Serious doubts re: that report of 14,000 US deaths from Fukushima…

Researchers Trumpet Another Flawed Fukushima Death Study
By Michael Moyer | December 20, 2011

In June I wrote about a claim that babies in the U.S. were dying as a direct result of Fukushima radiation. A close look at the accusation revealed that the data used by the authors to make the argument showed no such thing. “That data is publicly available,” I wrote, “and a check reveals that the authors’ statistical claims are critically flawed—if not deliberate mistruths.” The authors appeared to start from a conclusion—babies are dying because of Fukushima radiation—and work backwards, torturing the data to fit their claims.

Now the authors have published a revised study (PDF) in the International Journal of Health Services. A press release published to herald the article warns, “14,000 U.S. Deaths Tied to Fukushima Fallout.” This is an alarming accusation. Let’s see how the authors defend it.

First, the authors assert: “In the United States, Fukushima fallout arrived just six days after the earthquake, tsunami, and meltdowns.” They provide no evidence for this assertion, no citation to back up their facts. The authors then note that the U.S. Environmental Protection Agency monitored radioactivity in milk, water and air in the weeks and months following the disaster. Ah, here must be the data, the careful reader hopes. Alas, “the number of samples for which the EPA was able to detect measurable concentrations of radioactivity is relatively few,” the authors write. They then conclude, with evident disappointment, that “clearly, the 2011 EPA reports cannot be used with confidence for any comprehensive assessment of temporal trends and spatial patterns of U.S. environmental radiation levels originating in Japan.” In other words, the EPA didn’t find evidence for the plume that our entire argument depends on, so “clearly” we can’t trust the agency’s data.

Read more.

2 replies on “Serious doubts re: that report of 14,000 US deaths from Fukushima…”

Trying to confirm the reported significant increase of infant deaths in the 14 weeks after Fukushima relative to the expected number, I used an approach which allows for overdispersion in the data. A regression analysis of the data in Table 3 of the article yields a 10.5% increase in the observed number of infant deaths over the expected number. The estimated number of excess infant deaths is 259 in weeks 12 through 25 of 2011. This increase, however, is not significant (p=0.101) when a t-test with 52 degrees of freedom is applied. The authors of the study had reported a highly significant p-value of p<0.0002.

The main reason for this increase is that the data before and after Fukushima differ: After Fukushima, the authors included 119 cities in their analysis, before Fukushima only 104 cities. The excess infant deaths come from the 15 additional cities. A trend analysis of weekly infant deaths with the official CDC data from week 50, 2009, to week 25, 2011, yields no increase but a 1.3% decrease of infant deaths after Fukushima.

Yesterday I wrote a letter to the editor of journal theat published the stdy (IJHS) . I wonder whether it will be published since IJHS normally does not accept letters to the editor for publication.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.