Light drinking in pregnancy is not associated with poor child mental health and learning outcomes at age 11
- 1 Department of Obstetrics and Gynaecology, Aarhus University Hospital, Aarhus, Denmark
- 2 Institute of Public Health and Center for Healthy Ageing, University of Copenhagen, Copenhagen, Denmark
- Correspondence to
Dr Ulrik Schiøler Kesmodel
Department of Obstetrics and Gynaecology, Aarhus University Hospital, Brendstrupgaardsvej, Aarhus 8200, Denmark;
Health authorities in most countries recommend that pregnant women abstain from alcohol, but in England the current recommendation from the department of health suggests that intake of up to 1–2 UK units once or twice a week may be acceptable.1 Furthermore, many clinicians do not recommend women to abstain from alcohol in pregnancy. While daily alcohol drinking during pregnancy is potentially harmful, it is still controversial whether weekly intake of alcohol may have adverse effects on fetal development and later child development.2
Sayal et al used data from the Avon Longitudinal Study of Parents and Children, a well-described cohort of 14 541 pregnancies previously used for several studies on the association between alcohol intake in pregnancy and adverse pregnancy outcomes and child development. Inclusion was restricted to white European women with singleton deliveries.
Information on alcohol intake was self-reported in the mid second trimester, and was categorised in this study as: no intake (45%); <1 drink per week (39%); ≥1 drinks per week (16%). When the children were 11 years old, parents and teachers completed the strengths and difficulties questionnaire (SDQ), a validated measure of the children's mental health. Academic achievement was measured by the Key Stage 2 (KS2) examinations. All results were adjusted for a number of potential, relevant confounders.
Parent-rated SDQs and teacher-rated SDQs were available from 54% and 52% of the 12 286 women, who had provided information on alcohol intake. KS2 scores were available for 86% of children. Non-response to parent-rated SDQ was associated with younger age, high parity, smoking, use of illicit drugs, being unmarried, depression, attainment of a low level of education, rented home, low gestational age and child birth weight. Non-response to teacher-rated SDQ and KS2 was associated with use of illicit drugs, being unmarried, rented home, low level of education and, for KS2 alone, smoking and depression.
All the analyses comparing the SDQ and KS2 outcomes between children in the three alcohol-exposure categories were statistically insignificant, except for two results: in girls, total parent-rated SDQ score was statistically significantly worse among those exposed to <1 drink per week compared with abstainers (adjusted regression coefficient=0.38, 95% CI 0.01 to 0.74), but not in those exposed to ≥1 drinks per week (adjusted regression coefficient=0.13, 95% CI −0.39 to 0.65). Total teacher-rated SDQ score was statistically significantly better among those exposed to ≥1 drink per week compared with abstainers (adjusted regression coefficient=−0.45, 95% CI −0.89 to −0.01).
The results of any study can potentially be due to random error or bias. In the case of multiple statistical tests, one or more significant findings may be expected because of chance alone, even in the absence of a true effect. In this study, two statistically significant findings were described, one showing a negative and one showing a positive effect. Both effects were noted to be small. The finding of a negative influence of alcohol was described only in the group reporting intake of <1 drink per week, but not in the higher intake group, which is biologically implausible. Thus, the possibility of random error should be considered.
While a response rate of around 50% for the SDQ scores is not bad for a long-term follow-up, the response rate is a problem, because participants and non-participants differed in respect to a large number of important potential confounders. Selection bias is therefore a potential key problem in this study.
In addition, information bias is a possibility in studies of alcohol during pregnancy.3 In this case several alcohol intake groups were lumped together into one group comprising all levels of intake from 1 drink per week upwards. In most studies, and in real life, the majority of pregnant women who drink alcohol do so in small amounts. High, daily alcohol intake during pregnancy is reported only by a minority. Hence, in a population-based sample, as in this study, a potential effect of moderate intake might be overlooked, simply because a small group of women with a moderate intake is categorized together with a large group of women with a low intake of a few drinks per week, which may not be harmful. Even so, with several exposure categories, non-differential misclassification due to under-reporting could be expected to lead to bias away from the null value.2
Finally, although the authors adjusted for possible confounders, they were limited by the variables that were included in the data set. For some important potential confounders, including parental intelligence, the authors only included proxy measures, such as university degree. It is likely that there are additional confounders that were not accounted for in the model.
Despite these limitations, it is notable that there was no strong association observed between light drinking in pregnancy and the outcomes of interest. Drinking behaviour is a particularly interesting topic, because of diverse results in the scientific literature. As concluded by the authors, however, light drinking in pregnancy does not appear to be associated with clinically important adverse effects for mental health and academic outcomes at the age of 11 years.