Analytical method for evaluating (and correcting) the impact of outdoor radon concentration on the estimates of percentage of dwellings exceeding reference levels

Abstract

Outdoor radon concentration contributes to indoor radon levels, generally causing a shift from lognormal distribution of measured radon concentration data distribution, and it makes more challenging the estimation of radon distribution parameters on the basis of the lognormal assumption. In particular, lognormal assumption with no correction could lead to a significantly biased estimate of the percentage of dwellings exceeding a certain level, e.g. a reference level (RL), since this is based on biased estimates of geometric mean (GM) and geometric standard deviation (GSD) of radon concentration distribution.Subtracting to each measured data a constant outdoor radon level can usually compensate data distribution departure from log-normality (except for low radon levels), if the appropriate outdoor level value is chosen by means of a lognormal fit of the data. This approach – already (but not always) used in literature – cannot be applied in cases where all the data of radon concentrations are not available (e.g., for a review study). For these cases, this work presents an analytical method to quantitatively evaluate and correct the impact of outdoor on the lognormal distribution parameter estimates and, in particular, on the percentages of dwellings exceeding radon reference levels. The proposed method is applied to a number of possible situations, with different values of outdoor radon level, GM and GSD.The results show that outdoor radon levels generally produce an underestimation of the actual GSD parameter, which increases as the outdoor level increases, and in the worse cases, could lead to an underestimation higher than 50%.Consequently, if the outdoor contribution is not properly taken into account, the percentage of dwellings exceeding a certain RL is almost always underestimated, even by 80%–90% for RL equal to 300 Bq/m3. This could have implications for the classification of areas as regards radon concentration and for the estimation of avertable lung cancers attributable to radon levels higher than some possible RLs.