DA von R. Zwissler, 2002

Titel

Statistical Prediction of Chernobyl Childhood Thyroid Cancers from Latest Observations

(DA: D-MAVT)

Betreuung

Prof. Thomas (City University London) / Prof. Dr. W. Kröger / Dr. R. Mock

Abgabe

März 2002

Abstract

The latest datasets on childhood thyroid cancer cases have been analysed for the most affected cohort, the young (up to 17 years old), in the regions close to the scene of the Chernobyl accident in 1986, Belarus, the Ukraine and Russia. It has been assumed that the excess number of cancer cases have been caused by iodine-131 from Chernobyl fallout. The lognormal distribution has been confirmed to match well the data available for the region of Belarus. The resultant values for the mean latency period and standard ratio have been applied to datasets of cases in the Ukraine and Russia. A best prediction for the three regions has been worked out by minimization of the chi-squared sum at 4405 cases over all years with an upper bound of 7594 cases. Analysis of age dependent data could corroborate these predictions.

Since the cure rate for surgery and subsequent continuing thyroxin supplement treatment is expected to be better than 90%. If no further long-term, adverse health effects are found, the total number of deaths from childhood thyroid cancer would be expected to be limited to hundreds.

Cases on vCJD in the UK, 104 victims up to the end of 2001, have been analysed assuming these cases to be proportional to the collective dose of the BSE agent incurred as a result of eating beef. Human-to-human transmission of the infectious agent has not been considered.

Several parameter variations and source term types have been investigated.

A best prediction has been worked out, applying a normal distribution, at 149 cases over all years with an upper bound of 1007 cases.

Sensitivity of both the CTC and the vCJD predictions to their input vector have been investigated. It could be seen that the predictions, even though the currently best possible, have a range pertaining to a 95%-confidence that can spread widely putting only a few observations into the analysis.