Fukushima Voice has a report of the Tenth Report of Fukushima Prefecture Health Management Survey, which has just been released. These are outcomes of ultrasound examinations of 94,975 children residing in Fukushima prefecture, from April 1, 2012 through January 21, 2013 (fiscal year FYH24). These outcomes included the existence and size measurements of thyroid nodules and cysts in these children. A previous set of examinations of 38,114 children, from the previous fiscal year (FYH23), October 2011 through March 2012, also measured these outcomes. It was found that in the previous fiscal year, 3 children had already undergone surgery for thyroid cancer. 7 more were found to have an 80% chance of having thyroid cancer, based on biopsy results.
The rate of thyroid abnormalities in these children for April 2012 – January 2013 were compared to the rate of abnormalities for October 2011 – March 2012, in order to determine if this rate had significantly increased. Also, the rate of thyroid cancer for FYH23 was compared to a previously established rate of incidence of this disease. Both sets of measurements were made after the onset of the Fukushima nuclear catastrophe.
Fisher’s exact test (FISHER) and optimal discriminant analysis (ODA) were utilized to compare the rates of incidence of thyroid nodules and cysts. These tests were performed by Optimal Data Analysis for Windows (using the author’s modified copy). Binomial probabilities (BINOMIAL) were used to compare thyroid cancer incidence rates. These probabilities were computed by Excel Statistics Spreadsheets Helper. The prior rate of thyroid cancer (1-3 cases per million) was obtained from this link. The upper end of this range (3 cases per million) was used in the proportion comparison.
The significance level adopted for these analyses was P < 0.05. RESULTS
Large nodules were defined as those greater than 5mm in size, and large cysts greater than 20mm in size.
The effect size indicator ESPV (effect strength for predictive value) was used instead of the more familiar ESS (effect strength for sensitivity), due to the very small incidences of large thyroid nodules and cysts relative to the population, which rendered ESS meaningless in the context.
The use of Mid-P probabilities was discussed in a previous post. They were generated in ODA for Windows by using the ADJUST keyword of the MCARLO command. Mid-P overcomes the excess conservatism in Fisher’s exact test (and equivalently, ODA, in this context), and in all discrete statistics in general, avoiding the waste of statistical power.
The use of one-tailed statistics was also discussed in the previous post. The Fukushima Daiichi catastrophe released huge amounts of radioactive iodine-131, which was attested to by the Japanese government, MEXT, Tepco, the CTBTO, and many researchers. Iodine-131 is well established as a cause for thyroid cancer and abnormalities, as attested to by uncountable scientific articles, the ECRR, ICRP, IAEA, which used data from atmospheric nuclear testing, Sellafield, Chernobyl, Three Mile Island, radiotherapy, and many other nuclear accidents, dumping sites, and mishaps. Thus we are only interested if these thyroid abnormalities increased or not, and are not concerned with whether they decreased as a result of this catastrophe. There is no evidence that the release of iodine-131 protects anyone from thyroid cancer or abnormalities. Also, the precautionary principle in epidemiology argues that harm should be assumed, and the onus of proof is on those arguing that there is no risk from an environmentally harmful action. Therefore, Mid-P one-tailed statistics are both correct and ethically required in this case. Two-tailed non-Mid-P statistics have been included in the above table also, since they are overwhelmingly used in scientific literature, even though in this case it is both incorrect and unethical to use them.
It can be seen from the table that the rate of incidence of large nodules is significantly increased over the prior time period. The rate of large cysts has increased also, but the small number of children with these cysts has resulted in lack of significance, as evidenced by the higher ESPV. The rate of cysts of all sizes has significantly increased in a powerful way.
Since it is possible for a child to have both a nodule and a cyst, rates for having either of these conditions were included. Large nodules or cysts were significantly increased, as well the number of children with any nodule or cyst.
The thyroid cancer rate from FYH23 was also seen to significantly increase in a powerful way, relative to the rate from before the Fukushima catastrophe. This is true for both the small (3 cases) and the large (10 cases) estimates.
The social and ecological importance of these findings is obtained from the effect size, not P-values. Though the values of ESPV might seem low relative to an ordinary experiment, in this context with a large sample size, an extremely vulnerable population, and the likelihood that disease will increase over many years, the medical and ecological significance of these findings is huge.
The rates of incidence of thyroid cysts and large thyroid nodules in Fukushima children has been shown to be significantly increasing since the Fukushima nuclear catastrophe. The rate of thyroid cancer has also increased since the disaster. The social, medical and ecological significance of these results is unquestionable. Everything that is possible should be done to treat these children, and to prevent more thyroid disease from arising in Fukushima, in Japan, and all across the world.