Search This Blog

Monday, June 4, 2012

Mangano Fallacies

This post is a carry-over from the comment section associated with an op-ed on the blue fin tuna from the Forbes' website.

One commenter had posted that all the criticisms of Mangano had come from the nuclear industry.  That itself is a logical fallacy.  It's okay to be skeptical of what someone from the industry states, but unless you have evidence that the person has lied many times before (as in you KNOW she/he did, not you think so), then that person's claims should be fairly weighed against Mangano's.  Also the "industry" is composed of tens of thousands, if not hundreds of thousands of people.  In any large group of people there will be some bad folks and some good folks.  One shouldn't stereotype.

In any event, here is a link from the non-nuclear industry publication Scientific American properly criticizing his most recent "work".


When contacted by MedPage, which also isn't a part of the nuclear industry, he said:

"There are probably a variety of factors that could be linked to this excess of 14,000 deaths," he said. "But it does raise a red flag."

Yes, there probably are a variety of factors.  And no it doesn't raise a red flag, because spikes come and go regardless of nuclear accidents.  There is no known underlying biological mechanism  for very low doses of radiation to cause infant deaths.  This is analogous to the old days of Salem, MA when someone would give someone else the "evil eye", and then when something bad eventually happened to the person who received the "eye" (became ill, house caught fire, slipped and fell, etc.), the person who gave the "evil eye" was labeled as a witch.  That's of course nonsense...so is the connection between a nuclear accident and an eventual spike in infant mortality (low birth weights, accidents, SIDS, congenital malformations, etc.).

So, if we look for example at an early paragraph in that particular paper, we find:


"Shortly after the report was issued, officials from British Columbia, Canada,
proximate to the northwestern United States, announced that 21 residents had
died of sudden infant death syndrome (SIDS) in the first half of 2011, compared
with 16 SIDS deaths in all of the prior year. Moreover, the number of deaths
from SIDS rose from 1 to 10 in the months of March, April, May, and June 2011,
after Fukushima fallout arrived, compared with the same period in 2010 (2).
While officials could not offer any explanation for the abrupt increase, it coincides
with our findings in the Pacific Northwest."

Okay...how does radioactivity cause SIDS?  It's the "evil eye" causes illness effect.

At least there was a bit of honesty with the next sentence:

"Any comparison of potential effects of radiation exposure must attempt to
examine the dose-response relationship of the exposure of a population."

MUST, MUST, MUST....YES, YES, YES, but did they do that?  NO, NO, NO.

I could go on and on with details with this particular "study" but on a broader perspective related to all of his "studies"...


What a good epidemiologist does, is do epidemiology, not just wait for a nuclear accident, and then wait for a spike and tie the two together.  Mangano needs to look at the actual causes of deaths, estimate the radiation doses actually received by the infants from the accident, identify possible confounders, etc.  Instead he intentionally engages in the ecological fallacy, and intentionally avoids the fundamental Bradford-Hill criteria, which every epidemiologist is familiar with.

But even if there was some radiation phenomenon no one knows about, we can do a quick reality check.  Let's look at U.S. natural radioactivity levels and look at infant mortality rates.  If radiation is a strong factor in infant mortality, where the exposure is continuous, it's fairly uniform, and we can look at many data sets what do we get?  I'll just focus on what CDC calls the Mountain Region (high natural radiation) and the West Central Region (low natural radiation).  Let's just look at the last report for now, you can pick any dates you want..and we get the Mountain Region at 59 and the W.C. Region at 66.  Now that's anecdotal and also involves the ecological fallacy, but if radiation was a strong factor we expect the opposite by a lot!

Note that that study was published in The International Journal of Health Services.  That is not a health physics, epidemiology, or medical scientific journal.  That journal was chosen for a reason.   The reason is  that to the extent any peer review is done, it won't be done by people with strong epidemiology backgrounds.  Health services deals with legal, political and management issues in health care.  So rather than share this "important finding" with other experts in the field, he intentionally avoids them.

But once published somewhere, he can broadcast press releases and use phrases like "peer reviewed" and "health journal", which is really what he wants in order to fool people into thinking he's surpassed some kind of intellectual threshold, when he actually avoided it.  It's all about garnering public attention to his anti-nuclear agenda., not science.

Which is why when one looks at Mangano's website, it's full of links to coverage by the general media.

But notice there are no links associated with any of his "studies"to the American Pediatric Association, the American Medical Association, the Health Physics Society, the American Epidemiological Society,  or even the Centers For Disease Control & Prevention (from which the data from his last "work" originated)!

Is there a grand conspiracy against Mangano by this diverse array of employed professionals or is Mangano (who is seeking YOUR DONATION)  a con man?

(If you can't answer in favor of the latter by now, you might want to review the psychological phenomena known collectively as cognitive bias.)

8 comments:

  1. Well, Bob perhaps some day the mystery will be discovered as to why there are spikes in infant deaths in certain areas so soon after radiation leaks and accidents.

    I found this article from Counterpunch interesting, including what their statistician revealed: "This geographic concentration of radiation effects in the northernmost Pacific coast area is, in a sense, to be expected

    because radiation plumes, like pollution plumes and ash plumes, do not disperse uniformly with distance, contrary to the equations used by all atmospheric computer modelers. To the contrary, actual observations of radiation dispersal after Chernobyl, or volcanic dispersal after any notable eruption, including the recent Icelandic eruptions, always show that the particles disperse in unpredictable and quite concentrated plumes which touch down occasionally and with high concentration at great distances from the source."

    You have to scroll down to read the whole article: http://www.counterpunch.org/2011/06/17/definitely-bachmann-over-weiner/

    Have there been any measures of radioactive particles in infants who died during mortality spikes following nuclear accidents?

    ReplyDelete
  2. lol @ cognitive bias. It seems to be a very common trait among industry professionals. :)

    ReplyDelete
  3. I don't know of any measurements taken on infants after "spikes". The reason folks aren't rushing to do so is because there is no reason to. Infants tend to be well sheltered by their parents, and what little Fukushima radioactivity they might intake wouldn't be expected to cause their deaths so soon.

    The gold standard epidemiological study on radiation health effects is from the atomic bomb survivors. There was no infant mortality spike at low doses.

    There is no "mystery" as to why there are spikes. The science of epidemiology has developed to help minimize fooling ourselves with false patterns and recognizing patterns where they aren't obvious. When we don't follow its discipline, we will see patterns which don't exist. It's like the coin tosser who gets 3 heads in a row, and bets heavily on the next toss being a tail, when in reality the chance of a tail is 50%, just like it was in the previous throws. Mangano know this stuff and that's why he doesn't employ sound epidemiology.

    The article you mentioned was able to show no statistical significance (until they themselves cherry-picked the data).

    Regarding the non-uniform distribution of plumes, that is correct. That is why one can't use AVERAGE concentrations in epidemiology (or average effects like infant mortality). One has to estimate the actual dose an individual receives and one has to compare that to a pre-determined health effect (can't add health effects as you go along). Then the results from that group of individuals need to be compared with a similar, unexposed group to see if there are statistically significant differences.

    The late Christopher Hitchens said, "That which explains everything, explains nothing". That applies here. Using radioactivity to explain all the various health effects incorporated into infant mortality, explains nothing.

    ReplyDelete
  4. Cognitive bias is common among people. That's why strict, objective epidemiology is important.

    ReplyDelete
  5. Bob, I noticed that NEI Nuclear News accused Mangano of contradicting himself because he admitted that, as you note above, there could be a variety of factors for the spike in deaths. I don't think he contradicted himself. Nowhere have I seen that he said the spike was caused by radiation. Am I missing something?

    ReplyDelete
    Replies
    1. In a narrow sense he didn't contradict himself because in his paper he didn't specifically state he concluded that only Fukushima radioactivity was responsible.

      However, he didn't list other possible factors in his paper. A good epidemiology study lists potential weaknesses that the author(s) acknowledge. Not only should Mangano have listed the other factors, he should have explored them. And he should have listed other weaknesses like the cherry-picking of data, the ecological fallacy, etc. So in a general sense he has contradicted himself.

      Delete
  6. Bob, what do you think of this report from Stanford? It's from 2005 and shows that low level radiation increases risk of cancer.

    http://news.stanford.edu/news/2005/october26/abrams-102605.html

    Nothing yet has convinced me that infants aren't at risk of death from low levels of radiation and that the spikes after Fukushima and TMI aren't linked. Has any study been done to disprove this?

    ReplyDelete
    Replies
    1. Actually, that's not a report from Stanford. That Stanford article is about a Stanford professor who was part of the National Academy Of Sciences BEIR VII report. I agree with BEIR VII, see my page (look at upper right of screen) "LNT Model & Radiation Studies". And no study has been done to disprove it (BEIR VII, that is).

      BEIR VII is the gold standard for the reasons I describe there. As your article headlines, the significant radiation effect is cancer. Cancer contributes to infant mortality very, very, very, very, very infrequently because it takes a long time for cancer to manifest and result in death.

      Don't confuse infant mortality with cancer. Infant mortality is most associated with congenital malformations, low birth weights, SIDS, and accidents.

      Our understanding of cancer risks has led the ongoing BEIR committees to increase the risk of cancer relative to dose over many decades (the first report was called BEIR I). But the risk is still small. I expect the risk estimates will continue to climb until all the bomb survivors die.

      The BEIR committee is the antithesis of Mangano.

      Delete