Feature Data Briefing

Death clocks: How long have I got?

BMJ 2017; 356 doi: https://doi.org/10.1136/bmj.j346 (Published 26 January 2017) Cite this as: BMJ 2017;356:j346
  1. John Appleby, chief economist, Nuffield Trust, London, UK
  1. john.appleby{at}nuffieldtrust.org.uk

Knowing when you are going to die could help make life choices, but John Appleby finds that his life expectancy varies depending on who he asks

Data from the Office for National Statistics national life tables for 2013-15 show that over the past 33 years average life expectancy at birth for UK residents has been increasing by, on average, 13.1 weeks per year for boys and 9.5 weeks for girls.1 Period life expectancy (assuming no change in future death rates) at birth averaged across the three years of 2013 to 2015 are estimated to be around 79 years for boys and 83 for girls. The good news is that we are all living longer. The bad news is that we will all die … but when? Baby girls born in 2015 can’t all expect to live to 83. Some will die sooner, some later. Our genetic inheritance, lifestyle, wealth, employment, and consumption of everything—from education to recreational drugs and healthcare—are all potential causes of variations around the average life expectancy.

But as Rufus has summarised,2 many other factors can affect mortality and hence life expectancy. Being married can add over a year to life expectancy compared with being single, for example.3 Sleeping less (but not too much less) is associated with an improved life expectancy compared with sleeping more.4 One study suggests that optimists have a 55% lower risk of early death than pessimists.5 A typical Swiss man has a longer life expectancy (81 years at birth) than a typical Latvian man (69 years at birth).6 And of course, not only do the rich have more money than the poor, they also tend to enjoy a longer life to spend their wealth.3

So, while the ONS period life tables—based on fixed, age specific death rates—provide one population based view of life expectancy, for a more individual perspective you need to adjust these figures for personal characteristics and circumstances.

Based on my sex and the current mortality for my age group, ONS life tables suggest I will die around May 2040, about a month after my 82nd birthday.7 But as figure 1 shows, plugging a few more personal details—such as my marital status, income, and stress levels—into a random selection of online “death clocks” produces a range of predictions for my life expectancy from 67 (eek!) to 89 (yay!). Four required only details of age and sex but produce life expectancies ranging from 83 to 89.14151621 One probed 15 different aspects of my life to generate a life expectancy of 88.18 And one provided a slightly more cryptic and difficult to understand “3% risk of dying in the next five years” rather than a specific age at death.8

Figure1

Fig 1 Variables required by a selection of death clocks89101112131415161718192021 and predictions for the author’s life expectancy (bar chart). Note that the first model produces a risk of death in the next five years rather than an actual age of death

Some of the variation in predictions is due to differences in the basic life table data that the clocks use (some are based on non-UK data for example). Differences will also arise given the particular risk calculators (prediction models) used, the number of variables included, and the way they combine variables to produce individualised forecasts.

Variation also arises from treating what are essentially probabilistic forecasts as point estimates. One online predictor of death does, however, provide a nice pictorial representation of the uncertainty surrounding its forecast. Based on just sex and age, the “Years you have left to live, probably” calculator15 does what it says on the tin, and with an interactive chart builds up a distribution curve of probable life expectancies. (Fig 1 shows the average for this calculator, which is around 83 years).

So perhaps the only safe conclusion is that death clocks should come with a health warning: calculating your date of demise is somewhat sobering and the results should be taken with a pinch of salt.

Footnotes

  • Competing interests: I have read and understood BMJ policy on declaration of interests and have no relevant interests to declare.

  • Provenance and peer review: Commissioned; externally peer reviewed.

References

View Abstract