Show HN: Mortality Rates in the US Visualized | Hacker News Reader

Show HN: Mortality Rates in the US Visualized

matthewisabel.com

92 pointsmatthewisabel96 comments

One interesting thing I learned from this is how highly correlated mortality from heart disease and cancer are. They're quite different types of disease, but perhaps what they have in common is that they're diseases of civilization in the sense of being caused by de facto "toxins" (I don't mean this literally) of industrialized life like processed foods, lack of exercise, cigarettes, etc.

Is there local data about possible causes of mortality? If so that might be a good thing to add.

How would those look overlaid with poverty/education? Are poor eating habits and smoking significant factors in heart disease and cancer overall?

I didn't run the numbers but the correlation between poverty/education and cancer+heart disease seems relatively high. The mortality rates for those appear to be highest in the deep South, Appalachia and lower Midwest, which are some of the poorest and least-educated areas of the country.

I would love to do some overlays to put it in the context of other data. Something more granular would of course be nice too.

Hopefully I'll find more data sets that I can plug in.

cancer and blood flow are very closely related: http://en.wikipedia.org/wiki/Angiogenesis

They definitely are. Cancerous masses tend to ramp up blood vessel production because that's how they survive. In fact, there are tests, like angiography, that are done to identify areas of increased blood flow, which may indicate cancerous areas.

However, there hasn't really been a link shown between cancer and heart disease, like hypertension, atherosclerosis, heart failure, etc, -- i.e. the two diseases haven't been shown to correlate pathophysiologically. But there is data showing a correlation between chemotherapy and heart disease, which could certainly contribute to the similarity in distributions in mortality between heart disease and cancer. For instance, doxorubicin is a chemotherapeutic agent infamously known for causing dilated cardiomyopathy.

PG, what did you think of that "Can Google Solve Death?" Time magazine cover story from a month ago?:

http://content.time.com/time/magazine/article/0,9171,2152422...

or Google's project Calico in general?:

http://www.cnn.com/2013/10/03/tech/innovation/google-calico-...

Also note that strokes are highly correlated with those two most likely due to it being a consequence of cardiovascular (read: heart) disease. The concentration of heart disease/stroke mortality in that region seems to be highly correlated with obesity distribution, which isn't a surprise. Actually, the state with the highest rate of obesity [1], Mississipi, is also the one with the highest mortality due to heart disease, and 3rd highest for strokes.

As for cancer mortality, the distribution closely matches that of smoking prevalence by state, more so than the distribution for heart disease, again not a surprise [2].

What I find the most interesting, though, are the lone states that don't seem to be a part of any cluster and have a high prevalence of, like high respiratory mortality (and influenza/pneumonia) in Wyoming, or Alzheimer's in Washington and North Dakota, or high drug mortality in Arizona (and not in any of the other border states).

[1] http://www.cdc.gov/obesity/data/adult.html

[2] http://www.cdc.gov/vitalsigns/AdultSmoking/

I have no expertise here, but I wonder if the Alzheimer's rate in Washington and North Dakota reflect differences is how deaths are classified.

They're also statistically small numbers, so percentages are larger comparatively. But it definitely is one of the standout questions from this.

The cancer bumps in West VA and Kentucky (and to some extent the rest of Appalachia) may have something to do with the mining industry.

http://en.wikipedia.org/wiki/Mountaintop_removal_mining

"According to 21 scientific studies there has been major effects on the population in the Appalachia where MTM takes place including over 50% higher cancer rates, 42% higher birth defect rates, and $75 billion a year in public health costs from pollution."

New research suggests that low vitamin D levels may be factors in both CVD and cancer, which could explain part of the correlation.

The high rates in the states around Mississippi are because of the high fraction of African Americans, who experience higher (double) rates of CVD compared to whites, a fact not inconsistent with the vitamin D hypothesis.

Oxidative stress and the resulting inflammation has been (loosely) linked to both cancer and heart disease.

My totally unscientific pet theory is that if you somehow managed to survive long enough, eventually you will succumb to cancer or heart disease. So places with high cancer/heart disease rates might simply be places with the best healthcare system!

Average lifespan data can help answer that question. My guess is both conditions are preventable if caught early, so a good healthcare system would do a better job at reducing cancer and heart disease deaths than most.

That's a good observation but right at the top of the chart it says, "DATA AGE-ADJUSTED AND PER 100,000 PERSONS (2010)". If it's adjusted for age, then I think your theory wouldn't apply.

On first glance, I was shocked by how similar it looked to the electoral map from Presidential elections. 39/50 states in this chart are the correct color for the 2012 electoral map. For fun, I then took the correlation of the mortality rate and percentage of votes for Romney for each state and got at 49% correlation.

It goes without saying that correlation != causation and I do not mean this as a political statement, rather a quantification of this curious resemblance.

Put another way, different mortality causes could be reflected in different political needs.

Remote, low population areas are high for accidents and suicides.

The Alzheimer's chart was unusual, all things being equal, I'd have expected the statistic to be flatter.

Suicide, Alcohol and Parkinson's have a strange sort of correlation.

how come whenever I see these kinds of things, it's always for the U.S.? Is the data simply not available in other countries/areas? I've almost never seen similar visualization experiments for say...France. Is the U.S. just that* much better in collecting and distributing data about itself?

This is a fascinating tool you've put together. One can look at all sorts of interesting contrasts and/or patterns from it.
One that I found interesting is that Washington and Florida have among the lowest rates for the flu, on complete opposite corners of the map and with what most would say are very different climates.
I was also surprised that the flu kills more per 100,000 than alcohol, firearms, or vehicles. It doesn't seem to get the press that those others do.

It should say influenza and pneumonia for that label. I just realized that error. That number sounds more appropriate that label. Adjusting now.
Update: Fixed.

What tool did you use to generate these maps?

D3.js and a bit of jQuery. I really designed as I built. When I clean it up and add some other data it'll probably be a good place to use Backbone.

How does mortality data account for age? Given heart disease, cancer, and respiratory diseases were the top three causes of death in 2010, if in 2060 the average life span were 10 years longer, isn't it likely that these three things would still be the leading cause of death? The geographic distribution is interesting, but it seems to me that to breaking it down by age group would be more informative.

According to the site's heading, the data is age-controlled.

On page 110 of the CDC's PDF the formula for computing age-adjustment is discussed. You'll also find the original data there where you can look column-by-column and see the impact that adjustment has for various states and cases.
http://www.cdc.gov/nchs/data/nvsr/nvsr61/nvsr61_04.pdf

I thought that this would be a time-trend display, like the one shared a while ago by another HN participant from Scientific American:
http://www.scientificamerican.com/article.cfm?id=longevity-w...
We've made significant progress against most causes of death, at all ages, since I was born.

Once I get that data I'll be all over. I definitely think this data (and something more granular) is particularly fascinating when considered over time.

Funny, terrorism didn't even make the list ;)

That implies that the TSA is really good at their job right?

I was kind of hoping that they would all be 100% as a dark reminder that we will all die.

Anybody know anything about VT health stats? I'm struck by the fact that it:
* Has the lowest influenza/pneumonia rate (7.9)
* Has the lowest nephritis rate (6.7)
* Is in the "really low" rate color for HIV and homicides*
* But has the highest Parkinson's rate (9.6)
What's a contributing factor for Parkinson's?
* Edit: That appears to be the "no data" color actually. I just assumed it was very low because VT has few homicides (1.4 / 100k residents average rate for the past 3 years, or about 10 homicides annually)

From Wikipedia, Vermont is 98.12% white. Many statistics correlate with demographics.
http://en.wikipedia.org/wiki/Vermont#Race_and_gender

Some interesting patterns here.
I know several folks who live in Kentucky. Doesn't look like the state does too well. Florida, on the other hand, seems to hold up well across the board. Obesity kills in the deep south and especially the central Gulf states big time, and there's a string of alcoholism than runs along the Rocky Mountains.
I don't find the Connecticut/Hawaii good numbers too strange, as both have very high incomes and cost of living.

If the data is age adjusted it would be interesting to know if the Florida figures are influenced by retirees moving there.

The west virginia->smoking->nicotine->no parkinsons link seems apparent. High lung disease though.

West Virginia is also coal-mining territory, for what that's worth.

I always realy like these kind of data. Not in a morbid sense but just to see the real data instead of marketing driven FUD. Another thought that I always have seeing data like this is that it gives a clear view where the government should spend their/our money.

How did you incorporate data from Washington DC into this? I can almost guarantee that the reason Maryland has such a high rate on the HIV/AIDS chart is because of the District, and its outlying areas (Prince George's County, Anacostia, etc.)

I didn't include DC data. It does exist if you reference the data PDF put out by the CDC.
I originally had included DC but it's a major outlier since it's almost entirely urban. It's also not a state and doesn't visualize well because of its size, so it's not in the csv I used.

What about Baltimore?

Suicide seems to be highest in rural (lonely) cold (also less sunlight) states.

> cold (also less sunlight)
Yeah, this is also consistent with northern Europe and its higher suicide rates in cold, long, and dark winters. There's definitely a correlation between sun exposure and suicide, while it's not the only factor.

For the "lonely" hypothesis, this map is not precise enough to confirm it. You would probably need a map with dots representing the exact location of all suicides on the map, and overlay it with a density map to see if there is a proper correlation or not.

I think suicide is prevented by HIV – the two maps seem to be nearly complete opposites.

HIV correlates with sexual activity, which seems likely to have a negative correlation with suicide. People who are getting laid are happier.

I wouldn't say that. People who are in steady relationships are happier, and those have less HIV risk.
People who bounce for person to person might be having fun, but long term and not usually happier.

What would be behind the suicide strip just in from the West coast? Struggling farmers? (I'm not from the US, so not overly familiar with all the states and what they're known for.)

Social isolation? Three of the bluest states are Ca, Ny, Il, with the biggest cities.

Could be.
This suggests that farmers are concentrated in a central strip: http://www.slate.com/articles/technology/future_tense/2012/0...
The map top right here indicates correlation with population density, as you suggested: http://en.wikipedia.org/wiki/List_of_U.S._states_by_populati...
And the first map on this US history page suggests those areas were late to the state party. Would that have had a strong impact? http://en.wikipedia.org/wiki/History_of_the_United_States_(1...

Alaska and Nevada have some of the highest rates of suicide, but those areas are not suitable for farming. Another article posted earlier in this thread indicates population density isn't a factor, and the map seems to show that as well (Nebraska and Minnesota have low suicide rates and low density). Date of statehood is nothing more than a coincidence.

Some recent research suggests the primary factor is "residential instability":
"Put another way, the very traits associated with the West (and romanticized in American culture)—individualism and independence, stoicism and solitude—may also have deeply negative implications for its people."
http://www.rwjf.org/en/about-rwjf/newsroom/newsroom-content/...

Thanks for the link.
"They also guessed that the West's low population density might play a role in its high suicide rates, but the findings did not bear that out."
For others who don't read the article, residential instability is their suggestion leading to lessened social integration.

Availability of means to take one's life surely matters too in those states with higher suicide rates. The intermountain west is a place where everyone has access to firearms. In fact, I remember from a few months ago that there was a National Public Radio report about just that issue in connection with youth suicide in Wyoming.
http://www.npr.org/2013/03/19/174761612/a-turning-point-for-...
Most of the intermountain west is very arid, so rather than "farming" as that is usually understood, the occupation tends to be ranching (keeping cattle).
About the leading causes of death, they tend to correlate strongly with poverty, and it's not surprising to find high rates in poorer rather than richer states.

Access to firearms may play some part, but the rates of death due to firearm injury is high in the South which has relatively low suicide rates.
Perhaps religious/cultural attitudes play a part? That wouldn't account for Utah though.

Interestingly Alaska has the highest death rate by Firearms. Anyone can propose a rationale for that?

Guns are pretty useful in remote areas inhabited by large, aggressive animals.

Are you saying it's more likely to commit suicide just because you own a gun ?

Suicide by firearm is more likely to succeed than suicide by other methods (for instance poisoning). Per the American Association of Suicidology, "if a gun is used to attempt suicide, a fatal outcome will result 78% to 90% of the time."[1] Suffocation is also very likely to succeed but poisoning results in death less than 5% of the time and falls about 25% of the time.[2] One can make an impulsive decision to commit suicide with a gun while other methods (for instance hanging) may take considerably more thought and preparation. Suicide by firearm seems to be particularly problematic for youth.
[1] - [http://www.suicidology.org/c/document_library/get_file?folde...]
[2] - http://www.cdc.gov/violenceprevention/suicide/statistics/cas...

Good point, thank you. Is there any data on suicide attempts per state, then, available anywhere online ? Are suicide attempts even recorded officially ?

According to the Harvard Injury Control Research Center, there is some risk, particularly when considering younger demographics.
"The preponderance of current evidence indicates that gun availability is a risk factor for youth suicide in the United States. The evidence that gun availability increases the suicide rates of adults is credible, but is currently less compelling." [1]
There's a lot of good additional data in the original report. According to the table on page 40 [2], self-harm by firearm represented 19,392 of 38,364 suicides. This dwarfed the number of gun-related homicides which was 11,078.
Page 83 explores death by intent. Interestingly, there's also a category for legal intervention/war which represented 344 of all firearm deaths.
[1] http://www.hsph.harvard.edu/hicrc/firearms-research/gun-owne...
[2] http://www.cdc.gov/nchs/data/nvsr/nvsr61/nvsr61_04.pdf

in addition to your other replies, I think it's worth pointing out that in Canada the native populations have an elevated suicide rate. It wouldn't be at all surprising for that to be true in Alaska too.

In Alaska, there are a number of factors that contribute to a high suicide rate. The time of contact between Alaska Natives and others was much more recent in Alaska than elsewhere in the US. That contact brought with it a number of issues:
- disease: Flu and smallpox epidemics wiped out more than half the population in many villages. This was a traumatic event, that left many young people disconnected from their families and cultural groups.
- cultural suppression: Initial contact with Native groups often involved missionaries. While a few missionaries added to an existing culture without taking anything away, most missionary groups came in and told the Native population that their ways were of the devil, and they needed to drop what they were doing and take on Christian traditions. These efforts contributed to severe disconnects between generations.
- introduction of alcohol and other drugs: Combined with all of the other issues that led to young people feeling isolated, alone, and depressed, alcohol and other drugs have had devastating effects on villages, and on urban areas as well.
These are complex issues, and not issues that everyone feels comfortable talking about. They are issues I had no idea existed before I moved to Alaska. That's a really brief overview, but I'm happy to expand on any of those if anyone has any questions.

This is absolutely sluggish in firefox 24, but runs like butter in chrome. Does anyone have a similar experience?
It looks like New Yorkers look both way when they cross the street.

Works in Waterfox with no issue...

Looks like suicides and accidents are at least a little bit correlated as well. I wonder if that is due to incorrect attributions? It could go either way, too.

Both happen in states where people live long distances apart. Loneliness, boredom, and lots of driving maybe?

It's interesting, but I want to know, as a young individual what is likely to kill me? I want to know where I should put my efforts towards not dying.

Have a varied diet and be somewhat physically active regularly?

So it's actually even worse than that - you need to know not only what will stand a good chance of killing you, but what is sensitive to intervention too. You need a causal link that you can break, not just correlation and that's so hard to research that there's much less known about it.
For example, studies tell us that sitting more than three hours per day correlates with a reduced life expectancy: http://www.usnews.com/news/articles/2012/07/09/study-excessi... but that doesn't tell you that if you take someone who is sitting that much and make them stop it'll make any difference at all.

Right but it's evidence towards the hypothesis "standing for three hours will increase your life span". It seems less likely that people who live longer are inherently more likely to stand.
But yes I see your point. A lot of pop science "doing X will help you live longer" strikes me as complete bullshit.

You're quite right of course (although actually, it seems very plausible to me that fit people are more likely to stand more / choose jobs where they will stand more) My comment was mainly out of the frustration I have because so many interventions are very difficult to do and so I want good confidence that they will actually make a difference, but most research just gives you at best strong correlations.
Still, I've got a standing desk now...

Why is there such a high rate of Alzheimer in Washington state? There may be something to discover there related to its causes.

It does seem like quite an outlier. I found another article that mentions it and splits it by county. http://socialcapitalreview.org/washington-tops-u-s-in-12-yea...
It looks like a problem in the western part of the state, but I don't see any suggestion of why that is.
*Correction: Meant county instead of country.

This definitely deserves some kind of investigation. There's probably something to learn from such high levels of difference in very close counties : http://socialcapitalreview.org/wp-content/uploads/2013/03/Sc...

As a resident of Washington the first thing that comes to mind is the difference in climate between the west and east. The eastern part of the state is mostly semi-arid, nearly desert in some areas.
I've heard we also have one of the highest rates of multiple sclerosis in the US, but I'm having a hard time finding good data on that.

It could also be something innocent, like a large treatment center or palliative care facility.
That is, unless the data is meticulously groomed to report the patients' county/state of origin.

I contacted Dr Doraiswamy (a well known Alzheimer specialist) and he could not think of any other reason than "higher rate of APoE4 genotype due to greater Scandinavian heritage". He also suggested this could be coming from the way the death ratio is calculated "ie the denominator of deaths to all other causes is lower due to better diet".
I'll see what I can find.

You mean by county, I presume.
I would also be interested in county data, if only because upstate New York and New York City are essentially two completely different places :P

Reminds me of "A Fire Upon the Deep", because the warmer the climate the worse the mortality rate.

That's not accurate.
California, Arizona, Hawaii, Florida are among the best states for longevity. The mortality rates in Hawaii for example are the lowest, and California is #2. The numbers for Texas and New Mexico, while elevated, are still dramatically lower than the high mortality belt numbers.
Also, one of the biggest offenders is West Virginia, and having spent a lot of time there, it's not a particularly warm state about half the year (certainly not compared to Mississippi).

Is it only me, or is California suspiciously low in almost every single category? Why would that be?

Suspicious in relation to what? Deaths per 100,000 is a relative measure and wouldn't be higher just because CA's population is large.

What I actually meant by suspicious is that it's uniformly low across many categories which at first sight look unrelated (e.g. cancer and firearms)

Its either going to be above or below average in every category, whether they are related or not. With enough categories, its going to be similarly situated across several of them no in any case.
Is it suspicious that Texas sits right in the middle in most categories?

> Is it only me, or is California suspiciously low in almost every single category?
Well, its the second best overall, but "almost every single category"? 1. Heart, it is on the low side but near the middle of the pack;
2. Cancer, it is one of the best;
3. Respiratory, it is on the low side but not very low;
4. Stroke it is near the middle of the pack;
5. Accident it is fairly low;
6. Alzheimer's its is significantly worse than average;
7. Diabetes it is near the middle of the pack;
8. Nephritis it is quite low;
9. Influenza it is a bit worse than average, but near the middle;
10. Drug it is slightly better than average;
11. Suicide it is a little better than average;
12. Motor vehicle it is quite low but far from the best;
13. Firearms it is quite low but far from the best;
14. Alcohol it is notably worse than average;
15. Parkinson's it is a little better than average;
16. Homicide it is a hair better than average;
17. HIV it is better than average but far from the best.
No, I don't think that counts as "suspiciously low in almost every single category".

For Alzheimer, if you don't die from any other condition, you will likely live longer and therefore the neurodegenerative illnesses will be more prominent, it seems very logical actually.

Just guessing; maybe age distribution, income distribution, and climate?
Here's a table showing age distribution by state. This doesn't explain everything, but WV does have the highest percentage (16%) of 65-and-overs. That might explain their poor showing, but LA and MS are unremarkable there: http://www.census.gov/compendia/statab/2012/tables/12s0016.p...

California? What about Hawaii? They are by far the lowest in almost every single category.

state names on hover would be helpful

That, or you could learn your US geography. :)

Imagine how lost you would be if it showed the 22 régions of Metropolitan France! :)

Don't assume everybody is from the US.

That internet exists so that we don't have to remember anything.

In that case: Can you name me every county in South East England? :)

Neat. A world map would be interesting, even if the data isn't as readily available.

Reply on news.ycombinator.com