Skip to content
 

“Calm Down. American Life Expectancy Isn’t Falling.”

Ben Hanowell writes:

In the middle of December 2016 there were a lot of headlines about the drop in US life expectancy from 2014 to 2015. Most of these articles painted a grim picture of US population health. Many reporters wrote about a “trend” of decreasing life expectancy in America.

The trouble is that the drop in US life expectancy last year was the smallest among six drops between 1960 and 2015. What’s more, life expectancy dropped in 2015 by only a little over a month. That’s half the size of the next smallest drop and two-thirds the size of the average among those six drops. Compare that to the standard deviation in year-over-year change in life expectancy, which is nearly three months. In terms of percent change, 2015 life expectancy dropped by 1.5%… but the standard deviation of year-over-year percent change in life expectancy is nearly 4%.

Most importantly, of course, life expectancy in the US has increased by about two months on average since 1960. [see above graph]

Hanowell has the full story at his blog:

The media is abuzz about a small drop in life expectancy in 2015. Yet despite sensationalist headlines, human lifespan has actually risen globally and nationally for decades if not centuries with no signs of a reversal. Alarmist news headlines follow noise rather than signal, causing us to lose sight of what’s really important: understanding how human lifespan has improved; how we can maintain that progress; how social institutions will cope with a rapidly aging population; and trends in vital statistics more fine-grained than overall life expectancy at birth.

Don’t believe the hype. Life expectancy isn’t plummeting.

Hanowell then goes through the steps:

What Is Life Expectancy?

Fact: Human Lifespan Has Risen Globally for Over 250 Years

Then he gets to the main point:

Fact: There’s No Evidence American Life Expectancy at Birth Is Falling

Okay. So the human lifespan has been increasing over the last few centuries in the U.S. and other nations. There still could have been a recent slowdown or reversal, right? Well, yes, but there’s virtually no evidence for it. The 2015 annual drop in lifespan is a mere 1.2 months of life. That’s 50% smaller than the average among six annual drops since 1960. Yet between 1960 and 2015, life expectancy in the U.S. increased by about two months per year on average. In 1960, newborns could expect to live just over 71 years. Now they can expect to live just under 79 years.

If words aren’t enough to convince you, here is an annotated picture of the numbers.

And then he gives the image that I’ve reproduced at the top of this post.

What, then?

Hanowell continues:

Let’s Stop Crying Wolf About Falling American Life Expectancy

Here are some examples of sensationalist, alarmist headlines about life expectancy:

U.S. life expectancy declines for first time in 20 years (BBC News)
Drugs blamed for fall in U.S. life expectancy (The Times)
Dying younger: U.S. life expectancy a ‘real problem’ (USA Today)
Heart disease, Alzheimer’s and accidents lead to drop in U.S. life expectancy (Newsweek)
We’ve already seen that American life expectancy is probably not a “real problem.” Quite the opposite. There may be an explanation for this short-term drop. Maybe The Times is right and it has something to do with the so-called “opioid epidemic.” Maybe Newsweek is right and we should chalk it up to heart disease and Alzheimer’s (although probably not). Maybe it’s something else entirely.

By sensationalizing short-term trends without the proper long-term context, we lose sight of the progress we’ve made. That leaves us less informed about how we’ve come so far in the first place, and where to go from here.

What We Should Be Talking About Instead of Falling Life Expectancy

(Because It Isn’t Falling)

Falling American lifespan isn’t a pressing problem. What should we focus on instead? Here are a couple of ideas:

Understand How We Came This Far and How to Keep Going . . .

Improve Health and Quality of Life at Advanced Ages Without Overwhelming Social Institutions . . .

Pay Greater Attention to Trends in Finer-Grained Vital Statistics Than Overall Life Expectancy . . .

Hanowell concludes his post as follows:

Recent headlines about a drop in expected American lifespan are misleading. Although life expectancy dropped by a small amount between 2014 and 2015, the long-term trend shows climbing lifespan. Instead of worrying about a problem for which there is no evidence, we should be focusing on meeting the challenges that come with longer human lifespans, and understanding why lifespan differs by demographic characteristics.

And then he has a question for me:

How can we encourage journalists and the prominent scientists they quote that you can still make a story about steadily increasing life expectancy despite occasional faltering, and it won’t hurt your chances of it “going viral” or getting research funding next year? Because to me, steadily increasing life expectancy is a more interesting story once you take into account how we got here, and what we’ll need to do to keep up with our own needs while taking care of the elderly.

30 Comments

  1. Paul Alper says:

    Interpreting data is indeed tricky so here is a hilarious website by Brad Plumer, “These 31 charts will destroy your faith in humanity”

    http://www.washingtonpost.com/blogs/wonkblog/wp/2013/05/24/these-31-charts-will-destroy-your-faith-in-humanity/

    Number 7 showing the marked increased in U.S. life expectancy from 47 in 1900 to 77 in 1998 is interpreted as “More 77-year-olds are dying than ever before”
    Number 20 showing the fabulous rise in electrification of U.S. homes to almost 100% by 1965 is seen as “Electrification rates have stagnated since the 1960s”

  2. Tom Passin says:

    The 2nd graph is a good example of one that is indecipherable to someone like me who has some color blindness. I’m partially, though not totally, red-green color blind. I can easily distinguish the US and Ethiopia (yellow and blue). The others look nearly identical to me. India seems brighter, and I *think* I can distinguish it, but partly because it is isolated at the right. Two of the others I can barely distinguish, but I can’t match them to the legend.

    Which country has that single long record going back to 1543? I can’t tell.

    I’ve commented in this vein before; this is a pretty good example of how not to do it.

    OTOH, most of these details don’t really matter for this particular chart, if all you are interested is the general idea.

    • Dzhaughn says:

      That’s a bummer! It’s inconvenient, but the semi-good news here is that you could follow two links and get the raw data.

      More good news is that these days colors are generally chosen by algorithm, so it does not take another mass political movement to make a change, once you can describe precisely a better way to choose colors in general. Alas, it has to be a way that’s better for everyone, not just you. However, that way could involve letting individual viewers choose their own color scheme, although that actually requires broader awareness.

      Still, it could happen; around 4% of the population is color blind. Only 1% has celiac disease but look at all the Gluten Free options these days.

    • Ben Hanowell says:

      I apologize for my insensitivity to color blindness. I could have reproduced that graphic in a way more friendly to you. You should contact Our World in Data to ask if they could have a color blind friendly palette. I could have used small multiples to drive home my point that Western Europe and US and Japan came first followed by everyone else.

    • mpledger says:

      England has the long record going back to 1543.

      According to wikipedia, “Parish registers were formally introduced in England on 5 September 1538 following the split with Rome, when Thomas Cromwell, minister to Henry VIII, issued an injunction requiring the registers of baptisms, marriages and burials to be kept.”

      On the face of it getting worried about American life expectancy seems over the top… but there is lots of subtext… the USA spends the most per capita on health care so they shouldn’t be languishing behind most first world countries … and there a real differences in subgroup health between ethnicities and between different parts of the country.
      And there are real problems with access to health care and the economic catastrophe that can occur to people when they have an unexpected medical event.

  3. Jonathan says:

    Why would we want to stop people from talking about things? We’re a flock of birds twittering. Some of our tweets are important but most is just twittering. The idea that we can or should stamp out stories that aren’t accurate – or in this case, aren’t entirely accurate if you look at longer trends and other episodes – is not only absurd but dangerous: who decides what is to be eliminated? That guy? Since the data he points to tells its own story, then why did he bother to point out that there was a piece that amounted to nothing rather than ignoring it? Indeed, I’d argue that most research work could be eliminated by this standard – and I’m including published work – because on the scale of nothingness to ground-breaking most of it by far is by far closer to nothingness. So get rid of all writing and talking and we can all be silent to keep the accuracy police happy! Let a thousand duct-taped mouths keep their thoughts to themselves. Then maybe we can listen to the birds twittering.

    • Andrew says:

      Jonathan:

      Nobody’s saying to stop people from talking about things! Hanowell wants people to “stop crying wolf.” In no way is that equivalent to stopping people from discussing or writing about these issues.

      So, maybe you have a point, but whoever you’re arguing with, it’s not Hanowell and it’s not me.

      • Jonathan says:

        Andrew, if you expect people to stop crying wolf, then you’re expecting them to be quiet. Part of the twittering of our species is that process. BTW, I’m mostly arguing with Hanowell because the point made isn’t important at all: if someone wants to say “there’s a downturn!”, that doesn’t in any way affect the actual trend of the data and it would only be if the downturn becomes material that this would matter in any conceivable policy approach. So why point it out? People say silly things all the time but this to me is like correcting someone’s split infinitive. To the extent I’m arguing with you, which I’m not, it’s that this is stories, not studies, and apparently not even a projection and thus it’s entirely different than your usual concerns with accuracy and meaning.

    • Dzhaughn says:

      Oh please. Birds are smarter than you think, and therefore less tolerant of wrong tweeting than you suggest.

  4. Anoneuoid says:

    How accurate/relevant are those historical life expectancy numbers? Most of the time when I look up a historical figure from pre-1900 it seems they either died of violence or in their 80s.

    • Bill Jefferys says:

      It’s not particularly relevant, both for the reason you mention and for the fact that a lot of the increase in life expectancy is just due to much better care of infants and children so that people born actually become adults.

      There’s another perverse fact about these figures. Some, particularly Republicans, look at graphs like the first one with alarm and think that this means that programs like Social Security are unsustainable because they will “run out of money”. But programs like this are primarily aimed at the elderly (there’s a survivor’s insurance component, but it’s a minor part of the outlays), and although there has been some increase in life expectancy for individuals that actually attain age 60 or so, it has been very much more modest than you’d think by looking at that figure. Furthermore, the people that have experienced the greatest increase in life expectancy from age 60 are those that are better off financially. The result of this is that the people that are being most damaged by such things as the increase in full retirement age that took place during the Reagan administration are those that can least afford it, those that are worse off financially and will therefore lose the most. These are primarily blue-collar workers whose jobs have been by and large considerably more strenuous on average and who find themselves having to retire earlier than those in better financial situations…so this decrease in benefits has hurt them considerably more.

  5. elin says:

    Among the frustrating elements is that there is no real understanding of how it is that demographers calculate life expectancy or the impact of using the mean versus the median. Or (yet again) how the movement of cohorts through life tables may create many small changes (e.g. the cohorts where the birth rate was falling probably are older even within the one year intervals than the ones who were born when the birth rate was rising). As stated in the linked article, life expectancy is most importantly shaped by survival of young children and the elderly.

  6. Tom Passin says:

    Ben Hanowell said:
    “I apologize for my insensitivity to color blindness. I could have reproduced that graphic in a way more friendly to you.”

    Thank you for writing this. My comment was to emphasize the subject, which has come up here in comments several times before. The intent was mostly a consciousness raising rather than singling any one work out.

    To help people like me, I suggest using more cues than just color and saturation. I have found that only two or possibly three levels of saturation can be discernible, but more become hard. Colors taken from the palettes of different kinds of color-blindness are helpful: red with blue, green with yellow, etc. I can often distinguish colors from grays, but reds and greens don’t separate well from browns most of the time. Black, dark, and light gray usually work well together (I personally can usually distinguish light gray from light cyan, but I wouldn’t be so sure about other folks).

    Line width, texture (e.g., solid/dashed), open vs closed symbols, all are very helpful.

    To apply this information to the second graph, I’d try making the line that goes al the way to 1543 a thin, solid black line. Then the thicker (is it red?) line that starts right on top of it, before 1800, I’d make that a medium width medium-to-light gray. The thin black on top of the wider gray would be easy for nearly everyone to see.

    I would only use symbols to differentiate colors, not for time periods (I’m not sure if you did this or not). I’d make half the countries that use symbols (don’t use them for all the countries) closed and half open.

    Well, anyway, that could be a starting point. I’m basically emphasizing here that one should try to use more cues than just color.

    Thanks for this chance to follow up.

    • Rahul says:

      Are any client-side solutions possible? I mean, can the hues you cannot distinguish between be automatically remapped by a client side application or browser plugin?

      Just curious. Then it’d help get a customized solution to every color-blind person depending on his or her individual perceptual limitations.

    • A.G.McDowell says:

      I claim that good GUI design means following the rule quoted at the bottom of http://www.lukew.com/ff/entry.asp?11: “The golden rule is the designer must never rely on color alone, it must always be a redundant clue.” I do sometimes have problems negotiating the changes required by this past my boss, whose initial GUI designs always include color coding for everything, using a bizarre selection of incredibly garish colours. He is the biggest fan of colour coding I have ever come across. He is also red-green colour blind.

      PS There are sometimes options in web browsers to control colours. For instance, in this Firefox I see under Options>Content options to take back the selection of colours from the html. This may not help, because this may be used for different purposes. Somebody with other vision problems might be using this to increase the contrast (especially with the current vogue for low-contrast design). Somebody who has remapped colours to make the text more readable (and perhaps forgotten about this setting) may find that colour coding in graphics is incomprehensible, and not be able to find a setting which satisfies both requirements.

  7. Tom Passin says:

    A.G.McDowell said:
    ‘I claim that good GUI design means following the rule quoted at the bottom of http://www.lukew.com/ff/entry.asp?11: “The golden rule is the designer must never rely on color alone, it must always be a redundant clue.”’

    One way to approach it is to print the image on a black and white printer. Now can you discern all the information? In the says when most copiers were B&W, that’s what I did routinely. The common fate of most graphics back then was to be copied on B&W copiers. You’d be amazed how many graphs were copied in B&W and given out, despite being unintelligible.

    I’m not saying that all graphs should be gray-scale, but if they work in gray-scale, they usually work for almost everyone. Color can then be a nice (and useful) addition for the normally color-sighted folks.

  8. Tom Passin says:

    Rahul said:
    “Are any client-side solutions possible? I mean, can the hues you cannot distinguish between be automatically remapped by a client side application or browser plugin?

    Just curious. Then it’d help get a customized solution to every color-blind person depending on his or her individual perceptual limitations.”

    It’s a good idea. But it’s not the whole answer, because as I’ve been saying, one should use more cues than just hue. There are saturation, line width and texture, and symbol shapes just to name some of them. They should all work together to help distinguish the different data sets.

  9. Life expectancy is not the outcome that I care about. I’m much more interested in years-of-health. Check out the fantastic Health Canada video linked in this comment.

  10. Mary Beckman says:

    Another thing that is misleading about the chart from the OurWorldInData.org site is that most of the x-axis is taken up with datapoints from before any of the countries shown had any national health insurance systems. Indeed, half of it is from a time when a common medical treatment was leeching (the treatment that killed George Washington). A more meaningful comparison be the one presented in this report:
    https://www.oecd.org/unitedstates/Health-at-a-Glance-2013-Press-Release-USA.pdf
    which shows that the US now has a lower life expectancy than the average for OECD countries, and that the rate of increase has leveled off. The leading causes of death, too, are different between the US and other countries that have faster rates of increase. In particular, diabetes and diabetes-related complications, which are expensive to treat as well as not very conducive to a happy old age, are far more common causes of death in the US than in, e.g., Japan.

Leave a Reply