Epidemiological implications of spatial and temporal radon variations

Epidemiological studies require accurate assessments of total radon-daughter exposures. Short-term radon measurements taken in current dwellings may misrepresent past exposures. In the upper midwest, we have observed significant spatial variation, on a scale of several kilometers, in yearly average indoor radon concentrations. Thus, ecological studies using small samples, or case-control studies using only the current residences for exposure assessment, could misjudge the actual long-term exposures if the size of the geographical cluster used is larger than a county. Short-term measurements also may introduce unacceptable variation. In a comparison of two measurement techniques in 80 upper-midwest homes, a correlation was found between year-long alpha-track and 2-day charcoal canister measurements. However, the observed coefficient of variation (factor of two) between measurement protocols may introduce enough scatter to obscure weak correlations in small samples. A correlation that was observed between lung-cancer rates in 13 rural Minnesota counties and the annual-average radon concentration (median = 100 Bq m-3) estimated from year-long alpha-track measurements was not found when charcoal canister data were used to estimate the radon exposure. Year-to-year variations may also be important. In 12 homes that have been monitored for longer than 1 y, yearly variations ranged from 0 to 500%, with a median variation of 22%. We are investigating a retrospective detection technique that may improve long-term radon exposure assessments. We have found correlation between the surface alpha activity and radon exposure for glass surfaces exposed from 6 to 2000 kBq y m-3. We have developed an alpha-track detection system that measures both current radon concentrations and exposure history