andrewducker | So this is it...we're going to die

Yeah, but stats like that are complicated by the average life expectancy including infant mortality. I don't know how those ones were reached, mind you.

Much of the gain is in the chance of making it to 65 in the first place: the numbers you quote don't take into account infant mortality, deaths from pregnancy and childbirth, or most deaths in battle.

The mass media have very little apparent understanding that the huge changes in life expectancy are very much due to a hell of a lot more people surviving childhood rather than some miraculous prolonging of life.

A relevant article: http://www.significancemagazine.org/details/webexclusive/1050903/Human-life-expectancy--will-the-young-survive-to-100-years.html

Although it's not the one I was looking for, which was more of a critique of the pitfalls of writing about such matters.

AFAIK very little research effort has gone into attacking ageing itself. It's fair enough to consider someone like Aubrey de Grey as fringe, but where are the non-fringe researchers? How come it's not worthy of mainstream attention?

And violate God's will? Insanity!

I think the pensions crisis that would result from a cure for aging is of rather more concern :)

Apply the reversal test to that? http://en.wikipedia.org/wiki/Reversal_test

Nah, just debt-enslave all the oldsters and put them to work in the landfill mines for a couple of centuries until they've paid off the cost of the rejuvenation tech :-)

Slightly off-topic, but there was a post I wanted to share with you about mortality statistics, how they rule out several possible models of the ageing process, and some possible implications for life-extension tech. Unfortunately, I can no longer find it. Do you know the piece I'm talking about? If not, I'll look harder. One conclusion that I remember was that a simple "accumulate X hit points and then you die" model can't account for observed mortality patterns.

Was it this?
http://gravityandlevity.wordpress.com/2009/07/08/your-body-wasnt-built-to-last-a-lesson-from-human-mortality-rates/

(if so you'd have found it on http://delicious.com/AndrewDucker/death )

Yes! That's the one. Thanks!

No problem - you want me to point

ciphergoth at it, or would you like to?

Got it! :-)

Hurrah! Does it have any relevance to de Grey's programme? AIUI some of his ideas (eg, about telomeres) fail under that analysis, but it's been a while since I've looked into SENS.

The motherlode for this information for the UK is the very-excellent Office of National Statistics:
http://www.statistics.gov.uk/hub/population/deaths/life-expectancies/index.html

And the website there helpfully flags up that official figures are (almost) always period life expectancy, rather than cohort life expectancy. Basically, period life expectancy is how long you would expect someone at a particular age now to live if they experienced current mortality rates. So you take a notional person aged (say) 65 now, calculate their chances of dying aged 65 from the mortality stats for 65-year-olds, and then their chances of dying aged 66 from the mortality stats for 66-year-olds, and so on.

Cohort life expectancy is how long you actually expect a person to live. Of necessity this involves some guessing about likely future improvements - which is why period life expectancy is almost always used for comparisons, because it's robust to changes in future expectations: it's a construct based on what has happened, rather than on what we think will happen, and we have much better data about what has actually happened than on what will.

To summarise the UK stats there:
Period life expectancy 2008-2010:
Males at birth: 78.1
Females at birth: 82.1
Males at age 65: 17.8 (= die aged 82.8)
Females at age 65: 20.4 (= die aged 85.4)

Cohort life expectancy 2011:
Males at birth: 90.3
Females at birth: 93.8

The historical motherlode of info is the House of Commons Research Paper 99/111 "A Century of Change: Trends in UK statistics since 1900" http://www.parliament.uk/documents/commons/lib/research/rp99/rp99-111.pdf [PDF]
That only shows (period) life expectancy at birth (on p.8), but you can see it soar over the C20th from just under 50 in 1900 to about 80 now. You can see that most of this is driven by improvements in child mortality in the accompanying graph, which shows infant mortality plunging from about 140 per 1,000 births in 1900 to less than 6 by 1997.

We are not making similarly great strides in improving maximum lifespan. You can also see the top end of this in the stats about the oldest humans (http://en.wikipedia.org/wiki/List_of_the_verified_oldest_people) - there are only 6 living people in the top 100 (and the oldest is at number 15). That wouldn't be the case if we were making dramatic strides here.

You'd want decent odds on a bet that a newborn today won't see their 100th birthday, and staggering odds on a bet that any of us commenting here will see our 120th.

For all this looks gloomy to us who are well over 5 years old, the improvement in infant mortality is bloody brilliant and worth cheering loudly about. One thing that is particularly amazing is that it continues to fall (in the UK, at least) despite it being really pretty low already.

So, the key thing in the OP is that in 1900 once you had reached 65 you were probably good for about the same number of years as you are now if you reach 65

but

your chances of seeing 65 in 1900 were very much lower.

While commonly believed this is not true. Since 1970 changes in life expectancy in industrialised nations have been pretty largely driven by extensions in life for older people.

When we reach 65 we now go on for longer and it is this rather than anything pre 65 which is now driving increases in life expectency.

J.R. Wilmoth / Experimental Gerontology 35 (2000) 1111±1129

It's not true that in 1900 your chances of living to 65 were significantly lower than they are in 2011?

It is not true that you're good for approximately the same number of years from 65. Andrew's belief is based on taking the absolute number he should be looking at the percentage increase since then. The remaining life expected at 65 has increased by 50% in that period.

i.e. 19-12=7

7 / 12 = 58%

?

and that this effect is currently (in the West) the only significant factor in extending average life expectancy

i.e. almost everyone got to 65 in the 1970's and now they are living longer than 65 in increasing numbers?

Exactly my point yes... I would not say it is the only significant factor but it is the most significant one. I'm looking at a mortality curve for 1995 births the estimate is that less than 1 in 20 die before the age of 50 (and very few of those in the first year of life). Around about 85% make it to 65. Small (in absolute terms) increases in longevity for the over 65s now have a greater effect on mean longevity than a large reduction in death rates before 50.

Right - I think I get this now. Thanks for the explaination.

Sounds like we've done the easy work of getting most people up towards to the best we can do and we're now left with the harder job of nibbling away at the maximum.

Not the worst problem to have I suppose. In fact, almost by defination a first world problem.

we're now left with the harder job of nibbling away at the maximum.

Exactly... and as I comment elsewhere, that "nibbling" might not be such slow nibbling as some people think. Mean and maximum lifespans appear to be showing consistent linear increases where data can be reliably found.

Two thoughts occur to me – neither I fear particularly novel.

If the nibbling is actually closer to biting then I and my two year old son (more so) should prepare ourselves for a significantly longer life – not just pensions but careers, paying for housing costs and assumptions about the amount of change we might see in our lives. I’m expecting gerontology to get a huge boost when the Chinese and Indians get up to speed and have a larger aging population of their own and start spending their money on the study of aging.

Secondly, for populations that are currently still not very devoloped and having problems with child mortality etc they are going to get all of these medical and public health advances in a oner and from an average life expectancy in the 40’s with 50% child mortality to life expectancy in the 90’s with child mortality in the low 0.1% (hopefully sooner rather than later).

If the nibbling is actually closer to biting then I and my two year old son (more so) should prepare ourselves for a significantly longer life – not just pensions but careers, paying for housing costs and assumptions about the amount of change we might see in our lives.

Now this relates to a big lie we've been sold time and again -- that pension companies have no idea about what life expectancy will be. In fact, there is a vast amount of information about this and most of the ``data" about life expectancy that people discuss (life expectancy at birth right now for example) is ``projected". So, when we say something like, "life expectancy at birth in Kensington and Chelsea right now is 89.8 years" then that is actually a prediction, it's not based on the average ages at which people died but based on ``mortality" at each age.

http://www.ons.gov.uk/ons/rel/subnational-health4/life-expec-at-birth-age-65/2004-06-to-2008-10/statistical-bulletin.html#tab-Local-area-life-expectancy

(In fact, this particular model is ``data driven" in that it is not really a prediction about how long people born now will live but is a prediction about how long people born now would live if they experienced the mortality rates right now. There's no predictive element -- other models add predictive element to how those mortality rates will move.)

The model which actually tries to predict this is known as "cohort life expectancy" -- here's an actuarial table for it from a US source
http://www.ssa.gov/oact/tr/2011/lr5a4.html
You can see how they expect life expectancy at 65 to progress as well (they give three tables with upper and lower bound expectations).

In fact, for a pensions company, predicting how long someone will live is the ``easy" part of that sum. It's a well understood science, sure it's not easy in the sense that you or I could do it without months or years of training and data analysis but it's a heck of a lot easier than working out how the economy will go over the 40-100 year horizon to pay for it. So, for pensions projections, the part of that sum which is hard to do is not working out how long you will live but is working out how much you need to invest to get the lifestyle you want at the end of it.

Whenever you view any information about the pensions crisis and aging population then view it within the context that a lot of time and energy and brainpower has been spent predicting these things and the models have proved pretty good.

It could be that significant effects will knock the aging models out slightly... but an estimate of the progress of mean lifespan within 5% over 100 years (given no apocalyptic events) is not so difficult compared with working out the return on a £10,000 investment over the same timespan.

for populations that are currently still not very devoloped and having problems with child mortality etc they are going to get all of these medical and public health advances in a oner and from an average life expectancy

Sure -- life expectancy in the developing world is still driven by child mortality right now. This will increase faster than life expectancy in the west until they have the aging population problem that we have.

I did my accountancy traineeship with one of the world's largest Life Assurance and Pensions companies.

Some of these concepts are ringing bells with me.

Ah, I'm probably slightly preaching to the choir then. Your company would probably just work from a standard set of tables for estimated life expectancy which includes a model of how that grows over time.

Something like that.

I confess at the time I was there I was paying more attention to chatting up the actuarial trainees than listening to the discussions on actuarial methods from the old boys.

Thank you for that!

All that stuff is vitally important to remember whenever people talk about pensions being unaffordable and the need to extend the retirement age. Those discussions are often "illuminated" by entirely irrelevant data about huge increases in life-expectancy at birth, which were mostly driven by reductions in infant mortality.

there are only 6 living people in the top 100 (and the oldest is at number 15). That wouldn't be the case if we were making dramatic strides here.

Douglas Douglas.. That's because you're only on that list for a small number of years before you die. In fact unless you make the top ten, there's only two years before you entering that top 100 list and you dying. We would not expect many living people to be on that list, it's by definition not a list you can be on for long. There's an equally important problem: data for "living" people is also what we refer to as "right hand censored" -- the living ones are less likely to be near the top of the list because they're not yet dead. (I'm sure you can untangle that to get what I mean more clearly).

What is, actually, much more germaine is that the majority of that list (more than half I think) died in the 2000s. Almost none of them died in the 1980s and only one before then. Only one person who died before you or I were born makes that list and they are disputed -- ISTR my 1979 Guinness book had the oldest person recorded at 113 (disallowing the 1959 death) -- they would not make the top 100, probably not the top 1000.

In 1987 the oldest person recorded was 114. The oldest person now recorded is 122. That's eight years of extra life on the "oldest" record in twenty five years. To me that seems like frighteningly fast progress -- terrifyingly so. Indeed given the theoretical maximum of 1 year onto longest life per year, getting nearly 1/3 of a year onto longest life per year is pretty damned good.

A better reading of that list which reveals more than a count of the living is that unless I slip my count (wholly possible), 1 of them died in 1959 (but that is disptued) 4 of them died in the 1980s, 30 in the 1990s and 44 between 2000 and 2009 (inclusive).

Incidentally though, the startlingly fast growth in "oldest ever" I reported may be partly an issue of record keeping. As time goes on a higher proportion of people will be able to verify their age.

My rough conclusion though is that unless that is a major factor we should not be surprised if at least one person born since 2000 lives to 160. [Scratch that though because I looked back a little further into the oldest person records -- we got "oldest people" well documented born in 1788 and 1792 who lived to 110... this gives us a better projection I think -- which slows the rate massively, I thought that 1 in 3 was huge... given a 1875 well documented 122 this looks to me like around 13 years in a century -- which is still pretty good and would project our 2000 birth to around about 138 not 160. The discrepency seems to be that people born between 1825 and 1875 did not do so well as we might expect -- almost as though there were some traumatic events in the early and mid 20th century].

I think when we compare reduction in child mortality with increase in maximum lifespan it's really important to bear in mind that the comparision is hard to do. A tenfold decrease in child mortality is impressive but actually much easier than the lifespan extension for a simple reason -- if everybody stopped dying right now and became immortal, it would still take over 1100 years to produce a tenfold increase in maximum lifespan.

Edited 2012-02-23 01:15 (UTC)

I dunno if you saw this upthread:
http://gravityandlevity.wordpress.com/2009/07/08/your-body-wasnt-built-to-last-a-lesson-from-human-mortality-rates/

I saw something recently that I can't find now on mortality rates which showed that the shape is changing from a curve to a square - we're less likely to die early, by the upper end isn't changing much, so you get a long straight line of survival near 99% and then a sudden cutoff at the end when it all catches up with us.

I had a quick look around at that article but it does not look consistent with data from elsewhere. Maximum age data from sufficiently small places with sufficiently good record keeping seems to show a linear increase. Mean age data from industrial nations shows a "two regime" linear increase with a break where "older old people" took over from "babies die less often" as the driver.

Incidentally -- people who championed that theory about 50 years ago were putting the theoretical maximum mean age of death at birth at 90. This is already met (and I think exceeded) in some places.

You are right that the graph is "filling out" because fewer people die early though.

Actually, to be clearer, that article *is* consistent with data from elsewhere but does not look at how those trends move in time. So when he says "I can say with 99.999999% certainty that no human will ever live to the age of 130" he follows it with "Ignoring, of course, the upward shift in the lifetime distribution that will result from future medical advances".

That sentence is critical to your understanding of the essay.

I think life expentency increases have been driven by increased age for old people, not child mortality since around about 1970 -- I put a quote and link elswere on this thread.

Just occurred to me that the "longest lived person" metric is very susceptible to the changed sample size (the population of the earth is growing -- the population of the earth with good verifiable birth records even more so.)

Oh, meant to mention the huge disparities and injustices these data throw up. Places like Afghanistan have infant mortalities now that are up where UK rates were in 1900. And even within the UK things vary - (period) life expectancy for a newborn boy in Glasgow City is 71.6, but for a newborn girl in Kensington & Chelsea it's 89.8.

Yeah all that stuff about infant mortality and social class!

*entirely* off topic:

of all the advances that have increased life expectancy, 80% have been in civil engineering. Most particularly, sanitation and water treatment.

20% are medical.

that's an estimation by - if I remember rightly - a previous head of the Royal College of Physicians. Who was not a civil engineer.

[figures pulled from my unreliable memory]

Go, we civil engineers!

Dommy_nick

I thought that a more accurate way of casting the issue was to say that when you get to an age that could be considered "old", your quality of life will be better than in the past as you inch towards the inevitable. Well, if you live in a country with a fit-for-purpose healthcare system.

Good point - I'm pretty sure you can see this in QALY/DALY figures for the population. You're way more likely to make it to 65, you're likely to live a few more years if you do, and those years are likely to be considerably freer of disease and disability. (Entirely the opposite of the commonplace view that we're living longer but with lower quality of life.)

Hold on hold on... You're doing exactly the wrong thing with your stats here.

The life expenctancy (years left to live) at age 85 has increased 50% between 1900 and 2007. The life expectancy (years left to live) at age 65 has increased more than that. Seeing it as an absolute number is not that helpful. So yes... as an absolute number it has not gone up much... but if the absolute number had gone up a lot the effects on mean lifetime and population would be genuinely frightening.

People's expectations about this are not a good guide because in general people are not good at understanding survival functions (including myself and I've written a published paper involving them).

So this is it...we're going to die

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject