Comparison of biochemical and questionnaire estimates of tobacco exposure

Abstract

Measures of habitual tobacco exposure were examined in 142 men. Seventeen variables from a smoking questionnaire were tested for the magnitude and independence of their associations with two objective tests, the levels of expired air carbon monoxide (CO) and of serum thiocyanate (SCN). Of the 17, cigarette smoking frequency was the strongest predictor of the biochemical results; it accounted for 24% of the variance in an index that combined the SCN and CO values. Only two other questionnaire variables contributed additional information in a stepwise multiple regression analysis: the time elapsed since last smoking (6%) and the longest period the smoker had been off cigarettes in the past (6%). Conventional questionnaire estimates of dosage such as depth of inhalation, the amount of each cigarette smoked, and the use of filters did not contribute significant information to the multiple regression of SCN and CO, and this finding was confirmed by an analysis of covariance. The CO and SCN analyses provide an objective and practical addition to estimates of tobacco exposure from questionnaire responses.