Use of carbon monoxide breath validation in assessing exposure to cigarette smoke in a worksite population.

Assessed the usefulness of carbon monoxide (CO) breath validation of self-reported smoking status in a large worksite population (N = 4,647). CO assessment was performed as part of a baseline survey procedure. CO levels differed substantially in relation to self-reported smoking status and amount smoked. Correcting for ambient exposure (estimated by mean CO levels among never smokers) produced more satisfactory results than uncorrected CO levels. Striking company differences were observed in mean CO exposures among self-reported never smokers. An unexpected finding was that 17.1% of current smokers reported smoking less than daily. Although the CO measure was excellent in detecting moderate and heavy smokers, it was inadequate in detecting occasional and light smokers. If detection of occasional or lighter smoking is critical to the purposes of the study, the more expensive (but more accurate) cotinine measure is preferred.