A Bayesian hierarchical approach to account for left-censored and missing radiation doses prone to classical measurement error when analyzing lung cancer mortality due to γ-ray exposure in the French cohort of uranium miners.

Epidemiological data on cohorts of occupationally exposed uranium miners are currently used to assess health risks associated with chronic exposure to low doses of ionizing radiation. Nevertheless, exposure uncertainty is ubiquitous and questions the validity of statistical inference in these cohorts. This paper highlights the flexibility and relevance of the Bayesian hierarchical approach to account for both missing and left-censored (i.e. only known to be lower than a fixed detection limit) radiation doses that are prone to measurement error, when estimating radiation-related risks. Up to the authors' knowledge, this is the first time these three sources of uncertainty are dealt with simultaneously in radiation epidemiology. To illustrate the issue, this paper focuses on the specific problem of accounting for these three sources of uncertainty when estimating the association between occupational exposure to low levels of γ-radiation and lung cancer mortality in the post-55 sub-cohort of French uranium miners. The impact of these three sources of dose uncertainty is of marginal importance when estimating the risk of death by lung cancer among French uranium miners. The corrected excess hazard ratio (EHR) is 0.81 per 100 mSv (95% credible interval: [0.28; 1.75]). Interestingly, even if the 95% credible interval of the corrected EHR is wider than the uncorrected one, a statistically significant positive association remains between γ-ray exposure and the risk of death by lung cancer, after accounting for dose uncertainty. Sensitivity analyses show that the results obtained are robust to different assumptions. Because of its flexible and modular nature, the Bayesian hierarchical models proposed in this work could be easily extended to account for high proportions of missing and left-censored dose values or exposure data, prone to more complex patterns of measurement error.

Mortality analyses in the updated French cohort of uranium miners (1946-2007).

PURPOSE: The objectives are to analyze mortality risks in the extended follow-up of the French uranium miners' cohort and to examine their potential relation to occupational exposure to ionizing radiation (IR). METHODS: The total cohort includes 5,086 uranium miners employed in the CEA-COGEMA group and followed up from 1946 to 2007. Vital status, causes of death, and cumulative radon exposures were recorded. The post-55 subcohort includes 3,377 miners first employed after 1955, for whom long-lived radionuclides (LLR) and external gamma-ray exposure were also recorded. External mortality analyses were performed by computing standardized mortality ratios (SMR). Excess relative risks (ERRs) due to IR exposures were estimated from Poisson regression models. RESULTS: The miners included in the total cohort were followed up for 35.4 years and exposed to 36.6 working level months (WLM) on average. There was no evidence of a difference in overall mortality between miners and the general French male population. Miners had a statistically significant excess mortality rate from lung cancer (SMR = 1.34 [95% CI 1.16-1.53]) and from kidney cancer (SMR = 1.60 [1.03-2.39]). Cumulative radon exposure was significantly associated with lung cancer risk (ERR/100 WLM = 0.71 [0.31-1.30]) and cerebrovascular risk (ERR/100 WLM = 0.41 [0.04-1.03]). In the post-55 subcohort, this excess mortality from lung cancer remained associated with exposure to radon, and also with exposure to LLR and external gamma rays. CONCLUSIONS: The analyses in the extended follow-up strengthen the results previously observed among French uranium miners about their excess risk of mortality and its association with their occupational IR exposure.

A Bayesian hierarchical analysis to compare classical and atypical scrapie surveillance data; Wales 2002-2006.

We describe the application of Bayesian hierarchical models (BHM) to the analysis of risk of sheep scrapie using data from multiple surveillance sources. More specifically, we analysed data from the test results of three surveillance sources on classical and atypical scrapie in Wales for the period 2002-2006. For each form of scrapie, a BHM was fitted to assess the occurrence of spatial patterns of risk shared by the multiple surveillance sources and the association between covariates and disease. We defined a shared-component model whereby the two types of data sources: exhaustive lists (e.g. reports of clinical cases) and sample-based data sources (e.g. abattoir survey) shared a common spatial pattern of risks at parish level. This shared component was adjusted by a risk-gradient parameter that moderated the individual contribution of the datasets. For both forms of scrapie, the risk-gradient was not significantly different indicating that the sensitivity of the two types of dataset was similar for the two diseases. The spatial patterns of the combinations of data sources appeared similar within disease. However, our results suggest that classical and atypical scrapie differ in their spatial patterns and disease determinants. The joint approach permitted inference from all the available evidence and resulted in robust and less biased estimates of risk, particularly for atypical scrapie where the number of observations was very limited.