1. One cohort study included here36 presented detailed findings only in relation to hospital admissions up to five years of age, but tabulations by age at admission suggest a similar strength of association between maternal smoking and admission for bronchitis or pneumonia at all ages from birth to five years. The results for all ages are therefore included in the metaanalyses
2. The early report by Colley et al2 suggested that the effect of parental smoking on the incidence of bronchitis and pneumonia was most marked in the first year of life (odds ratio 1.96, 95% CI 1.30 to 2.99), declining thereafter with increasing age of the child to an inverse relationship in the fifth year. Results from the Dunedin, New Zealand cohort showed a similar pattern, with a slightly greater effect
3. Discussion The direction of the association between parental smoking and lower respiratory illness is generally consistent across different study designs, methods of case ascertainment, and diagnostic groupings (table 2). Only one study from Brazil41 found an inverse relation (with pneumonia), but another South American study from Chile56 found a highly significant doubling in risk of pneumonia in the offspring of mothers who smoked. The latter could not be included in the meta-analysis as no confidence intervals could be derived