How long have I got: mortality predictors from UK Biobank

Counting down...

Counting down…

A lot of research  gets reported in terms of whether something increases or decreases your chance of death. We’ve yet to find anything that reduces your chance of death to zero – we all die of something, eventually. Most of us would like to die of something as far into the future as possible.

Which is why timescale is so important. Studies using mortality as an outcome report the chances of death during that particular study and follow-up, which could be a couple of weeks or 20 years. Yet these are often cited as if they were absolute risks – reducing or increasing your overall chance of death by 15% or whatever.  And baseline chance of death differs for each study, depending on how old the participants are, whether they are healthy, whether they are being treated for a life-limiting disease and so on. Only when you’ve sorted all that stuff out can you make sense of the risk reduction or increase being studied.

I’ve been researching the interplay between inherited risks (those ‘in the genes’) and the environment. It’s complex – even if you have an inherited genetic susceptibility for one disease, that same gene mutation might protect you from something else, so the effect on mortality isn’t obvious. And a genetic susceptibility will only convert into disease if something in your environment triggers it. That something might be tobacco smoke, or ultraviolet light, or just the amount of time you’ve been alive. That’s one of the reasons why it’s so hard to say that one mutation, or one foodstuff, ’causes’ disease.

What science can tell us quite well is how predictive certain factors are of impending doom. All you need is a big bunch of data from a lot of people, and a few years of follow up. That’s exactly what’s happening with the UK Biobank, currently generating a stream of interesting research on the half-million participants who provided data including DNA samples, biometrics, lifestyle and family history. (I was one of them; see My Day as a Lab Rat for more.)

This summer, Swedish researchers published the results of crunching the Biobank data to look at the strength of 650 predictors of five year mortality among Biobank participants aged 37 to 70.  There were a few interesting results. Firstly, you don’t need any type of test to predict whether you’ll last another five years. You probably know everything you need to know already.

Have you been diagnosed with cancer? How do you rate your health? Do you smoke? Do you walk slowly  or briskly? (It doesn’t matter what other people think of your health or your walking pace – it’s your own assessment that matters.) People without a cancer diagnosis, who feel in good or excellent health, don’t smoke and walk fast have a great chance of seeing the end of 2020. There are lots of questions which have a smaller predictive value, but these four questions were the most highly predictive of mortality.

It’s important to stress these are only predictive factors – walking fast is an indicator of being healthy, not necessarily a cause of good health. But I was interested to see how low down on the scale family history of disease ranked. Lifestyle factors, primarily smoking, were way more important. If you’d like to put some science into your forward planning, you can discover your own five year death risk on the project website, here, or check out how the different predictive factors perform, here.

Image: from Mac(3)’s photo stream on Flickr with CCL.

%d bloggers like this: