Research Questions:

1. Is there an independent effect of maternal consumption of alcohol during pregnancy on childhood behavioural, mental health and learning outcomes?

2. Is there a dose-response relationship?

3. Do the patterns and timing of alcohol consumption during pregnancy influence outcomes?

4. What are the causal and mediating relationships in influencing these outcomes?

Background

There is robust evidence of an aetiological relationship between high levels of alcohol consumption during pregnancy and adverse physical and neuropsychological childhood outcomes. When specific criteria are met, these have been best characterized at the extreme end as Fetal Alcohol Syndrome (Sokol et al, 2003). However, this raises the question of whether there is a safe level of drinking during pregnancy or if lower levels of exposure are also independently associated with adverse outcomes. There are conflicting accounts as to whether there is a small independent effect from lower levels of exposure (Linnet et al, 2003; Sood et al, 2001). Policy recommendations have become more cautious in recent years, suggesting abstinence from alcohol throughout pregnancy. However, the extent to which these recommendations are based on evidence remains uncertain.

Previous studies addressing these questions have had methodological limitations such as the use of high-risk samples which do not readily generalise to the wider population, short periods of follow-up, limited measures of antecedent confounders (such as maternal mental health), and retrospectively collected data that risks recall bias relating to the exposure (Linnet et al, 2003). In contrast, the ALSPAC dataset is very well placed to overcome these methodological issues in a large representative epidemiological sample.

Our preliminary analyses focussed on the effects of occasional drinking during pregnancy (Sayal et al, 2007). This is the largest published study (n=9086) that investigates the effects of prenatal alcohol use on child mental health outcomes and the methodology and findings are of international relevance. After controlling for a range of prenatal and postnatal factors, we found that the consumption of less than one drink of alcohol per week during the first trimester was independently associated with clinically significant mental health problems in girls (assessed at three time points between the ages of 4-9 years using ratings from 2 different informants). We have also found that the consumption of four or more drinks in a day in the second or third trimesters was independently associated with higher total levels of mental health problems (especially hyperactivity and conduct problems) in girls at the age of 47 months and in both genders at 81 months (paper in preparation). The consumption of four or more drinks continued to carry risk for these outcomes in the absence of daily drinking. Based on these preliminary findings, we wish to investigate whether these associations and outcomes persist over time (till age 13) and generalise to other outcomes (such as child temperament, development, learning, and mood).

Key Measures

Exposure: Maternal (and paternal) use of alcohol during pregnancy.

Outcomes:

Pre-school: Temperament measures (using the Carey & EAS scales); SDQ at 47 months; development (using the Denver Developmental Screening Test - done at 18, 21, 30 months).

School age: child mental health (externalising and internalising problems as measured by the SDQ at later time points, parent and teacher DAWBAs, & Mood & Feelings Questionnaires); cognition and learning (IQ measures and school SEN information); and neuropsychological/attentional outcomes (such as the TEACH).

Confounder variables: The analyses will control for a range of pre-, ante-, and post- natal factors such as maternal and paternal pre- and post- pregnancy alcohol consumption; maternal smoking and drug use in pregnancy; gestational age; birth weight for gestational age; birth complications; child gender; family structure and maternal and paternal age, mental health, education, and socio-economic status;.

Analytical Approach: Following initial bivariable analyses, multivariable logistic regression analyses will assess the effect of prenatal alcohol use on child outcomes. As sample attrition may be related to both the exposure and outcome, response status at follow-up will be related to baseline variables. As there are repeated measures of outcomes, latent constructs will be developed approaches such as structural equation modelling will be used to explore overlapping predictor factors for dependent variables and to test for causal and mediating relationships. This will also quantify the role of other key risk factors, especially smoking during pregnancy.