Maternal and partner prenatal alcohol use and infant cognitive development

Highlights

• A large pregnancy cohort study assessed alcohol use by mothers and partners.

• No observable effects of prenatal alcohol exposure on infant cognition.

• Alcohol use by partners showed similar patterns of association.

• Findings were robust using multiple methods to control for confounding.

• Interaction between alcohol, social factors and infant development likely.

Abstract

Background

Teratogenicity of heavy prenatal alcohol exposure is established, but uncertainty remains regarding the impact of moderate alcohol exposure on cognitive deficits in infants. Separating in utero effects from environmental confounding is a challenge for observational studies; consideration of alcohol use by partners as well as mothers may help clarify this. This study examined associations between prenatal alcohol use by both mothers and their partners and infant cognitive developmental outcomes at 12-months.

Methods

Pregnant women (n = 1331) and their partners (n = 699) were recruited from antenatal clinics of three metropolitan public hospitals in Australia, and completed detailed interviews about alcohol consumptions throughout pregnancy. Infants were assessed with the Bayley Scales of Infant Development − Third edition (Bayley) at 12-months of age.

Results

Alcohol use during pregnancy was reported by 65.7% of mothers and 84.1% of partners. Using multiple methods to adjust for confounding factors, no evidence for impaired cognitive ability associated with alcohol use by mothers or their partners was observed. Children born to women who drank low-levels of alcohol had slightly higher Bayley cognitive scores than those born to abstaining women. There was some evidence for an interaction between sociodemographic factors and prenatal alcohol exposure on infant cognitive outcomes.

Conclusion

This finding corroborates existing evidence to suggest there are no detrimental effects to infant cognitive development at 12-months of age following low-level prenatal alcohol exposure. Future prospective studies involving families of a broad range of backgrounds would be informative to clarify interaction between alcohol exposure and environmental factors on developmental outcomes.

Introduction

Alcohol has been established as a teratogen and may produce central nervous system deficits at high levels of exposure, yet uncertainty remains regarding the association between low-level prenatal alcohol exposure and impaired cognitive development. Most large-scale studies fail to detect any observable effect (Alati et al., 2008; Forrest et al., 1991; Kelly et al., 2009; Kesmodel et al., 2012; O'Callaghan et al., 2007; Williams Brown et al., 2010). A recent meta-analysis found evidence for lower IQ scores among children exposed to binge-level alcohol exposure, yet this effect was no longer statistically significant when only studies considered ‘high-quality’ were included (Flak et al., 2014). The same meta-analysis found a small but significant counterintuitive effect in with low-level exposure was associated with higher IQ scores in childhood (Flak et al., 2014). No association was found between binge or moderate exposure and impairment in other cognitive domains including attention, verbal ability, mathematical ability at school, language development, or visuo-spatial ability (Flak et al., 2014 supplement). Some studies, however, have shown the apparent ‘positive’ effect of low-level alcohol exposure on IQ is mitigated with appropriate adjustment for confounders such as maternal socioeconomic status and education (Brown et al., 2010; Jacobson et al., 1993; Kelly et al., 2009).

A major challenge facing research in this area is that systematic differences exist between women who drink alcohol during pregnancy and women who are non-drinkers on factors that are independently associated with child developmental outcomes. Furthermore, comparison of two systematic reviews reveals divergent sets of characteristics between those whose children have a diagnosis of FASD compared with those who tend to drink alcohol in pregnancy − an “epidemiological puzzle” (Meurk et al., 2014). A systematic review found that apart from pre-pregnancy alcohol consumption, older age, higher educational attainment, higher parity, being employed, and higher income were consistent predictors of increased likelihood to drink during pregnancy (Skagerström et al., 2013). Many of these factors describe a population of overall advantaged women whose offspring are likely to be raised in enriched environments. In contrast, a systematic review found that lower educational attainment, being single during pregnancy, unemployment, residing in a remote area, and lower income were all associated with having a child with FASD (Esper and Furtado, 2014). This is in line with the common finding that Fetal Alcohol Spectrum Disorders (FASD) are concentrated in disadvantaged, low SES populations (Abel, 1995; Abel and Hannigan, 1995; Burns et al., 2013; Chudley, 2008; Kvigne et al., 2003; Lange et al., 2013; May et al., 2000; May et al., 2004; May et al., 2009; Meurk et al., 2014). Older maternal age, substance use, smoking, and experience of trauma or violence were common risk factors for both FASD (Esper and Furtado, 2014) and prenatal alcohol exposure (Skagerström et al., 2013). At an epidemiological level, these divergent sets of demographic predictors suggest some interaction between prenatal alcohol exposure and environmental factors in producing deficits. For this reason, it is critical that studies attempting to assess impacts of prenatal alcohol exposure collect comprehensive data on psychosocial and demographic factors and include these in models.

Random allocation to an alcohol exposure condition in pregnancy is impossible; therefore, confounding in observational studies is dealt with statistically by inclusion of covariates in regression models. Propensity Score Matching is an alternative technique, used when randomisation to conditions is not possible, which ensures baseline characteristics are consistent between groups and any differences between them attributable to ‘treatment’ or exposure effects (Austin, 2011; Rosenbaum and Rubin, 1985). The Millennium Cohort Study in the UK observed no difference in cognitive performance between children with low-level prenatal alcohol exposure and those not exposed, and this was regardless of whether ordinary least-squares regression or propensity score matching was used to account for bias (Kelly et al., 2013).

Another method by which the impact of shared postnatal family environment can be disentangled from that of the intrauterine environment is to compare magnitude of associations between maternal alcohol use and offspring outcomes with that of alcohol use by partners (Alati et al., 2008; Smith, 2008). Some studies using animal models suggest paternal effects of pre-conception alcohol use on offspring, including lower birth weight (Abel, 2004; Meek et al., 2007), and some suggest cognitive deficits in early life (Abel, 2004). In human research, one prospective cohort study found that while moderate alcohol use by mothers predicted poorer language ability in children, this effect was no greater in magnitude than the association with alcohol use by fathers (Alati et al., 2008). However, a subsequent follow-up of the same sample found a specific effect of frequent, heavy alcohol consumption by mothers and associated lower academic abilities in offspring (Alati et al., 2013). Few other prospective longitudinal studies have accounted for the effect of partner alcohol consumption when studying effects of prenatal alcohol exposure on child development.

The aims of this paper were to (1) examine the relationship between prenatal alcohol exposure across four time points during pregnancy on infant cognitive development at 12 months of age and (2) assess the relationship between alcohol consumption by partners and infant cognitive development.