What has happened to UK life expectancy and congratulations to more centenarians

Any discussion of life expectancy tends to grab people’s attention. How long we can expect to live for is obviously a really important issue. There are few things as important in our life. So when governments release their latest life expectancy statistics they are guaranteed to get a lot of coverage. The trouble is life expectancy is a slippery concept and the headlines tend to focus on a measure that isn’t really a good measure of life expectancy.

For instance, today the UK’s statistical authority, the ONS, released its latest ‘period’ measures of life expectancy based on mortality rates for the period 2020-2022. This is an update from those based on 2017-2019. It's not great reading. UK life expectancy on this basis has fallen from 83 years to 82.6 years for females and from 79.3 to 78.6 for males.

UK life expectancy trends aren’t great. Across high-income countries, in general, life expectancy gains are declining, e.g. they aren’t growing as fast as previously, but the UK has, in particular, had poor recent trends. But the latest data shows not just declining growth but an actual fall in life expectancy. Or do they?

The problem is that the headlines are about what is called ‘period’ life expectancy, and despite its name, it isn't really a concept that matches up with what in everyday language we would consider to be life expectancy. Instead, it is really an average age of death in a particular year. Let me explain…

To calculate life expectancy, you use a concept called survival rates. That is the probability of surviving a particular year. So, a newborn will have a given probability of making it to their first birthday. Their survival rate is 1 minus the mortality rate of children aged between 0 and 1. Similarly, a 1-year-old will have a survival rate for making it to their second birthday – that is 1 minus the mortality rate of children aged between 1 and 2. You can obviously construct survival rates for all ages but beyond a certain level they are zero. For instance, there is no one currently in the UK who is aged 120 and so the survival rate between 119 and 120 is currently 0.

Basically, life expectancy is the sum of all these survival rates. So far, so good, but here is where the problem comes. How do you work out those mortality rates? What the period measure of life expectancy does is choose a particular year (or set of years) and calculate ‘life expectancy’ on the basis of mortality rates in that year. So to calculate life expectancy based on 2023 data you use the mortality rate for 1 year olds, 2 year olds, etc. in that particular year.

The problem, of course, with this is that the mortality rate for an 80-year-old in 2023 is unlikely to be the mortality rate of a person born in 2023 when they reach 80 in 2103. Based on past data, we know that mortality rates tend to improve, and if that continues to happen, then a newborn today will face a better survival rate (or a different one, anyway) compared to an 80-year-old today. So period life expectancy measures will underestimate their likely life expectancy.

So, another option is to calculate ‘cohort’ life expectancy. That is to project forward what you think mortality rates will be in the future and, in that way, calculate the life expectancy of a newborn today. All sorts of sophisticated statistical methodologies exist to do this and they basically involve extrapolating past historical trends.

The trouble with cohort estimates of life expectancy is that they are projections. Who knows if they are true. So, in general, governments, when they release data, tend to emphasise the period measures of life expectancy. After all these are based on firm data – the mortality rates observed in that year.

But as the above shows, the problem is that this isn't really a measure of life expectancy for a newborn. Assuming continued improvements (in other words, the cohort estimate) a new born will live for probably significantly longer than the period estimate of life expectancy. Of course, the older you are, the more relevant the period of life expectancy is, as you will see less change in mortality rates as you have fewer years to live and will only benefit from reductions in mortality over fewer ages.

So that is why it is important not to read too much into the latest UK numbers showing falling life expectancy. Recognising that year-to-year fluctuations in mortality may be misleading, the ONS bases its period estimates on an average of three years. Every three years it shifts the window over which period mortality is calculated. So, the numbers announced this week are from 2020-2023, and the previous window was 2017-2019. In other words, the latest window includes two pandemic years during which time mortality rates were significantly higher, especially at older ages. The result is that period measures of life expectancy using these years show a decline on those based on 2017-19. But the main message to get from that is Covid was bad for life expectancy which is hardly a shock.

Of course, the ONS realise that. That is why in the midst of the report they say? “A fall in period life expectancy does not mean that a baby born in 2020 to 2022 will go on to live a shorter life; average lifespan will be determined by changes in mortality rates across their lifetime – if mortality rates improve, then period life expectancy will go back up”.

?

The trouble is that note won’t get much attention in the press coverage and headlines will focus on falls in life expectancy. I do wish that we could do one of two things. One is to focus on cohort not period measures of life expectancy when releasing data to the public. I understand the challenge. It's hard to forecast? future mortality and there is a lot of uncertainty. But when governments present a budget that doesn’t stop them forecasting future GDP. Further government projections of future fiscal policy are rightly based on cohort estimates of life expectancy. So why not just share those more widely with the main statistical release so that headlines will be about a more plausible estimate of life expectancy? The other option would be to stop calling these period measures ‘life expectancy’ and just rename it the average age of death in a particular year.

?

I realise that the above conversation has taken a subject that gets everyone interested (how long am I likely to live for?) into seemingly arcane statistical issues but this is important. As the TIIA show in their work on longevity literacy people are poorly informed about life expectancy. Only about a third of people surveyed could correctly identify their likely life expectancy given their age. One in three didn’t know, one in ten overestimated and one in four underestimated. To the extent that headlines about life expectancy are what people read about the focus on period measures will contribute towards people underestimating. Failing to appreciate how long you might live for leads to all sorts of problems in terms of financial and career plans.

?

The big question, of course, is what will happen when these numbers are recalculated using 2023-2025 data. To the extent that these years are not pandemic years we should expect a significant improvement unless there are other nasty surprise waiting for us. But how much of an improvement is right now unknown. Has Covid weakened the health of the population? Have delays to operations during the pandemic years worsened mortality? Have the adverse UK trends that were appearing before Covid worsened? We will have to wait three years to find out but it is highly likely they will show period life expectancy recovering. Just as the average age of death falls in a pandemic so it will likely rise after it.

?

As if to prove that point at the same time as announcing falls in (period) life expectancy the ONS also announced. Remarkably these show that despite the pandemic there is now a record number of centenarians – over 15,000 – in England and Wales. That’s a doubling since 2002. In 2022 over 6000 people celebrated 100 not out – that is about 18 each day. Some of that growth is just about population growth but most of it is about rising mortality rates at older ages. Turns out a 100 year life is becoming more of a reality for more and more people. The increase is particularly pronounced amongst men – the number of male centenarians has increased nearly fourfold since 2002 whilst doubling for women.

Of course, it isn’t just the number of people aged over 100 that have increased in number. As the chart below shows there are now a lot more people aged over 90 years of age. There are now over half a million people living in England and Wales aged 90 and above.?

If you are interested in longevity and the individual and societal challenges and opportunities that longer lives bring then do check out my new book – out in the UK in March (https://amzn.eu/d/24KzOCg) and in the US in April (https://a.co/d/iYO9EkI). Now we are living such long lives, life will never be the same again!

?#healthspan #lifespan #ageingwell #ukageing

?