Behavioral and Cognitive Differences in Early Childhood related to Prenatal Marijuana Exposure

Abstract

Prenatal marijuana exposure (PME) negatively impacts child development and behavior; however, few studies have examined these associations at early ages among children exposed to today’s highly potent marijuana. Using a prospective prenatal cohort (Columbus, Ohio, USA), PME was determined from maternal self-report, medical chart abstraction, and urine toxicology from prenatal visits and delivery. At age 3.5 years, 63 offspring children completed tasks assessing executive function (EF), visual spatial ability, emotion regulation, and aggressive behavior. Caregivers reported on children’s EF and problem behaviors. Logistic regressions and analyses of covariance controlling for key variables were used to examine associations between PME and child outcomes. Compared to non-exposed children, children with PME had more sleep-related problems, withdrawal symptoms, and externalizing problems, including aggressive behaviors and oppositional defiant behaviors. Children with and without PME did not differ in terms of executive functioning. Findings suggest behavioral problems associated with PME may manifest by age 3.5.

Marijuana is the most commonly used illicit drug in the United States, especially among young adults between the ages of 18 to 25 (SAMHSA, 2017). A particular demographic of concern is pregnant women, among whom the rate of marijuana use is rising steadily (Volkow et al., 2019; Young-Wolff et al., 2017). This is problematic given that psychoactive cannabinoids within marijuana products, such as Δ9-tetrahydrocannabinol (THC), can readily pass through the placental and blood-brain barrier, ultimately providing a mechanism for prenatal marijuana exposure (PME) to influence fetal development (Blackard & Tennes, 1984; Gaoni & Machoulam, 1964; Gomez et al., 2003; Little & Van Beveren, 1996). Findings from a limited number of studies suggest PME may negatively impact children. Specifically, PME is associated with deficits in executive functioning, which encompasses cognitive processes critical to the development of novel-problem solving skills (Fried & Smith, 2001), and increased behavioral problems across childhood and adolescence (Day et a., 2006; El-Marroun et al., 2011; Goldschmidt et al., 2000).

Executive functioning (EF) refers to a series of dissociable but interrelated adaptive, goal-directed cognitive processes that allow children to override more automatic or established thoughts and responses (Garon et al., 2008; Lezak, 1995; Mesulam, 2002). These processes include working memory (temporary storage and manipulation of information), cognitive flexibility, and inhibitory control (Lezak, 1995; Miyake et al., 2000; Weintraub et al., 2013). These higher order cognitive processes critically influence children’s problem-solving and ability to perform in school, relate interpersonally, and engage with the world around them. The prefrontal cortex is the region of the brain primarily associated with EF (Durston et al., 2006; Struss & Alexander, 2000; Struss & Benson, 1984). Research with rats and humans suggest cannabinoid receptors that interact with and are altered by exposure to THC are present in the pre-frontal cortex and other regions of the brain (Gray et al., 2005; Saez et al., 2014). These receptors have central roles in the development of the brain, cognitive functioning, executive functioning, control of a multitude of behavioral functions, and are present as early as 14 weeks in the human fetal brain (Biegon & Kerman, 2001; Park et al., 2003; Trezza et al., 2008; Wang et al., 2004). Additionally, the density of receptor bindings is greater in fetal brains compared to adult brains, which furthers concerns about the consequences related to prenatal exposure to marijuana. These concerns are further supported by structural and functional neuroimaging that reflect differences in development and functioning of the frontal part of the brain between children with and without PME (El-Marroun et al., 2016; Smith et al., 2006; Smith et al., 2010). In sum, changes to the cannabinoid system functioning influenced by prenatal exposure have the potential to explain disparities in EF observed between children with and without PME. Prior findings converge on PME negatively impacting children’s cognitive skills associated with EF as well as skills that engage EF. Specifically, pre-school and school-aged children with PME demonstrate worse short-term memory and inhibitory control, visual-spatial ability, and verbal and quantitative reasoning compared to their non-exposed counterparts (Day et al., 1994; Fried & Watkinson, 1990; Goldschmidt et al., 2008; Griffith et al., 1994; Leech et al., 1999). In contrast, a lack of relationship between IQ and PME is consistently documented (Day et al., 1994; Fried et al., 1992; Fried & Watkinson, 1990; Griffith et al., 1994). Many IQ tests place more emphasis on established knowledge and learned information rather than aspects of EF that constitute more fluid abilities. As a result, general cognitive assessments are often regarded as lacking the nuance to detect effects of PME, and assessments of EF may prove more sensitive to differentiating between exposed and non-exposed children (Fried, 1991). Taken together, the earlier research suggests PME is linked to deficits in EF processes as opposed to global cognitive abilities, often measured by IQ.

Assessments of children’s behavioral problems also differentiate exposed from non-exposed children. Children with PME are more irritable at birth and more likely to display sleep disturbances as neonates and toddlers (Dahl et al., 1989; Fried & Makin, 1987; Scher et al., 1988). Some evidence suggests children with PME are more aggressive as early as 18- months of age (El-Marroun et al., 2011); however, this relationship was only detected among girls. PME has also been linked to higher rates of conduct problems among school-aged children (ages 6-9) (O’Connell & Fried, 1991), as well as to increased hyperactivity, impulsivity, delinquency, inattention, other externalizing problems, and depression as early as age 10 (Goldschmidt et al., 2000; Gray et al., 2005). As mentioned, the effects of PME on the cannabinoid system and its role within the development of the pre-frontal cortex and other areas of the brain stand to influence behaviors in children (Farrington et al., 2017). Behavioral problems associated with PME may also be partially explained by altered activity within dopamine, endorphin, and serotonin systems associated with alteration in emotion regulation that can precipitate behavioral problems (Gray et al., 2005; Hurd et al., 2005; Saez et al., 2014). In sum, findings suggest PME negatively impacts children across multiple aspects of behavioral development; however, additional evidence is needed to confirm and explore potential neurobiological pathways for which PME impacts aggression and other behavioral problems.

Despite evidence of the negative effects of PME, significant knowledge gaps persist. First, prior studies among children ages 3-4 focused on the assessment of general cognitive processes but did not directly focus on the assessment of EF. This age window is critical as it represents the time during which specific components of EF begin to emerge (Anderson, 2010; Diamond & Taylor, 1996; Espy, 1997; Welsh et al., 1991). As a result, our understanding of how PME may impact EF in pre-school age children remains limited. Further, prior studies did not assess maternal/caregiver executive functioning, a strong predictor of both maternal substance use (Lee et al., 2012; Squeglia et al., 2014) and child executive functioning (Calkins, 2011; Cuevas et al., 2014), limiting their capacity to tease out the effects of PME. Additionally, prior studies may have failed to detect or underestimated effects both because they relied on maternal self-report of marijuana use during pregnancy and because marijuana is five times more potent now than it was 30 years ago (Chandra et al., 2019; ElSohly et al., 2016).

The current study used data from a larger, longitudinal cohort study designed to address the gaps noted above by examining group differences among children prenatally exposed to marijuana among a cohort of children between 3.5-7 years of age. In the larger study, prenatal exposure was assessed using maternal self-report, medical record abstraction, and urine toxicology during pregnancy. The current study represents an exploratory investigation of exposure group differences using data solely from the first study visit (3.5 years of age). Specifically, the study examined differences between children with and without PME in terms of aggression and other behavioral problems as well as aspects of EF. We hypothesized that children with PME would demonstrate poorer EF and more aggressive behaviors than their non-exposed peers.

Methods

Participants

The current report used data from the Lifestyle and Early Achievement in Families (LEAF) study, an ongoing longitudinal study examining the effects of PME on children’s EF and aggressive behaviors during early childhood (Klebanoff et al., 2020). While LEAF consists of assessments when children are age 3.5, 5, and 7 years of age, the current study reports on only the first postnatal data collection point, which occurred when children were 3.5 years of age. Given that the research questions pertain to EF and behavior during early childhood when EF is thought to emerge (age 3-4.5), only the data collection from the age 3.5 visit was utilized. The LEAF study recruited women who previously enrolled in the Ohio Perinatal Research Network’s Perinatal Research Repository (PRR; Klebanoff et al., 2021; Moorehead et al., 2015) and consented to be contacted for future research. The PRR enrolled women during their pregnancy at The Ohio State University Wexner Medical Center (OSUWMC, Columbus OH) beginning in 2010. OSUWMC operates a large delivery service for both high- and low-risk obstetric patients. Participation in the PRR ended at newborn discharge. The LEAF study design and methods have been published previously (Klebanoff et al., 2020).

LEAF Study Procedures

Potentially eligible families were re-contacted via phone calls during which the LEAF study was described and a brief screening questionnaire was administered. Families were eligible if their child was within the age window (42 months – 47 months, 29 days). Families were excluded if their child a) had any cognitive or physical impairment severe enough to preclude engagement in any study task, b) was known to be a ward of the state, or 3) was known to be deceased. In-person visits lasting approximately 2-3 hours were conducted with families who agreed to participate in the LEAF study. If biological mothers were unable to participate with their children due to loss of custody, current legal guardians were invited to participate. The current report is based on maternal substance use data from the PRR, combined with family demographics, child outcome measures, and maternal/caregiver characteristics and reports on child behavior data from the age 3.5-year old visit of the LEAF study.

Prenatal Marijuana and Tobacco Exposure

As part of the PRR, women self-reported substance use during pregnancy, and drug use information was abstracted from their obstetric medical record (Klebanoff et al., 2021). Women also provided one or more urine samples depending on a number of prenatal visits (up to 3 during the pregnancy). The samples were frozen, archived, and analyzed for metabolites of marijuana and other drugs in preparation for LEAF. Women were considered to have used marijuana during their pregnancy if they self-reported use, or if use was noted in the medical record, or if any urine specimen during their pregnancy had a 11-nor-carboxy-Δ⁹-tetrahydrocannabinol (Δ9-THC-COOH) concentration of >15 ng/ml, which is the concentration considered by Substance Abuse and Mental Health Services Administration to represent active use when employing mass spectrometry (SAMHSA, 2012). Children were classified as having or not having PME based on these criteria. Women’s tobacco use during pregnancy was assessed via maternal self-report (Klebanoff et al., 2001) and by notation of tobacco use in the obstetrical record (Klebanoff et al., 2021). Children were classified as prenatally exposed or non-exposed to tobacco if they met either criterion. We have previously demonstrated that in prospective pregnancy cohorts not specifically focused on tobacco use, self-report is sufficiently accurate for classifying women as active tobacco users (Klebanoff et al., 1998; 2001).

Socio-Demographic Characteristics

A brief demographic questionnaire was administered to participating mothers/caregivers during the 3.5-age visit. This questionnaire collected the following information used to characterize the sample and compare PME and non-PME groups: caregiver education, household income, employment, marital status, race, sex, and age.

Outcome Variables

Child Global Cognitive Ability and Executive Functioning

Children’s cognitive abilities, including emerging executive function, were assessed using both direct testing of children’s abilities and caregiver report. The NIH-Toolbox (NIH-TB) Early Childhood Cognition Battery was administered to participating children on an iPad to assess multiple domains of cognitive ability: inhibitory control and attention (Flanker), episodic memory (Picture Sequence Memory), cognitive flexibility (Dimension Change Card Sorting), receptive vocabulary (Picture Vocabulary), and processing speed (Pattern Comparison Speed Test). The NIH-TB took roughly 30 minutes for children to complete and was administered first to limit fatigue. Age-corrected standard scores were calculated for each sub-scale and the Early Childhood Composite score, per prescribed NIH-TB scoring procedures and national norms (Weintraub et al., 2013). Children also completed the WPPSI-IV Picture Memory subtest (Raiford et al., 2014), where children are shown a page with a series of pictures, then the page is flipped and children are asked to correctly select which pictures they saw on the previous page. The measure yields standardized age-adjusted scaled scores (1-19). Higher scores represent better performance on this task. The third measure of EF, the Behavior Rating Inventory of Executive Functioning-Preschool Version (BRIEF-P; Gioia et al., 2003) is a caregiver report measure that yields several t-scores for subscales related to parental perception of child EF: (i.e., inhibit, shift, emotional control, working memory, planning/organization) as well as three broader indexes (Inhibitory Self-Control, Flexibility, and Emergent Metacognition) and a composite Global Executive Functioning score. Higher scores indicate worse EF.

Child Visual Spatial Ability

To assess visual spatial abilities on tasks that also engage EF, such as planning and organizational skills, children were administered a puzzle task from the Bayley Scales of Infant and Toddler Development - III (Bayley-III) cognitive assessment (Bayley, 2005) and the Block Design test from the Weschler Preschool & Primary Scale of Intelligence test (WPPSI-III; Wechsler, 2004). From the Bayley-III, children completed a three-piece puzzle (with up to two trials) and were scored based on whether they completed the puzzle or not, and whether they completed the puzzle on the first or second trial (0=unable to complete; 1=completed on Trial 2; 2=completed on Trial 1). On the WPPSI-III Block Design subtest, children are presented up to thirteen, timed block designs and asked to construct the same design with a separate set of blocks. The measure yields a standardized age-adjusted score (1-19).

Child Planning Abilities

A disk transfer task, the Tower of Hanoi (TOH; Bull et al., 2004; Klahr & Robinson, 1981), was administered to assess children’s problem-solving and planning abilities. Children completed three practice problems and up to six test problems, each having a maximum of two trials. Test problems increased in difficulty with the first problem requiring a minimum of two moves and the sixth problem requiring a minimum of seven moves to achieve the desired goal state. Children failed an item if they broke a rule or did not achieve the goal state within 20 moves. The task was discontinued if the child failed two practice problems and the first test problem or failed two consecutive test problems. Test problems were assigned a point value derived from the minimum number of moves to solve a problem (i.e., a point value of 2 for problem 1 and a point value of 7 for problem 6). If a child correctly solved a problem on Trial 1, regardless of the number of moves implemented, they received the full point value for that item (i.e., 2 points for problem 1). If a child failed Trial 1, a second trial was administered. If the child correctly solved a problem on Trial 2, regardless of the number of moves implemented, they received half the full value for that item (i.e., 1 point for problem 1). The total number of points per item were summed to create an overall score ranging from 0 – 27, with higher scores indicating better planning abilities.

Child Emotion Regulation

The Toy Behind Barrier task (Berry et al., 2019; Williford et al., 2007) was administered to assess children’s emotion regulation. During this task, children play with a toy for 90 seconds, after which the toy is removed and placed into a locked, transparent box for 90 seconds and the child’s reactions are coded for various emotion regulation behaviors: seeking help, physical negativity, verbal negativity, global frustration and global regulation. These scales are scored on a 1-5 rating scale with lower numbers representing less of a given behavior. To ensure inter-coder reliability, 20% of tasks were double coded and inter-rater reliability (via K-alphas; Hayes & Krippendorff, 2007; Krippendorff, 2004) ranged from .80 to 1.00 across scales.

Child Aggressive Behavior

Children’s aggression was measured using a Bobo Interaction Task (Bandura et al., 1961; Bendersky et al., 2006), caregiver reports on the Leifer-Roberts Response Hierarchy Questionnaire Reinisch Revision (LRRHQ; Reinisch & Sanders, 1986), and the Child Behavior Checklist for ages 1.5-5 (CBCL; Achenbach et al., 2001). The Bobo activity was video-recorded and coded for various forms of interaction with the Bobo doll. In line with Bendersky et al. (2006), aggression was calculated as the percentage of fisted hits to the doll that were fisted hits to the doll’s face. This scoring approach is preferred as it reduces the confounding effects of children’s activity level and weights more aggressive hits (Bendersky et al., 2006). To limit the confounding effect of the child being off task, we coded and interpreted the first minute of the Bobo task. Children who did not hit the doll were scored as ‘0’. Participating mothers/caregivers completed the LRRHQ, which queried how they thought their child would react to hypothetical interpersonal conflictual situations. For example, “Your child is standing in line for a drink of water. A kid comes along and pushes him/her out of line. What do they do?” The child’s caregiver was then presented with a series of choices between two options. The LRRHQ measure yields four subscale scores, each score ranges from 0-18. Higher scores indicate higher frequency of engaging in a specific type of coping (i.e., withdrawal, non-aggressive coping, verbal aggression, physical aggression). The aggression subscale of the CBCL was also used to assess maternal report of child’s aggression.

Child Behavioral Problems

Child behavioral problems were assessed using caregiver report on the CBCL (Achenbach et al., 2001). In addition to aggression, the CBCL yields scales for internalizing behaviors (emotionally reactive, anxious/depressed, somatic complaints, socially withdrawn, etc.), externalizing behaviors (attention problems, aggressive behaviors), and sleep problems, which comprise a total problems scale, a stress problems scale, plus Diagnostic and Statistical Manual (DSM) oriented scales for affective problems, anxiety problems, pervasive developmental problems, attention deficit/hyperactivity problems, and oppositional defiant problems. Higher scores on the CBCL indicate more problematic behavior.

Developmental and Behavioral Diagnoses

Mother/caregivers were asked to report whether a doctor or other health care provider ever told them that the child had one or more of a list of developmental and behavioral diagnoses. These questions were adapted from the National Health Interview Survey Child Health Supplement. Responses were combined to form a binary variable indicating whether the child had been given a developmental or behavioral diagnosis (Adams et al., 1996; Barr et al., 2002; Radeos et al., 2009).

Maternal/Caregiver Cognitive Functioning

Participating caregivers also completed the NIH-TB Cognition Battery (Weintraub et al., 2013), which includes assessments of inhibitory control (Flanker), episodic memory (Picture Sequence Memory), cognitive flexibility (Dimension Change Card Sort), processing speed (Pattern Comparison Processing Speed), and working memory (List Sorting), culminating in two subscale scores for fluid and crystalized cognition, as well as a global cognitive functioning composite score. The fluid composite score was utilized as a control variable for maternal executive functioning within the model. In the five instances for which caregivers were family members other than the mother, these family members completed this testing and their fluid composite scores were used as a proxy for maternal EF. The NIH-TB typically took roughly 30-40 minutes for mothers to complete.

Statistical Analyses

Independent samples t-tests and chi-square tests were used to initially compare PME and non-PME groups on participant characteristics and on the reported outcomes, as well as to compare eligible children who were seen versus not seen for the 3.5-year assessment. ANCOVA and logistic regression models controlling for hypothesized confounders (i.e., child race, child age, child sex, prenatal tobacco exposure, maternal/caregiver marital status, household income, and maternal/caregiver executive functioning via fluid composite score on maternal NIH-TB performance) assessed the effects of PME on outcome variables. Child age was excluded as a covariate in models for which the outcome measure was an age-adjusted performance score (e.g. NIH-TH performance sub-scale scores). Adjustments for multiple comparisons were not made per published recommendations (Rothman, 1990; 2014). The possibility of Type I error inherently increases with each additional analysis; however, adjustment procedures seek to rectify this problem at the cost of Type II error, which may be more problematic (Rothman, 2014). Given the study’s design, hypotheses, and use of observational, rating, and test performance outcomes to reflect children’s cognitive abilities and behaviors, we believe that while multiple comparisons should be considered when interpreting the findings, they should not be adjusted for within analyses.

Standard Protocol Approvals, Registrations, and Patient Consents

The study protocol and procedures for the PRR and LEAF projects were reviewed and approved by the Nationwide Children’s Hospital (NCH) Institutional Review Board (IRB). All participating women provided separate written informed consent for the PRR and for themselves and their child for LEAF.

Results

Among women in the PRR, ninety-nine families were eligible for the age 3.5 LEAF visit. Sixty-three children participated in the age 3.5 visit. Five participant families did not include biological mothers and instead had child caregivers participate (one biological father and four grandmothers). Of the thirty-six families that were eligible but did not participate, we were unable to locate or contact 21, 13 refused, and two were excluded (1 due to age and 1 due to a diagnosis of Down Syndrome). Of those who participated, fifteen were classified as PME, while forty-eight were classified as non-PME. Among the 15 classified as PME, only two self-reported marijuana use (13.3%), indicating that a large portion of the PME group (86.7%) would have been misclassified as non-PME had maternal report been used in isolation. These two women also had marijuana use noted on their obstetrics record and urine screens. Five women’s use was identified solely through urine screens, two were identified solely through obstetrics record abstraction, six women were identified by their obstetrics record and urine. PME and non-PME groups did not differ in sex distribution or in prenatal exposure to tobacco. The PME group was significantly more likely to have mothers who identified as African American and not married when compared to the non-PME group (See Table 1).

Table 1.

Participant Characteristics by Prenatal Marijuana Exposure (n=63)

Characteristics	PME Group (n=15)		Non-PME Group (n=48)		p
	μ or %	SD	μ or %	SD
Child sex
Male	6 (40%)	-	20 (42%)	-	p =.91
Female	9 (60%)	-	28 (58%)	-
Child age (months)	44.5	2.1	46.3	3.4	p = .08
Child race					p = .02
African-American	12 (80%)	-	15 (31%)	-
White	0 (0%)	-	23 (47%)	-
Bi-racial	1 (6.6%)		1 (2%)
Native American/Alaskan Native	0 (0%)		1 (2%)
Other	1 (6.6%)	-	4 (8%)	-
Missing	1 (6.6%)		4 (8%)
Annual household income					p =.09
Less than $10,000	3 (20%)	-	6 (13%)	-
$10,000-$19,999	5 (33%)	-	11 (23%)	-
$20,000-$29,999	4 (27%)	-	6 (13%)	-
≥$30,000	3 (20%)	-	24 (49%)	-
Missing	0 (0%)		1 (2%)
Maternal age (years)	26.8	4.5	28.1	5.1	p =.37
Maternal/caregiver education					p =.74
<High school	1 (7%)		3 (6%)
Diploma/GED	6 (40%)		20 (42%)
Some college	6 (40%)		12 (25%)
Associate degree	2 (13%)		6 (13%)
Bachelor degree	0 (0%)		4 (8%)
Graduate degree	0 (0%)		2 (4%)
Missing	0 (0%)		1 (2%)
Maternal/Caregiver marital status					p =.007
Not married	14 (93%)		25 (52%)
Married	1 (7%)		22 (46%)
Missing	0(0%)		1 (2%)
Child exposed to tobacco prenatally	7 (47%)	-	15 (31%)	-	p =.27

Comparative analyses were used to examine potential differences between age-eligible families who participated in the age 3.5 LEAF visit (n=63) and those who did not participate (n=36). Non-participant families tended towards a higher prevalence of prenatal marijuana use compared to families who did participate (41% to 23%), but this was not statistically significant. Rates of prenatal exposure to other substances (alcohol & tobacco) did not differ among those who did and did not participate. Further, participant and non-participant families did not differ on maternal race, age, or marital status.

Child Global Cognitive Ability and Executive Functioning

Many children were unable to complete all NIH-TB tasks, as required to obtain an early childhood composite score, resulting in a large amount of missing data. Rates of completion for various NIH-TB tasks were compared across exposure groups (See Table 2). The Picture Sequence Memory task had the highest incompletion rate. Children with PME were less likely to complete the Picture Sequence Memory task compared to those without PME (40% completed vs 62% completed). However, both unadjusted and adjusted models yielded no statistically significant group differences. Unadjusted and adjusted odds ratios are presented in Table 2, and group differences in NIH-TB task performance and other task performances are presented in Table 3. In total, no differences in NIH-TB task performance were observed between children with and without PME. Additionally, significant differences between children with and without PME were not detected in children’s performance on the WPPSI Picture Memory subtest. No significant group differences in maternal/caregiver reports on children’s executive functioning on the BRIEF-P were observed when controlling for covariates in the model (see Table 4).

Table 2.

Associations Between Task Completion and Prenatal Marijuana Exposure, Logistic Regression Models (n=63)

Outcome Variable	PME group %completed (n=15)	Non-PME group %completed (n=48)	Unadjusted Model		Adjusted Model
			OR	95% CI	OR	95% CI
Flanker (Inhibitory control & attention)	87%	82%	1.41	.26 - 7.49	0.48	.01 – 12.48
Picture Sequence Memory (Episodic memory)	40%	62%	0.41	.12 – 1.34	0.33	.08 – 1.46
Picture Vocabulary (Receptive vocabulary)	100%	98%
Dimensional Change Card Sort (Cognitive flexibility)	73%	64%	1.52	.42 – 5.55	3.61	.49 – 26.64
Early Childhood Cognition Composite Score	27%	36%	0.66	.18 – 2.41	1.37	.24 – 7.72
Pattern Comparison Processing Speed Test (Processing speed)	100%	89%
Bayley Puzzle (Visual Spatial Ability)	20.00%	51%	0.24	.06 - .96	0.5	.08 – 3.29

^*Adjusted Models include child race, sex, age, prenatal tobacco exposure, household income, caregiver marital status, and caregiver executive functioning as co-variates.

^**The Early Childhood Cognition Composite Score, of the NIH Toolbox, is a composite score consisting of the Picture Vocabulary, Flanker, DCCS, and Picture Sequence Memory.

Table 3.

Exposure Group Differences in Task Performance, ANCOVA Analyses

Outcome Measure	PME mean (SD)	Non-PME mean (SD)		Unadjusted Model			Adjusted Model
			r2	β	95% C.I.	p	β	95% C.I.	p
Visual Spatial Ability (n=57)
WPPSI-III Block Design	7.3(3.3)	7.7(2.8)	<.01	−.46	−2.20, 1.28	.60	.64	−1.18, 2.46	.48
Bayley – Puzzle Task	0.3(0.7)	0.9(0.9)		−.58	−1.11, 0.0	.04	−.16	−0.76, .44	.60
Inhibitory Control and Attention (n=48)
NIH-TB-Flanker	97.5(12.6)	97.9(16.3)	<.01	−.40	−10.44, 9.63	.94	1.80	−9.36, 12.97	.75
Episodic Memory (n=32)
NIH-TB – Picture Sequence Memory	82.7(11.3)	90.8(19.4)	.03	−8.15	−24.95, 8.64	.33	−8.07	−28.26, 12.12	.42
Receptive Vocabulary (n=57)
NIH-TB – Picture Vocabulary	91.6(14.2)	89.1(14.7)	.01	2.49	−6.25, 11.22	.57	3.84	−5.75, 13.42	.42
Cognitive Flexibility (n=39)
NIH-TB – DCCS	101.3(7.0)	97.7(11.8)	.02	3.58	−4.13, 11.30	.35	5.16	−5.75, 16.07	.34
Processing Speed (n=53)
NIH-TB – Pattern Comparison Speed	77.4(11.6)	75.5(17.4)	<.01	1.91	−7.86, 11.68	.70	−.67	−11.15, 9.79	.90
Global Cognitive Ability (n=19)
NIH-TB – Early Childhood Cognition Composite Score	94.0(5.0)	91.6(17.6)	<.01	2.44	−16.63, 21.51	.79	7.06	−22.99, 37.11	.61
Working Memory (n=57)
WPPSI-IV – Picture Memory	7.1(4.2)	7.5(4.9)	<.01	−.43	−3.25, 2.38	.76	.60	−2.45, 3.64	.69
Planning Abilities (n=56)
TOH-Score	10.0(2.1)	14.1(4.0)	.01	−1.20	−4.18, 1.79	.42	−.42	−4.00, 3.15	.81
Emotion Regulation (n=56)
Locked Box - Seeking help	1.5(.9)	1.3(.5)	.03	.23	−.14, 0.61	.22	.19	−.25, .63	.39
Locked Box – Distraction	1.5(.5)	1.6(.8)	<.01	−.10	−0.57, 0.38	.68	−.12	−.70, .47	.69
Locked Box - Physical negativity	1.9(1.3)	1.6(1.0)	.02	.36	−.32, 1.04	.30	.39	−.39, 1.17	.32
Locked Box - Verbal negativity	1.5(.7)	1.3(.8)	<.01	.12	−.38, .61	.64	.21	−.37, .79	.47
Locked Box - Resignation	1.9(.9)	1.7(1.2)	.01	.23	−.48, .93	.52	.22	−.65, 1.08	.62
Locked Box - Global frustration	1.8(1.5)	1.5(.9)	.01	.30	−.71, 1.30	.56	.61	−.67, 1.89	.34
Locked Box - Global regulation	4.3(1.5)	4.6(.9)	.01	−.33	−1.34, .68	.51	−.66	−1.93, .61	.30
Locked Box - Time resignation (seconds)	12.6(15.6)	15(17.7)	.01	4.25	−5.58, 14.07	.39	5.74	−6.16, 17.64	.37
Locked Box - Time frustration (seconds)	21.6(29.3)	3.4(10.4)	.04	10.90	−4.31, 26.09	.16	12.30	−5.68, 30.28	.17
Aggression (full sample; n=56)
Bobo-% of closed hits to face	34%	17%	.04	.12	−.07, .40	.16	.08	−.17, .33	.54
Bobo- fisted hits to body	0.1 (0.5)	0.6 (1.6)	.03	−.52	−1.38, .34	.23	−.56	−1.45, .33	.21
Bobo – fisted hits to face	2.6 (5.5)	3.1 (7.9)	<.01	−.47	−4.88, 3.94	.83	−.90	−4.83, 3.03	.65
Bobo – open-hand hits to body	0.3 (0.8)	0.5 (2.2)	<.01	−.20	−1.37, .97	.74	−.35	−1.73, 1.03	.61
Bobo – open hand hits to face	0.2 (0.4)	0.4 (1.2)	.01	−.29	−.94, .37	.38	−.25	−1.03, .53	.52
Aggression (subsample of children who engaged the doll; n=19)
Bobo-% of closed hits to face –	85%	57%	.11	.29	−.13, .71	.17	.74	0.26, 1.23	.01
Bobo - fisted hits to body	0.3 (0.8)	2.2 (2.4)	.16	−1.82	−3.98, 0.34	.09	−0.37	−6.24, 5.50	.89
Bobo – fisted hits to face	6.5 (7.3)	10.2 (11.8)	.03	−3.65	−14.75, 7.44	.50	3.65	−16.10, 23.39	.68
Bobo – open-hand hits to body	0.2 (0.4)	1.5 (3.9)	.04	−1.37	−4.80, 2.05	.41	−5.65	−16.76, 5.47	.26
Bobo – open hand hits to face	0.5 (0.5)	1.2 (1.3)	.07	−0.65	−1.87, 0.56	.27	0.45	−1.80, 2.71	.64

^*Adjusted Models include child race, sex, prenatal tobacco exposure, household income, caregiver marital status, and caregiver executive functioning as co-variates.

^**Performance scores that were not age-adjusted (Tower of Hanoi, Locked Box & BoBo) also included child age as a co-variate in adjusted models.

^***The Early Childhood Cognition Composite Score is a composite score consisting of the Picture Vocabulary, Flanker, DCCS, and Picture Sequence Memory, of the NIH Toolbox.

^****WPPSI-III = Wechsler Preschool and Primary Scale of Intelligence-III; NIH-TB = NIH Toolbox; WPSSI-IV = Wechsler Preschool & Primary Scale of Intelligence - IV; TOH = Tower of Hanoi.

Table 4.

Exposure Group Differences in Maternal Reports on Child Development and Behavior, ANCOVA Analyses

Outcome Measure	PME mean (SD)	Non-PME mean (SD)		Unadjusted Model			Adjusted Model
			r2	β	95% C.I.	p	β	95% C.I.	p
Maternal Report on Executive Functioning
BRIEF-P - Inhibitory Control	56.5(15.6)	52.6(10.9)	.02	3.84	−3.32, 11.01	.29	3.96	−3.73, 11.64	.31
BRIEF-P - Cognitive Shift	51.5(9.9)	52.2(12.4)	<.01	−.70	−7.74, 6.34	.84	.15	−7.55, 7.85	.97
BRIEF-P- Emotional control	53.4(13.1)	53.8(13.4)	<.01	−.43	−8.30, 7.44	.91	2.63	−5.04, 10.31	.49
BRIEF-P - Working memory	57.5(15.0)	58.5(14.7)	<.01	−.95	−9.68, 7.78	.83	−.39	−10.11, 9.33	.94
BRIEF-P - Planning / organizing	52.9(11.6)	54.3(13.5)	<.01	−1.47	−9.25, 6.30	.71	−.14	−8.10, 7.81	.97
BRIEF-P - Inhibitory Self Control Index	55.7(14.5)	53.4(11.8)	.01	2.30	−5.07, 9.66	.54	3.85	−3.69, 11.40	.31
BRIEF-P – Flexibility Index	53.1(11.1)	53.2(13.1)	<.01	−.10	−7.60, 7.41	.98	2.17	−5.51, 9.85	.57
BRIEF-P - Emergent metacognition Index	56.1(14.5)	56.8(14.3)	<.01	−.74	−9.24, 7.76	.86	−.04	−9.04, 8.95	.99
BRIEF-P - Global Executive Composite	56.1(14.2)	54.9(13.5)	<.01	1.19	−6.93, 9.32	.77	2.08	−6.15, 10.31	.61
Maternal Report on Child Aggression
LRRHQ - Non-aggressive	6.5(3.2)	7.1(2.7)	.01	−.53	−2.21, 1.16	.53	−0.78	−2.76, 1.20	.43
LRRHQ - Walkaway	5.9(3.0)	5.8(2.9)	<.01	.12	−1.60, 1.83	.89	.40	−1.61, 2.41	.69
LRRHQ - Verbal aggression	7.2(2.0)	7.5(2.0)	<.01	−.26	−1.42, 0.91	.66	−.08	−1.48, 1.32	.91
LRRHQ - Physical aggression	4.3(4.3)	3.8(3.4)	<.01	.53	−1.76, 2.81	.65	.18	−2.34, 2.70	.89
CBCL - Aggression	13.9(10.4)	9.2(6.8)	.06	4.70	0.08, 9.28	.05	5.73	0.89, 10.58	.02
Maternal Report on Child Problem Behaviors
CBCL - Emotional reactivity	3.3(3.4)	2.5(2.6)	.02	.81	−.85, 2.47	.33	1.22	−.55, 2.99	.17
CBCL - Anxiety	4.2(3.8)	3.7(3.3)	.04	1.69	−.36, 3.74	.10	1.37	−1.02, 3.77	.26
CBCL - Stress	2.5(2.3)	1.9(1.9)	.02	.68	−.52, 1.88	.26	1.07	−.23, 2.37	.10
CBCL – Somatic complaints	1.3(1.2)	1.9(2.0)	.02	.54	−1.69, 0.47	.27	−.49	−1.79, 0.82	.46
CBCL – Attention	2.7(2.4)	2.5(1.9)	<.01	.25	−0.96, 1.47	.68	−.07	−1.46, 1.33	.92
CBCL – Affect	3.7(3.9)	2.3(2.2)	.05	1.44	−0.18, 3.06	.08	1.58	−.38, 3.55	.11
CBCL – ADHD	5.0(3.7)	4.4(3.0)	<.01	.56	−1.32, 2.44	.55	.64	−1.38, 2.65	.53
CBCL – Withdrawal Symptoms	3.8(3.8)	1.7(2.6)	.09	2.07	0.33, 3.81	.02	2.07	.28, 3.86	.02
CBCL - Sleep Problems	5.1(3.8)	3(2.7)	.08	2.04	0.28, 3.81	.02	2.08	−.08, 4.24	.05
CBCL - Internalizing problems	12.5(10.9)	9.19(6.8)	.03	3.35	−1.91, 8.61	.21	4.93	−.84, 10.69	.09
CBCL - Externalizing problems	16.6(12.6)	11.7(8.0)	.05	4.93	−0.55, 10.42	.08	6.74	.76, 12.72	.03
CBCL - Total problems	46.1(32.9)	32.4(23.0)	.05	13.69	−1.47, 28.86	.08	14.54	−1.98, 31.05	.08
CBCL - Developmental problems	5.6(5.6)	2.9(3.6)	.07	2.68	0.20, 5.16	.03	2.25	−.46, 4.96	.10
CBCL - Oppositional defiant behaviors	4.6(3.7)	3.1(2.6)	.05	1.52	−0.21, 3.25	.08	2.07	.24, 3.90	.03

^*Adjusted Models include child age, race, sex, prenatal tobacco exposure, household income, caregiver marital status, and caregiver executive functioning as co-variates.

^**BRIEF-P = Behavior Rating Inventory of Executive Function-Preschool Version; LRRHQ = Liefer-Roberts Response Hierarchy Questionnaire; CBCL=Child Behavior Checklist

Visual Spatial Ability

Children with PME performed significantly worse on the Bayley puzzle task in unadjusted models. Specifically, children with PME were 31% less likely to be able to complete the task compared to children without PME (OR=.24, 95% CI: .06-.96). Among those who completed the task, children with PME performed significantly worse on the Bayley puzzle task, β= −.58[−1.1,0.0], p=.04. However, when child age, race, sex, prenatal tobacco exposure, household income, maternal marital status and maternal executive functioning were controlled in the models, no significant differences were detected between exposure groups (See Table 3). Further, no significant group differences were observed in performance on the WPSSI-III Block Design subtest within unadjusted or adjusted models.

Planning Skills and Emotion Regulation

PME and non-PME groups did not differ on planning task (TOH) performance scores or number of errors made during these trials. Further, no significant differences across PME groups were detected on child emotion regulation on the Locked Box task. Associations remained non-significant within the adjusted models as well.

Aggressive Behavior

Children with PME were not significantly more likely to engage the doll during the Bobo task compared to peers without PME (66% vs 50%). Among the full sample, children with PME did not engage in more aggressive behaviors compared to peers without PME. However, among the sub-group of children (with and without PME) who engaged with the doll, children with PME displayed significantly more aggressive behaviors. Specifically, among children who engaged with the doll, a substantially higher proportion of hits to Bobo were fisted hits to the face among children with PME compared to those without PME (85% vs 57%). These significant differences in aggression persisted within in the adjusted model, β=.74 [0.3,1.2], p=.01. While children with PME showed a higher percentage of more aggressive hits, they did not hit Bobo more times than children without PME.

Maternal/caregiver reports of children’s aggressive behaviors varied. On the CBCL, a significant difference based on PME status was found for child aggressive behaviors that persisted when controlling for child age, race, sex, prenatal tobacco exposure, maternal/caregiver marital status, household income, and maternal/caregiver executive functioning, β=5.73 [0.9,10.6], p=.02. However, no significant group differences were observed on the LRRHQ measure based on PME status.

Other Child Problem Behaviors

Maternal/caregiver reports also indicated that children with PME displayed more sleep-related problems, β = 2.08 [−0.8,4.2], p=.05; withdrawal symptoms, β=2.07[0.3,3.8], p=02; externalizing problem behaviors, β=6.74[0.8,12.7], p=.03; and oppositional defiant behaviors, β=2.07[0.3,3.9], p=.03, compared to children without PME, within controlled models.

Developmental and Behavioral Diagnoses

Children with PME did not differ from those without PME in terms of the likelihood of having a developmental or behavioral diagnosis from a health professional (adjusted OR=.95, 95% CI: .05-17.21).

Discussion

Some of the current findings are consistent with prior evidence of associations between PME and children’s development as early as 3.5 years. Children with PME demonstrated increased aggression and increased rates of a variety of behavioral problems compared to their non-exposed peers. However, little evidence of deficits associated with PME were observed regarding various components of EF, emotion regulation, and cognitive ability. Given the use of validated measures of EF and the age of these children, this information helps to clarify when deficits in EF may emerge following PME. Although the findings must be considered within the context of the small sample size, they align with prior work noting increased aggression and behavioral problems associated with PME, and also may suggest an earlier onset of behavioral problems related to PME than previously reported.

PME and maternal marijuana use during pregnancy has almost exclusively been determined by maternal report in previous studies (Fine et al., 2019; Fried, 1991; Richardson et al., 2002). Researchers have pointed out maternal self-report underestimates the proportion of children exposed by roughly 50% (Young-Wolff et al., 2017). Within the current sample, the use of only maternal report to determine PME classification would have resulted in 86.7% of the children with PME being misclassified as non-PME. Further, all women who self-reported (n=2) also had a positive urine screen and marijuana use documented on their obstetrics record. Eleven women had a urine screen positive for marijuana during pregnancy and six of those women also had marijuana use noted in their obstetrics record. The remaining two women who used marijuana during pregnancy were identified strictly through obstetrics record abstraction. Inconsistent reporting is likely influenced by maternal fear of potential loss of custody associated with self-reporting marijuana use during their pregnancy; however, we do not have data to confirm this. Regardless, misclassification is problematic, as it is likely to bias effect estimates towards the null, resulting in a failure to observe important effects. Families that we were unable to locate or engage into LEAF had higher rates of maternal marijuana use during pregnancy. Possibly, mothers who used marijuana during pregnancy were less likely to participate in follow-up for fear of losing custody or other penalties related to disclosing marijuana use while pregnant. However, the LEAF study recruitment scripts did not focus particularly on PME, but rather broadly on multiples aspects of child and maternal well-being; this suggests that other factors associated with marijuana use during pregnancy may have influenced mothers’ willingness to participate or led to increased difficulty in locating families.

PME is hypothesized to impact EF as well as skills that engage EF (Fried & Smith, 2001). Worse short-term memory and verbal reasoning associated with PME have been observed at age 3 (Day et al., 1994); however, these differences in cognition were not apparent after controlling for environmental factors (Fried & Watkinson, 1990). Among 4-year-old children in the same study, differences in EF persisted within controlled models (Fried & Watkinson, 1990), suggesting that as EF processes develop and become more complex, deficits associated with PME may become more pronounced. One prior study reported that exposed children within their sample demonstrated worse short-term memory, working memory, attention, inhibitory control, language comprehension, and verbal reasoning between ages 6 and 12 as EF develops (Fried et al., 1998; Goldschmidt et al., 2000; 2008; Leech et al., 1999; O’Connell & Fried, 1991). The current study found no significant difference between children with and without PME in working/episodic memory, inhibitory control, cognitive flexibility or processing speed when adjusting for key covariates at age 3.5. In previous research, the earliest EF differences detected within models that accounted for environmental confounders have been among 4-year-old children (Fried & Watkinson, 1990). Findings from prior studies taken together with current findings may suggest that differences in EF associated with PME emerge slightly later than age 3.5. The inclusion of maternal/caregiver executive functioning within models as a covariate may help to explain the lack of observed relationship between PME and EF. However, current findings do not provide support for this hypothesis as group differences were negligible in adjusted and unadjusted models. But the absence of differences may reflect the inability of some children to complete the EF battery and limited performance variance among those did complete the battery.

Significant growth in EF occurs between the ages of 3½ and 4½ therefore, children in this cohort may have been too young for differences in EF to be detected. Different components of EF are thought to develop along different developmental trajectories, with varying components typically developing between ages 3 and 5. Among the general population, the developmental literature suggests that children at age 3.5 begin to demonstrate the ability to inhibit behaviors with some errors, exhibit incrementally better processing speed, and continue to struggle with cognitive flexibility in situations with more complex rules, demonstrating difficulty with planning and organizing behaviors (Anderson, 2010; Diamond & Taylor, 1996; Espy, 1997). Children raised in low-income families tend to have later onset of and poorer overall EF compared to their peers (Espy et al., 2004; Farah et al., 2006; Noble et al., 2007). Based on the high proportion of low-income families in the current sample, the children may tend towards later onset of EF, which could also explain null findings. The NIH-TB was developed among a sample of children from relatively highly educated and high-income families, and this may partially account for why the current sample of children had challenges completing various NIH-TB tasks.

Current findings were mixed regarding differences in visual spatial abilities between children with and without PME, with significant differences on one of two tasks in unadjusted models. Differences in visual spatial abilities among children with PME have been reported as early as one week of age in the form of poorer habituation and less responsiveness to visual stimuli (Fried & Makin, 1987). Prior studies also report PME is associated with worse abstract/visual reasoning at age 3 (Day et al., 1994; Fried & Watkinson, 1990; Griffith et al., 1994). Associations between PME and worse visual discrimination, scanning, and sequencing memory persist among children ages 6-9 (O’Connell & Fried, 1991), albeit not in controlled models. In contrast, for children between 9 and 12 years of age (Fried et al., 1998; Fried & Watkinson, 2000), as well as 13- to 16- years of age (Fried et al., 2003), the association between PME and worse visual spatial ability was observed even in controlled models. In line with this trajectory, children with PME in the present study tended to be less likely to be able to complete the Bayley puzzle task, and those that could complete it performed worse; however, these differences were not significant after adjustment for key covariates. Further, no differences were detected on the Block Design subtest in the 3.5-year old participants. Possible differences in visual spatial ability and EF processes associated with PME may manifest themselves more distinctly at older ages when children can perform more complex tasks, thereby allowing for improved sensitivity for tasks to differentiate between exposure groups. As discussed above, limited variance in visual spatial ability among children of this age reduces the likelihood of observing significant differences, especially within controlled models. However, differences in visual spatial ability may become more pronounced as this cohort grows older. The inclusion of a familial component, maternal executive functioning, in the model may represent an alternative mechanistic explanation for understanding these visual spatial disparities.

In the current sample, children with PME were reported and observed to be more aggressive (as reported by caregivers on the CBCL and observed on the Bobo doll task). Children were also reported by their mothers to have more sleep-related problems, social withdrawal symptoms, externalizing problems, and oppositional defiant behaviors (as reported by caregivers on the CBCL); collectively, these results suggest that behavioral problems associated with PME emerge sooner than age 6. These findings are consistent with prior studies that report children with PME exhibit more aggression and worse attention as early as 18-months of age (girls only; El-Marroun et al., 2011) and poorer sleep quality with more sleep disturbances at age 3 (Dahl et al., 1989; Fried & Makin, 1987; Scher et al., 1988). Analyses stratified by sex were not possible in the current study due to the limited sample size. Instead, child sex was incorporated into the model as a covariate. Prior research has suggested a behavioral trajectory among children with PME in which exposed children are more hyperactive and impulsive starting at age 2; have more problems with sleep at age 3; have more behavioral problems starting at age 6; more delinquency, poorer school performance, and increased externalizing problem behaviors at age 10; and higher rates of substance use initiation in adolescence (Dahl et al., 1989; Day et al., 2006; El-Marroun et al., 2019; Frank et al., 2011; Fried et al., 1992; Goldschmidt et al., 2000; O’Connell & Fried, 1991). These differences might be explained by the aforementioned alterations to the cannabinoid system and receptors following PME (El-Marroun et al., 2016; Gray et al., 2005; Hurd et al., 2005); however, family and environmental risk factors (e.g., poverty, stress, conflict, abuse, intimate partner violence, parental substance use and mental health conditions) that may be disproportionately common among families in which maternal prenatal marijuana use occurs require consideration as well (Farrington et al., 2017; Flouri & Midouhas, 2017; Fong et al., 2017; Ibabe & Bentler, 2016). Both of these factors are likely influential, and require consideration within the conceptualization of relationships between PME and child behavioral development. Current findings suggest that problem behaviors associated with PME may emerge sooner than previously reported. Identifying the precise age at which behavioral problems associated with PME emerge is not only critical to understanding how PME may affect child development, but also has important implications for the delivery of timely and effective interventions that mitigate long-term risk.

Limitations

The LEAF study was powered to examine relationships between PME, EF, and aggression across all study time points rather than at each time-point cross-sectionally; therefore, the current study is limited by a relatively small sample size and number of children with PME. Given this study’s focus on a specific age window coupled with manuscript size limitations and on-going data collection and preparation of future timepoint, the current study utilizes data from when children completed the age 3.5 years study visit. Specifically, sample size limited the capacity for exploration of moderating and mediating effects on the relationships between PME and outcome variables. This limitation was further exacerbated within analyses of EF by children’s inabilities to complete EF performance tasks. However, we expect that findings for these very young children will provide a baseline to evaluate their progress as they are re-evaluated at older ages, when we will be able to determine whether inability to complete tasks predicts subsequent development. We did not employ multiple imputation or other missing data strategies because we were concerned that inability to complete the performance tasks was related to the child’s true, but unknown score on the test. Therefore, the true score would be “missing not at random”, which violates the assumptions of multiple imputation. Further, findings from this study provide insights about the development of EF among these children at an age at which EF may be beginning to develop. Child’s race was controlled for in adjusted models; however, a larger proportion of children in the PME group identified as African American compared to the non-PME group. Due to sample size constraints, we were unable to explore additional strategies for parsing out effects of child’s race from those of PME status. Another limitation is that this study was designed to successfully identify ‘users’ and non-users of marijuana during pregnancy; however, it did not collect maternal marijuana dosage and timing of exposure during pregnancy. This precluded exploration of how these factors may impact the relationship between PME and children’s developmental and behavioral outcomes. Prior studies primarily identified significant effects of PME among heavy maternal users; however, the substantial increase in THC potency in contemporary marijuana is likely to offset some of this usage effect, as light users are exposed to more THC than 30 years ago, when the previous cohorts were assessed. Also, residual confounding by unmeasured or incompletely controlled factors may be a possibility especially in the areas of family and environmental risk factors. Finally, except for aggression, child behavior problems were strictly measured through caregiver report. In five instances, current caregivers participated instead of biological mothers, following mother’s loss of parental custody. At future time points, child problem behaviors are also assessed via teacher report, which will provide a more nuanced perspective of children’s behavior. Given that we examined the association of PME with numerous outcomes, the likelihood of observing a statistically significant association due to chance is increased, although the findings regarding increased aggression converged across both direct observations and ratings. As detailed in the statistical analysis section, adjustments for multiple comparisons were not made per published recommendations (Rothman, 1990; 2014).

Strengths

Almost all prior studies have determined PME using only maternal report, which has been shown to underestimate the proportion of PME by about 50% (Young-Wolff et al., 2017). Some prior studies asked women to recall their marijuana use many years in the past (Fine et al., 2019). The present study determined PME prospectively using urine screens and medical chart abstraction in addition to maternal report, reducing the potential bias of misclassification. Within the current study, maternal report would have correctly identified only 13.3% of children classified with PME, highlighting the importance of utilizing multiple assessments of PME. Additionally, EF among children and their mother was measured using the NIH-TB, a psychometrically validated assessment tool. Consistent with suggestions that EF is the primary aspect of cognitive functioning affected by PME, this study’s robust and direct assessment of EF, both of the child and the caregiver, is a significant strength. For many outcomes of interest, multiple assessment tools were used that combined objective testing and observational assessment by blinded research staff’s or maternal report. Unlike previous research, this study measured and incorporated caregiver EF as a control variable, which is important given the strong relationship between caregiver and offspring EF (Calkins, 2011; Cuevas et al., 2014) and the observation that EF deficits predispose to marijuana use (Squeglia et al., 2014). Last, the current study represents the first time-point of the only prospective North American cohort specifically examining PME among a sample of children exposed to contemporary potent marijuana products. The current study provides evidence that behavioral differences associated with PME may become apparent earlier than previously reported.

Conclusions

Information regarding the potential effects of PME on children is urgently needed given changing legislation and softening attitudes regarding marijuana use, as well as increasing potency of marijuana itself. Policy-makers and expectant parents alike are looking for high-quality information to inform their decision-making around the use of marijuana products during pregnancy. The current findings are consistent with prior research identifying increased behavioral and developmental difficulties among children exposed to marijuana prenatally. This study provides new indications that some of these disparities, particularly in specific behavioral domains, may emerge sooner than previously observed. Given the current state of the literature, women should be cautioned against the use of marijuana during pregnancy due to the potential harmful effects on child development. Continued follow up of the current cohort and other cohorts that accurately differentiate children with and without PME is needed to better understand the onset and trajectories of developmental and behavioral difficulties associated with exposure. Future research should seek to expand our understanding of the potential mediating and moderating relationships of family and environmental factors on the association between PME and child outcomes in larger samples. Understanding how family environments and parenting intersect with relationships between PME and child outcomes may help identify parenting practices that can be targeted by interventions to mitigate the risk associated with PME.

Highlights.

• Children with PME were more aggressive than children without PME

• Differences in executive functioning between PME groups were not observed.

• Children with PME had more behavioral problems compared to children without PME.

• Results strengthened by comprehensive assessment of marijuana use during pregnancy.