« How to tell harassment from the P Game | Main | Of mice and men Friday open thread »

June 23, 2011


With even a slight acquaintance with American population density patterns, the top map tells you that most Americans do not live in the counties with declining life expectancy. The color coding of the map for anything but declining life expectancy is atrocious, (Shades of pink and gray???) but it appears to me that the washed out shade of gray that codes the highest increases in life expectancy is right in the middle of high population density areas, too. So I'd say, based on that map, that US life expectancy went up, and perhaps substantially.

I wonder if a similar map of other developed countries would show pockets of declining life expectancy, rather than uniform increases? Perhaps not, the US is a comparatively heterogeneous country. But maybe so.

"But it's also apparent that these local issues are part of some national problem that's affecting much of the rural Midwest and South -- and is moving even many white people out of First World America."

You know, that was my first guess: That we were looking at counties with net out-migration, with the sicker people staying behind. But, no, http://www.forbes.com/2010/06/04/migration-moving-wealthy-interactive-counties-map.html>this map doesn't seem to confirm that theory.

All I can really say is that they went to considerable lengths to make the situation look more dire than it really is. Remarkably terrible map design, that first one. It's hard to imagine it wasn't deliberate.

I may be mistaken, but the first slew of articles about this seemed to talk about this as a Southern phenomenon. It might be because my google news feed is tilted towards that region, but I'm wondering if we should consider it a regional phenomenon, or take it as you do, as a national phenomenon.

About Jackson and Marathon counties, Jackson only has 7,000 households, while Marathon has 47,000. One thought is that the population density allows one county to have more medical facilities, which then makes a difference in a lot of ways. Still, I don't know if that would extend to other places. However, a lot of life expectancy gains might derive from the presence of fast emergency service, which can deal with cardiac cases, so the availability of such care might factor in.

I've been pouring over these maps



Fixed the formatting...pesky html.

One thing I've noticed from the article is that the study authors used a different methodology for small locations than for the largeer ones in order to get meaningful numbers. This makes me wonder if the trend you found is actually a statistical error of this method and not an actual signal at all.

The only way to really compare the numbers is to use the same method across all population groups otherwise it's apples and oranges.

Reagan-era federal budgets made cuts to community-based social programs (programs that serve the community as a whole as opposed to individuals), including public health and mental health clinics. These cuts were supposed to be made up by funding from private philanthropy, but in the vast majority of situation the cuts were greater than local foundations and corporate giving programs could replace. So clinics that had formerly provided free services cut back on staff, or added fees, and people who couldn't afford the fees had to go elsewhere. I know this because, at the time, I was working as a research associate at a foundation that was studying the effects of the cuts.

But one of the requirements of the cuts was an end to certain kinds of recordkeeping -- as it was explained to me by a mental health nurse, they were no longer allowed to ask if the person had received help before, or where. The ban on keeping records meant there was no way to track the full effect of the cuts upon the population; without this information, it was possible to make people think the cuts had little effect on public health or wellbeing. I would have liked to study this further but -- surprise! -- couldn't get funding to do so.

I wouldn't be surprised if this contributed to what you're seeing, a generation down the road.

Unusually in relation to a Doctor Science post, I have major reservations about the cogency of the argument here. Starting with the first line:

The United States is no longer a First World country, not all the way through.

This suggests that by whatever criteria you define a First World country, we once were one “all the way through”. But I don't see one word in the post (or in a quick scan of the linked paper) that supports that contention. Do we have data by county for 10, 20, 40, 60 years ago? If not, then how do we know that the US hasn't been about the same on a fine-grained (i.e. by county) analysis all along?

Even worse is this:

We're not a First World country because we no longer have one of their most important characteristics: a high and steadily-increasing life expectancy. A study released this week shows that in significant areas of the US life expectancy is no longer increasing, and in a shockingly large part of the country the life expectancy for women is actually decreasing.

This restatement of the linked article is framed in absolute terms. “US life expectancy is no longer increasing” and “life expectancy for women is actually decreasing.”

But the linked article repeats, again and again, that what the authors studied is US life expectancy measured against life expectancy in selected other countries:

Between 2000 and 2007, life expectancies in more than 80% of United States counties fell in standing against the average of the 10 nations with the best life expectancies in the world, according to new research by IHME, in collaboration with researchers from Imperial College London. [my bold - jm]

This is not the same as saying that absolute life expectancy in the US has fallen. We have fallen behind in a particular ranking. Even more specifically, some counties have fallen behind in the ranking.

This tells us nothing, one way or the other, about whether life expectancy in the US, among American women, or in American counties is falling in absolute terms. It may be, but it may also simply be growing at a slower pace than it once was in relation to the average life expectancy in the 10 nations with the best life expectancies in the world.

I am as ready to get in a lather about depressing topics as anyone else. But not about crises concocted out of what looks like sloppy reading and careless logic. “Falling behind” isn’t the same as “falling.” At least let’s get in a lather about the right things.

Second guessing myself (with apologies for the snark), I looked more carefully at the map in the post, and went again to the original article and again found stuff like this:

The researchers found that women in 1,373 counties – about 40% of US counties – fell more than five years behind the nations with the best life expectancies. Men in about half as many counties – 661 total – fell that far.

This still doesn't say anything about absolute changes in life expectancy. It's comparative.

The map is from the newspaper. The newspaper itself appears to have misinterpreted the article, whether deliberately or out of innumeracy who knows.

Or am I the one who’s delusional? (Going off caffeine may have that effect.)

If three countries, A, B, and C, have life expectancies of A 90, B 90.3, and C 90.7 years, and some time period down the road the numbers are A 91.5, B 91, and C 90.9, C's life expectancy has actually increased, even though it has declined in relation to A and B. This seems to be what the original article is about.

The fact that we are not the best in the world (any more? at anything except military hardware?), or even just that other countries are improving at a faster rate than we are, may or may not be something we should be worrying about in any given instance. But it's not the same as if we were getting worse absolutely, i.e. as if life expectancies were actually falling, which is a different sort of problem. IMHO.

Finally (for now), to echo john j's point: to compare the US by county with entire countries, instead of comparing e.g. the worst-performing American counties with the worst-performing equivalent civil divisions in other countries (along with best to best) does not inspire confidence as a way to draw conclusions.

One of the immediate questions is: are we talking about life expectancy at birth here? Or life expectancy some later point? Because a change in infant mortality can seriously skew the results, even if the life expectancy at, say, age 5 is unchanged. And I would look in very different places for causes and cures, depending on what age sees the big changes in mortality.

Echoing JanieM:
We're not a First World country because we no longer have one of their most important characteristics: a high and steadily-increasing life expectancy.

US life expectancy is steadily increasing, see here. Note that this allows easy comparison with LE changes in Sweden, the UK, etc.
US life expectancy is relatively high: at 78.3, we don't compare that unfavorably with Denmark (78.4), Finland (79.3), Germany (79.4), the UK (79.4). Unless these countries are also falling out of the First World...

There is a problem when some areas of the US have declining LE (unless it is due to eg migration patterns). But I think we could examine that problem without the hyperbole about who is in the First World.

"We're not a First World country because we no longer have one of their most important characteristics: a high and steadily-increasing life expectancy."

I'm not commenting on the accuracy of this post, but life expectancy has been increasing in most countries, poor and rich, badly ruled or not, in most places for the past 100 years. It nearly doubled under Mao. link

There is a problem when some areas of the US have declining LE (unless it is due to eg migration patterns). But I think we could examine that problem without the hyperbole about who is in the First World.

Totally agreed. If we had some reliable information, we could examine a bunch of problems implied or mentioned in the post: life expectancy disparities between genders and races, disparities between genders and races in the rate of life expectancy changes up or down, etc.

But the study on which the newspaper material and this post are based doesn't seem to give us reliable information of this sort.

It has to do not with life expectancy declining in some areas of the US, but with that fact that LE in some areas of the US is declining in rank in relation to the countries with the top 10 life expectancy #'s in the world.

This doesn't tell us whether any given county's LE is actually increasing, staying the same, or decreasing. It just tells us something about a change, over a certain period of time, in where that county falls in a ranked list of countries.

The passage I quoted earlier about the male/female disparity doesn't tell us anything absolute either. Maybe the number of counties where women's LE declined in the ranking was greater than the number of counties where men's LE declined in the ranking because women's absolute LE rose less than men's did, and maybe that was because the men's was lower to begin with and had further to rise. Or not.

My point is that we can't figure any of that out from this study.

I just hate it when people draw sweeping conclusions from bad math. Someone told me a couple of weeks ago, as a horribly ominous climate change fact, that an area of Arizona was burning that was a big as the area between where I live in Maine and New York City. I hadn't been paying close attention to the news, so I hadn't been hearing any details beyond that there was a fire. Googling it, I found numerous references to a fire that was at that time spread over 600 square miles. I don't know if my acquaintance didn't know the difference between 600 square miles and 600 miles square or what (and it wasn't a situation where I thought it was politic to "quibble"). (Never mind that where I live is less than 400 miles from NYC.)

(Never mind that where I live is less than 400 miles from NYC.)

Sure, but how wide is that distance?

1.5 miles

90% of statistics are made up on the spot. Or maybe twice that.

You thought the USA was a first world country recently? how charmingly naive :)

90% of statistics are made up on the spot. Or maybe twice that.

You mean there is a probability that 80% of the statistics we observe cannot possibly exist, much less "be true"?

heh :)

I'm either statistically illiterate, or I was making a joke. Take your pick.

Even if, in the aggregate, the US is still a solidly First World nation, doesn't it suck that that is not the reality for such readily identifiable regions and populations?

In fact, doesn't it suck especially hard *because* the US is such a solidly First World nation in the aggregate?

What Russell said. In fact, that's my default comment in most threads.

To address some of JamieM's concerns:

The study mostly focuses on how different parts of the US fare relative to the international standard of best practices. That is, LE is increasing almost *everywhere* in the world, so just saying "lifespans are still increasing" doesn't mean that "the US health system is great!" Mere increase isn't enough, you have to be keeping up with the Joneses (and their friends the Suzukis, the Müllers, and the Jónsdóttirs).

The data displayed at the researcher's site charts how US counties compare to this international standard, it is about *relative* success/failure. The point there is that despite spending twice as much money per person on health care, US outcomes on the most basic level are getting worse compared to those in other countries, falling further behind. It's like a race where you're still running and getting ahead of where you were, but falling behind the other runners.

The the Washington Post data is *not* relative, it shows the actual gain or loss in LE by county. Women in Jackson County, WI, and both sexes in Pike Co., KY, have a *lower* LE than they did in 1987. These counties have fallen out of the First World into the Second.

Maybe I'm misreading Russell's point, but it seems to me that the US has always been a first world nation in spite of itself, which is to say that it has never been a first world nation thru and thru (but is any first world nation one thru and thru? Nor really sure about that). Johnson's War on Poverty and RFK's visits to Appalachia, the discovery, during conscription efforts for various wars, of the relatively pitiful state of a lot of the nation's citizenry all seem to suggest that our belief in first world status is a form of American exceptionalism.

That is, LE is increasing almost *everywhere* in the world, so just saying "lifespans are still increasing" doesn't mean that "the US health system is great!" Mere increase isn't enough, you have to be keeping up with the Joneses (and their friends the Suzukis, the Müllers, and the Jónsdóttirs).


The point there is that despite spending twice as much money per person on health care, US outcomes on the most basic level are getting worse compared to those in other countries, falling further behind.

That’s the point? Why didn’t you say so? The original post didn’t even mention health care, except to quote a researchers saying that in relation to the phenomena under discussion "It’s not the health care system that’s having the biggest impact on health; it’s the community".

Life expectancy numbers might just as easily be cited in support of a thought train about the appallingly huge food portions taken for granted in American restaurants, or the depressingly unhealthy characteristics of a lot of those foods (processed fats, processed carbs, enough salt to choke an elephant), or smoking, or drinking, or lack of exercise, or the crap wages that so many Americans get, if they get any wages at all these days.

I certainly never said that my doubts about how the statistics were being used supported any conclusions one way or the other about the US health care system, or anything else. What I object to is manipulation by dishonest use of numbers. (I’m not saying that’s what you did, DS, but on the face of it it certainly seemed like someone did in this sequence of reportings.)

And I don’t take it as a given in any general sense that “you have to” keep up with the Joneses, although decent health care might be an exception.

how US counties compare to this international standard, it is about *relative* success/failure.

But that's not a sensible comparison. It compares the low end of the geographical distribution of LE in the US to the average for other countries.

A comparison of LE distributions, either geographical or general, would indeed be interesting, but the comparison here doesn't tell us much. I'm going to guess that the worst areas of Sweden have a lower LE than the US average. What would that mean?

Out of curiosity I wonder how well or badly the smoking rates in women in the 1970s-80s track with the lowered life expectancy in counties now? I know that's had an impact on what women are dying of and presumably it would also impact on mortality.


The US has always had fairly high numbers of women who smoke, so that's one of the first things the study authors thought of. It turns out that the pattern of smoking prevalence (by Congressional District) doesn't really map very well to areas where women's LE is declining. North Carolina, the Upper Penninsula of Michigan, and upstate New York all have very high smoking prevalence without showing up on the Second World chart.


That is not the case. For 2000-2005, the worst-off counties in Sweden were significantly better than the US national average. This is in spite of the fact that only 81% of the Swedish population are "ethnically Swedish" -- they are far more diverse than e.g. the Japanese.

Dr. S.,

Maybe. OTOH, it looks, from the link, as if Swedish "counties" are substantially larger, and hence not really comparable to, US counties. They seem to have an average population of about 400,000.

But that's not really relevant. My point is that the comparison given is not really useful, whatever the numbers.

I'm not arguing any sort of position on health care here, just suggesting that the analysis could be better.

Likely to do with the maternal mortality rate. It has been going up instead of down the past ten years or so.

I'm still not sure what point you (Doc) are trying to make. You interpreted my objections to your statistics as saying "the US health system is great!" -- which was not remotely my meaning, and which I never remotely suggested. Maybe you thought that was what I was doing because you were trying to say (without ever actually saying it) that "the US health care system isn't great," and since I seemed to be disagreeing with you, you thought I was disagreeing with you about that.

I wasn't, because I didn't even realize that that's what you were getting at.

Then you wrote, The point there is that despite spending twice as much money per person on health care, US outcomes on the most basic level are getting worse compared to those in other countries, falling further behind.

But I still don't know what connection you're claiming between the health care system and life expectancy as an "outcome." Or is it really poverty and life expectancy you want to examine? I can't tell.

The whole set of topics is so interconnected that I don't see how any of us could "prove" anything, but I'll venture some thoughts.

If you're suggesting that the inadequacy of our health care system is the reason our life expectancy rates are not where you would like them to be, I think it's much more complicated than that. If anything, the causation is the other way around, but that’s only from a position of such oversimplification as to be meaningless. The food system, the educational system, and our lifestyles in general play major roles. Among other things.

Americans eat too much bad/empty food and not enough good food. It takes a lot of effort to construct a daily diet of really good, fresh, healthfully grown/raised food, prepared without too much sugar/salt/fat. (Particularly if you're not supposed to eat wheat. And now, salt.) I am well educated; I have enough money to live comfortably if I keep it modest; my kids are grown so I have only myself to worry about; and I have a lot of food sensitivities and restrictions (i.e. extra incentive). I find it a pain in the a$$ to try to eat well. I eat out a lot because that’s my main social life, and also because I travel a lot, and it’s hard to feed myself healthily if I don’t stay home, buy as much as I can from local/organic sources, and prepare it all myself. I mention all this to suggest that even someone who takes a lot of personal responsibility for eating well doesn't find it easy in this country.

Look http://dhss.delaware.gov/dhss/dph/dpc/files/usobesitytrends_2009.pdf>here for a series of maps showing US obesity rates from 1985 to 2009. Scroll past the text, then watch the map change as you click from year to year.

The obesity trends were not caused by the health care system. Nor can we expect the health care system to magically (or even by the expenditure of large amounts of money) make the effects of unhealthy eating/lifestyles (not to mention the effects of poverty) go away.

For another angle, http://www.nationmaster.com/graph/hea_obe-health-obesity>here's a chart of obesity rates by country. Funny, look where the Suzukis, the the Müllers, and the Jónsdóttirs are. Look at the Italians: famous for wonderful food (I think the Italians would be appalled by the food that Olive Garden purports to have modeled on food served in Italy), yet the obesity rate is well under a third of ours, and they're 12th on the life expectancy list, whereas we're 36th, right there with Cuba and, interestingly, Denmark. The US tops the obesity list at 30.6%, 126.4% of the next highest country, our neighbor Mexico.

I guess unlike Bernard I am trying to make a very general (and inevitably oversimplified) point about the health care system. And that is: if our life expectancy numbers are bad (I'm still not convinced by the analysis here, but setting that aside), it's not because, or only because, of our health care system, and it's not up to our health care system alone to fix the problem.

The problem is systemic in a more big picture way than that. You can't even talk about the health care system (e.g. maternal mortality) without talking about our politics and our economic inequality. In fact, I would love to construct a slide show to illustrate the increasing gap between the richest earners and everyone else in this country in the last 30 years. The echo of the obesity maps would be uncanny.

Now they even outsourced call girls to India ;-) (referring to the spam above).

Danes put too much sugar into everything. My memories are of coke that defies the laws of chemistry (at least thrice saturated solution), sausage that tastes sweet (sugar instead of salt?) and sweet dark bread (which should be selfcontradictory). Admittedly the other Scandinavian countries also were plagued with that bread (often imported from Germany where you would not find it anywhere) but the Danes seems to be the worst.

Thanks, thebewilderness. This is the kind of "threat" that women need to be worried about: "The report added to previous concern about maternal mortality in the U.S. raised by an inquiry in California suggesting that maternal mortality rates had tripled there in the previous decade."

It would be interesting to look at the correlation between maternal (and infant) mortality and
a) lack of health insurance (and, therefore, prenatal care)
b) illegal immigrant populations (again, a lack of prenatal care, due to a desire to avoid contact with authority as much as possible)

I don't know that either of those are significant factors. But they are a couple of possible factors that leap to mind. Could be a little difficult to disentangle poverty and either one though....

I'm only concerned with my own life expectancy. And, right now it's not looking very good.

I do not want to sound callous but maybe to reach the end of one's natural line now is not the worst given the rather dire predictions for the not to far future ("Go while it's still good!").

Hartmut, are you suggesting that I should drop dead?

Jackson County, Wisconsin, is really really freaking small. It's so small that having a very small number of young people die unexpectedly could result in a significant dip in "life expectancy" calculations. It could very well be a blip with no meaning whatever.

Shorter version: I want to see the error bars on their estimates of the decline in LE, especially in some of the smallest counties.

Old Soldier, that's exactly not what I mean.
I assumed that you meant that you expect to die in the not too far future from natural causes at a reasonably old age. If I were in the US and not rich, I'd rather be 80 than 65 (I am neither) since I expect the situation for seniors to massively deteriorate in the coming decades. In that scenario the older person dying while the situation is still tolerable has imo drawn the better lot than the one that has to live through the decline and fall. I fear we do not live any longer in a "will I live long enough to see [insert something marvelous]?" world but in a "I hope I will not have to see [insert something awful] in the few years I have left" one.
That does not mean "head for the exit now" but just not objecting to letting go when the time comes.

Whenever health seems to line up with geography, I start to wonder if pollution is having an effect. In all the talk about health care, poverty, and individual diet, it's easy to overlook. (There are big businesses actively trying to make people overlook the widespread effects of pollution.)

Pollution exposure tracks with poverty to a certain extent. If you and your neighbors have money, local industry will take you more seriously when you want them to use emission controls. Or when you want them to put their new factory in some other neighborhood. Having more money means you don't have to live next to the railroad, or the highway interchange. People in poor rural areas are more likely to be exposed to agricultural chemicals

The comments to this entry are closed.