Assessment of uncertainty associated with measuring exposure to radon and decay products in the French uranium miners cohort

Abstract

The reliability of exposure data directly affects the reliability of the risk estimates derived from epidemiological studies. Measurement uncertainty must be known and understood before it can be corrected. The literature on occupational exposure to radon ((222)Rn) and its decay products reveals only a few epidemiological studies in which uncertainty has been accounted for explicitly. This work examined the sources, nature, distribution and magnitude of uncertainty of the exposure of French uranium miners to radon ((222)Rn) and its decay products. We estimated the total size of uncertainty for this exposure with the root sum square (RSS) method, which may be an alternative when repeated measures are not available. As a result, we identified six main sources of uncertainty. The total size of the uncertainty decreased from about 47% in the period 1956-1974 to 10% after 1982, illustrating the improvement in the radiological monitoring system over time.

Impact of measurement error in radon exposure on the estimated excess relative risk of lung cancer death in a simulated study based on the French Uranium Miners’ Cohort

“…Due to changes in the methods of measurement, the characteristics of ME for radon exposure in the FUMC varied over time. Previous work (Allodji et al 2012) determined the most plausible characteristics of ME in the FUMC for radon exposure for the years 1956-1999, based on literature about uranium mines, annual reports of radiological monitoring of French uranium mines, as well as on information from interviews with experts specialized in radiation monitoring (from the company Alpha Gamma Delta, called Algade) and with those responsible for uranium exploitation in France (from AREVA NC). Annual individual measurements were not available for the first period (1946)(1947)(1948)(1949)(1950)(1951)(1952)(1953)(1954)(1955).…”

“…The type of radon exposure in the FUMC varied over time from Berkson to classical, reflecting changes in the process of radon exposure measurement over time in the French uranium mines. Indeed, ME associated with radon exposure retrospectively reconstructed for the years 1946-1955 and ME associated with ambient measurements of radon in mines for the years 1956-1982 were determined as of Berkson type (Allodji et al 2012) since they reflected a group's average assigned to each individual suiting the group's characteristics (Buzas et al 2004;Carroll et al 2006;Muirhead 2008). In contrast, ME associated with individual alpha dosimeters for the years 1983 through 1999 was determined as classical (Allodji et al 2012) since radon exposure was then measured individually.…”

“…Indeed, ME associated with radon exposure retrospectively reconstructed for the years 1946-1955 and ME associated with ambient measurements of radon in mines for the years 1956-1982 were determined as of Berkson type (Allodji et al 2012) since they reflected a group's average assigned to each individual suiting the group's characteristics (Buzas et al 2004;Carroll et al 2006;Muirhead 2008). In contrast, ME associated with individual alpha dosimeters for the years 1983 through 1999 was determined as classical (Allodji et al 2012) since radon exposure was then measured individually. Bender et al (2005) reported that due to the skewed distribution of some exposures (radon in their study), the application of usual additive ME assuming normally distributed ME may be problematic.…”

“…Because radon exposure in the FUMC presented asymmetric yearly distributions tending toward log-normal (LN) distributions, the following LN fluctuation was assumed (Bender et al 2005;Küchenhoff et al 2007;Allodji et al 2012): LNðÀr 2 ME =2; r 2 ME Þ with r 2 ME the ME variance. The FUMC data supported the hypothesis that relative ME size (i.e., the ratio of ME size to exposure level) did not vary with radon exposure level, that is, it was constant across all yearly exposure measurements within prespecified periods (Allodji et al 2012).…”

“…To make the simulation study as realistic as possible, data from the French Uranium Miners' Cohort (FUMC) study guided the design of the present simulation experiments (Vacquier et al 2008(Vacquier et al , 2009. Similarly, the detailed characteristics of ME previously analyzed in this cohort were used to determine the ME to be considered in generating radon observed exposure (Allodji et al 2012). A sensitivity analysis was performed to examine the impact of specific characteristics of ME (type, size, structure, and distribution) on the estimated ERR.…”