Nitrogen dioxide and respiratory illness in children. Part I: Health outcomes.

We have carried out a prospective cohort study to test the hypothesis that exposure to nitrogen dioxide increases the incidence and severity of respiratory infections during the first 18 months of life. Between January 1988 and June 1990, 1,315 infants were enrolled into the study at birth and followed with prospective surveillance for the occurrence of respiratory infections and monitoring of nitrogen dioxide concentrations in their homes. The subjects were healthy infants from homes without smokers; they were selected with stratification by type of cooking stove at a ratio of four to one for gas and electric stoves. Illness experience was monitored by a daily diary of symptoms completed by the mother and a telephone interview conducted every two weeks. Illnesses with wheezing or wet cough were classified as involving the lower respiratory tract; all other respiratory illnesses were designated as involving the upper respiratory tract. Exposure to nitrogen dioxide was estimated by two-week average concentrations measured in the subjects' bedrooms with passive samplers. This analysis is limited to the 1,205 subjects completing at least one month of observation; of these, 823 completed the full protocol, contributing 82.8% of the total number of days during which the subjects were under observation. Incidence rates for all respiratory illnesses, all upper respiratory illness, all lower respiratory illnesses, and lower respiratory illness further divided into those with any wheezing, or wet cough without wheezing, were examined within strata of nitrogen dioxide exposure at the time of the illness, nitrogen dioxide exposure during the prior month, and type of cooking stove. Consistent trends of increasing illness incidence rates with increasing exposure to nitrogen dioxide were not evident for either the lagged or unlagged exposure variables. The effect of nitrogen dioxide exposure on illness occurrence during at-risk intervals of two weeks' duration was examined using the generalized estimating equation approach. In these multivariate analyses, none of the odds ratios was significantly elevated for unlagged nitrogen dioxide exposures, lagged nitrogen dioxide exposures, or stove type. Duration of illness was assessed in relation to the same exposure variables; illness duration and nitrogen dioxide exposure were not associated. We have found that indoor exposure to nitrogen dioxide is associated with neither the incidence nor the duration of respiratory illnesses. The study was designed to have sufficient power to detect effects of nitrogen dioxide exposure of magnitudes previously reported and in a range relevant to public health concern; the lack of association cannot be attributed to potential bias from misclassification of outcome or exposure.(ABSTRACT TRUNCATED AT 400 WORDS)