Providing tools that don’t acknowledge uncertainty will erode the public’s trust in the health service
“Why are so many of us heading for an early grave?”, asked the Guardian last week. The NHS online heart age test made headlines when Public Health England revealed that 81% of the 1.9 million people who have taken the test had a heart age above their real age. If only the public would exercise, eat well, and stop smoking. If more people knew their heart age, perhaps they would?
Unfortunately, the evidence to support risk calculators (in the form of heart age, lifetime risk, or any other form of risk) being used as standalone tools for behaviour change is lacking. In my own qualitative research on this subject I was struck by how people respond in different ways, both positively and negatively, to seeing their heart age on a screen. But even those who didn’t like what they saw felt it was a good thing to do—and perhaps this is the point: it’s good to know your heart age because it’s good to take more control of your health. The problem is that the NHS heart age tool isn’t up to scratch.
One major flaw of the tool is it’s obsession with cholesterol. It tells everyone from the age of 30 to get their cholesterol checked (in large shouty red capital letters), despite there being no national guidelines to this effect. And it seems that anyone who doesn’t know their cholesterol or blood pressure gets a high heart age calculation. As Ben Goldacre pointed out on Twitter, if everyone in their 30s took the heart age test and followed the advice to check their cholesterol, that would be 8.7 million visits to the GP for a blood test. If even a fraction of these tried to book an appointment in the wake of this publicity it could cripple general practice—anyone with actual symptoms of a heart problem could be waiting months to see their GP. GPs are predictably cross about it. I’m sure I wasn’t the only one fielding calls last week from people panicking that their cholesterol might be high. But will there be a spike in cholesterol testing? And if so, what will the consequences of this be? It seems an important question. But it may provoke further awkward questions: if there is a spike in demand, GPs will be justified in demanding remuneration.If cholesterol testing doesn’t rise, then why not?
The use of postcodes is also a problem. Socioeconomic factors have a far greater effect on our health than cholesterol being a bit high. A baby girl born in Richmond is likely to live an extra 17.8 years in good health than a baby girl born in Manchester. But that doesn’t mean that a woman living in Richmond will have a younger heart age than one living in Manchester. There can be huge differences in wealth, education, and opportunity within postcodes—lumping everyone in a postcode area together and using the average for heart age estimates doesn’t make sense.
Take the heart age calculator at face value and it’s an incredible thing: answer 16 questions and the programme will work out your heart age. This is presented with great confidence: “your heart age is 38”. My heart age isn’t “about 38”, or “38, based on what you’ve told us”, or “38 plus or minus a couple of years because, you know, this is just a clever bit of programming and we can’t really look at your heart or see the future, obviously”. It’s a fact: my heart age is 38.
It’s certainly has more impact to use this wording, but at what cost? For those already suspicious that their calculation is wrong this is the final straw. For those who believe it, they are being misled about the accuracy of the estimate—like the health secretary Matt Hancock who tweeted that his heart risk is “a month older than me. Must do more exercise!” What is the chance that the heart age calculation will be inaccurate? What is the 95% confidence interval? I’ve looked at the supporting documents and can’t find any answers. Perhaps they’re there if you dig deep enough through the JBS3 website (which the tool points you towards), but that’s not the point. If I can’t find the answer after 20 minutes of looking, who else is going to bother? The public deserve to know how accurate these estimates are. Presenting uncertainty as fact is not what the public, or our patients, want or need. Are we scared that if we share uncertainty people won’t do what we want them to? Maybe they won’t, but providing tools like this that lack credibility and don’t acknowledge uncertainty will erode the public’s trust in the health service and stop important public health messages getting through.
Tom Nolan is a GP in London and a freelance clinical editor, The BMJ.
Competing interests: TN is a GP partner and elected member of Lambeth LMC.