Prevalence and Patterns of Prenatal Alcohol Exposure in Australian Cohort and Cross-Sectional Studies: A Systematic Review of Data Collection Approaches

Abstract

This study sought to determine data collection approaches in Australian cohort studies and explore the potential impact on reported prenatal alcohol exposure (PAE) prevalence and patterns. Inclusion criteria were that studies related to a general Australian antenatal population where PAE was assessed and reported. Studies were excluded if they were not peer reviewed, examined the prevalence of PAE in pregnancies complicated by alcohol-use disorders, or were published in a language other than English. A systematic search of five electronic databases (PubMed, Embase, CINAHL, Web of Science, and Scopus) was conducted. Risk of bias was assessed using the Effective Public Health Practice Project quality assessment tool. Results were synthesised using MetaXL. Data from 16 separate birth cohorts (n = 78 articles) were included. Included cohorts were either general cohorts that included alcohol as a variable or alcohol-focused cohorts that were designed with a primary focus on PAE. PAE prevalence was estimated as 48% (95% CI: 38 to 57%). When subgroup analysis was performed, estimates of PAE prevalence when self-administered surveys and interviews were used for data collection were 53% (95% CI: 41% to 64%) and 43% (95% CI: 28% to 59%), respectively. Use of trained assessors was an influencing factor of the prevalence estimates when data were collected via interview. Alcohol-focused studies reported higher prevalence of PAE, regardless of method of survey administration. Where interviewer training is not possible, self-administered questionnaires will likely provide the most reliable PAE estimates. No funding sources are relevant to mention. Review was registered with PROSPERO (CRD42020204853).

Keywords: alcohol; data collection; interview; pregnancy; prenatal; questionnaires; surveys; systematic review.

Figure 1

PRISMA flow diagram.

Figure 2

Forest plots displaying prevalence estimates with 95% confidence interval (CI) and subgroup analysis for alcohol exposure at any time during pregnancy for cohorts in which information on PAE was collected using self-administered survey-based methods and interview-based methods. I², between study heterogeneity; Q, Cochran’s Q; p, statistical significance. For each cohort, the prevalence rate reported by the publication using the largest sample of that cohort is shown. An overall prevalence value was not given for the NDSHS 2010 cohort, where the value prior to pregnancy recognition was used. The Triple B study used trained assessors to collect interview information. The Hunter New England Cohort and the NDSHS Cohort from 2001 to 2007 were excluded due to only reporting prevalence after awareness. The NDSHS Cohort from 2010 and the Barwon Infant Study were excluded due to small sample size reporting overall prevalence (n < 1000, despite overall cohort size n > 1000). Triple B and AQUA cohorts differentiated between pre- and post-recognition of pregnancy. For the interview subgroup, open triangle represents doctor, closed triangle represents midwife, open circle represents research assistant/research midwife, and closed circle represents trained assessor. Boxes in the interview subgroup represent studies where the assessor was unclear. Results suggest that while overall self-administered surveys generated higher prevalence estimates than interview-based methods, and the use of trained assessors in interview settings resulted in high PAE estimates.