On epidemiological studies of high background radiation areas

Until recently, the epidemiological studies of high background radiation areas (HBRA) have 'received relatively little attention', and sometimes even ignored, e.g. UNSCEAR omitted the studies in Kerala from a table of epidemiological studies of radiation and cancer. Authoritative reports on the effects from ionizing radiation such as those from the National Research Council, UNSCEAR, and ICRP instead rely on other sources of information - most notably the A-bomb survivor cohort, medically irradiated cohorts, cohorts of miners exposed to high levels of radon, and the residential radon case-control studies. No doubt this is because HBRA studies typically have had relatively large potential for bias and confounding, low statistical power, and limitations relating to dosimetry. However, great efforts have been made recently to improve HBRA studies, particularly in Kerala and the Yuangjiang area of China. It is anticipated that some of these studies may be able to provide important direct information on health effects from chronic low-dose radiation exposures, a primary concern for radiation protection. In fact, the width of 90% confidence intervals for ERR/Gy reported for incidence from the Kerala study (-0.58, 0.46), and mortality in Yangjiang (-0.67, 0.69), is already comparable to that for some prominent studies of low-dose rate health effects such as of incidence in the Techa River Cohort (0.3, 1.9) and of mortality in the 15-Country Nuclear Workers Study (0.27, 1.80). However, it should be stressed that these intervals do not account for non-sampling sources of uncertainty such as errors in dosimetry and residual confounding. This presentation will focus on ways to further improve HBRA studies around the world, with particular emphasis on the ones in India and China. This will include ways to increase statistical power - such as ways to combine information from different HBRA studies, and possible extensions of the studies such as those designed to better investigate effects relating to leukemia and childhood cancers. Also to be discussed are what needs to be done with respect to potential confounders such as bidi smoking, and ways for reducing and characterizing errors in dose estimates. The latter includes methods that incorporate results from cytogenetic studies. Guidance will also be given on how results from HBRA studies should be interpreted, such as how one might account for differences in the organ-specific cancer rates in the HBRA as compared to the A-bomb survivor cohort. Also to be considered are possible refinements to power calculations for determining when it might be possible to draw conclusions from the HBRA studies about low-dose rate effects. (author)