Robert Matthews argues that the language used by researchers is only adding to confusion
Once and for all: is cod liver oil good for you?
The idea of staying healthy by popping pills rather than watching what we eat has always been appealing. But now researchers seem to have debunked that idea in the case of fish oil.
For years, it’s been claimed there is good evidence that taking supplements of the stuff can help ward off heart problems.
Studies of people who eat fish – especially the oily variety like mackerel – on a regular basis are found to be less likely to die from coronary heart disease.
In 2012, an analysis of international studies involving more than 300,000 people found that just a few servings of fish a week can cut the risk of heart-related death by 15 to 20 per cent.
The key ingredient is thought to be so-called omega-3 fatty acids, which are found in relatively high amounts in fish. So supplements containing this are a convenient way of getting all the benefits without actually bothering to eat fish.
That, at least, was the idea – one backed by, among others, the American Heart Association, which recommends the supplements for people with heart disease who may not be getting enough omega-3 via their diet.
But this advice is now being challenged by a study which pulls together evidence from studies involving almost 78,000 people. Published in JAMA Cardiology, it found no significant benefit from taking the supplements.
Needless to say, advocates of fish-oil are not taking this lying down. In the US, the Natural Products Association insisted there remains a “strong body of evidence” that people who don’t consume enough seafood can benefit from taking omega-3.
Cynics will view this as standard industry bluster. After all, while smaller than the international study of fish consumption, the new claim is based on 10 large randomised controlled trials, widely regarded as the gold standard in assessing whether an effect is real.
In addition, the trials specifically looked at the effect of supplements rather than fish.
And the outcome was clear: the researchers found no benefit.
At least, that’s what has been widely reported. But it’s not true. The results actually point to a benefit of around a 7 per cent cut in the risk of coronary heart disease death.
So what’s going on?
The apparent double-think is symptomatic of an ongoing crisis blighting scientific research. It’s been brewing for decades, but has only recently begun to get the attention it deserves.
That is because it centres on the principal goal of science: drawing reliable conclusions from observations.
It’s now clear that statistical methods relied on to turn data into insight are fundamentally flawed.
Yet despite calls to tackle the problem, researchers continue to use the techniques – and add to the confusion.
In the case of the omega-3 debate, the arguments centre around the so-called statistical significance of the findings. The researchers do not actually say they failed to find an effect. Instead, they talk of finding “no significant association” between omega 3 and heart risks.
It’s in the meaning of that word “significant” that all the trouble lies.
In everyday use, it’s a synonym for worthwhile or meaningful. But in science it has taken on a very specific meaning, one which researchers themselves routinely misunderstand.
Given the prevalence of cardiovascular disease – it’s responsible for an astonishing 30 per cent of deaths in the UAE – even a modest risk reduction is worthwhile.
But researchers are not dismissing the 7 per cent reduction they found as pointless. Rather, they are saying their finding lacked statistical significance, which is something entirely different.
It is widely held to be a measure of the chances of a finding being a fluke. Researchers have been taught to regard findings lacking statistical significance as having too high a risk of being a fluke. Some even claim it implies there’s no effect at all.
Yet for decades statisticians have been warning that this is a dangerous mis-understanding. They point out that statistical significance – or its absence – reveals surprisingly little about whether a result is genuine. As such, it’s a very unreliable basis on which to assess findings – let alone change practices.
In 2016, the American Statistical Association issued an unprecedented public statement calling on scientists to switch to more sophisticated ways of assessing findings, so far without success.
As a result, there’s been no let-up in the flow of questionable research claims entering the public arena.
In the case of the omega-3 study, the researchers have focused simply on whether the findings are statistically significant, while playing down the fact that the findings suggest the most probable figure for the benefit from omega-3 is a 7 per cent reduction in risk.
Unlike many researchers, they don’t quite go as far as arguing there’s no benefit at all. They even concede that the uncertainty surrounding their findings means they cannot rule out some level of benefit from the supplements.
Even so, their focus on statistical significance leads them to claim there is “no support” for recommendations like those made by the AHA – despite the fact that there is.
Fortunately, this is unlikely to be the last word. There are four more large studies of the link between omega-3 supplements and heart health currently under way. These will add substantially to the weight of evidence one way or another.
The need for more data is all the more urgent given recent research suggesting a link between omega-3 and increased risk of prostate cancer. But that claim, too, is based on statistical significance, and so cannot be taken at face value.
So where does this leave people looking for clear guidance ?
The usual advice is that if you’re thinking of stopping – or starting – some medication, check first with your doctor.
But when it comes to omega-3 supplements, don’t be surprised if your doctor is as confused as you are.
Robert Matthews is Visiting Professor of Science at Aston University, Birmingham, UK