Prenatal tobacco and tobacco-cannabis co-exposure: Relationship with attention and memory in middle childhood

Abstract

We examined associations between prenatal tobacco exposure (with and without cannabis exposure) and children’s performance on laboratory measures of sustained attention, attentional set shifting, and working memory in middle childhood (9–12 years of child age). Participants were recruited in the first trimester of pregnancy and oversampled for prenatal tobacco exposure; with a smaller sample (n = 133; n = 34 non-substance exposed, n = 37 exposed to tobacco only, n = 62 co-exposed) invited (oversampled for co-exposure) to participate in the middle-childhood assessment (M age = 10.6, SD = 0.77; 68% Black, 20% Hispanic). Results for sustained attention indicated lower attention (percent hits) at the first epoch for tobacco only exposed compared to non-exposed and co-exposed; a trend (p = .07) towards increases in impulsive responding across time (a total of 8 epochs) for tobacco exposed (with and without cannabis) compared to non-exposed children; and a significant association between higher number of cigarettes in the first trimester and greater increases in impulsive responding across epochs. However, children prenatally exposed to tobacco (with and without cannabis) demonstrated greater short-term memory compared to children not prenatally exposed, and this difference was driven by higher scores for children prenatally co-exposed to tobacco and cannabis compared to those who were non-exposed. Overall, results suggest that prenatal tobacco exposure, especially in the first trimester, may increase risk for impulsive responding on tasks requiring sustained attention, and that co-use of cannabis did not exacerbate these associations. The higher short-term memory scores among children who were co-exposed compared to non-exposed are perplexing and need replication, particularly in studies with larger sample sizes and samples exposed only to cannabis to examine this more closely.

1. Introduction

Tobacco is one of the most commonly used substances during pregnancy, with prevalence of 0.8% to 18.2% nationally (Martin et al., 2023) and disproportionately high rates among underserved, less educated, low-income women. Tobacco is often co-used with cannabis, with rates of co-use about 2–3 times higher than use of cannabis alone (Coleman-Cowger et al., 2017; Ko et al., 2015). However, the literature on prenatal co-exposure effects on child outcomes, particularly studies using prospective design and measurement, is limited. This is a critical gap given recent changes in cannabis legalization across a number of states in the U.S. (as of October 2023 cannabis is legal for medical and recreational use in 27 states), tripling of cannabis potency in recent years (ElSohly et al., 2021), and perceptions that cannabis is safe to use in pregnancy (Bayrampour et al., 2019; Chandra et al., 2019).

Both nicotine and cannabinoids are delivered to the fetus via the placenta (USDHHS, 2014). Fetal exposure may result in persistent changes to neural development and impact higher order functioning such as attention (e.g., sustained attention and set shifting) and working memory. Although there is some evidence in the literature that prenatal tobacco exposure may impact these processes as noted below, co-exposure to tobacco and cannabis may exacerbate these associations but remains understudied.

1.1. Importance of working memory and attention

Working memory (i.e., the capacity to maintain temporary active representations of information and be able to manipulate that information) and specific attentional abilities such as sustained attention (i.e., the ability to attend to a specific task for an extended period of time), and set shifting (i.e., cognitive flexibility) are critical aspects of cognitive and self-regulatory processes (Ilkowska and Engle, 2010). Working memory is a function critical for planning, problem-solving, and decision-making (Hambrick and Engle, 2003), while sustained attention has implications for academic success (Isbell et al., 2018), language abilities (Ebert and Kohnert, 2011), and even safety (Zhang et al., 2017). Finally, set shifting/cognitive flexibility is crucial for the ability to appropriately adjust behavior in response to environmental changes and is associated with reading abilities and resilience (Engel de Abreu et al., 2014; Genet and Siemer, 2011). Given the critical function of these processes in the context of child development, it is important to understand how they may be affected by prenatal tobacco and cannabis co-exposure.

1.2. Prenatal tobacco exposure

Prenatal tobacco exposure results in increased norepinephrine, dopamine, acetylcholine, and serotonin concentrations in the developing brain (Lichtensteiger et al., 1988; Slotkin et al., 2015). Recent brain morphology studies indicate that children prenatally exposed to tobacco experienced cortical thinning, especially in the superior frontal and superior parietal cortices (El Marroun et al., 2016), increasing the possibility of deficits in cognitive function related to the frontal lobes, such as attention and working memory. In addition, the effects of prenatal tobacco exposure may extend across development with evidence of lower self-regulation and attention to stimuli across the neonatal period (Stroud et al., 2018). Indeed, in a meta-analytic review, Froggatt et al. (2020) reported a medium pooled effect size for the association between prenatal tobacco exposure and measures of attention during the first year of life.

1.3. Extant studies on prenatal tobacco exposure

However, only a small number of prospective cohort studies of prenatal tobacco exposure with assessments beyond infancy have used laboratory based or behavioral measures of attention and working memory, and results have been mixed. For instance, in a prospectively recruited cohort exposed to tobacco with some children co-exposed to both tobacco and cannabis, authors reported no significant associations between prenatal tobacco exposure and cognitive aspects of self-regulation that consisted mainly of attention and working memory tasks. However, they did report lower effortful control among children prenatally exposed to tobacco at preschool (Wiebe et al., 2015) and early school age (Clark et al., 2016). In contrast, a study recruiting for asthma and using self-report measures of smoking during pregnancy reported significantly lower executive functions and working memory at 4 years of age (Julvez et al., 2007). However, there were significant demographic differences between smokers and non-smokers that may have at least partially accounted for these differences, despite statistical controls. Similarly, in a sample of adolescent mothers, maternal self-reports of tobacco use in pregnancy were associated with lower scores on a cognitive interference task that required both selective attention and response inhibition compared to non-exposed children at 10 years of child age (Cornelius et al., 2011). Results from a study using a sibling comparison design with discordant smoking across two pregnancies and examining within family processes (Knopik et al., 2022) reflected a more complex pattern of results at 10–12 years. Prenatal tobacco exposure was associated with lower visual working memory but not attentional set-shifting or auditory working memory (Knopik et al., 2022). However, the study included only self-reported smoking measures and no reported assessment of other substance use.

Taken together, the literature on prenatal tobacco exposure and attention/working memory using behavioral measures is fairly limited, with conflicting results, despite robust evidence from neuroimaging and animal models indicating nicotine effects on central nervous system development (De Genna et al., 2019; El Marroun et al., 2016; Slotkin et al., 2015). These variations in results may be due to variations in methods used to ascertain exposure, with previous studies indicating that about 15% to 25% of women assumed to be non-smokers using single item self-reports only were determined to be active smokers based on more intensive calendar-based interviews and biomarkers (Shisler et al., 2017). In addition, mixed results may also be due to co-exposure to cannabis. Few studies recruiting cohorts on the basis of prenatal tobacco exposure examined the role of other substances such as cannabis that are commonly used with tobacco and may have additive or synergistic effects.

1.4. Extant studies on prenatal cannabis exposure

There are few recent prospective cohorts with data on prenatal cannabis exposure (see Porath and Fried, 2005, sample recruited in the 1970s; Day et al., 1994, recruited in the 1980s; Jaddoe et al., 2012, recruited in the early 2000s), and the literature on child outcomes is limited and inconsistent (see Huizink, 2014, review). However, prenatal cannabis exposure has the potential to impact development of the fetal endogenous cannabinoid system, with the potential to alter neurodevelopment (Richardson et al., 2016). Studies of children’s brain morphology indicate that preadolescent children exposed to cannabis prenatally have thicker frontal cortices, indicating a deleterious effect on neurodevelopmental maturation in regions which are commonly involved in higher order processing, including the development of cognitive functions such as attention and memory (El Marroun et al., 2016), resulting in poorer executive functioning. Functional imaging studies of young adults with prenatal cannabis exposure suggest alterations in neural functioning during these tasks (Smith et al., 2006; Smith et al., 2016). Specifically, Smith et al. (2006) reported dose response relationships between prenatal cannabis exposure and increased activation in left brain areas but decreased activation in the right prefrontal cortex, but no differences in cognitive performance on a visuospatial 2-back test. Authors suggested that the imbalance of right and left-brain activity may potentially reflect increased engagement of the left brain to compensate for regions of the right brain compromised by prenatal exposure. These results were replicated in their 2016 study (Smith et al., 2016) where there were no differences in cognitive performance on any of the four tasks (visupoatial n-back, go/no go, letter n-back, and counting Stroop). Similarly, behavioral studies with children indicate significant associations between prenatal cannabis exposure and decreased performance in attention and memory (Fried et al., 1998, Fried et al., 2003; Richardson et al., 2002) including a dose-response relationship with omission errors in four-year-olds (Noland et al., 2005) suggesting problems with sustained attention. However, other studies have found no association between prenatal cannabis exposure and attention in infants (de Moraes Barros et al., 2006) or executive functioning at preschool age (Murnan et al., 2021).

1.5. Potential impacts of prenatal co-exposure to tobacco and cannabis

Despite the high prevalence of co-use of tobacco and cannabis among pregnant smokers and the potential for both substances to impact attention and working memory among their offspring, little is known about the effects of co-exposure on these cognitive/regulatory processes. This is a critical gap in our knowledge as these substances may have synergistic effects on child outcomes. For instance, co-exposed neonates had lower scores on self-soothing and attending to stimuli compared to non-exposed infants, and co-exposure was associated with a 42% greater decrease in attention (as measured by behavioral observations using the NICU Network Neurobehavioral Scale; NNNS) than tobacco exposure alone (Stroud et al., 2018). It is unclear if these associations last beyond the neonatal period, although previous results from the current sample are suggestive. For instance, co-exposure was predictive of less optimal autonomic regulation in infancy which was prospectively associated with lower emotion regulation at toddler age (Eiden et al., 2018). Co-exposure was also associated with higher maternal affective dysregulation, which in turn was predictive of higher externalizing behavior problems at toddler age (Godleski et al., 2018). These aspects of infant physiological and maternal functioning related to co-exposure may also increase the risk for lower attention and working memory in later childhood.

1.6. The current study

The purpose of this study is to examine the association of prenatal tobacco exposure alone and combined tobacco-cannabis co-exposure with sustained attention, set shifting, and working memory in a sample of school age (9–12 years of age) children recruited prenatally and assessed across development. Importantly, this sample included prospective, multi-method assessments of exposure and included demographically similar non-exposed children, thereby controlling for potential demographic disparities by design. While extant literature has reported some regulatory deficits associated with tobacco and cannabis co-exposure, there is a paucity of evidence related to cognitive correlates. Based on this literature and the significant gap in our understanding of tobacco-cannabis co-exposure effects on cognitive correlates of self-regulation, we examined prenatal tobacco and tobacco-cannabis co-exposure associations with attention and working memory in middle childhood. We examined two aspects of attention that have been implicated in human prenatal exposure studies: sustained attention requiring vigilance across longer periods of time and attentional set-shifting that involves flexibility of attention and responses when the rules or contingencies change. We hypothesized that 1) prenatal substance exposure (any exposure vs. none) would be associated with lower sustained attention, lower cognitive flexibility in a computerized set shifting task, and lower scores on a computerized working memory task; 2) children in the co-exposed group would have poorer performance on these tasks compared to tobacco exposed or non-exposed children; 3) there would be dose-response associations such that children exposed to higher average cigarettes or joints/day during pregnancy, with additive effects of exposure to higher doses of both substances, would have lower performance on these tasks; and 4) given higher use in the first trimester before pregnancy recognition (Eiden et al., 2013), there may be stronger effects for first trimester exposure.

2. Methods

2.1. Participants

We recruited women in their first trimester of pregnancy from a local hospital at their first prenatal appointment. Eligibility criteria included <20 weeks pregnant at recruitment, having a singleton birth, aged 18 years or older, no illicit drug use other than cannabis, no heavy alcohol use (defined as <4 drinks per occasion or did not average > 1 drink per day), and the ability to complete the self-report screening form in English. All study participants provided informed consent at their first appointment. The initial goal of the study was to examine effects of maternal prenatal tobacco use on developmental mechanisms and outcomes of exposed offspring; thus, tobacco users were oversampled so that the closest eligible nonsmoking woman (matched on maternal age and highest educational attainment) was recruited for every two smoking women. This allowed for a full range of light to heavy smokers and accounted for our anticipation of greater attrition among tobacco smokers.

We enrolled a total of 258 mother–infant dyads into the study, all of whom completed the assessment at 2 months child age. Based on a priori exclusion criteria, one mother–infant dyad was excluded from the study once infant meconium revealed fetal exposure to methamphetamine, two infants were excluded because they had hydrocephaly, two were excluded because of a later diagnosis of autism, one was excluded because of maternal binge drinking during pregnancy, and one participant was excluded due to low maternal cognitive functioning which inhibited her ability to understand and respond to survey questions. Finally, four participants were excluded because they were smoking moderate amounts of cannabis during pregnancy but no tobacco (there was no cannabis only group recruited for this study), resulting in a final sample size of 247 (69 nonsmokers, 81 tobacco-only smokers, and 97 tobacco-cannabis co-users). These children were assessed at 2-, 9-, 16-, 24- and 36-months child age, and again at early school age (at least 3 months after they entered kindergarten). At kindergarten age we recruited an additional 50 participants. Of these 50 mother-child dyads, 16 were women who had completed at least one prenatal assessment (of a potential 3 assessments) in the initial phase of our study, but had not been included in final enrollment due to scheduling conflicts at the first postnatal assessment. Thus, we had prospective prenatal exposure data for these 16 participants (the remaining 34 dyads recruited at kindergarten age were not included in the middle-childhood assessments as they only had retrospective prenatal exposure data). For the middle childhood follow-up (child age 9–12 years), by research design, only a subset of participants were invited to participate, oversampling for heavier tobacco and tobacco-cannabis co-use, all with prospective pregnancy data. This resulted in a sample of 133 participants for the middle childhood assessment (34 with no substance use, 37 with tobacco use, 62 using both tobacco and cannabis). All substance using women smoked combustible cigarettes and smoked cannabis either as joints or in the form of blunts (cannabis rolled in cigar paper and smoked).

2.2. Measures

2.2.1. Prenatal tobacco and cannabis exposure

Use of tobacco, alcohol, cannabis, and illicit substances (e.g., methamphetamine, cocaine, opioids) were assessed using several indicators, including maternal self-report and biological assays. First, mothers self-reported their substance use with the Time Follow-Back Interview (TLFB; Sobell and Sobell, 1992) towards the end of each trimester, resulting in daily substance use data (e.g., number of cigarettes/day, number of joints/day, and number of standard drinks/day). This calendar-based method uses important events (e.g., conception date, birthdays, holidays, funerals, etc.) as memory anchors to aid in accurate recall. The TLFB is a reliable and valid method of obtaining daily data on patterns of substance use, including tobacco and cannabis (Robinson et al., 2014). This method has good test–retest reliability and is highly correlated with other intensive measures of self-reported substance use (Brown et al., 1998). Data were collected prior to the emergence of electronic cigarettes, thus all mothers who used tobacco smoked combustible cigarettes and none were e-cigarette users.

In addition to maternal self-reports, we collected maternal saliva samples towards the end of each trimester and stored them in a – 80 degree freezer until shipping (in dry ice) to the US Drug Testing Laboratory (USDTL; Des Plaines, IL) for analysis. Specimens were analyzed for cotinine, the primary nicotine biomarker, and for THC, the primary cannabinoid responsible for intoxication and subjective feelings of euphoria and well-being. USDTL conducted cotinine assays with enzyme-linked immunosorbent assay (ELISA) or liquid chromatography-tandem mass spectrometry (LC-MS/MS) at 10 ng/mL cutoff and ranged from 0 to 569 ng/mL. Assays for THC were conducted with immunoassay screening (4.0 μg/L cutoff) and GC–MS confirmation (4.0 μg/L cutoff).

In addition to maternal saliva, we also collected samples of infant meconium at the hospital across several days after delivery until the appearance of milk stool. These samples were assayed for several substances (nicotine, cannabis, cocaine, opioids, amphetamines) with a validated LC–MS/MS method (Gray et al., 2010a) at 2.5 ng/g nicotine, 1 ng/g cotinine, and 5 ng/g OHCOT, and with a validated two-dimensional GC–MS analytical method for THC, 11-hydroxy-THC; 8,11-dihydroxy-THC; 11-nor-9-carboxy-THC (THCCOOH), and cannabinol (CBN; Gray et al., 2010b). Limits of quantification for cannabinoid meconium assays were 10 ng/g for all analytes, except 11-hydroxy-THC at 15 ng/g.

Based upon the results of all substance use indicators, mothers were assigned to one of three groups (see Shisler et al., 2017 for details regarding concordance among the different measures). Mothers were assigned to the tobacco smoking group if they self-reported smoking tobacco at any time during their pregnancy (either on the screener or the TLFB; 0 missing), or if saliva samples were positive for cotinine at any time in the prenatal period (0 missing), or if infant meconium was positive for cotinine, nicotine, or trans-3′ hydroxycotinine (OHCOT; 28 missing). Mothers were assigned to the co-exposed group (tobacco and cannabis smokers) if, in addition to meeting the criteria for the tobacco group, they also self-reported cannabis use during pregnancy, if maternal saliva was positive for Δ9-THC in any of the three trimesters, or if infant meconium was positive for cannabinoids. Mothers were assigned to the control group if all of the above criteria were negative during each trimester of pregnancy. This three-group variable was used for all models testing group differences. In models testing dose-response relationships, we utilized the daily data from the TLFB, averaged either across pregnancy, or across the first trimester, depending upon which hypothesis was being tested.

2.2.2. Setting for child assessments

The middle childhood assessment took place in our lab. Consent (mom) and assent (child) were obtained, and mom and child completed their tasks in separate rooms across the hall from each other. The child interview room consisted of a table and 2 chairs that were used when the child assessor was interviewing the child and a desk with a laptop that the child used when working on a computer task on their own.

When the children entered the lab, the child assessors asked if they had a cell phone, and, if they did, asked them to turn it off and place it in a basket for the entirety of the appointment to eliminate distractions. The child assessors went over the rules to follow throughout the appointment with each child. The rules were also displayed on a poster on the wall and included following directions, staying in their assigned area, using materials appropriately, and trying their best. Children were told if they followed the rules, they would earn a sticker for each activity that they could then trade for a prize at the end of the appointment.

2.2.3. Sustained attention

We measured sustained attention with the widely used A-X Continuous Performance Test (CPT; e.g., Bubnik et al., 2015; Huang-Pollock et al., 2006; Rosvold et al., 1956; see also Cooper et al., 2021). In this task, the child was seated at the computer and presented with a series of letters which appeared on the screen one at a time. We instructed the participant to press the spacebar on the keyboard as quickly as possible when they saw an ‘X’, but only if it was immediately preceded by an ‘A’, and to not press the spacebar for any other letters. Children completed a short practice to confirm understanding of the task, during which time the research assistant ensured they were only pressing the spacebar for the correct targets. Once they successfully completed the practice they began the task, which lasted approximately 24 min, presenting stimuli across 8 epochs consisting of 100 trials each (150-ms stimulus presentation; 1500-ms blank screen response interval). The primary outcome for the CPT was target detection, or percent “hits” in each epoch ((number of targets identified/number of targets presented) * 100; there were 10 targets/epoch). The percent of false alarms (responses to any non-target letter; there were 90/epoch) was also examined to ensure that children were not responding indiscriminately to ensure obtaining hits.

2.2.4. Attentional set shifting

We measured attentional set shifting (i.e., cognitive flexibility) using the Dimensional Change Card Sort Test (DCCS), a validated measure of executive functioning recommended for use with subjects aged 3 to 85 years (DCCS; NIH Toolbox, 2016, Zelazo, 2006). During this computerized task, children were seated at the iPad and given a standardized set of instructions followed by a short practice session. Two pictures were presented that varied along two dimensions (i.e., shape and color). We asked the children to match a series of test pictures to the target picture, first according to their shape, and then according to their color. Finally, switch trials were employed in which the participant must change the dimension being matched (e.g., the participant may be asked to match on shape in one trial, then on color the next). We instructed the children to answer as quickly as possible without making mistakes with a total of 40 trials. Using the NIH Toolbox scoring algorithm, total scores were based on a combination of both accuracy and reaction time, with the computed score having a possible range of 0 to 10, and higher scores indicating a greater level of cognitive flexibility.

2.2.5. Working memory

We used the Spatial Span task as our measure of children’s working memory (i.e., the capacity to maintain temporary, active representations of information and to manipulate that information). Building on work by Tannock and colleagues (e.g., Bedard and Tannock, 2008), the current version of the task (e.g., Shiels et al., 2008; Hawk Jr et al., 2018) is a computerized adaptation of Corsi’s block tapping task (Milner, 1971) and spatial span subtest from the WISC-IV Integrated (Wechsler et al., 2004). The task was programmed in E-Prime 2.0 (Psychology Software Tools, Pittsburgh, PA; available by request). During the task the child was seated at the computer, and an array of ten white squares on a black background was presented on the screen. For each trial, a yellow smiley face appeared in a sequence of squares at a rate of one square per second. For forward span trials, we instructed the children to use the computer mouse to click on the squares in the same order in which the smiley face had just appeared. For backward span trials, we asked the children to click on the squares in the reverse order in which the smiley face appeared. Forward span trials were always administered prior to backward span trials, as is done in the WISC-IV Integrated (Wechsler et al., 2004). For both directions (forward and backward), each level of difficulty consisted of two trials. The task began with the easiest level (two-location sequences) and advanced in difficulty to a maximum of eight-location sequences. The task terminated when both trials within a difficulty level were incorrect. Forward span measures short term storage of visual-spatial information while backward span measures the ability to manipulate such information (e.g., Shiels et al., 2008).

2.2.6. Analytic strategy

Repeated measures ANOVA were used to examine group differences in sustained attention with the eight epochs as the within-subjects factor and tobacco exposure status as the independent variable. Following previous studies (Bubnik et al., 2015), time effects were examined using orthogonal polynomials of linear, quadratic, and cubic trends. To address our first hypothesis, we examined group differences by any exposure (either tobacco only, or tobacco and cannabis co-use) vs. no exposure. To address our second hypothesis, we examined group differences for the three groups (no-exposure, tobacco exposure only, and co-exposure groups). We then used orthogonal contrasts to examine differences between co-exposed and non-exposed groups, and between the co-exposed and tobacco only groups. Next, in the final model, dose-response associations were tested using multilevel regression in two parts. In the first part, we tested dose-response associations (hypothesis 3) using the average number of cigarettes/day and average joints/day (scores for joints/day were transformed using square root transformation to address skewness and kurtosis) during the entire pregnancy followed by first trimester specific effects (hypothesis 4) given the significant declines in use for all women after pregnancy recognition (Eiden et al., 2013).

ANCOVAs were used to examine associations between prenatal substance exposure and children’s score on the DCCS (for attentional set shifting) and forward and backward spatial span (for working memory). Hierarchical multiple regression analyses were used to examine dose-response associations, with covariates entered in the first step followed by the main effects of average cigarettes and average joints/day in step two, and the interaction term of the two in the final step.

For all analyses, maternal IQ (measured at the middle childhood appointment via the WASI-II, Wechsler, 2011), maternal self-identified race/ethnicity, prenatal stress (using the Perceived Stress Scale; Cohen et al., 1983), breastfeeding duration, maternal prenatal depression (using the Beck Depression Inventory – II, Beck et al., 1996), maternal prenatal anger/hostility (using the Buss-Perry Aggression Questionnaire, Buss and Perry, 1992), prenatal alcohol exposure (number of standard drinks per day), child chronological age at the middle childhood assessment, current child exposure to secondhand smoke (measured via child cotinine assays) and child biological sex at birth were considered as covariates. These covariates were selected on the basis of previous literature, but only variables that were significantly associated with child attention or working memory at the bivariate level (p < .10) were included in the final models.

2.2.7. Missing data

Out of 133 participants at the middle childhood wave, none were missing working memory or set shifting data and six participants (4.51%) were missing sustained attention data; five were missing due to technical difficulties and one child was unable to complete the task due to frustration. Given the low rate of missing data, we used listwise deletion for analyses of sustained attention data. There were no significant differences in demographic characteristics (e.g., mothers’ age, education, living status with partner) between participants with missing vs. non-missing in child sustained attention. There was no missing data on the prenatal TLFB or maternal saliva, and 28 missing meconium samples (either because mom changed hospitals, or the sample was insufficient for testing). However, as described above, in the absence of meconium, smoking group status was assigned based on TLFB or saliva, thus there were no missing smoking group assignments.

3. Results

3.1. Descriptive statistics

Demographic characteristics of the whole sample and of each substance exposure group are shown in Table 1. Maternal age ranged from 18 to 39 (M = 24.47, SD = 5.39) at the first trimester appointment, and the majority of mothers were Black with. Approximately 44.4% of women were married or living with their partner at the time of their first appointment, and just over 54% had a high school diploma or less. On average, child age was 10.62 years old (SD = 0.76) at the middle childhood visit and about 51% were female . There were no significant group differences (p’s > 0.10) in these maternal and child characteristics by prenatal substance exposure group status (using either the two level or three level grouping variables).

Table 1.

Participant characteristics.

	All	No-use	Tobacco only	Co-use
	(N = 133) M (SD)	(n = 34) M (SD)	(n = 37) M (SD)	(n = 62) M (SD)
Maternal age (years)	24.5 (5.4)	24.7(5.8)	25.4(5.8)	23.8(4.9)
Child age (years)	10.6 (0.77)	10.8 (0.75)	10.6 (0.78)	10.5 (0.75)
1st Trimester: Average # Cigarettes/Day	5.8 (6.0)	0.0 (0.0)	7.0 (6.3)	8.3 (5.4)
1st Trimester: Average # Joints/Day	0.7 (1.6)	0.0 (0.0)	0.0 (0.0)	1.5 (2.1)
Pregnancy: Average # Cigarettes/Day	4.0 (4.4)	0.0 (0.0)	4.8 (4.6)	5.8 (4.1)
Pregnancy: Average # Joints/Day	0.3 (0.7)	0.0 (0.0)	0.0 (0.0)	0.7 (0.9)
	n (%)	n (%)	n (%)	n (%)
Maternal Race/Ethnicitya
Black	78 (58.6)	21 (61.8)	19 (51.4)	38 (61.3)
White	39 (29.3)	6 (17.7)	15 (40.5)	18 (29.0)
Hispanic	19 (14.3)	9 (26.5)	3 (8.1)	7 (11.3)
Other	7(5.3)	1 (2.9)	2 (5.4)	4 (6.5)
Maternal education
Less than high school	34 (25.6)	9 (26.5)	8 (21.6)	17 (27.4)
High school/GED	38 (28.6)	8 (23.5)	8 (21.6)	22 (35.5)
Some college w/o degree	42 (31.6)	13 (36.2)	15 (40.5)	14 (22.6)
Associate degree	15 (11.3)	4 (11.8)	5 (13.5)	6 (9.7)
Bachelor’s degree	4(3.0)	0 (0.0)	1 (2.7)	3 (4.8)
Relationship w. partner
Married/Cohabiting	59 (44.4)	17 (50.0)	17 (46.0)	25 (40.3)
Living apart	45 (33.8)	10 (29.4)	14 (37.8)	21 (33.9)
Single	28 (21.1)	7 (20.6)	5 (13.5)	16 (25.8)
Divorced	1 (0.8)	0 (0.0)	1 (2.7)	0 (0.0)
Child Sex
Female	68 (51.1)	18 (52.9)	15 (40.5)	35 (56.5)
Male	65 (48.9)	16 (47.1)	22 (59.5)	27 (43.6)

^aParticipants were asked to to check all that apply for race/ethnicity questions.

With the exception of child age and sex, none of the potential covariates considered in analyses (see above) were significantly associated with child attention or working memory and were not included in the final models. Older child age was associated with higher number of trials correct in both forward (r = 0.20, p = .02) and backward span (r = 0.22, p = .01) and was included as a covariate in analyses of working memory data. A higher number of standard drinks per day across pregnancy was associated with a higher number of trials correct in backward span (r = 0.17, p = .047) and was included as a covariate in those analyses. There was a marginal sex difference in percent hits, F (1, 125) = 3.20, p = .076 but no sex by time interactions. Females had marginally higher percent hits overall compared to males. Thus, child sex at birth was retained as a covariate in analyses of sustained attention data.

3.2. Associations between prenatal substance exposure and children’s sustained attention in middle childhood

Children’s sustained attention was measured by total hits and false alarms across 8 epochs using the CPT task. In order to examine group differences at epoch 1, ANOVA with percent hits at epoch 1 as the dependent variable and two group (tobacco exposure vs. non-exposed) status as the independent variable indicated no significant group differences. However, ANOVA with the three-group status as the independent variable indicated a significant group difference for percent hits at epoch 1, F (2, 124) = 3.10, p = .049. Simple contrasts indicated that children exposed to tobacco only (M = 93.33, SD = 8.16) had lower percent hits at epoch 1 compared to children who were non-exposed (M = 88.33, SD = 12.53) and children who were co-exposed (M= 93.27, SD = 9.43). Results from repeated measures analyses with the two-level smoking group variable (tobacco exposure vs. non-exposed) indicated that as expected, there was a significant linear decrease in percent hits over time, F (7, 125) = 26.44, p < .001 (see Fig. 1A). There was also a marginal between subjects effect on overall percent hits, F (1, 125) = 3.29, p = .07, with children exposed to tobacco (with or without cannabis) having lower percent hits compared to non-exposed children. However, contrary to expectations, there was no significant smoking group by time interaction, F (7, 125) = 0.61, p = .69. Group differences across the three groups were examined next to test the second hypothesis. There were no significant group differences or group by time interaction effects for percent hits (see Fig. 1B). There were no significant dose-response associations between prenatal tobacco or tobaccocannabis co-exposure and total hits or sex differences in these associations. Results from multilevel models examining associations between the continuous dose-response measures across pregnancy (number of cigarettes and joints/day and the interaction term) and the doseresponse measures for the first trimester did not yield any significant associations (p > .05 for all analyses).

Fig. 1.

Percent hits across epochs by the 2-group prenatal substance exposure status (Panel A) and 3-group exposure status (Panel B).

There were no significant group differences in percent false alarms at epoch 1, no significant changes across time, or any group by time interaction effects for percent false alarms (p > .05, see Fig. 2). Results from the dose-response analyses indicated that the average number of cigarettes per day in the first trimester was significantly associated with the linear slope of false alarms across eight epochs (β = 0.23, p < .001). Simple slope analysis indicated that children with greater than average exposure to tobacco in the first trimester (1 SD above the mean) displayed a statistically significant increase in false alarms over the eight epochs (dy/dx = 0.22, SE = 0.08, p = .006) while children exposed to the average or lower amounts of tobacco in the first trimester (1 SD below the mean) had a statistically significant decrease in false alarms over time (dy/dx = −0.23, SE = 0.09, p = .007) (Fig. 3).

Fig. 2.

False alarms across epochs by the 2-group prenatal substance exposure status (Panel A) and 3-group exposure status (Panel B).

Fig. 3.

Association between average number of cigarettes/day in the first trimester and children’s false alarms across epochs.

3.3. Associations between prenatal substance exposure and children’s attentional set shifting in middle childhood

The DCCS scores ranged from 2.75 to 9.75 (M = 6.70, SD = 1.19). ANOVA with the two (tobacco exposed vs. non-exposed) or three (tobacco only, co-exposed, non-exposed) level grouping variables indicated no significant differences in set shifting (see Table 2). Similarly, results examining dose-response associations yielded no main or interaction effects for substance use variables across pregnancy or in the first trimester (p > .05).

Table 2.

Group differences in set shifting and working memory by prenatal exposure group.

	No-exposure (n = 29)	Tobacco only (n = 34)	Tobacco and cannabis (n = 60)
Set shifting	M (SD)	M (SD)	M (SD)	F	p
DCCS total	6.66 (1.06)	6.78 (1.37)	6.68 (1.16)	0.16	0.89
Forward span	5.73 (1.97)a	6.40 (2.30)	6.56 (1.89)b	2.99	0.05
Backward span	5.03 (1.91)	4.92 (2.13)	5.32 (1.24)	1.18	0.31

Note. Means with different superscripts were significantly different at p < .05.

3.4. Associations between prenatal substance exposure and children’s working memory in middle childhood

The number of trials correct for forward span ranged from 0 to 10 (M = 6.31, SD = 2.04) and for backward span ranged from 0 to 8 (M = 5.13, SD = 1.90). Results from ANCOVA with the two level grouping variable (tobacco exposed vs. non-exposed) and child age as a covariate indicated a significant effect of smoking group status on number of trials correct for forward F(1,130) = 5.61, p = .019 (M= 5.61, SD = 1.97 and M = 6.55, SD = 2.04, for non-exposed and tobacco exposed respectively). There were no group differences for backward span. Results from ANCOVA with the three-level grouping variable (tobacco exposed, co-exposed, non-exposed) indicated a marginal effect of the three-level grouping variable for forward, but not backward span (see Table 2). Simple contrasts indicated a significant group difference between the non-exposed compared to children co-exposed to both tobacco and cannabis for forward span but no significant group difference between non-exposed and tobacco only exposed or between the tobacco only vs. co-exposed groups. Results from regression analyses examining dose-response associations indicated no significant main or interaction effects for either forward or backward span and dose-response associations with across pregnancy or in the first trimester substance use variables.

4. Discussion

The main goal of the study was to examine the relationship between prenatal tobacco or tobacco-cannabis co-exposure and offspring attention and working memory in middle childhood. Specifically, we examined sustained attention, attentional set shifting and working memory. We hypothesized that children prenatally exposed to tobacco (with or without cannabis) would have lower sustained attention, lower cognitive flexibility in a computerized attentional set shifting task, and lower scores on a computerized working memory task; that performance would be lowest for the co-exposed compared to the other two groups; and that there would be dose-response associations, particularly for first trimester exposure. Results indicated lower initial attention (at the first epoch) for the tobacco only compared to non-exposed and children co-exposed to tobacco and cannabis, and marginally lower overall attention in the sustained attention task for tobacco exposed (tobacco only and co-exposed combined) compared to non-exposed children, although there were no differences in sustained attention across time. Children with greater first trimester tobacco exposure had more impulsive responding based on increases in the number of false alarms across the sustained attention task. However, contrary to expectations, children co-exposed to tobacco and cannabis had higher scores on the forward span indicating greater short-term storage compared to non-exposed children. This perplexing finding is potentially spurious and discussed further below. There were no group differences between children who were co-exposed compared to tobacco only exposed on any of the attention tasks or in working memory with the exception of the first epoch of the sustained attention task.

As we noted in the introduction, the literature on prenatal tobacco exposure and attention/working memory using task performance measures is limited and results are mixed. We hypothesized that one explanation for conflicting results may be the frequent co-use of tobacco with cannabis. However, we found no support for this hypothesis. In addition, tobacco exposure effects were sparse. Although results regarding the general lack of group differences need to be viewed with caution given the limitations of sample size, particularly for the tobacco only group, the overall pattern of results may reflect the literature indicating that tobacco exposure may have specific effects on behavioral regulation, but not cognitive correlates of self-regulation such as attention and working memory. For instance, results from a study (Wiebe et al., 2015) using laboratory measures of attention, working memory, and effortful control indicated no group differences (between any tobacco exposure vs. none) in measures of attention and working memory, but significant group differences in what the authors termed “motivational self-regulation” that included behavioral measures of effortful control. This may be similar to our finding of dose-response associations with first trimester tobacco exposure and increases in impulsive responding in the sustained attention task. Results regarding the general lack of associations with sustained attention are also similar to those reported by Cornelius et al. (2011) at 10 years of child age. These studies also used a prospective cohort design and demographically similar comparison group, with the study by Wiebe et al. (2015) also using multi-method assessments of exposure and including children who were co-exposed to cannabis. The lack of association between any tobacco exposure vs. no substance exposure and measures of attention is also consistent with previous results using sibling comparison design and examining within-family associations indicating no direct associations between any prenatal tobacco exposure and sustained attention or set-shifting (Knopik et al., 2022; Wiebe et al., 2015). However, Knopik et al. (2022) reported significant tobacco exposure associations with decreased performance on visual working memory using the spatial span task, although there were no associations with auditory working memory. This study, using a sibling comparison design, enabled researcher to distinguish within family processes from between family processes, and also had a primarily white, middle-income sample with at least some college education. It is possible that these design and demographic differences from the current sample may have accounted for differences in results regarding working memory. Finally, these results are also similar to those obtained in a smaller sample of children prenatally exposed to tobacco indicating no significant tobacco exposure effects on these cognitive aspects of performance but significant differences in behavioral self-regulation tasks that involved effortful control (Huijbregts et al., 2008). Taken together with the results from the current study, this pattern of consistent findings indicates that any tobacco exposure may not have a teratological association with attention and working memory, but that tobacco effects may be specific to aspects of performance that require motivational self-control or emotion regulation (see Wiebe et al., 2015).

The current results for the children co-exposed to tobacco and cannabis suggest that co-exposure did not have additive effects on child attention. Despite previous studies reporting that co-exposure may have double the impact of tobacco exposure alone on neonatal regulation (Stroud et al., 2020), that may have cascading effects on attention and working memory (Esterman and Rothlein, 2019), co-use effects have seldom been examined. Problems with sustained attention are critical because the ability to attend to a specific task for an extended period of time is intrinsic to academic performance (Isbell et al., 2018), and is implicated in myriad real-world tasks such as safe driving and fruitful communication. Extant empirical research indicates that suboptimal arousal may be implicated in attentional lapses (Esterman and Rothlein, 2019). Indeed, a study of 2-year-olds found that prenatal co-exposure was related to poor autonomic regulation during infancy, which was predictive of poor toddler emotion regulation at two years of child age (Eiden et al., 2018). Studies using these same two cohorts have also found association of co-exposure with children’s cortisol levels during both baseline and stress response (Eiden et al., 2020; Stroud et al., 2020). Both autonomic regulation and stress responses are related to arousal and are implicated in attentional challenges. This may be particularly salient for substance exposed children as chronic stress can permanently alter stress and arousal processes (Marin et al., 2011; McEwen, 2017), leaving substance exposed children especially vulnerable to impairments in attention. Many previous studies of tobacco exposure either do not measure or do not report other substance use, or statistically control for other substances in analyses when samples are polysubstance exposed. While our lab-based measures were conceptualized as measures of attention, attentional set shifting, and working memory, these behavioral tasks have a significant self-regulatory component as well including regulation of feelings of frustration or boredom and not responding impulsively. Given prior literature on significant tobacco exposure effects on effortful control and impulsivity (e.g., Wiebe et al., 2015), it is possible that future studies that include behavioral or affective coding of frustration or boredom during these tasks may be able to discriminate between attention and impulsivity or frustration.

We also hypothesized that there would be dose-response associations such that children exposed to higher average cigarettes or joints/day or the interaction of the two variables would have lower performance on the three tasks. Relatedly, we hypothesized that given higher levels of substance use in the first trimester before pregnancy recognition (Eiden et al., 2013), there would be stronger effects for first trimester exposure. Indeed, children with higher doses of prenatal tobacco exposure in the first trimester had greater increases in impulsive responding across time, supporting previous studies indicating tobacco exposure effects may be specific to behavioral regulation (Wiebe et al., 2015). Previous studies have also reported that 10-year-old children exposed to 10 or more cigarettes per day during the first trimester had higher behavior problems, including higher aggression and delinquency (Cornelius et al., 2011), and stronger first trimester associations between prenatal tobacco exposure and initiation of multiple substances by age 16 years (Goldschmidt et al., 2012). These results are not consistent with previous studies reporting second and third trimester tobacco exposure predicted sustained attention deficits at early school age, although the sample consisted of mothers who did not change the amount of smoking during pregnancy and had low levels of alcohol and cannabis use (Leech et al., 1999).

There was no association with the amount of cannabis exposure, or the interaction between tobacco and cannabis exposure on child outcomes. These results are similar to the conclusions from a recent systematic review of cannabis exposure on child cognitive outcomes (Torres et al., 2020). Authors concluded that both cannabis and non-cannabis exposed children performed within the normative range and the differences, while statistically significant across studies, were clinically non-significant. Results are also similar to those obtained by Cioffredi et al. (2022) indicating that in a large sample of 9–10-year-old children, prenatal cannabis exposure was associated with maternal and teacher reports of attention problems but was not associated with cognitive performance on behavioral tasks of brain activation using functional imaging. These results taken together with the current findings also suggest that perhaps child behaviors such as impulsivity and hyperactivity may be driving the differences reported in earlier studies using maternal or teacher reports of behavior problems. One limitation of the current study was our inability to examine effects of cannabis exposure alone. Although we used statistical models to examine potential dose-response associations with cannabis, we did not have mothers in our sample who used cannabis in the absence of tobacco. Thus, these results regarding cannabis dose-response need to be viewed with caution. It is also possible that because the women in this study were recruited primarily on the basis of tobacco use in pregnancy, they were not heavier users of cannabis and samples that include cannabis use alone or in combination with tobacco and reflect heavier use patterns may yield different results.

Contrary to expectations, children co-exposed to tobacco and cannabis had higher scores on short-term memory performance compared to non-exposed children. This result is puzzling, and given its contrary nature, we are hesitant to speculate. This result needs further replication. It is important to note that to our knowledge, although there are a number of studies on parent or teacher reported attention and behavior problems among children prenatally exposed to cannabis (see Eiden et al., 2023), there are few studies of cannabis exposure examining working memory performance using laboratory-based measures among children co-exposed to tobacco and cannabis. Recent results from the Adolescent Brain and Cognitive Development (ABCD) study indicated no differences between adolescents who were cannabis exposed and those who were non-exposed on working memory (EN-Back task) performance or any differences in patterns of brain activation during this task (Cioffredi et al., 2022). However, studies examining acute effects of specific cannabinoids (e.g., cannabidiol or CBD vs. tetrahydrocannabinol or THC) among adults suggest that CBD may have positive effects on cognitive performance, especially in the context of other clinical conditions including chronic cannabis use (see Morie and Potenza, 2021), while THC may have a negative effect on visual working memory performance (Adam et al., 2020). Future studies of prenatal exposure that include biomarkers of specific cannabinoids present in children at the time of testing may be helpful in this regard.

5. Strengths, limitations, and conclusion

Our use of multiple measurements (i.e., self-reports, infant meconium, maternal saliva) to determine tobacco and cannabis smoking during pregnancy, including lab measures of attention and working memory, the prospective design, and having demographically similar comparison groups are strengths of this study. However, there are limitations as well. One limitation of this study may be the potential lack of generalizability. This sample was restricted to pregnant mothers with low levels of alcohol use, and no illicit drug use other than low to moderate levels of cannabis (illegal at the time of recruitment). Heavy poly-drug users may show different patterns of results. In addition, the sample for this study was drawn from a larger cohort, and compared to those who were not selected for the middle childhood phase (based on being heavier tobacco and/or cannabis users), mothers included in the middle childhood phase had slightly higher levels of education (12.56 years compare to 12.06 years, p = .03), though there were no differences on any other demographic variables. Future studies might also seek to disentangle the effects of prenatal exposure and postnatal exposure, although this is practically difficult to do given the strong associations between pre- and postnatal smoking. Perhaps if paired with biomarkers or current postnatal cannabis exposure, future studies could also test for potential interactions between prenatal and postnatal co-exposure. We were also not adequately powered to fully examine group differences, with smaller group sizes in the tobacco only and non-exposed group. Thus, we were unable to examine sex differences in exposure effects, or exposure by sex interactions. Extant literature suggests that females may be more vulnerable to hyperarousal, especially when stress-induced, while males are more likely to have cognitive deficits, including deficits in attention (Bangasser et al., 2019). Future studies with larger sample sizes are needed for replication and extension into examining if there were sex differences in exposure effects.

Another limitation of this study is that we did not use a genetically informed design, which precluded us from disentangling prenatal exposure from genetic transmission. While there is evidence of familial transmission in existing literature, we did not have parental performance on the cognitive tests. Thus we were unable to explore how parental cognitive functioning in the specific areas of attention and working memory may predict or attenuate child outcomes. Future research may be better placed to examine such facets by utilizing a genetically informed design and capturing attention and memory related outcomes for both child and parents. It is also challenging to compare results across studies using varying assessment of executive functioning. Some of the inconsistent results found among studies may indeed be related to variations in outcome measurement. Finally, we did not have a cannabis only group, so while we were able to examine tobacco only exposure, and co-exposure to tobacco and cannabis, we were unable to examine these groups in comparison to cannabis only exposure. Future studies may benefit from including this additional group in recruitment efforts, and this recruitment process may become easier as both the social and legal acceptability of cannabis use continues to grow.

Data availability

Data will be made available on request.