Why the Government's war on alcohol will fall on deaf ears
Many public health studies distort data evidence in order to further what they see as righteous ends, writes Eoin O'Malley
'There's no fire alarm without fire." That will never make it as a popular aphorism because as anyone who works in a large building knows, it just isn't true.
Most of us experience fire drills on a regular basis. Sometimes, for me, it's at the start of a class, and you know that the planned class is ruined.
Sometimes you're in the middle of a problem, and the fire alarm goes off, you're tempted to lock your door and continue working. Because at this stage I know that fire alarms don't mean fire.
We're now so used to false alarms that we'd want to see flames, or at least smell smoke, before getting remotely concerned.
The same is true of the alarming headlines generated by nutrition research. Nutritional research - the stuff that tells you that chocolate wards off cancer, or that coffee gives you heart attacks - is notoriously unreliable.
One week wine is good for you, the next it kills. What are we to do?
Some of the problems with nutritional research come from small sample sizes - there are just too few people in the study to be certain that the effect is real.
So when a massive new study on the medical impact of alcohol consumption based on data from nearly 600,000 people was published last week in The Lancet, many were interested in what it said.
Its findings were pretty stark, according to the authors, who argued that it should encourage big policy changes along the lines the Irish Government proposes in the Public Health (Alcohol) Bill.
That bill is coming under fire from European colleagues and the drinks industry for the proposed regulations it wants to impose, including health warnings and minimum unit pricing.
With a study this size you can't do experiments - give one group a particular treatment (say a high alcohol intake) for a period of time, watch what happens to them compared with a controlled group who don't receive the high-alcohol treatment. Instead you do an observational study.
That involves looking at a lot of people, asking them some questions about themselves and their lifestyles and then, in this case, looking at follow-up health changes - big ones, such as dying - in the following years.
This study found that for a man aged 40 drinking well over the UK recommended maximum amount of alcohol, he could expect to die about two years earlier.
This might have been intended to shock, though my initial reaction was that it seemed a fair trade off.
For those at the current recommended amount, you might expect to die six months earlier than you otherwise would.
If that means dying aged 81-and-a-half instead of aged 82, again it hardly seems a terrible hardship for enjoying a drink.
It is one thing dying, but most of us would prefer to die healthy. What about chronic diseases?
Well here the news for drinkers is even better. Unless you are an alcoholic, there appears to be a slight drop in the likelihood of heart disease for moderate drinkers.
Even the heaviest drinkers in the study only saw modest rises.
The news wasn't so good for strokes. Heavy drinkers were more likely to get strokes. But don't worry, the data shows a stroke is still very unlikely, just not as unlikely as it had been.
All this assumes the study is a good one. It should be. Remember 600,000 participants.
But actually it is not. Because it is an observational study, saying there is a causal relationship between drinking and health outcomes is harder than the authors of the study admit.
There are two main problems. One big one is with what are called confounders. Confounding variables are things that influence both those things we're studying, here health outcomes and drinking.
So what else could cause people to have heart disease or die early? Well smoking obviously would, and they control for that, which is good.
But what about the type of job you do, or your diet, or your social class?
It could be that people who work in stressful jobs and have a poor diet are heavy drinkers.
The alcohol might not be causing the slightly earlier death rates recorded. They didn't control for this, which isn't so good. Another problem is that the data is based on people's reported drinking patterns.
As most of us will privately attest, we drink more than we'd like to admit to, so relying on self-reporting will underestimate how much drink is actually consumed.
That means that any negative effects observed might actually be for much higher levels of alcohol consumption that this study reports.
Some types of drinking are harmful to the drinker's health. This is verifiable from research. Unfortunately many in the public health lobby drop the 'some types of' and just announce that 'drinking is harmful to the drinker's health' and leave it at that.
They fixate on costs, but deliberately ignore the benefits of alcohol.
Many public health studies suffer from what is called White Hat bias - that is where researchers distort evidence or make greater claims than a fair reading of the data would allow, in order to further what they see as righteous ends. We've seen examples in public health, but also in climate change research.
The problem for well-meaning researchers is that when they do this they can have the opposite to their intended effect.
Despite the stark claims of the public health lobby, the current recommended alcohol limits seem unrealistically low, and if the Government does go ahead with its war on alcohol, insisting on warning labels on wine bottles and reducing the recommended intake guidelines, most of us will not notice them.
That's the problem with so many false alarms, they make us deaf.
- Eoin O'Malley is director of the MSc in Public Policy at the School of Law and Government in DCU