Background: Recent studies have found that cotinine is a better predictor of birthweight than the number of cigarettes smoked in pregnancy. In this paper we test this hypothesis and use cotinine to explore the effect of environmental tobacco smoke (ETS) on birthweight.
Methods: In all, 1254 white women were interviewed at booking, 28 and 36 weeks about the number and brand of cigarette smoked. Cotinine was assayed from blood samples taken on the day of interview. The outcome was birthweight for gestational age.
Results: There was good agreement between self-reported smoker/non-smoker status and maternal cotinine with 1.3% women mis-reported as non-smokers at booking, 0.6% and 1.8% mis-reported at 28 and 36 weeks respectively. Among smokers, cotinine was more closely related to birthweight than the number of cigarettes smoked at all three time points (r = -0.25 versus r = -0.16 at booking). A reduction in cotinine between booking and 28 weeks was associated with increased birthweight but the effect was not statistically significant. Among non-smokers the association between birthweight and cotinine was not statistically significant after adjusting for maternal height, parity, sex and gestational age. Difference in mean birthweight between non-smokers in the lower and upper quintiles of cotinine was 0.2% (95% CI: -2.4, 2.8). Pooling the results of 10 studies plus our own gave an estimated difference in mean birthweight between women unexposed and exposed to passive smoke of 31 g (95% CI: 19, 44).
Conclusions: Cotinine is a better predictor of birthweight than the reported number of cigarettes smoked. If biochemical analysis is impossible, then self-reported smoking habit should be obtained prospectively using a structured approach. Any effect on birthweight of maternal passive smoking during pregnancy is small compared with the effects of maternal active smoking.