atavism. From time-to-time he'll be analysing some science stories that have been in the news.