CLASSIFICATION OF RESIDENTIAL EXPOSURE TO NITROGEN-DIOXIDE

In epidemiological studies of indoor nitrogen dioxide (NO2) reported to date, only a limited number of measurements were made to estimate personal exposures. The accuracy of exposure estimates depends on the measurement error inherent in the samplers and in the sampling scheme. If categorical measures of exposure are used, the consequences of measurement error vary with classification rules. The effects of alternative measurement strategies on exposure classification were evaluated using NO2 diffusion sampler measurements from 653 residences with gas cooking ranges. A time-series of consecutive 2-week integrated measurements was obtained for each home. During the winter season, the standard deviation was 26% of the mean. The probability of misclassification, a function of the mean and the standard deviation per home, was examined for three concentration classes (< 25, 25-50, greater-than-or-equal-to 50 ppb). The average probability of misclassification was 10% for three 2-week integrated measurements across the winter and decreased to under 5% for continuous sampling (i.e. 13 consecutive 2-week samples). The sensitivity and predictive value were improved by increasing the number of samples obtained. The misclassification is determined by the number of concentration classes and by the relationship of class boundaries to the distribution of true means.