Background International guidelines have not reached consensus on safe alcohol recommendations for pregnant women. Accurately measuring exposure to alcohol use during pregnancy is important in estimating the fetal effects of gestational alcohol consumption to inform alcohol policy. However, the prevalence of alcohol use during pregnancy is often conflicting. We aimed to compare the prevalence and predictors of alcohol consumption during pregnancy in three contemporary studies.

Methods We used log linear binomial regression in STATA to compare the prevalence and predictors of alcohol use in two Irish retrospective studies and one multi-centre prospective cohort study (Ireland, UK, New Zealand, Australia). Growing up in Ireland (GUI) is a population based cohort study which sampled women from the national state child benefit register and collected data on alcohol consumption through face to face interviews 9 months after birth. SCOPE is a prospective multi-centre cohort study which used convenience sampling to interview 5628 nulliparous women on their alcohol consumption at 15 and 20 weeks gestation. PRAMS is a cross sectional study which used random sampling of delivery records at a large urban obstetric unit in the South of Ireland to administer postal surveys to 718 women 2–6 months after birth.

Results SCOPE and PRAMS participants were more advantaged than GUI participants (89% had tertiary education in SCOPE compared to 82% in PRAMS and 56% in GUI) though GUI participants were more representative of the Irish pregnant population. Alcohol consumption during pregnancy in Ireland ranged from 20% in GUI to 80% in SCOPE and from 40% to 80% in the UK, Australia and New Zealand. Levels of exposure among drinkers also varied substantially ranging from 70% consuming more than 1–2 units per week in the first trimester in SCOPE Ireland to 46% and 15% in the retrospective cohorts. Smoking during pregnancy was the most consistent predictor of gestational alcohol use in all three cohorts RR (95% Confidence Interval): GUI 1.50 (95% CI 1.36, 1.65), SCOPE 1.17 (95% CI 1.12, 1.22), PRAMS 1.42 (95% CI 1.18, 1.70) while other characteristics including age were inconsistently related to alcohol use in each cohort.

Conclusion Alcohol use during pregnancy may be as common as 80% but methods to improve exposure measurement are required due to conflicting estimates. The observed social patterns of gestational alcohol use may also vary across studies depending on data collection methods. Future research should address both under-reporting and reporting biases which reduce both their internal validity and policy translation.