Previous studies have found that mean-residential-radon (Rn) levels for U.S. counties are negatively associated with age-adjusted county rates of lung-cancer mortality (LCM), after adjustments for potentially confounding factors. Those results may be due to (a) confounding unaddressable by any county-level (‘ecological’) study design, or (b) county-level factors such as Rn/smoking or age/radon correlations or exposure misclassifications from the use of disparate data sources. Possibilities (b) were addressed by comparing age-specific LCM rates for white women in 2821 U.S. counties who died in 1950–54 at age 40+ (∼11% of whom ever smoked), or at age 60+ (∼% of whom ever smoked), to county Rn levels newly estimated from U.S. Rn, climatic and geological-survey data. Significant negative LCM v Rn trends were found for both age groups, after adjusting for age and subsets of 21 county-level socioeconomic, climatic and other factors. Negative trends were largest for counties with ≤ 100 Bq m−3 Rn (p ≤ 0.00087; 420 analyses). Adjusted relative risk (RRadj) for LCM was significantly elevated (1 150 Bq m−3v 65–100 Bq m−3 Rn, most involving adjustment for climate- and education-related factors likely to have influenced exposure to indoor air contaminants such as Rn and cigarette smoke. Though inconclusive due to potential ecological-fallacy-related confounding that could not be controlled, results from this ecological study are most consistent with a U-shaped dose-response relationship between 1950–54 LCM risk and U.S. residential radon in white women who predominantly never smoked.
Read full abstract