A Statistical Method For Testing Epidemiological Results, As Applied to the Hanford Worker Population

Publisher Website
Google Scholar
Abstract
Some recent reports of Mancuso, Stewart and Kneale claim findings of radiation-produced cancer in the Hanford worker population. These claims are based on statistical computations that use small differences in accumulated exposures between groups dying of cancer and groups dying of other causes; actual mortality and longevity were not reported. This paper presents a statistical method for evaluation of actual mortality and longevity longitudinally over time, as applied in a preliminary analysis of the mortality experience of the Hanford worker population. Although available, this method was not utilized in the Mancuso-Stewart-Kneale paper. The author's preliminary longitudinal analysis shows that the gross mortality experience of persons employed at Hanford during 1943–70 interval did not differ significantly from that of certain controls, when both employees and controls were selected from families with two or more offspring and comparisons were matched by age, sex, race and year of entry into employment. This result is consistent with findings reported by Sanders (Sa78). The method utilizes an approximate chi-square (I.D.F.) statistic for testing population subgroup comparisons, as well as the cumulation of chi-squares (I.D.F.) for testing the overall result of a particular type of comparison. The method is available for computer testing of the Hanford mortality data, and could also be adapted to morbidity or other population studies.

This publication has 2 references indexed in Scilit: