After removing duplicates, we identified a total of 567 records. Based on the inclusion criteria, we included 11 publications in the further analysis. We combined two publications that presented the same study (France et al., 2013, 2014); thus, we included 10 studies in the systematic review. An overview can be seen in the following Fig. 1.
We found a huge variety in the study designs, including four randomised controlled trials (RCTs), a secondary data analysis, an observational comparative study, a case control study, a formative evaluation, a non-randomised pre-post intervention study, and an experimental randomised controlled trial. Seven out the 10 studies had follow-ups (Bazzo et al., 2015; Caley et al., 2010; Driscoll et al., 2018; Ingersoll et al., 2013, 2018; Tenkku et al., 2011; Wilton et al., 2013), while Cil (Cil, 2017) compared administrative data over multiple years. Only one study investigated FASD incidence rates (Cil, 2017).
Most studies were conducted in the USA (n = 8), while the other two took place in Italy (Bazzo et al., 2015) and Australia (France et al., 2013, 2014). Four studies consisted of samples of AEPRG (Ingersoll et al., 2013, 2018; Tenkku et al., 2011; Wilton et al., 2013), four further studies of women in general (Cil, 2017; Driscoll et al., 2018; France et al., 2013, 2014; Yu et al., 2010), following one including both healthcare professionals and alcohol exposed pregnancy (AEP) risk groups (Bazzo et al., 2015) and one of solely healthcare professionals (Caley et al., 2010). Four studies only included non-pregnant women (Ingersoll et al., 2013, 2018; Tenkku et al., 2011; Wilton et al., 2013), one included only pregnant women (Bazzo et al., 2015), and three addressed both pregnant and non-pregnant women (Driscoll et al., 2018; France et al., 2013, 2014; Yu et al., 2010).
The intervention and prevention types employed in the studies were six on a universal level (including one on a structural level) and four primary preventions for the specific target group of AEPRG. The structural level intervention in the study by Cil (2017) analysed the effect of alcohol warning signs or labels (AWS) placed on alcoholic beverages on birth outcomes on a statewide level. Interventions on a universal level implemented message framing effects, i.e., the effects and influence of the presentation of information on our decision making, (France et al., 2013, 2014; Yu et al., 2010), awareness and educational posters in women’s restrooms (Driscoll et al., 2018), multilevel FASD awareness programs (Bazzo et al., 2015), and educational workshops on screenings and interventions for healthcare professionals (Caley et al., 2010). The interventions for the specific target group AEPRG included web-based and/or mail-based interventions (Ingersoll et al., 2018; Tenkku et al., 2011), video, brochure, or motivational interviewing (Ingersoll et al., 2013), and telephone or in-person interventions (Wilton et al., 2013). The outcomes assessed were drinking and contraceptive behaviour, FASD and prenatal alcohol consumption awareness, knowledge, beliefs and opinions, self-efficacy, symptoms of mental illness and drug use, as well as intervention implementation behaviour.
Self-reports (Bazzo et al., 2015; Driscoll et al., 2018) and Timeline Followbacks (TLFB; Sobell & Sobell, 1992) (Ingersoll et al., 2013, 2018; Tenkku et al., 2011; Wilton et al., 2018) were used as the method for measuring alcohol consumption. The TLFB was also used to assess contraceptive behaviour, although not every study that measured this outcome used a TLFB. In addition to the TLFB, Tenkku et al. (2011) and France et al., (2013, 2014) also implemented the Alcohol Use Disorders Identification Test (AUDIT), while Ingersoll et al. (2013) used the MINI Module J to screen for alcohol use disorders.
The study that received the highest quality rating was Ingersoll et al. (2018) with 90.4%, followed by Cil (2017) with 90.0%, and Wilton et al. (2013), Ingersoll et al. (2013), and Driscoll et al. (2018), with 87.5%, 81.8%, and 78.6%, respectively. The lowest rated study was by Caley et al. (2010) with 52.9%. A brief description of the studies we included in this review can be found in Table 1.
Table 1 Characteristics of the included studies Bazzo et al. (2015) used an observational comparative study design to evaluate an FASD health campaign that took place in Treviso by comparing healthcare professionals’ and pregnant women’s knowledge of and opinions about prenatal alcohol consumption to that of those in Verona, where the campaign did not take place. Additionally, they investigated the sources and kind of information that the groups either provided or received. The study showed that the campaign had long term positive effects on healthcare professionals’ knowledge and practice as they provided more information to their patients, although no significant difference was found among the pregnant women. The authors concluded that providing information alone is not an effective prevention strategy, as integrated and specialised approaches are needed. We identified the study’s limitations as being specific to population samples that limit the generalisation of the outcomes, the small sample size, and that the confounders were not controllable as it was purely an observational study.
Caley et al. (2010) evaluated the effectiveness of a workshop that focused on implementing FASD interventions using a case control study. Their sample consisted of health and human service professionals and took place four months after a workshop on FASD interventions. They found that 61% of the professionals initiated interventions, of which most were primary (59%). We found limitations in the low response rate (37%) and because no baseline was measured, no comparisons could be made prior to the workshop.
Cil (2017) evaluated the effectiveness of alcohol warning signs on decreasing prenatal alcohol consumption, prenatal binge drinking, and birth outcomes using a secondary data analysis by comparing data from national natality statistics and national surveys on behavioural risk taking. They found that AWS laws led to an 11% decrease in prenatal alcohol consumption odds and a 75% decrease in prenatal binge drinking odds. Furthermore, a change in prenatal alcohol consumption was largest amongst primipara and women who were over 30. Significant differences after the implementation of AWS laws in FASD birth outcomes or incidence rates could not be found. Cil theorised that the lack of a significant correlation between AWS laws and FASD birth outcomes was due to lack of diagnoses at birth. We found that limitations of the study included the lack of comparability of the data, as the author compared samples with different sources and years of origins. Additionally, the apperception of AWS was not measured, i.e., whether or not individuals were even aware of AWS, so evidence for a direct link between AWS and lower prenatal alcohol consumption was lacking, as was the role that other factors may have played in the observed decreases.
France et al., (2013, 2014) evaluated the efficacy of different message types regarding their persuasiveness, including their influence on the intention to and confidence in a person’s ability to abstain from or reduce alcohol consumption or prenatal alcohol consumption. Results of this RCT showed that there was a general significant increase in intention and confidence in abstaining or reducing alcohol consumption, although threat concept messages were the most effective in increasing behavioural intentions and confidence in possible modification. The authors concluded that threat messages should be implemented in preventive messages and campaigns, while adding that self-efficacy concepts in communication helped decrease potential reactions, emotions, and cognitions. Thus, a combination of concepts was found to be the best way to develop preventive messages. Limitations of the study included difficulties in generalizing from the findings, as women with a lower socio-economic status were not included in the study.
In a pilot RCT, Ingersoll et al. (2018) evaluated the efficacy of an automated, individually tailored intervention on AEP risk versus a static educational website. The study focused on risky drinking and contraceptive behaviour. Significant outcomes included reductions in unprotected sex, risky drinking, and AEP risk amongst the specialised intervention website group targeted. There was no significant change amongst the static educational website group. The authors concluded that if the participants utilised the program more, they were more likely to have experienced a change in their behaviour. Additionally, participants were more likely to finish all core modules in the individualised website. A limitation of the study was that the statistical power was weak and that larger sample sizes would be needed to further validate the findings.
Using an RCT, Yu et al. (2010) evaluated the effect of different message framing types (statistic vs. exemplar appeals, loss vs. gain appeals) on the intention to prevent FASD. The messages were designed to look like public service announcements found in newspapers. Results showed that exemplar appeals that were loss-framed elicited significant levels of fear, whereas gain-framed messages increased efficacy. Yu et al. (2010) concluded that each message frame had advantages and that message goals should be considered when implementing or creating awareness campaigns utilising framing effects. The authors concluded that a limitation of the study was the selective sample, as the students did not view themselves as at risk of pregnancy.
Tenkku et al. (2011) evaluated the effectiveness of a web-based intervention in reducing AEP risk using a nonrandomized, pre-post intervention study. The sample consists of AEP at-risk women who self-selected into two delivery methods, either web- or mail-based. The interventions consisted of four modules based on motivational messaging that were specifically tailored to the individuals’ answers and needs. The study had a follow-up of four months after baseline. Outcomes showed a significant decrease in AEP risk, although no significant difference could be found between the groups. The authors concluded that a self-guided intervention utilising motivational interviewing techniques showed effectiveness in preventing AEP. We determined that limitations of the study included the self-reports, low follow-up rates, and insufficient, unbalanced sample groups.
Ingersoll et al. (2013) used an RCT to evaluate the efficacy of a one-session motivational AEP prevention intervention. For this, women were assigned to three groups, the first of which consisted of an individual, face-to-face, 60-min single counselling session that utilised motivational interviewing (EARLY). Participants in the second group watched an informational video, received a 5-min briefing from a counsellor, and were offered brochures. The third group only received informational brochures. Results showed that all intervention conditions effectively reduced AEP risk, and that the individualised face-to-face condition showed the highest reduction rates in ineffective contraceptive behaviour and AEP risk. The brochure condition proved to be more efficient than the video condition. The authors theorised that this could be due to possible stereotyping and stigmatisation that resulted from the contents of the video, as well as differences in general contents between the two conditions. This could explain why the motivation to change was lower, as the women in that condition group did not view their behaviour as extreme as that of the women portrayed in the video. The authors suggested a lack of non-TLFB control group and possible cross-contamination due to the same counsellors implementing all conditions, as limitations to their study.
Using an RCT, Wilton et al. (2013) compared two sessions of an in-person and telephone administration of a brief AEP risk reduction intervention. Results showed a significant but small reduction in alcohol use, as well as a large and significant increase in effective contraceptive behaviour. In general, the intervention resulted in a significant reduction in AEP risk and differences between administration types were not found. The authors concluded that brief telephone interventions could be used successfully and were more cost-effective than in-person interventions. Additionally, the study showed that it might be easier to increase effective contraceptive use, as women of childbearing age might not be interested in decreasing their alcohol consumption due to their current lifestyle. The authors suggested the small sample size, funding limitations that lead to change in counsellors in follow-up with possible rapport differences, and possible recollection errors in the TLFB due to the nature of self-reports, as limitations to their study.
Driscoll et al. (2018) evaluated the effectiveness, acceptability, and feasibility of FASD prevention messages in women’s restrooms in a formative evaluation study. The authors placed health communication messages on pregnancy test dispensers or as posters in women’s restrooms in establishments that served alcohol. These health messages included information on outcomes of FASD, as well as numbers for telephone carelines while addressing contraceptive behaviour. Pregnancy tests and condoms were distributed for free or for a small charge in the establishments. Results showed that pregnancy tests that were free were more likely to have been used. Both health message distribution types were effective in increasing knowledge, but the dispensary type proved the most effective. Although prenatal alcohol consumption was lower at follow-up, it was still prevalent (< 20%). We found limitations of the study in the lack of baseline data collection and control groups. Additionally, it was unclear exactly how participants were recruited for baseline; the only information given was that recruitment took place in the communities in which the messages were distributed.