Prenatal alcohol exposure and pregnancy and childhood outcomes: a systematic review of alternative analytical approaches

Abstract

Listen

Translate article

BackgroundInvestigation of the effects of prenatal alcohol exposure in observational studies using standard analytical approaches such as multivariable regression can produce results that are biased by residual confounding or measurement error due to self-reported alcohol consumption. We aimed to determine the causal effects of maternal alcohol consumption in pregnancy on long-term outcomes in offspring. MethodsWe conducted a systematic review of studies that used alternative analytical methods. We searched Medline, Embase, Web of Science, and PsycINFO (Jan 1, 1950, to Feb 11, 2015) using a combination of MeSH terms and text words to reflect outcomes (eg, pregnancy, fetal development), exposure (alcohol), and relevant study designs, limiting the search to publications in the English language. Study designs including Mendelian randomisation, negative control, and other family-based comparisons, and quasi and natural experiments were included. Pregnancy outcomes (such as birth weight), and features of fetal alcohol spectrum disorders (FASD) related to alcohol exposure during pregnancy were examined. Data were extracted by one of four reviewers and were checked by the lead reviewer. This study is registered with PROSPERO, number CRD4201501594. FindingsOf 751 full text articles, ten reported effects of prenatal alcohol consumption on offspring outcomes (six Mendelian randomisation and four quasi-experimental, three of which were negative control studies and one a natural experiment). One Mendelian randomisation analysis reported evidence that prenatal alcohol consumption was associated with an increased risk of “a FASD-like construct”, and two other analyses conducted on the same cohort showed evidence of lower childhood intelligence quotient (IQ) (mean difference per additional risk allele −1·2 points, 95% CI −1·89 to −0·52) and lower school results for the group genetically predisposed to consuming more alcohol. A family-based negative control study (maternal–paternal comparison) in the same UK cohort suggested intrauterine effects on IQ and school results for maternal compared with paternal binge drinking but not for moderate alcohol use. Analyses of a Swedish natural experiment also showed evidence of worse educational and occupational outcomes in offspring caused by exposure to higher levels of alcohol consumption proxied by increased availability of liquor (eg, 24% decrease in earnings, 95% CI 19 to 29). InterpretationWe found a causal effect of prenatal alcohol consumption on offspring outcomes. Further studies are needed and women should be made aware of existing findings. FundingMRC Integrative Epidemiology Unit, National Institute for Health Research Collaboration for Leadership in Applied Health Research Care (NIHR CLAHRC) West.