German Study finds Nuclear Energy Causes Leukemia… or maybe not…
January 13th, 2008
|
| Share |
This is something which all the anit-nuclear anti-science pages out there have jumped on big time. Apparently a German Study which was released not long ago found “statistically significant” increases in cases of childhood cancer, especially childhood leukemia amongst the population which lives within five kilometers of nuclear power plants in that country. Not surprisingly, this study has been cited in articles with headlines like “Nuclear radiation may increase risk of child leukemia” or simply “Nuclear Power Causes Leukemia.” My favorite headline however is “Childhood leukemia in the vicinity of the Geesthacht nuclear establishments near Hamburg, Germany.” Not surprisingly, there’s also allegations of a government conspiracy to keep this quite.
There have been other studies like this done by the Germans and they have often found increased rates around nuclear power plants.
Well lets get something straight here. The study did not conclude that the nuclear power plants were the reason for the apparent increase in leukemia rates, and in fact the researchers went out of their way to stress that they could find any causal relationship with the nuclear power plants.
Actually there are plenty of good reasons to think that the nuclear plants did not have the first thing to do with the “clusters” of cancer which were reported amongst these:
- There was no detectable increase in radiation in the areas.
- No direct relation to other radiation-related health problems was found.
- None of the plants had any unplanned discharges or accidents which would cause containment concerns
- Other larger studies have been done which have established no connection between cancer rates and proximity to nuclear reactors.
So why on earth would they find these clusters?
Well, there are a couple of things that are important to consider here. First, the actual number of cancer patients which makes up one of these “clusters.” It turns out the number of childhood leukemia cases found around nuclear plants in Germany in the most recent study was 37, while the statistically expected rate in the general population would be 17. So although this qualifies as “statistically significant” only a relatively small number of cases are required to tip the balance. Thus, the data is not of the highest quality.
But there is something much more important to notice from this: The study compared those living within five kilometers of nuclear plants to the general population. What it did not do was compare them to a demographically similar control group. Why does this matter? Because those who live near a nuclear power plant are not a representative sample of the German population. This is pretty easy to see as soon as you look at the locations of nuclear plants in Germany.
Germany has a population which is distributed in both urban and non-urban areas, but power plants are not generally located in major cities like Hamburg or Munich. They may be located close to cities or major population centers, but never directly in such large cities. The reasons are obvious: land is expensive in cities and nobody wants to live next to be big industrial complex like a power plant. They are however, not located in the middle of nowhere either. Rather these power plants tend to be located in what one might call “industrial hamlets.”
These are the working class towns which were once the industrial muscle of Germany. They’re often industrial ports and tend to contain a lot of factories, warehouses, refineries and alike. They’re not the richest parts of the country in general. To compare them to elsewhere. In the United States, think of a combination between Allentown Pennsylvania and Jersey City. In Brittan, think of some small industrial port city like Liverpool, but a bit smaller.
It’s no surprise that they are located in such areas. This is where the power is needed. This is where the transmissions lines are. And most importantly, these are the industrial areas where stuff like power plants and other big industrial facilities tend to be built. So if you live near a nuclear power plant in Germany, it basically means you live on the outskirts of a rather blue-color community. You live in an industrial area amongst factories and warehouses and within sight of some big steaming cooling towers. Hence, you probably do not come from the most well-off family.
So what might be the actual cause?
There’s little doubt that economics and the avaliablity of quality health care is going to have a huge effect on all manner of health, especially in children. Additionally, there are enviornmental factors to consider. Living in an area with heavy industry is likely to have lower air quality, less open space and exercise and other activities. This has been observed many times and is entirely logical.
These small “clusters” have been observed many times before in other areas which are not located near nuclear plants. This is one reason for the scare over power lines which occurred after studies found modest increases in some conditions around high voltage transmission lines, but did not take into account that the location would skew the sampling. Despite lack of an established connection, this was a major scare back in the late 1980’s and early 1990’s.
In addition to the possibility that those who live very close to such plants and the industrial areas which surround them might be of a lower income and social class in general, one should take a closer look at the history of some of these sites. One area which was claimed to have the largest “cluster” turns out to be the Geesthacht area, which is the location of a major nuclear power plant. However, what they might not mention is that Geesthacht is also the location where Alfred Nobel built one of his first German explosives factories, establishing it as the heart of the German chemical industry at the turn of the century. It turns out that the power plant is located less than a mile from both the old chemical factory and from a major materials testing and development laboratory.
Indeed, a 2002 flood of the area around Geesthancht raised major concerns about stirring up contaminated sediments from past contamination of the area. Much of the chemical industry was targeted by the Allies in the Second World War, and before that it had been the location of synthesis of everything from paint thinner to mustard gas.
One thing to bare in mind is that I’m not offering a conclusion here, but only a hypothesis which I think is entirely reasonable. Given the poor demographic controls of these studies, which index the cancer rate only to the general population average and the lack of any established connection to the plants (despite a lot of effort to find out) and the existing studies, this seems like a much more reasonable explanation for the modest increase observed. Cancer rates have already been conclusively linked to income and the availability of good nutrition and health care and it’s well known that many chemicals, including benzene, a common industrial solvent have been linked to increased cancer risk, including leukemia.
But unfortunately, this has hit many anti-nuclear pages and has even been showing up in the general media, where it is presented with limited context. It is one example of how scientific credibility is sometimes claimed by a movement which really has about zero.
This entry was posted on Sunday, January 13th, 2008 at 7:35 pm and is filed under Bad Science, Enviornment, History, Nuclear, Politics. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.
View blog reactions
January 13th, 2008 at 8:27 pm
These results are not significant. Studies of an association between childhood leukemia and the proximity of reactors have been inconclusive.
One 1991 study of 107 counties near 62 nuclear facilities by the American National Cancer Institute found that the childhood leukemia rate for the reported areas dropped slightly after the reactors started operating. Another report of the same study indicated that one childhood Leukemia cluster was associated with the Millstone Power Plant located in New London, Connecticut. Three of the studied facilities had significantly fewer leukemia cases than were expected. Windham County, Vermont, where the Vermont Yankee reactor is located was reported to have only 9% of expected childhood leukemia cases.
Repeated studies failed to uncover an association between The Three Mile Island accident and childhood leukemia. Even more remarkable, studies have failed to uncover a relationship between exposure to radiation from Chernobyl and childhood leukemia.
Two previous German studies showing the same thing were shown to be invalid, so of course they are going to try again. Nothing coming out of that coal controlled government can be trusted, they are out to destroy the nuclear sector in that country and they will stoop at nothing to get the job done.
Quote Comment
January 13th, 2008 at 11:33 pm
Yeah, well according to this particular report they are “statistically significant” but that’s a rather deceptive term out of context. When I posted before about New London County Connecticut it was reported to be “Significant” because the levels were more for a certain study by slightly more than one standard deviation. Given that there are six counties in Connecticut, it’s hard to make that kind of statistical judgment.
Yes, the millstone study I’ve looked at before. I see no reason to think the “cluster” was due to the reactor since the geographic distribution was no way an indicator of this and the “cluster” was very small. I’m just trying to make a point that “higher than average” around a nuclear plant hardly means much.
If you look at Windham County for one thing, it’s rather sparsely populated, so a small change can throw off the numbers. Anyway with something like childhood cancer you can find some kind of “cluster” if you look at enough numbers.
I simply see no reason why “higher than the national average” means anything in this context.
There’s every reason to think that it’s not.
Quote Comment
January 14th, 2008 at 12:17 am
The apparent connection is weak. The study is not fully controlled and is indexed to the population in general. There is no apparent mechanism.
This all sounds suspiciously like the leukemia/powerlines bull or the cell phone towers/cancer or other BS studies where they just survey something over and over until something shows up in the numbers.
I think you’re right about one thing: if there is any connection, it’s probably just demographics. And to respond to DV82XL, I highly doubt that the power plants prevent cancer. I know we could go into the whole low-level dose thing, but they’re actually lower in radiation than plenty of stuff people are exposed to from living at a high altitude or near a uranium deposit. I think it just goes to show that these numbers fluxuate and that the enviornmental factors are numerous and chaotic.
There is no relationship and these idiots need to get over it.
Quote Comment
January 14th, 2008 at 1:01 am
I wonder if they plan on doing any similar studies around coal plants. There’s really no point though. I think it’s well established that coal burning has numerous detrimental health effects on those in the area so I’d be surprised if it didn’t contribute to childhood cancer. But doing yet another study on coal would never get anyone’s attention. It’s ridiculous. They sit around debating whether maybe possibly nuclear could effect the health of those in the area while ignoring coal which we KNOW is killing thousands by respiratory ailments and other factors.
Quote Comment
January 14th, 2008 at 3:43 am
You can bet that there will be no coal plant studies done in Germany Thinker, however there have been plenty done in the States that talked of thousands of deaths, not tens.
Dave, I’m not suggesting that living near a nuclear plant is good or bad, I was just pointing out that when you are dealing with such a small number of ‘victims’ any real trends get lost in the error bars. The statistical noise from any number of confounding variables render a study like the German one useless for anything except propaganda.
And for the record: I do not buy into the Radiation Hormesis Theory uncritically, I only brought it up (and will bring it up) as a response to the Linear-Non-threshold Dose-Response Model for the purpose of showing the latter’s lack of epidemiological foundation.
I was never suggesting we all start taking radium pills….
Quote Comment
January 14th, 2008 at 4:21 am
I think that DV82XL is right on the money. I don’t care whether the results are declared “statistically significant” (especially since it is only a risk factor of just over 2, which was calculated using a flawed control group as DrBuzz0 points out), these results are not really significant.
Let’s look at what we’re talking about here. The results of this study are based on only 20 extra cases of leukemia, or about 1.25 cases per plant over 23 years! That is the source of the brouhaha in the media. With numbers as small as these, this could be well within the statistical noise.
By the way, this is a good explanation of the potential contributing factors around German nuclear plants. Lots of good stuff there.
Quote Comment
January 14th, 2008 at 5:37 am
“There are three kinds of lies: lies, damned lies and statistics” -Mark Twain
Quote Comment
January 14th, 2008 at 10:31 am
“Statistically significant” is a deceptive term when dealing with these kind of numbers. You can’t really apply the same kind of statistical analysis when you are dealing with such small quantities. As it is stated, only a few are needed to throw off the entire study. I think this is a circumstance where you have to take into account how small the numbers are to begin with. Also, the fact that it’s compared to a demographically flawed control completely negates it anyway.
Quote Comment
January 15th, 2008 at 3:22 pm
My own reviews of Reactor-cansor research, shows that a the reactors cause leukemia hypothesis has failed so far See:
http://nucleargreen.blogspot.com/2007/12/morning-thoughts.html
http://nucleargreen.blogspot.com/2007/12/out-of-pile-excursions-by-transuranium.html
I consulted “OUT-OF-PILE STUDIES OF FISSION-PRODUCT RELEASE FROM OVERHEATED REACTOR FUELS AT ORNL, 1955 -1965. G. W. Parker, C.J. Barton, G.E. Creek, W. J. Martin and R. A. Lorenz,” (C.J. Barton is my father.) These findings indicate that something very nasty and destructive has to occur inside a light water reactor if radioisotopes are to escape from the core. There is no evidence of reactor damage in the case of the Kruemmel reactor. One hypothesis was that the Elbmarsch cluster was caused by releases of tritium. A study was made comparing the Elbmarsch area with the area of the sevannah River reactors where known releases of tritium had taken place. No evidence of a leukemia cluster was found in the Savannah River area was found. Further there were no extra Leukemia cases associated with either Three Mile Island or Chernobyl, cases where there were large scale fission product releases from reactors.
Several commentors have already noted the negative findings associated wityh a past attempt to identify leukemia clusters in associations with 62 American reactors. The weight of ebidence supports the following conclusions:
1. Releases of fission products, including massive releases in the event of a major reactor accident, cannot be shown to cayse Leukemia clusters.
2. Repeated releases of radioactive gasses like tritium cannot be shown to cause Leukemia clisters.
3. Living in the vacinity of reactors has not been shown to cause Leukemia clusters.
4. To date no cause of the Elbmarsch cluster has been found.
5. No evidence of a large scale fission product release from the Kruemmel reactor has been found.
6. No evidence has emerged that individuals in the Elbmarsch Leukemia clusters were exposed to fission products.
What we have left then is a case of German anti-nuclear paranoia.
Quote Comment
January 16th, 2008 at 11:21 am
I don’t think anything here is “statistically significant”; they would have quoted a p-value if it were. And anyway, this study tries to find a correlation (which will then be presented as causation, of course) between risk of leukaemia and proximity to a NPP. So why didn’t they simply plot number of leukaemia case against distance to the nearest NPP, fitted a curve and reported the correlation coefficient?
Because it doesn’t work. With so little data, the curve jumps around wildly, you can fit everything and the correlation will always be piss-poor, simply because everything is swallowed by random noise. So instead they opted to select an arbitrary cutoff of 5km. Why not 2 or 20 or 30? Wouldn’t it make sense to check and report whether there’s an effect at different cutoffs? Of course it would, so why did the study omit this?
Again because it doesn’t work. There very probably is no effect at different cutoffs, and even though the study claims they selected those 5km before collecting data, I bet they actually plotted the aforementioned curve and then simply set the cutoff just right of the highest peak observed. Which also means it isn’t significant, even if the (unreported) p-value for this single observation is below 1/20.
BTW, the Elbmarsch leukaemia cluster appears to be real (that alone accounts for 9 or so excessive leukaemia cases near a NPP), but it is pretty much clear that the cause wasn’t the Krümmel NPP, but maybe the GKSS (nuclear marine propulsion research facility) nearby. One study reported the finding of radioactive particles, but since they contained thorium they cannot have come from Krümmel, where no thorium was ever used.
Quote Comment
September 16th, 2009 at 10:04 am
This could be another reason
“Increased deposition of polychlorinated biphenyls (PCBs) under an AC high voltage power line “
Seems to be “in press” at the journal Atmospheric Environment so there is no volume information
http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6VH3-4X6FNRV-1&_user=952952&_coverDate=09%2F10%2F2009&_rdoc=14&_fmt=high&_orig=browse&_srch=doc-info%28%23toc%236055%239999%23999999999%2399999%23FLA%23display%23Articles%29&_cdi=6055&_sort=d&_docanchor=&view=c&_ct=145&_acct=C000049223&_version=1&_urlVersion=0&_userid=952952&md5=2456480c3c1da0cbc557f63696a794db
maybe there could be increased deposition of other chemicals than PCBs?
Quote Comment
September 16th, 2009 at 10:45 am
Mark said:
In the abstract they say they found the average deposition of PCBs under the power line was almost double however they do not give values. Without this, and without some data on exposures at these levels, nothing can be said. The doubling of background from 0.001 ppb to 0.002 ppb may be said to be statistically significant it may well be toxicologically meaningless.
At any rate there is no evidence of problems to begin with, so I have issues with a study that starts off saying: “epidemiological findings are not supported by a clearly defined mechanism,” when there are no meaningful findings anyway.
Quote Comment