We are not yet approaching any maximum human lifespan, according to an examination of human mortality over time and across 19 countries

 

How long can people live? As medicine moves the bar, researchers turn to data analysis to test whether the data are consistent with the idea that we have reached a maximum possible human lifespan

 

The question of whether there is a maximum possible human lifespan – and what it is – has been a subject of human curiosity for millennia.  The writers of the Psalms pegged it at 80; ancient Roman thinkers estimated 100 or maybe 110 years.  But for 25 years, no-one has outlived Jean Louise Calment, the world’s oldest recorded human, who died in 1997 aged 122 years and 164 days.  This has led to speculation that we may, indeed, have reached a biological limit. 

 

“If you read the popular press, you will have seen that people are worried about the fact that life expectancy in the U.S., and in some other countries, such as the U.K., has stopped increasing,” McCarthy said. “But there are big generational differences that these reports often obscure.  In the U.S., for instance, mortality probabilities have indeed risen for people of middle age. But in recent decades, mortality rates of older people in the U.S. have actually been improving faster than they have at any time since the decade following the introduction of Medicare.”

 

McCarthy’s paper “Mortality postponement and compression at older ages in human cohorts” uses a novel approach to investigate this question, and was published in PLOS ONE this week.

 

The paper analyzes the mortality of older people in 19 countries across the world, and examines how the rate of increase of mortality by age differs between cohorts born in different years.  “If we are reaching a biological maximum, we would expect that as mortality at younger old ages improves, the rate at which mortality worsens by age should rise, to preserve the biological maximum.  But this is not what we found.” says McCarthy.  Instead, the data show that for cohorts born in the first part of the twentieth century, the rate of mortality increase by age has actually fallen along with falls in the mortality rates of those at younger old ages.  This suggests that the maximum age at death will actually increase dramatically in the coming decades as surviving members of these cohorts reach advanced old age.

 

“Standard life expectancy calculations, called period-based calculations, are really just summary measures of the mortality rates of a given population in a given year.  They will only be the actual life expectancy of real people if mortality rates never change again”, says McCarthy.  “Demographers have used these summary measures to show that average lifespans increased because more people live longer — not because the oldest people are living longer.  Cohort-based calculations, on the other hand, which are used in this paper, look at the change in the mortality real people can expect over their lives, as mortality changes in the future.  The paper shows that changes in cohort mortality rates are consistent with what demographers call ‘mortality postponement’, where the maximum age at death increases, rather than ‘mortality compression’, where this age remains fixed.  Our results show that longevity records have not increased because those people who are old enough to have broken longevity records were members of cohorts that did not experience mortality postponement.  As newer generations reach these advanced ages, we can expect that longevity records will indeed be surpassed,” says McCarthy.

 

“Having said that, we do note that mortality compression has been the primary pattern of mortality improvement throughout much of recorded history”, says McCarthy.  “For instance, it is striking that the maximum age at death in Sweden – the country with the longest series of reliable mortality data – was almost the same for males born in 1900 as it was for males born in 1780, four generations earlier.  But for cohorts born after this time – coinciding with the rise of modern medicine in the period following the end of the Second World War – the data show that mortality postponement becomes dominant.”

 

Indeed, it is striking how long some individuals have managed to maintain physical robustness into their eighties and beyond.  McCarthy points to the example of Johanna Quaas, born in 1925, who is the world’s oldest competitive gymnast, at 97.  “While people like Johanna Quaas will probably never be typical, it is likely that enough people like her will cause mortality records to be broken in the coming decades if the patterns we highlight in the data continue to hold”. 

 

McCarthy and his former graduate student, assistant professor Po-Lin Wang of the University of South Florida, analyzed vital statistics records from 19 industrialized data from 1890 and 1970 and found similar patterns in each country. More people are living longer, but statistically they found the oldest among us today will likely set new longevity records.  McCarthy and Wang projected a more substantial mortality postponement for those born between 1910 and 1950. The oldest of these cohorts may regularly live to 120 or beyond. 

 

“As these cohorts attain advanced ages in coming decades, longevity records may increase significantly,” McCarthy said. “Our results confirm prior work suggesting that if there is a maximum limit to the human lifespan, we are not yet approaching it.”

​

On this web page

​

The published paper contains graphs for all countries, available by clicking the button above on the left.  The code and data can be obtained by clicking the button above on the right.  Individual results for each country can be obtained in the country file for that country by clicking the appropriate button to the right.   

 

Each country file contains four items.   

​

(1) A table that shows 95% confidence intervals for the age at which we estimate the average mortality hazard rate of surviving people in each birth cohort will first reach 2/3 (this corresponds to an annual death probability of around a half), and 95% confidence intervals for the age at death of the longest-lived person in each birth cohort.  95% confidence intervals account for uncertainty in our estimates, and, if the model is correct, will contain the true value 95% of the time.  

​

(2) A table that shows our estimates of changes in the remaining life expectancy at age 50 of each birth cohort relative to the cohort born ten years before.  We divide changes in remaining life expectancy into two parts: that due to mortality postponement (which leads to an increase in maximum possible lifespans) and that due to mortality compression (where mortality at younger ages improves but there is no increase in the maximum possible lifespan), along with 95% confidence intervals for each quantity.

​

NOTE: All Bayesian estimates are combinations of a Bayesian prior (chosen by the investigator) and the data.  Some figures are greyed out in the tables because these results were not robust to changes in the prior - in other words, their magnitudes depend more on modelling assumptions made by us than on observed mortality data.  We have included them for completeness, but with this caution. 

​

(3) A set of figures.  Each figure is shown separately for females (on the left) and males (on the right). 

​

The first row of figures shows our estimates of the rate of increase in mortality by year of age for each birth cohort (on the vertical axis), plotted against the estimated mortality hazard of 50-year olds in that cohort (on the horizontal axis).  Birth cohorts are shown in boxes on the graph.  In general, points move upwards and towards the left in successive birth cohorts, meaning that mortality rates at age 50 have fallen, but the rate at which mortality rates increase with age has risen.  Mortality postponement occurs when the increase in the rate at which mortality rises with age does not fully compensate for reductions in mortality at younger old ages, leading to an increase in the maximum possible lifespan.  

​

The second row of figures shows our estimate of how the remaining life expectancy at age 50 (in years) has changed across birth cohorts at 10-year intervals.  For instance, a value of 1 on the graph means that individuals born in that year who reach age 50 can expect to live one year longer after age 50 than individuals born ten years earlier.  We split the change into the portion due to compression and the portion due to postponement (described above), and show confidence intervals for each portion.  For the cohorts that are already extinct, we show the actual change in remaining life expectancy at age 50 for that cohort.  Confidence intervals widen for more recent cohorts - that is, our estimates become much less certain - as less data is available upon which to base our estimates.

​

The third row of figures shows our 95% confidence intervals for the longest-lived person in each birth cohort (grey) and the age at which we project the mortality hazard rate for surviving members of each birth cohort will first reach 2/3 (this corresponds to an annual death probability of around one half).  Black dots indicate the longest-lived person in each cohort available from our data sources.  Green dots indicate that the longest-lived person in each cohort is still alive.  The diagonal line separates observed data from projections: anything to the right of the diagonal line represents a projection, and anything to the left is observed.

​

(4) The final section in each country file is a list of the data we have available for each of the individuals indicated with points in the third set of figures above, and the data source for that person.

​

Dr David McCarthy would be happy to provide any further information or answer any questions at dmccarth@uga.edu