Aims: To estimate the effect of maternal cannabis use on birth weight.
Design: Meta-analysis of published observational studies adjusted for cigarette smoking. Separate analyses were performed for studies of low birth weight and mean birth weight. We used fixed and random effects models, but in all cases the results were identical.
Setting: From the Medline database, we identified 10 studies in which the results were adjusted for cigarette smoking. In seven studies, information on cannabis use was collected prenatally. Five studies reported results for differences in mean birth weight associated with maternal cannabis use.
Participants: 32,483 women giving birth to live-born infants.
Measurements: Mean birth weight and odds ratio for low birth weight.
Findings: Three analyses of the studies on mean birth weight were conducted to avoid double-counting women from one study. The largest reduction in mean birth weight for any cannabis use during pregnancy was 48 g (95% confidence interval (CI) 83-14 g), with considerable heterogeneity among the five studies. Mean birth weight was increased by 62 g (95% CI 8 g reduction-132 g increase; p heterogeneity 0.59) among infrequent users (< or = weekly) whereas cannabis use at least four times per week had a 131 g reduction in mean birth weight (95% CI 52-209 g reduction; p heterogeneity 0.25). From the five studies of low birth weight, the pooled odds ratio for any use was 1.09 (95% CI 0.94-1.27, p heterogeneity 0.19).
Conclusions: There is inadequate evidence that cannabis, at the amount typically consumed by pregnant women, causes low birth weight.