Study: Living Near Fracking Correlates With Increased Hospital Visits

New submitter Michael Tiemann writes: An article published in PLOS One finds increased hospital admissions significantly correlate with living in the same zip code as active fracking sites. The data comes from three counties in Pennsylvania, whose zip codes mostly had no fracking sites in 2007 and transitioned to a majority of zip codes with at least one fracking site. While the statistical and medical data are compelling, and speak to a significant correlation, the graphical and informational figures flunk every Tufte test, which is unfortunate. Nevertheless, with open data and Creative Commons licensing, the paper could be rewritten to provide a more compelling explanation about the dangers of fracking to people who live within its vicinity, and perhaps motivate more stringent regulations to protect them from both immediate and long-term harm.

Comments

Re: Happy Fag Marriage! Nope!

[Anonymous Coward]

Failtroll. Diesel doesn't light on fire like that.

What's a Tufte test?

[thegarbz]

What the heck is a Tufte test? Normally I would Google it but.... actually I did Google it and I still can't figure it out.

Re:What's a Tufte test?

[alvinrod]

It's probably a reference to Edward Tufte who wrote The Visual Display of Quantitative Information. If you follow the second link and look at some of the charts used, they're not very useful because they completely fail to convey the data in a useful and meaningful way.

Also, I wouldn't call the statistics overly compelling either. They ran enough tests that they were likely to come up with at least one positive result. What they should do is use the few positive results that they've recorded here and verify them by conducting the same experimental procedure in different locations where fracking is also occurring to see if the same results are being seen.

Re:What's a Tufte test?

[thesupraman]

It means that these people are trying to RAM smoke up your arse, but cannot make it look convincing enough even then,
so want help to polish the turd.
They are trying hard to pull a 'correlation is causation' scam, but dont even have the ability to do that it seems.

Perhaps they need to ban Icecream first.
Killer Icecream

The obvious problem here is that there is almost certainly a correlation between these locations and poorer communities,
which also have a very well established correlation with increased health issues.

The scary thing is that this is even being reported. Congratulations Slashdot. It almost but not quite makes it to satire!

Re:What's a Tufte test?

[manu0601]

They are trying hard to pull a 'correlation is causation' scam

Except that in this case we have an intervention study, as some areas started fracking activity while other did not. Therefore looking at data versus time will tell us something.

And I also not we have explanations for causation. I see two obvious: chemical leaks, and nocebo effect.

Re:What's a Tufte test?

[Citizen of Earth]

Sounds like the XKCD Significant test.

Re:What's a Tufte test?

[darkain]

Here, I found an entire web site dedicated to helping find these correlations! http://www.tylervigen.com/spur...

Re:What's a Tufte test?

[ganjadude]

the other thing they forget is the crew working on the rig (i assume, having not RTFA)

move in more people, there will be more hospital visits

Re:What's a Tufte test?

[ganjadude]

one causation is working on a fracking rig. its probably dangerous, and there are probably a lot of new people in the area who were not prior. making hospital visits go up when populations, especially those in dangerous jobs go up

Re:What's a Tufte test?

[Anonymous Coward]

Or, the economy in the area improves and now more people can afford to get their health problems looked at. Unless specific billing codes are increasing, this is the second most likely explanation. The most likely explanation is that they did a bunch of tests and didn't correct for multiple testing in their stats.

Re:What's a Tufte test?

[PopeRatzo]

The obvious problem here is that there is almost certainly a correlation between these locations and poorer communities,

That's because rich people don't want fracking operations anywhere near their families.

Re:What's a Tufte test?

[funwithBSD]

Not Vermont's fault.

It is Bernie Sanders throwing the curve off.

Re:What's a Tufte test?

[PopeRatzo]

You mean like the Beverly Hills Oil Field?

Exactly. The oil was found and drilling started in 1895 before their were rich people there. But no fracking on the Beverly Hills Oil Field in 2015, even though 2/3 of the oil underneath will require it.

And that's because the people who live there have the means to prevent it. You will also note that the oil rigs are hidden behind soundproof walls.

Re:What's a Tufte test?

[cavreader]

Extraction sites tend to be located in under developed rural areas where income levels are lower than in highly populated urban centers. Although there are oil pumps scattered throughout some of the wealthy suburbs of LA that are hidden by clever landscaping. They could discover vast oil and gas fields under any large city but the costs to get at such a resource is astronomical. Energy resource extraction has always been dangerous, dirty, expensive, and controversial. But the fact is even those complaining the loudest against exploiting fossil fuels directly benefit from the oil,,gas, and petroleum products produced. The environment also takes a major hit when extracting the rare earth elements needed for building all of our modern day electronics. Everything from computers, cell phones, and cruise missiles. The toxicity emanating from these mining areas is almost as dangerous as a open nuclear waste landfill. The US closed down almost all of the rare earth element mines because the cost of satisfying the EPA requirements made it cheaper just to buy the rare earth elements mined in foreign countries.

Re:What's a Tufte test?

[swillden]

Or, the economy in the area improves and now more people can afford to get their health problems looked at. Unless specific billing codes are increasing, this is the second most likely explanation. The most likely explanation is that they did a bunch of tests and didn't correct for multiple testing in their stats.

Another very likely explanation is that the correlation is cherry-picked. A good way to achieve a study like this is to look for correlations across numerous statistics (e.g. health costs, mortality rate, days of work missed due to illness, etc., etc., it's easy to come up with a dozen proxies for "health"), and if you cast a net wide enough you're statistically guaranteed to find at least one with a correlation that exceeds the standard threshold for statistical significance. The definition of "statistical significance" ensures it. Then you publish that one while discarding the rest.

Moreover, you can achieve this same effect merely by having many research teams tackle aspects of the question. The negative results will go unpublished, or published in obscure journals and receive no mainstream press attention, while the one that "hits" shows up on slashdot, and not even the researchers will believe they've done anything wrong.

Or maybe the correlation is real, in which case we'll eventually find a cause. Time will tell, but it general goes against the sky-is-falling types.

Re:What's a Tufte test?

[gzuckier]

They are trying hard to pull a 'correlation is causation' scam

Except that in this case we have an intervention study, as some areas started fracking activity while other did not. Therefore looking at data versus time will tell us something.

And I also not we have explanations for causation. I see two obvious: chemical leaks, and nocebo effect.

But I can't see where they really looked at data versus time, in the way we need to see; ie. we'd like to see increase in fracking (in any of a number of metrics) associated with increase in hospitalization rate. Ideally it would be a nice continuous function. At minimum, four numbers: baseline hospitalization rate before (b) and after (a) fracking, in places with (1) and without (2) fracking. We'd need to see that (a1-b1)> (a2-b2). I do not see that anywhere in the paper. In fact, they state explicitly that the hospitalization rate within any zip code did not change much over the 5 years, and that the fracking increased greatly. Given that time sequence, it's more accurate to say the hospitalization rate causes the fracking than the other way around.

Before and after

[tompaulco]

Since there are a lot of things that correlate with location of fracking sites, such as lower income, better chance of hurting oneself on drilling equipment, rural areas, it would lend more credence to the study to list if there were also more hospitalizations in those zip codes compared to other zip codes BEFORE fracking started.

Re:Before and after

[Rei]

That's controlled for by the randomness of the counties involved - both changes before and after drilling, and with no-drilling areas in the same region as controls (the control county had a drilling ban because it was in the Delaware River watershed). The admissions were largely not due to accidents - cardiology admissions were the strongest correlated. However, the authors don't identify the particular causative factors. They speculate, for example, that it might be diesel exhaust from all of the work vehicles that could be a causative agent. Another speculation is that the development of the industry has changed the demographics of drilling areas.

We really shouldn't be surprised that living next to industry in general isn't good for one's health, just from these sort of factors alone. Exhaust from heavy work vehicles, noise, dust, etc aren't famously conducive to good health. Even living next to a busy road is correlated with negative health effects.

A real problem with the study is, as they wrote, "Given that our modeling approach cannot account for within zip code demographic changes over the study period,". Curiously, while there were positive correlations between wells and health problems in most fields, there were negative correlations in gynecology and orthopedics. They remark "However, within the medical categories of gynecology and orthopedics, inpatient prevalence rates are expected to decrease each year by around 13–14% and 3–4%, respectively. Despite this surprising result, it is unclear why gynecology and orthopedics inpatient prevalence rates are decreasing each year. It is unlikely that these decreasing rates are related to the increased hydro-fracking activity." I'm surprised that they were allowed to get away with this - you shouldn't be allowed to credit increases to an industrial effect while just dismissing data (quite significant data) that doesn't match your hypothesis. There could be actually very useful information about the validity of their overall study and their conclusions in the reason for why gynecological inpatient cases are declining. For example, perhaps the demographics are changing to a lower percentage of women due to the arrival of the drilling industry. Men have shorter average lifespans and in particular a higher rate of cardiovascular disease.

To me, this is a really big hole in their study, and again I'm surprised it passed peer review with it there. But apart from that, I see no problem with the study, so long as people don't overinterpret the results. It's a very broad, generalized study focused entirely on correlation and not causation.

Re:Before and after

[PopeRatzo]

Since there are a lot of things that correlate with location of fracking sites, such as lower income

Why do you think rich people don't want fracking sites near their homes? I'm sure it's just coincidence and has nothing to do with the fact that they're making people sick.

I am sure that rising rates of health insurance

[Crashmarik]

from people being employed in the areas had nothing to do with more people going to hospitals.

Re:I am sure that rising rates of health insurance

[epine]

from people being employed in the areas had nothing to do with more people going to hospitals

The proper experimental control is to take three regions that went from no fracking to fracking, and three other regions that went from no fracking-like revenue to a fracking-similar amount of fracking-like revenue so as to match the upticks in net employment.

Obviously for natural experiments, this is not always easy to pull off (and your detractors will necessarily claim you didn't succeed no matter how far you go).

So why don't you just cut to the chase and declare that all natural experiments are moist excrement? Is there any standard for a controlled natural experiment you'd actually accept? From the structure of your comment I suspect not, as you never once mention the caliber of controls actually used (which is, for maximal troll-seed efficiency, entirely beneath the notice of those who reject the entire category).

Done right, I view this as a form of agile econometrics. First you see what clears the fence under modest controls, before gold-plating round two.

On the other side, blanket cynicism is a crutch of the anti-progressive mindset.

War is hell...and so is an oil boom.

[Anonymous Coward]

I worked in the Bakken oil/gas fields during the first two years of the boom, having grown up in the area.

It wreaks utter havoc on a community. Huge numbers of strangers move in, heavy equipment swarms all over the countryside, traffic goes from light to gridlock, all social and other services are crushed under the load, and the local economy turns upside down. In short, it's very stressful for pretty much everybody. It would not surprise me at all if a boom in local oil/gas development raises stress related problems (like heart attacks and mental health issues.)

But I don't see any evidence (or rational) supporting an assumption that the fracking portion of this larger whirlpool of human activity and chaos is in of itself the cause. The correlation makes sense for less exotic reasons.

Re:Living near hospital trolls

[ShanghaiBill]

Also, receiving a monthly royalty check from the gas company increases your disposable income, and means you can spend more on things like health care.

Less McDonalds, more hospitalizations?

[mveloso]

Note that during that study period sales at McDonalds across the nation were dropping. We thus can conclude that reduced sales at McDonalds leads to higher number of hospital visits.

Re:Less McDonalds, more hospitalizations?

[rsilvergun]

Um... the reduction in McDonald's is nationwide, while the study makes the point that there are more hospital visits near fracking, which is a localized thing. You're point isn't relevant.

I bet

[penguinoid]

I bet living near hospitals correlates to more hospital visits too.

Re:I bet

[Opportunist]

I'd sure as hell try to steer clear of hospitals. Most people in the developed world die in hospitals!

Effect sizes are microscopic

[Neuronaut137]

The two "significant" effects, for cardiology and neurology, are increases of 0.07% and 0.06%, respectively. Not 7% and 6%, but 0.07% and 0.06%. These are the smallest effect sizes you will ever see published. Effect sizes of that tiny size can easily be explained by decisions on which data to use, how to analyze it, etc. Even if those effects were real, those effect sizes are too small to care about. Nothing to see here. Move along.

.06% "Compelling"

[cluge]

"...the graphical and informational figures flunk every Tufte test, which is unfortunate" -- Says so much about the author of the post. .06% increase in a data set of this size is compelling? It stinks when the data doesn't fit one's preconceived notions.That's one of the the beauties of science and why healthy scepticism is required.

Active fracking?

[Anonymous Coward]

Active fracking is an activity measured in days. Most of the life of a well is simply pumping...

Both apply

[SuperKendall]

1) An incredibly small population.

Which describes all fracking sites exactly - they are in remote communities.

2) An incredibly dangerous working environment.

No, it does not have to be "incredibly" dangerous. Just MORE dangerous than work in the surrounding area, which if you are in some remote mostly bedroom or farming community is absolutely going to be true for any complex mechanical complex which has frequent shipping and operation. "Hospital visits" is extremely vague and can include something like a small cut which most people would just patch up but which a company has to send to a doctor for examination for legal reasons.

Bullshit

[Karmashock]

http://journals.plos.org/ploso...

Look at their graph and compare the first year with the last year. They're so f'ing similar.

Beyond that, consider they're not showing you how many illnesses anyone had... just hospital visits. Thus they could be going to the hospitals because idiots in the media scared them and it is causing a clearly very small uptick in hypochondria.

If I had a super power... it would be to urinate in the faces of people that push this shit.

Please contradict me. I would love to be wrong. I really would be... No really. this garbage depresses me with how dumb it is and if I just made stupid mistakes then that would be on me. I'd much prefer that. No really.

Until that happens... I'm going to be exposing myself to gamma rays and letting odd radioactive insects bite me on the off chance that I'll get the super power the world both needs and deserves.

Re:Bullshit

[KGIII]

You might actually want to look into DDT and the scientific studies that have been done since the faulty studies done in the '60s and '70s. There is a reason that the WHO has brought back the use of DDT. Not using DDT resulted in more deaths than using DDT ever could. Then again, deaths from malaria are a third world problem and not glamorous so few people actually pay attention to it. Oddly the Bill and Melinda Gates Foundation seems to think it is a priority and, even stranger, they seem to think Viagra is handy for medically treating those who have contracted malaria but I digress.

their data says admissions did NOT increase

[raymorris]

The study authors say there was no fracking in 2007, and lots in 2011.

They then say (quoting):
"The inpatient rates are relatively stable from 2007â"2011 Indeed, the average overall inpatient prevalence rates for 2007â"2011 are, respectively, 15.18, 15.30, 14.86, 14.00, 14.25"

So the introduction of fracking did NOT increase hospital admissions. Indeed, over the four year that the economy in the area got a boost from fracking, people got healthier, according to the numbers in the study.

Then then do a bunch of gymnastics to discover, then obscure, the fact that oil wells tend to be located in more rural areas, and people's health tends to be slightly worse in those areas.

Re:Living near hospital trolls

[gzuckier]

Also, receiving a monthly royalty check from the gas company increases your disposable income, and means you can spend more on things like health care.

Similarly, as they said, fracking areas may have more temporary residents employed doing the fracking, which would raise the hospital admissions but not the denominator they used, which was the number of PERMANENT residents, thereby raising the admission rate.