Hacker News Reader

Comment by abeppu | Hacker News Reader

Zooming out from this specific town and article, it's kind of interesting to look at regional differences in cancer incidence. Like, I get that people live very different lives in different places, and that states have different resources to bring to public health problems, but the areas of red in the midwest, northeast and gulf coast are quite striking.

https://statecancerprofiles.cancer.gov/map/map.withimage.php...

View on HN

Beware of the sample size issue. The Midwest has lots of low-population counties, for which a single cancer case plus or minus makes a great difference in the incidence rate.

See also: Bayesian Data Analysis, Gelman et al., 2nd edition, section 2.8.

Wyoming and Idaho should have the same problem but appear pretty blue.

That's the plus or minus part.

Thank you for the reference. Imagine where we would be if everyone provided sources to their statements like that...

But we should expect the plus or minus to be fairly evenly distributed if there weren't an underlying reason besides randomness and small sample sizes. That would lead to regions with lots of blue/red/yellow speckling. Instead we see fairly homogeneous regions of red and blue with transition zones between them. This definitely doesn't look like a noisy undersampled data set

I've done a lot of cross-country exploring of the USA by motorcycle in the past, and it always shocked me how obviously harmfully polluted the air can be in residential areas neighboring industrial activity.

I don't understand how it continues unchecked. Sometimes I'd spend as little as an hour breathing the stuff and my nose and throat would be sore for days afterwards.

My assumption is the residents are too poor so they lack representation and likely are employed by the same companies behind the pollution. The government should be more actively monitoring these industrial activities and protecting its citizens from their negligence.

> The government should be more actively monitoring these industrial activities and protecting its citizens from their negligence.

Local and state politicians don’t want to be seen as killing industrial jobs. The Federal government really needs to take action here, but this is obviously a non starter in the current administration.

Pittsburgh’s air (while still f’ed up) is infinitely better than it was 30 or 40 years ago, largely because of EPA regulations.

> I don't understand how it continues unchecked.

Profit and lack of regulation and lack of effective penalties

I would be curious to see this alongside / with some relative derived metric for life expectancy. To my knowledge cancer can definitely be influenced by environmental/lifestyle factors but its incidence would also be increased by the relative age of the people in an area. For example if everyone dies by 50/60 of alcoholism, addiction, diabetes, or work-related injuries the cancer incidence would be lower than expected even though the people there are definitely unhealthier than the general populace

Actually, I see that the stat is age adjusted. So not sure what effect that would have on the comparison that I described

That map makes me think of another very interesting map:

https://www.npr.org/sections/krulwich/2012/10/02/162163801/o...

Not that they are related- but the mysterious & provoking bands evoke eachother.

That was such a fascinating read. Thank you for sharing!

The vast differences between the SW corner of Virginia and the surrounding states suggest there's some sort of difference in reporting or screening going on.

Look at the mortality data as well.

I feel like the coloring might make it appear more striking than it is? The difference in cancer rate thresholds between the blue and red colors is just over 20%. We don’t know the actual rates, of course, just noting the coloring range is surprisingly narrow.

Seems to overlap somewhat with smoking: http://time.com/3832031/smoking-map/

The mortality data are possibly more telling. High incidence may be a matter of screening and early detection. Dead bodies have stories to tell and are harder to hide.

The outline of the Ohio-Mississippi river valleys is particularly clear.

https://statecancerprofiles.cancer.gov/map/map.withimage.php...

Both those rivers coincide with state borders, so I suspect the cancer incidence map is showing artefacts of reporting differences.

The mortality map shows equivalent death rates across borders. Indiana, Kentucky, Illinois, Arkansas, Mississippi and Louisiana. And is markedly higher in non-urban (and desperately poor) counties. By contrast, incidence rates (clinical detection preceding mortality) clusters more toward urban and wealthier counties.