Among U.S. women, self-reported past-month cannabis use during the first trimester of pregnancy, when fetuses may be most susceptible to harm, increased from 5.6% in 2002–2005 to 8.1% in 2014–2017, with the highest use in the first month, before many women are aware of their pregnancy (Alshaarawy & Anthony, 2019). In 2017, 9.8% of recently delivered U.S. women reported using cannabis in the 3 months before pregnancy and 4.2% during pregnancy (Ko et al., 2020). Women may continue using cannabis during pregnancy to treat pregnancy-related symptoms (Bayrampour et al., 2019) or because they perceive little risk from use (Bayrampour et al., 2019; Jarlenski et al., 2017; Westfall et al., 2006).

The endocannabinoid system plays an important role in fetal brain development (Helliwell et al., 2004; Richardson et al., 2016). Cannabinoid receptors are widespread in the fetal cerebral cortex, hippocampus, and basal ganglia (Jutras-Aswad et al., 2009). Further, Δ9-tetrahydrocannabinol (THC) and its metabolites freely pass the placental and fetal blood–brain barriers (Little & VanBeveren, 1996). There is concern that perinatal exposure to exogenous cannabinoids could permanently alter neurodevelopmental processes in the offspring. In some animal models, prenatal cannabis exposure results in persistent adverse changes in cognitive performance, behavior, and stress response (Roncero et al., 2020). Comprehensive reviews of human studies have not identified consistent effects on global developmental level (Grant et al., 2018; Roncero et al., 2020; Warner et al., 2014). However, longitudinal human studies have reported associations between prenatal cannabis exposure and behavioral regulation deficits in the offspring, including increased externalizing behavior, impulsivity, hyperactivity, and delinquency, starting at ages 3–4 years and continuing into adolescence (Grant et al., 2018; Huizink, 2014; Metz & Borgelt, 2018; National Academies of Sciences, 2017; Richardson et al., 2016; Roncero et al., 2020). Some studies have documented increased anxiety and depressive symptoms in the offspring associated with first trimester cannabis use (Day et al., 2011; Gray et al., 2005; Leech et al., 2006). Disturbed sleep patterns during infancy and childhood have also been reported with prenatal cannabis exposure (Dahl et al., 1995; Day & Richardson, 1991).

One large retrospective birth registry study found that prenatal cannabis use recorded at the first prenatal visit (median 79th gestational day), versus no exposure, was associated with a significantly increased likelihood of being diagnosed subsequently (median follow-up 7.4 years) with autism spectrum disorder (ASD) (Corsi et al., 2020) although a smaller case-control study of children ages 3–5 found no association of ASD with cannabis use in the 3 months before pregnancy, during pregnancy, or in any trimester (DiGuiseppi et al., 2022). Persons with ASD commonly demonstrate deficits similar to those observed with prenatal cannabis use (as described above), such as attention deficits, anxiety, depression, and sleep disturbances, as well as intellectual disability (reviewed in Lyall et al., 2017). Associations of prenatal cannabis use with cognitive and behavioral outcomes might differ in children with symptoms of ASD compared to children without such symptoms, due to several possible mechanisms. Peripregnancy cannabis use could alter epigenetic processes relevant to neurodevelopment (Smith et al., 2020; Vassoler et al., 2014) that could influence the occurrence of both ASD symptoms (Smith et al., 2020) and behavior problems, or amplify the severity of behavior problems that often co-occur with ASD (Bacherini et al., 2021; Kanne & Mazurek, 2011; Mazurek et al., 2013; Richdale & Schreck, 2009; Wiggins, Levy, et al., 2015). Another potential mechanism is maternal psychopathology, which is associated with cannabis use (Defoe et al., 2019; Galéra et al., 2010; Galéra et al., 2013; Jacobus et al., 2016; Pedersen et al., 2001; Pingault et al., 2013), and is also associated with having offspring with ASD (e.g., Chen et al., 2020; Chien et al., 2022; Jokiranta et al., 2013) and with more severe behavioral problems in their children with ASD (Baker et al., 2011; Ekas & Whitman, 2010; Firat et al., 2002; Machado Junior et al., 2016). Another possible mechanism is the resumption or persistence of maternal cannabis use in the post-partum period, which can adversely affect maternal caregiving (Dunn et al., 2002; Neger & Prinz, 2015) and, when combined with parental stress associated with parenting a child with symptoms of ASD (Crowell et al., 2019), might intensify problem behaviors in children with ASD symptoms but have less impact on those without such symptoms.

The aim of this cross-sectional analysis of data from the Study to Explore Early Development (SEED) was to examine and report effect modification by ASD symptoms of the associations of maternal cannabis use shortly prior to conception or during pregnancy with child behaviors and cognitive functioning.

There were 6256 children who completed the clinic visit and were classifiable into one of two study groups (i.e., children with and without ASD symptoms). Mothers of 6188 (98.9%) children supplied information on peripregnancy cannabis use and 6182 (98.8%) provided data on use by time period (i.e., preconception and trimester). Among those with information on peripregnancy cannabis use, CBCL data were complete on 5998 (96.9%) children and MSEL data were complete on 6090 (98.4%) children.

Mothers of 322 (5.2%) children reported any peripregnancy cannabis use, while 174 (2.8%) reported use during pregnancy. Of mothers of children with ASD symptoms, 164 (6.0%) reported any peripregnancy cannabis use versus 158 (4.6%) mothers of children without ASD symptoms (p = 0.012) (Table 1). Among mothers reporting peripregnancy cannabis use, use declined markedly over the course of pregnancy, from 94% in the pre-conception period to 15% in the third trimester, with similar declines in use by trimester in both study groups (Table 2).

Table 1 shows sociodemographic characteristics and maternal smoking, alcohol and other illicit drug use during pregnancy for each study group. Children with and without ASD symptoms differed statistically significantly (p < 0.001) for all sociodemographics and maternal substance use during pregnancy except child age (p = 0.181) and other illicit drug use during pregnancy (p = 0.408).

Table 1 shows mean CBCL t-scores for each syndrome scale and mean MSEL domain scores by presence or absence of ASD symptoms. Relationships of maternal cannabis use with child emotional and behavioral problems and cognitive functioning among children with and without ASD symptoms are presented below and in Figures 1 and 2 and Table 3. Figures 1 and 2 show adjusted mean differences (AMD) in CBCL and MSEL t-scores, respectively, between children whose mothers reported use in the specified time period versus no use, by presence or absence of ASD symptoms. Table 3 shows adjusted odds of having clinical/borderline CBCL syndrome scale t-scores (i.e., ≥ 65) and below average MSEL domain t-scores (i.e., <40), between children whose mothers reported use in the specified time period versus no use, in children without and with ASD symptoms.

Although a previous analysis of SEED data did not find an association between peripregnancy cannabis use and ASD (DiGuiseppi et al., 2022), in our current analysis we observed that associations of maternal cannabis use during peripregnancy with selected child emotional and behavioral problems varied in the presence of ASD symptoms. We did not observe associations of peripregnancy cannabis use with developmental indicators, regardless of ASD symptoms.

Among children with ASD symptoms, aggressive behavior problems t-scores were on average more than 5 points higher, and the odds of having clinical/borderline aggressive behavior problems t-scores was twice as likely, in offspring of mothers who used cannabis only during preconception compared to offspring of mothers who did not use cannabis. In a Dutch population-based birth cohort, cannabis use only before pregnancy was similarly associated with significantly increased externalizing (but not internalizing) problems at age 8 years (El Marroun et al., 2019). In contrast, we did not observe an association between preconception-only cannabis use and aggressive behavior problems in offspring without ASD symptoms. We also found that maternal preconception-only cannabis use was associated with sleep problem t-scores that were nearly 3 points higher, and emotionally reactive t-scores that were more than two points higher, compared to no maternal cannabis use, but only in children with ASD symptoms.

Several potential explanations for the associations of preconception-only cannabis use with more aggressive behavior, emotional reactivity, and sleep problems in children with ASD symptoms but not children without such symptoms may warrant further research. As described in Sajdeya et al. (2021), preconception cannabis use could potentially lead to maternal hormonal, immunological, neurological, or genetic changes that interfere with neurodevelopment of the fetus. Animal evidence suggests that preconception cannabis exposure alters epigenetic processes, potentially resulting in heritable changes in genes and molecular pathways relevant to neurodevelopment (Smith et al., 2020; Vassoler et al., 2014). Such changes may mediate a relationship between preconception exposure to cannabis and occurrence of ASD (Smith et al., 2020), and may similarly mediate a relationship between preconception cannabis exposure and behavior problems.

Preconception cannabis exposure might also act to amplify regulatory and behavior problems that often co-occur with ASD. Although not primary features of ASD, aggressive behavior, emotional reactivity, and sleep problems are common in children with ASD (Bacherini et al., 2021; Kanne & Mazurek, 2011; Mazurek et al., 2013; Richdale & Schreck, 2009; Wiggins, Levy, et al., 2015). Epigenetic modifications (Smith et al., 2020), or maternal distress in early pregnancy (Weinstock, 2008) associated with cannabis withdrawal syndrome (Bahji et al., 2020; Graves et al., 2022), might mediate a relationship between preconception exposure to cannabis and increased severity of symptoms that co-occur with ASD, resulting in children with ASD symptoms having on average higher behavior problem scores than children with ASD symptoms who were not exposed to cannabis. In contrast, among children without ASD symptoms, such amplification may not have been observed because co-occurring behavior problems were less common or milder to start with.

Rather than direct or indirect effects of preconception cannabis exposure on neurodevelopment, our findings could be explained if a common underlying etiology results in mothers both using cannabis and having children with problem behaviors and ASD symptoms. Maternal psychopathology is potentially one such etiology. Having externalizing behavior and conduct problems increases the likelihood of early initiation and use of cannabis, which is in turn associated with impaired mental health and problematic cannabis use (Defoe et al., 2019; Galéra et al., 2010; Galéra et al., 2013; Jacobus et al., 2016; Pedersen et al., 2001; Pingault et al., 2013). Aggressive behavior and antisocial problems in the mother also predict externalizing behaviors in the offspring (Davies et al., 2012) and are associated with maternal substance use problems, affective disorders, and low socioeconomic status, which can further contribute to child behavior problems (Antunez et al., 2018; Davies et al., 2012; Emery et al., 1999). The association between depression and substance use is also well-established (Lev-Ran et al., 2013), which may reflect shared neurocognitive mechanisms, common underlying risk factors, or self-medication of depressive symptoms with cannabis (Kuhns et al., 2022). Persistent depressive symptoms in the mother have been associated with greater emotional and behavioral difficulties in the offspring at age 5 years (van der Waerden et al., 2015). Parents with psychiatric diagnoses are also more likely to have offspring with ASD (e.g., Chen et al., 2020; Chien et al., 2022; Jokiranta et al., 2013), and their offspring with ASD are more likely to have more severe behavioral problems (Baker et al., 2011; Ekas & Whitman, 2010; Firat et al., 2002; Machado Junior et al., 2016). Thus, maternal psychopathology may underlie greater likelihood of maternal cannabis use, ASD symptoms and behavioral problems in the offspring. Adjustment for mother having ever been diagnosed with depression did not materially change our results. However, we had no measures of current depression symptoms or symptom severity, nor of maternal aggressive behavior or antisocial problems.

Another possible mechanism is effects of maternal cannabis use in early childhood on caregiving. Many women who achieve abstinence during pregnancy will resume cannabis use post-partum. Preconception cannabis use may thus be a marker for maternal cannabis use in infancy and early childhood (De Genna et al., 2022; Fried et al., 1985; Thomson et al., 2023). Some studies have shown that parents with substance abuse spend less time interacting with their children, are less satisfied with their parental relationships, have problems with attachment to the child, and exhibit low bonding behaviors starting in infancy (reviewed in Dunn et al., 2002; Neger & Prinz, 2015). Negative effects of maternal cannabis use during infancy and early childhood on caregiving and attachment, when combined with strains on parent–child interactions and parental stress associated with parenting a child with ASD (Crowell et al., 2019), may exacerbate problem behaviors in the child, compared to effects of maternal cannabis use on caregiving of children without ASD symptoms.

It is possible that preconception-only cannabis use is not associated with child behavior differently in children with ASD symptoms, but rather that mothers who use cannabis preconception differentially report both ASD symptoms and problem behaviors in their offspring. Whether cannabis use specifically, or substance use in general, leads to inaccurate reporting of emotional or behavioral problems in children has not been established (Chilcoat & Breslau, 1997; Olino et al., 2020). However, studies have demonstrated over-reporting of externalizing, and in some studies, internalizing, problem behaviors by parents experiencing depression, anxiety, or stress, with the strongest, most consistent evidence for parents with depressive symptoms (reviewed in De Los Reyes and Kazdin (2005)). Mental health disorders are associated with substance use, as noted above, raising the possibility that over-reporting could account for the associations of maternal cannabis use with child emotional and behavioral symptoms observed in our study. Systematic reviews have demonstrated high proportions of depressive, anxiety and obsessive-compulsive disorders diagnosed in mothers who have a child with ASD (Schnabel et al., 2020) and higher levels of stress among parents of children with ASD (Hayes & Watson, 2013), which could have similarly led mothers of children with ASD symptoms to over-report their child’s behavior problems compared to mothers of children without ASD symptoms. Thus, we might have found stronger associations of pre-conception cannabis use with behavior problems in children with ASD symptoms compared to those without ASD symptoms due to inaccurate (i.e., over) reporting.

Mothers of children with both ASD symptoms and problem behaviors may also differentially recall or report their peripregnancy cannabis use. Mothers of offspring with ASD symptoms may be more likely to accurately recall cannabis use around the time of pregnancy than mothers of typically developing children, for example, if mothers of children with ASD symptoms have already thought carefully about their past exposure history in an effort to understand why their child might have developed ASD symptoms (i.e., recall bias). Since children with ASD symptoms are also more likely to have behavioral problems, this could result in observing an association between cannabis use and behavior problems only in children with ASD symptoms.

This study has several potential limitations. As noted above, maternal cannabis use was self-reported. While self-reported prenatal cannabis use collected 1 year after delivery correlates moderately well with data from antenatal interviews (Jacobson et al., 2002), recall may be less accurate 4–5 years later. Women may have underreported cannabis use during pregnancy due to social stigma associated with substance use while pregnant. In addition, some SEED sites were in states without legalized cannabis use during data collection periods; perceived legal ramifications may have contributed to under-reporting. Women who used cannabis during pregnancy and had offspring with ASD may be particularly reluctant to report its use due to feelings of guilt and blame regarding their child’s diagnosis (Elder, 1994; Gray, 1995; Mercer et al., 2006). We lacked information on route of ingestion, dose, or frequency of cannabis use. We included broad measures of deficits via the CBCL and MSEL but were unable to examine more nuanced indicators of executive functioning deficits. Although our sample included more than 6000 participants, relatively few mothers reported peripregnancy cannabis use, limiting power for some analyses. Sample reductions after stratification by exposure period precluded adjusting for a wider array of confounding variables, although we were able to adjust for indicators of socioeconomic status, other substance use, time trends (i.e., SEED phase), and in supplementary analyses, depression. Lifestyle and health-related behaviors of participants who agreed to take part in SEED’s intensive research protocol may differ from those not represented in this study, affecting generalizability of our findings.

Study strengths include research-reliable administration of standardized instruments to evaluate behavior and cognitive development and to classify children with and without ASD symptoms, identification and inclusion of children not previously diagnosed with autism (perhaps reflecting milder symptoms), and ability to adjust for multiple covariates associated with cannabis use and child behavior (Schendel et al., 2012).

With rising cannabis use among pregnant women, our findings support the need for further research about its potential effects on the offspring. Future studies can examine a range of developmental risks associated with the timing and patterns of cannabis use prior to conception as well as during pregnancy. Another promising direction is the exploration of potential explanations for emotional and behavioral problems associated with maternal preconception cannabis use among children with ASD symptoms but not without ASD symptoms, for example, examining interactions between underlying genetic or epigenetic risk and cannabis exposure.