How a typo tricked the media: half a beer a week won’t harm your health

Eggs were on faces yesterday after some astonishingly bad science reporting in sections of the media. In articles that have since been taken offline, the Telegraph and Mirror announced that half a pint of beer a week is sufficient to harden the arteries and cause heart disease. Given the almost homeopathic quantities of alcohol involved — not to mention the fact that moderate drinking is known to reduce heart disease risk — this was a rather surprising finding, but it had supposedly been published in the Journal of the American Heart Association and was therefore considered legit.

Alarm bells started ringing when the Telegraph came up with this eyebrow-raiser:

‘The UK study defined consistent long-term heavy drinking as equivalent to drinking one serving of alcoholic spirit, half a pint of beer or half a glass of wine per week.’

The idea that drinking half a pint of beer once every seven days constitutes ‘long-term heavy drinking’ is patently ludicrous. There are some strange ideas floating around in the world of ‘public health’ these days, but things are not quite that mad. You have to wonder how that sentence got written, let alone approved and printed, without somebody at the Telegraph saying ‘surely that can’t be right’?

The rest of the story hinged on this basic error. The study itself is pretty good. Its authors set out to see whether heavy drinking stiffens people’s arteries, because arterial stiffness is a predictor of cardiovascular disease. They used a database of civil servants stretching back to the 1980s and measured something called pulse wave velocity (PWV) to gauge the state of their arteries. The higher the number, the harder the arteries.

The authors note that previous research has found the relationship between PWV and alcohol consumption to be J-shaped, which is to say it is relatively high for non-drinkers, lower for moderate drinkers and high again for heavy drinkers. This is significant because the relationship between cardiovascular disease and alcohol consumption is also J-shaped, ie moderate drinkers have a lower risk than both those who abstain and those who drink heavily.

The authors set out to test this and succeeded. Defining heavy drinkers as anyone who consumed more than 14 units a week, and measuring in metres per second (m/s), they found PWV levels among men of 8.8 m/s for non-drinkers, 8.3 m/s for moderate drinkers and 8.7 m/s for heavy drinkers. Among women the readings were 8.6 m/s, 7.9 m/s and 8.3 m/s.

Over the years, the PWV levels rose, but it was the non-drinkers and ex-drinkers who saw their levels rise the most. By contrast, the authors note that ‘stable moderate drinkers have the lowest PWV values throughout the study period’.

This seems to confirm the J-curve, but how does it support the claim that drinking half a pint a week gives you heart disease? It doesn’t. The study found the exact opposite of what the Telegraph and Mirror claimed. It showed that moderate drinkers, including those who limit themselves to a swift half once a week (if such people exist), have a lower risk of heart disease than those who never drink at all. So how did the press get it so wrong?

If you are going to rely on reporting science by press release, you have to be confident in the press release. Unfortunately for the fourth estate, this one was a stinker. The study’s authors defined anyone who drank more than 112 grams of alcohol a week as a ‘heavy drinker’. There are eight grams of alcohol in a unit, therefore they were drinking more than 14 units, but when the press release tried to explain this, it all went wrong:

‘Consistent long-term, heavy drinking was defined in this UK study as more than 112 grams (3.9 ounces) of ethanol per week (roughly equivalent to one serving of alcoholic spirit, half a pint of beer, or half a glass of wine); consistent moderate drinking was 1-112 grams of ethanol per week.’

The first part is true and so is the last bit. It is the middle section that’s the problem. 112 grams is by no means ‘roughly equivalent to one serving of alcoholic spirit… [etc]’. It is 14 servings of alcoholic spirit. I can’t be sure how this error crept in. My hunch is that whoever wrote it meant to put ‘eight grams of ethanol is…’ just before ‘roughly equivalent…’, but failed to do so.

Whatever the reason for this slip, it was dutifully repeated by sections of the media and the bizarre notion that ‘long-term heavy drinking’ is defined as one unit a week was born. And since the study found that heavy drinkers have harder arteries, this turned into ‘Just half a pint of beer a week increases risk of heart disease — new study’ (you can read the now deleted Telegraph article here thanks to the Wayback Machine).

I hesitate to call anything a new low in health reporting but this is definitely in the same postcode as the nadir. I think there are two lessons that can be learned from it.

Firstly, if proof were needed that some reporters do not read the research they write about, this is it. It is glaringly obvious that the journalists in this instance did not give so much as a cursory glance to the abstract, let alone to the tables. This is worth bearing in mind next time you read a health report in a newspaper.

Second, it says something about the outlandish claims made by ‘public health’ academics that a journalist would find it perfectly believable that they have not only started to define ‘heavy drinking’ as one small drink a week, but that they view this as a potentially lethal dose.

Daft as this may seem, it has to be put in the context of previous assertions by the ‘public health’ lobby which have been accurately reported under such headlines as ‘Cancer risk of two beers a year’, and ‘No safe level of drinking, health chiefs warn’. If you spend your working hours reading press releases from people who think that roast potatoes and buttered toast are the new asbestos, it can’t be easy to separate fact from fantasy.

The Telegraph and Mirror did such a bad job of reporting this particular story that they made the rest of the media look like Pulitzer prize winners, but the truth is that everybody covered it pretty badly. The Sun ran with ‘Men who drink ONE pint a day are “increasing their risk of having heart disease or a stroke”‘ and the Daily Mail went with ‘How just one pint a day can increase the risk of heart disease by prematurely ageing the arteries’.

While both newspapers avoided the ‘half a pint a week’ booby trap, they used the lightest of ‘heavy drinkers’ as their example to make a claim that is not supported by the study. On average, the people in the study who drank more than 14 units a week had stiffer arteries than those who drank less, but averages can be misleading. There is nothing in the data to suggest that 14 units is the threshold at which risk increases. It could be 30 or 40 units for all we know.

More importantly, every newspaper ignored the crucial finding that the heavy drinkers only had an ‘increased risk of heart disease’ if you compared them to moderate drinkers. It was the non-drinkers – both ex-drinkers and lifelong teetotallers — who had the stiffest arteries of all, as the graph below shows. The moderate drinkers are shown by the line at the bottom, the non-drinkers are at the top.


This is biological evidence which supports the enormous quantity of epidemiological evidence showing the benefits of moderate drinking to the heart. It is the only reason anyone in academia would be interested in it and yet it is the one finding that went without mention yesterday.

Off the back of a typo in a press release, a study which found that moderate drinking protects against heart disease turned into a story about tiny quantities of alcohol causing heart disease. It’s a good job nobody takes this stuff seriously any more otherwise the public could become confused.


  • Airshod

    Two things are apparent here; the telegraph is just as crap as the mirror and we are obsessed with alcohol.

    • JonFrum

      These articles were no worse than the flood that fill every other newspaper – including the ones you read and rely on. Autism rates increase 10x in decades? Absurd on the face of it, but such has been reported repeatedly over the years. The truth is that newspaper reporters and editors are not trained in critical thinking – they only think they are. They are a remarkably ignorant bunch, happy to serve as conveyor belts for crap.

  • Chris Oakley

    It is hardly surprising that the Telegraph published this. It employs nobody who has any clue about science and operates a healthist editorial policy with a particularly strong anti-alcohol bias. I no longer read it because its editorial prejudices combined with journalistic incompetence make it even less credible than the Daily Mail. I am not sure who and what keeps it going. Why do allegedly quality newspapers believe that they can continue to get away with producing weapons grade junk?

    • Ans.: Because Leftist weapons-grade idiots like to have their prejudices confirmed.

      • Chris Oakley

        But the Telegraph is a right wing newspaper and the Tories are not short on authoritarian idiots either. Liam Fox springs to mind and of course the frequently disengenuous Sarah Wollaston.

        • Correction: the Telegraph WAS a CONSERVATIVE newspaper. Britain has not had a visible right wing since Oswald Mosley roamed the streets — and even then, there was no danger that they would woo anybody but the odd malcontent.

          THE LEFT is the enemy of freedom: in this country, and in every other one on the planet.

  • thammond

    I have my doubts about the paper from basics. The difference in PMV is less than 5%, and I’m sure the differences are variable, i.e. not every teetotaler has a PMV of 8.8. I’m also sure that the measurements have some error margin, that might be as large as 2-3%. It might be a lot more.

    And then the association between PMV and heart attack is pretty low – you can draw nice J-shaped graphs but if your relative risk is below 2, then all that means is you have lots of noise and no signal. There are so many confounding variables (genetics, exercise, diet, environment, smoking…) and perhaps some are not independent – those who drink a lot don’t exercise and eat a poor diet for example.

    Finally, the study uses self-reporting: it doesn’t measure how much alcohol the people actually drank, but uses what they say they drink. This means the underlying data is highly unreliable – people lie, they over and underestimate. So what you actually have is data that may have significant errors, mapped against unreliable data that may well be very wrong, with huge amounts of noise from confounding factors.

  • Clactonite

    I have two observations.
    1. The Telegraph has a very serious issue with anything numeric. They cannot even report income tax rates properly – something which is very readily available in Gov.uk and will take less than one second to determine.
    2. As a person in the medical profession Christopher, you may be familiar with the IT phrase which has a surgical bent “cut and paste”. No thought required or even proof reading. Ctrl+C, Ctrl+V. Job done, move on.

    Cheers

  • SteveW

    Another take home from this is that the consequence of the revised guidelines for alcohol consumption (down to 14 units a week) is enough to position the newly defined ‘heavy drinking’ group as a whole, as being at lower risk of cardiovascular disease than the teetotal group.

    Public health doing what public health do best.

  • Re: the headline: Good lord.What sort of hysterical ninny would imagine that it did?! It’s even stupid if we change the last word to ‘DAY’!
    Incidentally, just as it’s not true for health and good functioning that ‘a calorie is a calorie’, it is equally unscientific to assert that spirits = wine = beer = cocktails = any other alcoholic beverage.