Don’t worry, there is such a thing as ‘safe drinking’
You might have seen some startling headlines the other day. Alcohol is “never good for people under 40,” according to the Guardian. The Daily Mail reports that “under-40s should avoid ALL alcohol for the sake of their health,” but … ”a small glass of red can cut risk of heart disease, stroke and diabetes in older adults”.
What the hell’s going on? Alcohol is bad for you if you’re 39 but good for you if you’re 41? How does this make sense?
Life begins at 40, apparently
These stories are based on a paper published last week in the Lancet, a “meta-analysis” which aggregates the findings of several previous studies. It’s carried out by the Global Burden of Disease (GBD) team at the University of Washington, Seattle. And it does indeed say that the evidence supports ”stronger interventions, particularly those tailored towards younger individuals”, to reduce alcohol intake. In fact, the lead author of the study, Professor Emmanuela Gakidou, goes so far as to say in the press release that “Our message is simple: young people should not drink, but older people may benefit from drinking small amounts.”
But, to cut a long story short, this is a really strange piece of research that lumps together lots of things that probably shouldn’t be lumped together, gives advice that is of no help to individuals or to policymakers, and – most bizarrely of all – implicitly assumes that the only reason people drink alcohol is for the good of their health. Sir David Spiegelhalter, the Winton Professor for the Public Understanding of Risk at the University of Cambridge, told the i that “Everything’s wrong with this. Absurd logic, inappropriate interpretation of the results, completely unjustified recommendations for policy.”
To explain what the problem is, we need to look at why all those earlier studies talked about a protective effect. People who drink loads and loads are more likely to die in any given year than people who don’t drink at all: their “all-cause mortality” is higher. But, interestingly, people who only drink a small amount have lower all-cause mortality. On a graph, you see the risk dip down at first and then swoosh back up, in (sort of) the shape of a J. This “J-shaped curve” is a consistent finding, and it’s usually associated with cardiovascular disease – heart failure, stroke, high blood pressure, that sort of thing.
So there you go, right? Moderate drinking is good for you? Well – probably, but it’s still controversial. The main objection is that in societies like Britain, people who don’t drink at all are unusual. Perhaps a large percentage of them don’t drink because they’ve got some medical reason not to. There’s still an ongoing row in alcohol research about whether the J-shaped curve is real. “If you want to start a fight at an alcohol research conference, cardiovascular protection is the way to do it,” says Colin Angus, an alcohol policy researcher at the University of Sheffield.
In fact, the GBD team’s last, also widely reported, major meta-analysis got lots of headlines in 2018 for saying that the J-shaped curve isn’t real and concluding that there’s no safe level of alcohol for anyone.
This new paper contradicts that entirely – according to this, the J-shaped curve is real, but it’s much bigger for older people.
How they get that conclusion, though, is really strange. What they do is lump together all the bad things that alcohol can cause – from heart disease and stroke, to cancer, liver disease and pancreatitis, to drink-driving injuries, violence and suicide. And then they combine those into one big risk chart.
And when you do that, you see that, indeed, the apparent protective effect of alcohol is far greater among older people than younger people. But that’s extremely obvious, because – mercifully – young people don’t tend to die of cardiovascular disease, and in fact they don’t tend to die at all. But when they do die, it’s often from injury – and alcohol often causes that, whether by violence, suicide, drink-driving, or just falling over while drunk.
So, naturally, the protective effect of alcohol appears far smaller, in relative terms, for younger people, because the thing alcohol is believed to protect against doesn’t happen very often. It’s a bit like complaining that shark repellent doesn’t work very well in Wolverhampton because people who use it there are no less likely to be attacked by sharks than people who don’t.
The study also assumes that all forms of drinking are equally bad, but obviously a 22-year-old who drinks one drink a day is at much less risk of dying in an accident than one who drinks nothing for six days then has seven drinks on the Friday.
A theory of relativity
Stay with the sharks-in-the-Midlands analogy. Imagine you’ve just been told that wearing red swimming shorts raises your risk of being eaten by a shark in the Wolverhampton Swimming and Fitness Centre by 50 per cent.
Hang on, you might think. My risk of being eaten by a shark in Wolverhampton was pretty low. Let’s say it was one in three billion before. If you raise that by 50 per cent, it’s about one in 2 billion now. The relative risk has gone up quite a lot – 50 per cent! – but the absolute risk, one in three billion, is tiny, so you don’t care that much. But if you don’t know the absolute risk, then you just see the scary “Your risk has gone up by 50 per cent” statistic.
That’s why it’s really important to report the absolute risks. And in fact the Lancet’s own guidelines make it clear that papers should always give absolute risk, not just relative. But the GBD paper doesn’t.
Which is a shame, because – obviously – older people are at much higher absolute risk of dying from alcohol-related issues. Roughly speaking, according to the GBD’s own data on their website, about 10 in every 100,000 people aged 20 to 24 die of alcohol-related causes each year. About 140 people aged 70-74 do. “For a young person, a five per cent increase from bugger all is still bugger all,” says Angus. “While a one per cent increase for an older person might be lots bigger. They need to compare the absolute risks, and they could have done, but they just didn’t.”
Of course, if a 22-year-old dies, many more years of life are probably lost than if a 72-year-old does. But the paper doesn’t talk about that. And according to Angus, when you do dig out the GBD’s own data, it suggests that when you do look at “disability-adjusted life years lost (DALYs), “they skew younger, inevitably, but they still don’t suggest we should worry about the under-40s more than the over-40s”.
Here’s a sense of the scale of the risk. In 2018, the last time the GBD published a meta-analysis, the Lancet’s press office made up for the scientists’ oversight and gave absolute risks. Writing at the time, Spiegelhalter pointed out that you’d need to stop 25,000 moderate drinkers – people drinking one drink a day – from drinking in order to prevent a single serious health problem. Most of us would consider that level of risk “safe”.
A Jägerbomb a day keeps the doctor away
Picture the J-shaped curve. If you’re a non-drinker, you have a certain level of risk of the various diseases we call “alcohol-related” – things like cancer and heart disease. (People get those things even if they don’t drink, of course.)
If you’re a moderate drinker, your risk is slightly lower – either because of a real protective effect, or some confounding caused by non-drinkers being unhealthy for some reason.
And as you look at people who drink more, the risk climbs back up, until it reaches, and then passes, the risk for non-drinkers.
The point at which they meet is called the “non-drinker equivalent”. And that’s what the GBD considers the “safe level” of drinking: when your risk is the same as someone who doesn’t drink.
But we would never think in those terms for anything else we enjoyed: if we like scuba-diving or skiing, we don’t think the safe level is “how much scuba-diving I could do before my risk rises above that of someone who never scuba-dives”. Obviously, that would be zero, because scuba-diving has small, but real, risks. Instead we think how much risk am I comfortable taking, for a given amount of fun?
And you can’t do that if you don’t, firstly, acknowledge the absolute risks – I might be willing to trade a one-in-10,000 chance of death for a scuba-diving holiday, for instance – and secondly, acknowledge that these things have value to people, beyond whether or not they extend our lives.
“You’d think the point of research like this is to help people make decisions about how much or whether to drink,” says Kevin McConway, an emeritus professor of statistics at the Open University. “Or to help public health bodies make recommendations.” But this study – which, again, got lots of attention in the press – doesn’t help with that. “All it does is shout at people: ‘If you’re young, drinking is worse for you than not drinking.’ But people don’t drink because it’s good for them, they do it because they like it.”
To be clear – drinking lots and lots of alcohol isn’t good for you. But the idea that there is “no safe level” of alcohol for young people involves a definition of the word “safe” that we wouldn’t use in any other context – there’s no “safe level” of driving, or eating peanuts. “It’s absurd logic,” says Spiegelhalter. “By that thinking, people should not do anything that’s harmful to their health. Mountain-biking? Shouldn’t do that. Should we even get out of bed? Life has risks.”
#Dont #worry #safe #drinking