The Geography of Cancer Risk

I came across a map this morning that visualizes the per county risk of getting cancer the US. Except for Kansas… maybe they don’t have cancer there?

I don’t have any commentary on it, but I thought it was interesting and that other people might like to see it and/or have thoughts to share about it.

[URL=“http://www.visualizing.org/visualizations/geography-cancer-risk”]Here is the map.

(There is a little arrow button on the bottom left corner of the picture that you’ll need to click in order to activate the interactivity of the map)

Hmm. That is interesting, but I wish there was more explanation provided. Some areas that I would normally think of as healthy, like the Seattle area or much of Montana, have surprisingly high rates. Is that due to better reporting? Environmental factors? Something else? It’s impossible to say from the map.

Notice how dense it is on the St. Louis area.

My high school–which remains one of the highest-population high schools in the state of Missouri (927 people in my graduating class in 1985) sat on this large tract of land, various portions of which were either wildlife preserve owned by the Busch Brewery, or restricted/zero access sites owned by the state and federal government. Turns out there was a nice munitions plant just a mile away from the school during WWII, and apparently they did some post-Manhattan Project munitions work there involving uranium…and just dumped radioactive waste into the ground. At the end of our football field–just 20 yards from our track–was this big pond. It was surrounded by a double-strength chainlink fence topped with barbed wire. It had radioactive waste warning signs posted all over it. None of my classmates who were smart EVER drank from the drinking fountains at our school. That area of St. Charles County now has an absurdly high rate of cancer per thousand persons, even though it is still very lightly populated. I know of dozens of former classmates who have had cancer.

Also, in north St. Louis county there were apparently unregulated toxic and dioxin-laced chemicals used to a great extent in the late 1960s and early 1970s and there have been stories in local media about the high cancer rates in the various townships in that area.

Kinda makes you wonder what’s skewing the stats in the northeast, too, no? Delaware, NJ, NY, Connecticut, Mass…yikes.

I wonder what happens to the map if you overlay average age and smoking rates.

I have to wonder if the analysis is taking proper account of population density, given the giant swath of nothing across the Southwest and the opposite in the Northeast.

Yeah that would be kinda important to know. It would be pretty pointless if it were just a map of cancer cases, since that would probably just be a copy of a population density map.

If you mouse over the map, you’ll see that the authors are showing new cases (by county) per year per 100K population.

Well, click on it, then mouse over it. I missed the first step, so it seemed non-interactive. That method has its own population-related problem, which is that data for sparsely populated areas is going to have high variance due to random noise. I looked around til I found one: Borden County, Texas. It’s the dark red spot south of the panhandle. Its cancer rate: 507.1 new cases per year per 100000 people. Its population: 641. So that works out to around 16 people over the course of the survey. Also, looking at the numbers related to the colors shows that the variance is pretty tight. Everything up to 400 is the lightest color, and somewhere around 500 is the darkest (it’s not exactly 500 because I found a 496, so I’m not sure what the cutoff is). There are adjoining counties in NJ with the aforementioned 496 and 569, the highest number I’ve found, and they share a color so that 440 and 450 can have distinct shades.

In short, this map feels less useful than a sortable list of numbers would be.

Fuckin’ radon, man.

Air pollution, water pollution, and other environmental factors are probably at play there.

I’m not so sure. There are comparable dark areas along the coast of Alaska, the Pacific northwest…areas that are relatively low in pollution.
I think it’s primarily lifestyle, with maybe some external environmental factors at work.

This map needs additional info…like smoking, drinking and obesity rates for the same areas.

Well for Alaska a lot of employment is in the oil industry and like Funkula said, it looks more like a population density map more than anything else. And in areas with sparse population a few additional persons can turn an area from light to very dark.

That site has been shown to have wildly misrepresented data maps in the past. Though I like what they’re trying to do, the scientific value of what they’re providing is mostly crap and designed to get page hits. The biggest thing they don’t account for is access to healthcare based on wealth. The dark coverage in Vermont is not because it’s a cancer cluster, and the lack of color in much of Texas is not because it is safer than the rest of the country. It is because the average income and educational level in Vermont is wildly higher than what you see in most of Texas (excluding Austin). Texas has the most uninsured citizens of any state in the country and Rick Perry has basically said F* You to the poor of his state. In fact, the map should be showing some of the greatest cancer clusters in the state of Texas as they have the worst water, so, and air pollution standards in the U.S. To get an idea of the real cancer clusters read this article http://www.nrdc.org/health/diseaseclusters/files/diseaseclusters_issuepaper.pdf

You will of course not see Vermont nor Maine on the list.

Phew, being Canadian means I’m cancer risk free.

I agree, and said much the same thing in this thread’s second post. Numbers without context don’t have much value.

Story on the North County cancer cluster in St. Louis: http://www.ksdk.com/video/2133138617001/1/Cancer-map-may-show-enormous-St-Louis-clusters

I’m a medical oncologist and this map is a good example of trying to be too clever by half. I’m sure it took quite some time to put that map together, but whoever did it should have saved their time and studied some cancer epidemiology before releasing something like this into the wild. This map is less useful than the underlying data sources and suggests a conclusion (i.e. darker color = higher risk) that is not supported at all by the underlying data.

The title of the map is very misleading, it should not be called “Geography of Cancer Risk”, the more appropriate label would be “Geography of Cancer Incidence” which is a very important distinction. It suggests that the chance of developing cancer are related to your location in the county. Maybe it is, maybe it isn’t, but the underlying data sources don’t suggest such a thing. Funkula’s points are absolutely right on. Location in the country as an variable for cancer risk independent of age or smoking status or other already known risk factor would be difficult to prove. I’m not saying it isn’t true, but the data sources for this map don’t establish this.

I would highly suggest you send the author, Chris Walker, an email stating this. My concern is someone who would use this information to sell their state as safe, when it is absolutely the opposite. Rick Perry has been advertising in Michigan like crazy, trying to convince people to close their businesses here and move to Texas citing various advantages, most of which are false. For people who care about their family health, this kind of diagram is extremely deceiving.

jpinard, this isn’t the P&R forum man. We get that you don’t like Texas. I posted it here specifically because I thought it was interesting and wanted to avoid the kind of comments you’re making. Instead I was looking for the kind of stuff that Hiredgoons posted.

After looking at the points people made and reexamining the map, I really appreciate the “Geography of Cancer Incidence” point you made. I think the geography of incidence is still incredibly interesting, but I agree the map is kind of misleading. I’d second jpinard’s idea of emailing the guy who made the page to suggest the name change.