Abstract

From 1958 to 1983, the Apollo nuclear facility produced low-enriched uranium, highly-enriched uranium, and thorium fuels. A history of poor regulatory compliance, poor materials accountability, inadequate building construction, absence of monitoring, and several accidents (including a fire that is believed to have released kilogram quantities of highly-enriched uranium into the atmosphere) suggest potential for occupational and environmental radionuclide exposures.

This analysis investigates all-cancer incidence surrounding the Apollo facility from 1990–2010 using indirect standardisation methods. It addresses many of the limitations of prior studies by improving the accuracy of population estimates and assessing the magnitude of bias due to residential misclassification. We also evaluated whether changes in population distribution over time, for example due to ageing and migration of workers after closure of the facility, affected incidence rate ratios. We fit a lognormal Poisson model with random effects to account for person-time heterogeneity of age across calendar periods.

We identified excess cancer incidence in the Apollo community relative to the number of expected cancers based on Pennsylvania statewide rates. Between 1990 and 2010, the observed rate of cancer incidence among males in the Apollo area was 1.56 (95% CI: 1.47–1.65) times the expected statewide incidence and among females the cancer rate was 1.38 (95% CI: 1.29–1.47) times the expected statewide incidence. The excess observed vs. expected rates was reduced by adjustment for prior estimates of geographic misclassification of cases. Differences in incidence ratios by sex could reflect occupational exposures. Estimates from the random effects model did not differ significantly from the standard estimates, indicating that the amount of heterogeneity in incidence ratios by changes population distribution was minimal.