Repeated validation of parental self-reported smoking during pregnancy and infancy: a prospective cohort study of infants at high risk for allergy development.

Exposure to environmental tobacco smoke (ETS) during fetal life and infancy is closely related to the smoking habits of the parents. Estimates of exposure to ETS require valid and detailed information on changes in cigarette smoking over time. The objective was to test the validity of self-reported smoking among parents during pregnancy and early childhood in a cohort of children at high risk for allergy development by measurement of exhaled carbon monoxide (CO). The cohort comprised 117 families enrolled from the general population of pregnant women at admission to antenatal care. Data on parental tobacco smoking were obtained by interview and exhaled CO was measured (Micro-Smokerlyzer(R)) in parents twice during pregnancy and when the child was 6 and 18 months old. The median (range) exhaled CO levels were 3 (0-10) parts per million (ppm) for non-smokers and 15 (1-39) ppm for smokers (P < 0.0005). A receiver operating characteristic (ROC) analysis was performed at each examination. The areas under the ROC curve were high for both mothers (between 0.88 and 0.99) and fathers (between 0.87 and 0.89), indicating exhaled CO as a good diagnostic tool for determining smoking status. Comparing the ROC areas obtained for mothers from late pregnancy and during infancy with the area from early pregnancy showed no statistical differences (P = 0.21, 0.43 and 0.44 respectively) and the same was true for fathers during infancy (P = 0.81). The level of 8 ppm was used as the cut-off between smokers and non-smokers, based on data from a pilot study. Using CO as a diagnostic tool for smoker status showed very high specificity (between 97 and 100%), indicating that very few persons claiming to be non-smokers had CO levels higher than 8 ppm. In conclusion, the validity of interview-obtained self-reported smoking among parents during pregnancy and early childhood was high. Repeated interviews and CO measurements in a prospective study design did not change the validity, indicating a low risk of information bias. A structured interview combined with measurement of exhaled CO is a valid and reliable method for estimating ETS exposure to the fetus and young infant.