Prenatal alcohol exposure (PAE) has been linked to many adverse neurodevelopmental outcomes, such as impaired motor skills, reduced cognitive function, and behavioral problems in children (Chu et al. 2022). Children with PAE are at risk for developing fetal alcohol spectrum disorders (FASDs), which involve a broad range of impairments in physical, behavioral, cognitive, mental, and adaptive functioning (Maya-Enero et al. 2021).

Autism spectrum disorder (ASD) is a heterogeneous developmental disorder characterized by deficits in social communication and interaction as well as the presence of restricted and repetitive behaviors, interests, and activities (American Psychiatric Association 2013; Lord et al. 2020). Children with ASD exhibit wide variability in the prevalence of intellectual disability or co-occurring behavioral, developmental, and medical conditions (Levy et al. 2010; Wiggins, Levy, et al. 2015; Wiggins et al. 2022; Tian et al. 2020). While there is no single underlying cause of ASD, available evidence indicates that both genetic and environmental risk factors play an important role in ASD etiologies (Miles 2011; Kim and Leventhal 2015; Lyall et al. 2017; Hertz-Picciotto et al. 2018). Genetic factors could influence an individual’s predisposition to autism, while environmental factors might modify the phenotypic expression of this disorder through epigenetic alterations (Varadinova and Boyadjieva 2015). Some studies suggest that the embryonic period is a critical exposure window affecting the risk of developing ASD (Arndt et al. 2005; Packer 2016; Schieve, Tian, et al. 2018).

PAE has been investigated as a potential environmental risk factor for ASD, but its effect, if any, on ASD development remains unclear (Carpita et al. 2022; Luo et al. 2022). Individuals with ASD and FASDs share some characteristics, such as cognitive delays, behavioral problems, and social skill deficits (Stevens et al. 2013; Carpita et al. 2022). The clinical overlap of these conditions suggests possible shared etiologic mechanisms. However, a recent meta-analysis of observational studies suggested that currently available data do not support an independent association between PAE and ASD (Luo et al. 2022). These observational studies investigating the link between PAE and ASD typically assessed ASD outcomes using a binary measure, comparing those meeting the study’s ASD criteria to a control group without ASD (Luo et al. 2022; Eliasen et al. 2010; Gallagher et al. 2018; Singer et al. 2017). Given the significant variability in ASD symptoms and severity, conducting a risk assessment solely based on a classification of ASD versus no ASD may obscure meaningful effects of PAE on the development of ASD in offspring. Furthermore, it remains unclear whether PAE has varying effects on neurodevelopmental outcomes, such as early learning ability, behavioral problems, and social engagement skills, in children with ASD compared to those without ASD. Assessing the associations with neurodevelopmental outcomes could enhance our understanding of the heterogeneity of this disorder, potentially providing insights into its etiology, diagnosis, treatment, and prognosis (Georgiades et al. 2013).

The main objective of this analysis was to examine the association between PAE during the first trimester (PAE-FT) and various aspects of child development, including early learning ability, behavioral problems, and the severity of autistic traits in preschool children with and without ASD, using data from the Study to Explore Early Development (SEED), a large case–control study of risk factors for ASD in the United States. We sought to shed light on the relationship between PAE-FT and ASD, including core and co-occurring phenotypical symptoms and their severity. Additionally, we aimed to examine whether PAE-FT is differentially associated with these outcomes in children with ASD compared to those without ASD.

Of 2714 children who were assigned a final study classification of ASD or POP, 2571 (94.7%) had data on maternal alcohol use, early learning ability, autistic traits, and behavioral problems and were therefore included in the analysis. Of these, 1237 children were classified as ASD, and 1334 children were classified as comparisons without ASD (POP).

Maternal, child, and family characteristics of the study sample are provided in Table 1. Mothers of children in the ASD group vs. the POP group differed in maternal age at birth, race-ethnicity, education, and household income relative to the federal poverty threshold. Mothers of children with ASD were more likely to be foreign-born, to smoke or be exposed to household smoking during pregnancy, and to have a psychiatric condition. Children in the ASD group vs. the POP group were more likely to be firstborn and had lower median scores in early learning ability and higher median scores in autistic traits (both SCI and RRB) and behavioral problems (both internalizing and externalizing behaviors).

The timing of alcohol use by study group is presented in Table S1. Maternal alcohol use was more common during the first trimester than during the second and third trimesters in both study groups.

Maternal alcohol use was also more prevalent among children in the POP group compared to those in the ASD group during the preconception period and the second and third trimesters. However, the difference in the prevalence of alcohol exposure during the first trimester (20.2% vs. 18.5%, p = 0.269) was not statistically significant.

Among mothers of children with ASD, 80.7% reported no alcoholic drinks in the whole period of pregnancy; 10.5% reported an average of fewer than 1 alcoholic drink per week during the first trimester, 4.6% reported an average of 1–2 drinks per week during the first trimester, and 4.2% reported an average of 3 or more drinks per week during the first trimester (Table 2). Among mothers of children from the POP group, 77.8% reported no alcohol use in the whole period of pregnancy; 12.7% reported an average of fewer than 1 alcoholic drink per week during the first trimester, 6.2% reported an average of 1–2 drinks per week during the first trimester, and 3.4% reported an average of 3 or more drinks per week during the first trimester (Table 2). There were no differences in the distribution of alcohol quantity consumed per week between study groups (p = 0.072).

Unadjusted bivariate analyses showed that in both the ASD and the POP groups, when compared to children whose mothers did not consume alcohol during pregnancy, children whose mothers consumed 3 or more alcoholic drinks per week in the first trimester had more externalizing behavior problems (Table 2). Children in the POP group whose mothers consumed 3 or more alcoholic drinks per week had a lower median score in early learning ability, although the difference was not statistically significant. There were no statistically significant differences in median values across PAE-FT levels for any of the other outcome measures.

The maximum VIF was 1.6, indicating there was no severe collinearity problem between the explanatory variables. Adjusted quantile regression analyses demonstrated that in both the ASD and the POP groups, children whose mothers reported 3 or more alcoholic drinks per week during their first trimester had more externalizing behavior problems (Table 3). Specifically, the median CBCL externalizing behavior score of children whose mothers consumed 3 or more drinks per week during their first trimester was 2.87 points higher for children with ASD and 5.89 points higher for children in the POP comparison group than children whose mothers abstained from alcohol throughout the entire pregnancy.

Moreover, among children in the ASD group, PAE-FT was associated with SCI deficits and early learning delays depending on the number of alcoholic drinks consumed per week. The median SRS SCI score of children in the ASD group whose mothers consumed 3 or more drinks per week during their first trimester was 5.13 points higher than that of children whose mothers abstained from alcohol throughout the entire pregnancy (Table 3). The median MSEL ELC score of children whose mothers consumed 1–2 drinks per week during their first trimester was 6.48 points lower than that of children whose mothers abstained from alcohol during pregnancy. PAE-FT was not associated with SCI deficits or early learning delays among children in the POP group.

We used a large community-based sample with standardized data collection to examine associations between PAE-FT and measures of early learning ability, as well as severity of behavior problems and autistic traits in children aged 2–5 years, with and without ASD. Results revealed that PAE-FT of 3 or more alcoholic drinks per week was associated with an increase in externalizing behaviors in both groups of children. This finding aligns with previous research on the relationship between FASDs and ASD in children (Carpita et al. 2022). It also aligns with previous studies involving children or adolescents from diverse populations (Khoury et al. 2018; Paley et al. 2005; Sood et al. 2001) and suggests that PAE-FT at this level can disrupt neurodevelopment in children with and without clinically significant manifestations of ASD. Healthcare providers can use this finding in support of educating women on the potential harms of PAE-FT to the developing fetus.

Our findings also revealed that PAE-FT, with an average consumption of 1–2 drinks per week, was associated with a decrease in early learning ability among children with ASD, but not those without ASD. However, we did not observe the same association for those who consumed 3 or more drinks per week. This finding could be due to a high level of variability in cognitive scores among children with ASD whose mothers consumed 3 or more drinks, as shown by the wide confidence interval for the median (ranging from 51 to 80, with a median of 58). Additionally, the number of children in this group was small. Nonetheless, children with ASD exposed to as few as one alcoholic drink per week during the first trimester could benefit from being monitored for early learning delays and provided educational support, if needed.

PAE-FT of 3 or more alcoholic drinks per week was associated with increased SCI impairment among children with ASD, but not among children without ASD. Genetic factors play critical roles in ASD, and researchers have identified hundreds of genes that contribute to the significant deficits in communication, social cognition, and behavior commonly observed in individuals with ASD (Qiu et al. 2022; Gonatopoulos-Pournatzis et al. 2020; Rylaarsdam and Guemez-Gamboa 2019). Alcohol-induced modifications in epigenetic processes have been considered to underlie the enduring effects of alcohol and the passing down of autistic traits across generations (Varadinova and Boyadjieva 2015). Our findings, along with others, suggest that in addition to PAE neurodevelopmental effects, such exposure might interact with underlying genetic factors, possibly exacerbating SCI impairment in children with ASD. This finding has important implications for future research investigating the link between ASD and FASD and supporting pregnant mothers with genetic risk for ASD in their children.

Conversely, we found no significant associations of PAE-FT with early learning delays and SCI impairment among those without ASD in the population comparison group. A recent meta-analysis of six prospective cohorts (Jacobson et al. 2024) examining the effects of PAE on cognitive function found no evidence of PAE effects among the offspring of women who drank lightly (~1 drink per occasion) during pregnancy. This lack of effects at low exposure levels aligns with our findings of no association with early learning ability among children without ASD. However, the meta-analysis observed relatively strong effects in the offspring of mothers who consumed moderate (~2–3 drinks per occasion) or heavy (~3 or more drinks per occasion) amounts, which conflicts with our findings. Differences between the meta-analysis and our study may largely be attributed to confounding factors and differences in sample sizes. The meta-analysis included six cohorts with participants of unknown ASD status, whereas our study specifically examined children without ASD in the population comparison group. This allowed us to isolate the effects of PAE without the confounding influence of ASD-related risk factors, as the absence of genetic risk factors for ASD in the POP group may have reduced susceptibility to alcohol-induced genetic modifications. Additionally, by pooling data across six cohorts, the meta-analysis had larger numbers of participants in the moderate and heavy drinking categories, yielding greater statistical power. In contrast, our study had smaller sample sizes for individuals consuming an average of 1–2 drinks or 3+ drinks per week during the first trimester. These factors may also explain why we observed no association between PAE-FT and early learning ability in our non-ASD group, despite significant associations in the ASD group. Similarly, they may account for why an association between PAE-FT and SCI were observed in children with ASD but not in those without ASD.

The absence of an association between PAE-FT and the severity of RRB in children with or without ASD is congruent with findings from prior research, even though these earlier studies employed a binary measure (ASD versus non-ASD) as their outcome [Eliasen et al. 2010; Gallagher et al. 2018]. We also did not find an association between PAE-FT and internalizing behavior, as has been previously observed (Sood et al. 2001; Ichikawa et al. 2018). However, it is possible that the age of SEED participants limits the assessment of internalizing behaviors, like anxiety and depression, that are more likely to develop in school-aged children or adolescents (Liu et al. 2011).

This analysis could be limited by unmeasured confounding, exposure misclassification, and selection bias. First, although the SEED caregiver interview asked about many different health conditions and other health-related questions, we could not account for genetic, dietary, and other maternal and paternal characteristics that might potentially confound the associations. Second, we relied on retrospective reports of alcohol consumption collected 2–5 years after the birth of the child; reporting and recall errors may have led to misclassification of the timing and/or amount of alcohol consumption. Because of uncertainty related to the timing of conception, it is possible that some women reporting only preconceptional alcohol use may have actually consumed alcohol during the first trimester of pregnancy but may have been considered unexposed in this analysis. This analysis also used broad categories for quantifying alcohol consumption due to the limited sample size of mothers consuming 3 or more alcoholic drinks per week, on average, which may not have captured the associations with higher levels of alcohol or with varying levels of consumption per episode of drinking. Finally, selection bias may be a concern. Analyses from one SEED site found non-response to be associated with younger maternal age at birth, lower maternal education, and non-white race (Schieve, Harris, et al. 2018). Further, other lifestyle and health-related behaviors of participants who agreed to participate in SEED’s intensive research protocol may differ from those not represented in this study, which could—result in a sample that may not fully represent all children with PAE, with ASD, or without ASD, and might influence the generalizability of our findings.

This study has several strengths. First, we used data from a large case–control study that employed standardized instruments to evaluate and classify children with and without ASD. The data also included detailed information on alcohol use and covariates (Schendel et al. 2012). Second, we considered the heterogeneity of ASD and expanded previous analyses by examining continuous measures of early learning ability, behavior problems, and autistic traits in both children with and without ASD. This approach allowed for a more comprehensive evaluation of the relationship between PAE-FT and specific neurodevelopmental outcomes within each group, as well as identifying variations in associations between the two groups. Also, this study employed quantile regression to analyze the association between various levels of PAE-FT and continuous outcome measures. This method is more flexible in modeling non-normal distributions and is more robust against outliers compared to conventional linear regression. Additionally, alcohol exposure during the second or third trimester was excluded when assessing the associations between PAE-FT and the outcomes. By doing so, the comparison group consisted solely of mothers who abstained from alcohol throughout pregnancy, thus avoiding potential confounding effects of PAE during later stages of pregnancy.

In conclusion, we found that exposure to 3 or more alcoholic drinks per week during the first trimester of pregnancy was associated with increased externalizing behaviors in children with and without ASD. Exposure to 1–2 alcoholic drinks per week during the first trimester of pregnancy was associated with early learning delays in children with ASD, but not those without ASD. Additionally, first trimester prenatal exposure to 3 or more alcoholic drinks per week was associated with increased SCI deficits in children with ASD, but not those without ASD. These findings suggest that PAE-FT may adversely affect the behavioral development of children regardless of ASD status and may exacerbate autism-specific developmental problems and learning difficulties in children with ASD. These results are consistent with recommendations that women of reproductive age avoid alcohol entirely when pregnant, attempting to conceive, or could become pregnant (CDC 2016; ACOG 2011). Additionally, these data also point to the importance of obtaining a PAE history for all children (AAP 2024), including children with any neurodevelopmental disorder. This information can contribute to a better understanding of developmental trajectories, thereby aiding in informed decisions regarding interventions and support tailored to each child’s needs.

Additional supporting information can be found online in the Supporting Information section.