Preconception and prenatal cannabis use and the risk of behavioural and emotional problems in the offspring; a multi-informant prospective longitudinal study Free

Abstract

Background

Studies of the long-term consequences of maternal cannabis use on child development beyond the neonatal period are sparse. In the current study, we use a multi-information approach to assess the association of prenatal cannabis exposure and child behavioural and emotional functioning. To explore the possible causal nature of the association, we investigated whether maternal tobacco and paternal cannabis use during pregnancy were also associated with child problems.

Methods

The study population included children of a population-based birth cohort in The Netherlands (n = 5903). Information on parental cannabis use was collected using questionnaires; urine of mothers was analysed for the presence of cannabis metabolites. Child behavioural and emotional problems at approximately 7–10 years were measured using validated teacher-, child- and mother-reports.

Results

Our findings show associations of maternal cannabis use during pregnancy with offspring externalising problems (B = 0.53; 95% CI: 0.29−0.77), but not with internalising problems (B = −0.10; 95% CI: −0.31−0.11). However, maternal cannabis use before pregnancy was also associated with offspring externalising problems (B = 0.27; 95% CI: 0.02−0.52). Further, cannabis use by the father was associated with child externalising problems (B = 0.36; 95% CI: 0.22−0.49) but not internalising problems.

Conclusions

Prenatal exposure to maternal cannabis use is specifically associated with offspring behavioural problems, but not emotional problems. This association is probably not due to an effect of intrauterine cannabis exposure on fetal development, because both maternal and paternal cannabis exposure during pregnancy were related to offspring externalising problems. Our findings suggest that the association can be explained through residual confounding, most likely through shared genetic vulnerabilities for parental cannabis use and offspring behavioural problems.

Introduction

Increasingly, pregnant women use cannabis, with reported prevalence of 2–10%.^1–4 The psychoactive ingredients of cannabis, including Δ-9-tetrahydrocannabinol (THC), can cross the placenta and reach the fetus.⁵ Prenatal cannabis exposure has been associated with adverse birth outcomes.^6–10 However, studies on long-term consequences of maternal cannabis use and child development are sparse. Only two longitudinal studies have addressed this,^11,12 and these demonstrate childhood cognitive deficits, impairments in inhibitory control, hyperactivity, impulsivity, attention problems, conduct problems, adolescent delinquency and increased risk of drugs abuse later in life.^13–20 A recent review emphasizes the importance of genetic factors, gene-environment interactions and comorbidity, as these may confound the association between gestational substance use and offspring behaviour problems.²¹ Previously, in a population-based prospective cohort, we showed that prenatal cannabis exposure was related to fetal growth retardation, decreased fetal blood flow^22,23 and behaviour problems in toddlerhood.²⁴ Furthermore, using neuroimaging we reported that prenatal cannabis exposure was associated with cortical thickness differences.²⁵

Assuring that associations between maternal substance use and child psychopathology are not confounded by other factors is difficult. A method that supports causal inference includes comparing the associations for maternal and paternal substance use during pregnancy.^26,27 Yet, to date very few studies on prenatal substance use and offspring development have applied this approach. Moreover, it is also important to explore the effects of cannabis use exclusively preceding pregnancy. Disentangling the effects of prenatal cannabis exposure from prenatal tobacco exposure is challenging, as prenatal exposure to tobacco can also negatively influence child development. Therefore, in the current study, we compared several non-overlapping groups: (i) pregnant women who used cannabis; (ii) women who used cannabis before pregnancy only; (iii) pregnant women who smoked tobacco only; and (iv) non-using pregnant women (reference). Finally, the importance of multiple informants for assessing child behavioural and emotional problems has long been emphasized.²⁸ Correlations between different informants are generally low, and may reflect variations in children’s behaviours across diverse settings and relational circumstances^28–30 or informant error.³¹ Summarized, we aim to investigate the association of prenatal cannabis exposure with child behavioural and emotional functioning using a multi-information approach.

Methods

Design and setting

The present study is embedded in the Generation R Study, a population-based birth cohort in Rotterdam, The Netherlands.³² All children were born between April 2002 and January 2006 in the city of Rotterdam, and follow-up is ongoing. The Medical Ethics Committee of the Erasmus Medical Centre, Rotterdam, has given ethical approval of the study. Written informed consent was obtained from all participants. Only children participating in the prenatal assessment and the follow-up examination between 5 to 10 years of age (n = 7641) were considered (Figure 1). We excluded 476 children, as information on maternal cannabis was unavailable. Information on behavioural and emotional functioning was obtained in n = 5903 (77.3%) children; this information was reported by teachers (n = 3953) at 7/8 years, mothers (n = 4200) and children (n = 3726) at 9 years.

Maternal cannabis use

To optimize the cannabis assessment, two information sources were used: (i) maternal self-reports of cannabis use; and (ii) maternal urinary THC levels. In the first trimester, mothers indicated whether they used cannabis before or during pregnancy, and whether they were still using cannabis.⁴ Information about the product used and frequency of use (daily, weekly, monthly) was available. Urine samples were collected in early, mid and late pregnancy; the first available sample was used for urinalysis in a subset of the cohort.⁴ Urine samples were tested for 11-nor-Δ9-THC-9-COOH using the DRI^® Cannabinoid Assay (Microgenics) with a cutoff value of 50  μg/l as recommended. Agreement between self-reports and THC levels (Yule’s Y = 0.77) was substantial.⁴

Maternal tobacco smoking

Maternal tobacco smoking was assessed by postal questionnaires in each trimester.³³ Maternal smoking was categorized, using these questionnaires, into ‘no smoking during pregnancy’, ‘until pregnancy was known’ and ‘continued during pregnancy’. Women who used tobacco until the pregnancy was known were included in the reference group, because previously we could not document an association with adverse neurodevelopmental outcomes.^33,34

Paternal smoking and cannabis use

In the first trimester, mothers provided information on paternal tobacco smoking and cannabis use. We used maternal reports of tobacco smoking and cannabis use of the father of the child, because not all fathers completed a questionnaire. Maternal report of paternal tobacco or cannabis use was highly correlated to paternal self-reports (r_{tobacco use} = 0.85; P <0.001, and r_{cannabis use} = 0.80; P <0.001).

Child behavioural and emotional functioning

At age 9 years, children were asked to complete the validated Brief Problem Monitor (BPM), to obtain the child’s self-report of behavioural and emotional problems.³⁵ The BPM contained 19 items and was rated on a 3-point scale (0 = not true, 1 = somewhat or sometimes true and 2 = very or often true). Mothers filled out the school-age version of the Child Behavior Checklist (CBCL/6–18) when their child was approximately 9 years. The CBCL measures the degree of children’s behaviour problems.³⁵ We collected teacher ratings of child behavioural and emotional problems at 7/8 years using the Teacher Report Form (TRF/6–18).³⁵ Both CBCL and the TRF contain 118 problem items based on the preceding 2 months, rated on the same 3-point scale.

A broadband internalizing/emotional problems (i.e. anxiety/depressiond, withdrawal and somatic complaints) and externalizing/behavioural problems (aggressive behaviour and rule-breaking behaviour) scales were computed. Additionally, we used clinical cutoff scores (80th percentile) to classify children as having problems in the clinical range. Good reliability and validity have been reported for the CBCL, TRF and BPM.³⁶ In line with literature on cross-informant agreement,³⁷ correlations between parent, teacher and child-self-reports were modest (r_CBCL-TRF = 0.30, r_CBCL-BPM = 0.45 and r_TRF-BPM = 0.17 for the total problems score). Further, we found moderate correlations between the internalizing and externalizing scales in the same informant (r_{TEACHER REPORT} = 0.36, r_{SELF REPORT} = 0.33 and r_{MATERNAL REPORT} = 0.53).

Covariates

We considered several maternal and child characteristics as potential confounders.³⁸ Maternal characteristics were: maternal age at intake, ethnicity, educational level, pre-pregnancy body mass index, and alcohol use. Child characteristics were: age at assessment, gender, gestational age at birth and birthweight. Maternal ethnicity was defined according to the classification of Statistics Netherlands. Maternal educational level was categorized into primary (no or primary education), secondary (lower and intermediate vocational training) and higher (higher vocational education and university) education. Information on maternal alcohol use was collected using questionnaires in each trimester, as described above for smoking. Maternal psychopathology during pregnancy was assessed using the validated Brief Symptom Inventory in mid-pregnancy.^39,40 Gestational age at birth, weight at birth and gender of the child were extracted from medical records.

Statistical analyses

Descriptive statistics of the four groups were compared using analyses of variance (ANOVA) for parametric data, Kruskal-Wallis tests for non-parametric data and chi-square tests for categorical data. Supplementarily, we provide descriptive information on the frequency of tobacco and alcohol use per trimester (Supplementary Table 1, available as Supplementary data at IJE online). To test the associations of maternal and paternal cannabis use and childhood behavioural and emotional problems, both linear and logistic (for problems in the clinical range) regressions were performed. For the continuous outcomes, these were square root transformed to approximate a normal distribution. Covariates were selected based on previous literature and on the 5%-change-in-estimate criterion.⁴¹ We performed additional adjustment for internalizing problems if outcome was externalizing problems, as recommended by Achenbach and colleagues, as internalizing and externalizing problems are not fully independent of one another.⁴² The results using cutoffs for behavioural and emotional problems are only presented in the Supplementary material, available as Supplementary data at IJE online. Additionally, we tested the association of maternal alcohol use during pregnancy and childhood behavioural and emotional problems.

Missing information on the covariates was between 0% and 5%, with the exception of maternal psychopathology (18.1%). To avoid the bias of complete case analysis, we accounted for missing information on the covariates by using a multiple imputation technique, and generated 10 imputed datasets.

Results

Descriptive information

Women using cannabis during pregnancy were younger, had higher psychopathology scores, had lower educational levels and more often used alcohol during pregnancy (Table 1). Children exposed to cannabis during pregnancy had a lower birthweight (Table 1).

Cannabis use and smoking

Table 1 demonstrates that, of the women who used cannabis during pregnancy, only 9.9% never smoked any tobacco during pregnancy. Pregnant women using cannabis consumed this regularly: 51 (30.2%) women reported daily consumption, 48 (28.4%) women reported weekly and 21 (12.4%) women reported monthly cannabis consumption; 49 (29.0%) cannabis users did not report frequency of use. Pregnant women consumed either marijuana (n = 86) or hashish (n  =  55); in 28 women the cannabis product was unknown. Further, 118 women consumed cannabis until pregnancy was known and 23 women continued cannabis use even after pregnancy was known.

Substance use by the father

Most partners of cannabis-consuming pregnant women also used cannabis (79.3%, see Table 1) and tobacco (79.9%). Paternal alcohol consumption was highest (89.8%) in the group of mothers who used cannabis before pregnancy.

Maternal cannabis and tobacco use in relation to child behavioural and emotional problems

Table 2 demonstrates that maternal cannabis use during pregnancy was associated with teacher-reported externalizing problems in children (Model I: B = 0.55, 95% CI: 0.30 to 0.81), but not with internalizing problems (Model I: B = −0.06, 95% CI: −0.16−0.28). Maternal cannabis use before pregnancy and tobacco smoking during pregnancy were related to externalizing problems in offspring (Model I: B_{cannabis before} = 0.26, 95% CI: −0.00 to 0.52, B_tobacco = 0.25, 95% CI: 0.12 to 0.37). These associations remained when additionally adjusting for internalizing problems (Model II). The effect estimates (Model II) of cannabis use during pregnancy and externalizing problems were higher than the effect estimates of tobacco use during pregnancy and child externalizing problems (B_difference = 0.34, 95% CI: 0.07 to 0.60).

Next, we examined the associations of maternal cannabis and tobacco smoking and child self-reported behavioural and emotional problems. Consistent with the results based on the teacher reports, prenatal cannabis exposure during pregnancy was related to higher scores of externalizing problems (Model II: B = 0.48, 95% CI: 0.09 to 0.87), but not with internalizing problems. Again, maternal cannabis use before pregnancy (Model II: B = 0.37, 95% CI: 0.01 to 0.74) and tobacco smoking during pregnancy (Model II: B = 0.26, 95% CI: 0.06 to 0.45) were related to externalizing problems in offspring. There was no association between maternal tobacco smoking and internalizing problems. Self-reported externalizing problems did not differ between tobacco-exposed or cannabis-exposed children (Model II: B_difference = 0.22, 95% CI: −0.22 to 0.66). The results of the logistic regression models were similar (Supplementary Table 2, available as Supplementary data at IJE online).

Paternal cannabis use and child behavioural and emotional problems

Paternal cannabis use was associated with offspring teacher-reported externalizing problems (Table 3; Model II: B = 0.36, 95% CI: 0.22 to 0.49), but not with internalizing problems (Model II: B = −0.07, 95% CI:−0.19 to 0.06). When analyses were based on child self-reports, we found that paternal cannabis use was associated with neither externalizing problems (Model II: B = −0.21, 95% CI:−0.01 to 0.43) nor internalizing problems (Model II: B = 0.15, 95% CI: −0.09 to 0.39).

In sensitivity analyses of paternal cannabis use, we excluded the mothers who used cannabis before or during pregnancy. Results were very similar; paternal cannabis use was associated with teacher-reported offspring externalizing problems (Model II: B = 0.31, 95% CI: 0.14 to 0.48), but again not to internalizing problems (Model II: B = −0.07, 95% CI:−0.22 to 0.08).

Maternal reports of child behavioural and emotional problems

Supplementary analyses were performed using the maternal reports of child behavioural and emotional problems (Supplementary Table 3, available as Supplementary data at IJE online). In these analyses, the mother reported on both exposure and outcome. We found no associations with maternal cannabis use during pregnancy or tobacco smoking and child behavioural and emotional problems when taking into account the confounders, although the sample size for this analysis was larger than in other analyses.

Maternal alcohol use and child behavioural and emotional problems

We found that prenatal exposure to alcohol use was associated with less teacher-reported internalizing problems (B = −0.20, 95% CI: −0.29 to −0.12, P <0.001), whereas we found no association with teacher-reported externalizing problems (B = 0.08, 95% CI: −0.02 to 0.18, P = 0.10). Additionally, when children’s self-report was used, prenatal alcohol exposure was not related to internalizing (B = −0.03, 95% CI: −0.19 to 0.12, P =  0.68) or externalizing problems ((B = −0.04, 95% CI: −0.17 to 0.10, P =  0.62).

Discussion

Main findings

The aim of this study was to investigate the association between prenatal cannabis exposure and child behavioural and emotional problems, using a multi-informant approach. To study causality of this association, different contrasting or negative exposures, including tobacco exposure, cannabis use before pregnancy and paternal cannabis use, were used. Overall, our findings consistently show that maternal cannabis use during pregnancy is associated with child externalizing problems. However, maternal cannabis use before pregnancy and tobacco smoking during pregnancy were similarly related to an increased risk of offspring externalizing problems. Moreover, cannabis use of the father was to a comparable extent associated with more child externalizing problems.

Interpretation

Our findings suggest that cannabis exposure is specifically associated with offspring externalizing problems, but not internalizing problems. Importantly, both maternal and paternal cannabis exposure during pregnancy were related to externalizing problems. This was also observed for paternal cannabis use without any maternal cannabis use, and thus the association is most certainly partly due to shared familial or genetic confounding factors. Shared familial confounding factors, such as socioeconomic position and parental behaviours (e.g. poor diet) associated with both parental smoking and offspring behavioural problems, could confound the association of prenatal cannabis use and offspring behaviour.⁴³ In our analyses we adjusted for several socioeconomic indicators, and thus it is more likely that the association is due to common genetic factors. Smoking cannabis may be a marker for underlying psychiatric problems in parents (e.g. conduct disorder), and risk factors that predispose to smoking cannabis could be the same that predispose offspring to behavioural problems. For example, externalizing behaviour, smoking and substance use initiation have been shown to have common genetic and environmental origins.⁴⁴

Further, the estimate of maternal cannabis use during pregnancy and offspring externalizing problems is somewhat higher than the estimate of paternal cannabis use and offspring externalizing behaviour. This may reflect an accumulated genetic predisposition, because 80% of the fathers use cannabis when mothers do so. Assortative or mating in substance use disorders and other psychiatric disorders could be a plausible mechanism that results in an increased risk for externalizing problems in offspring of cannabis-using parents.⁴⁵ Also, maternal cannabis use before pregnancy was associated with offspring externalizing problems, which further supports that these susceptibilities co-occur in parents and their offspring. This may also implicate involvement of preconception effects, including epigenetic transgenerational inheritance of co-occurring substance use and psychiatric disorders.⁴⁶

Second, the associations of maternal cannabis use and externalizing behaviour were more prominent when the teacher reported on child problems. The teachers’ perspective is valuable and additive, since teachers observe child behaviour in task-oriented situations requiring concentration. Moreover, teachers can compare a particular child’s behaviour with that of a large group of classmates at the same developmental level. Last, teachers are unaware of any parental substance use during pregnancy. Importantly, the associations of prenatal cannabis use and externalizing problems were consistent across teacher- and child-reports, despite their low correlations.

Finally, residual confounding factors are certainly possible. It is known that mothers who use cannabis during pregnancy have different characteristics from those who do not,⁴⁷ and these characteristics could be related to increased risk for offspring psychiatric problems. Although we have adjusted for multiple confounding factors, residual confounding may still be present. Indeed, the analyses contrasting the findings of prenatal cannabis exposure with other exposure groups with similar confounding patterns (e.g. maternal tobacco smoking; maternal cannabis use exclusively before pregnancy) are compatible with the possibility of residual confounding.

Previous studies

Previous studies have demonstrated that maternal cannabis use during pregnancy was related to childhood externalizing problems, impulsivity and delinquency, assessed through mother-report, teacher-report and neuropsychological tasks.^19,48 Although these studies were able to take into account many confounding factors, these studies were not able to contrast the findings with the potential effects of paternal cannabis use. This is important, as contrasting exposure can give insight in whether the observed effect is a direct biological intrauterine effect or whether the effect could be due to shared confounding or genetic factors.^26,27

The seemingly contradictory finding that maternal alcohol use was related to less internalizing problems and no externalizing problems is a known phenomenon, as light alcohol consumption is a marker of relative socioeconomic advantage,^49,50 also in the current population.⁵¹

Strengths and weaknesses

Despite several strengths, such as the prospective nature of the study, the multiple informants of the outcome, information on various relevant confounders and the availability of contrasting exposures, some limitations should be discussed. First, it was not feasible to obtain clinical diagnoses. However, the instruments used are suitable for the general population as they map a continuum of psychopathology with good reliability and validity.³⁶ Moreover, cannabis use was assessed only once during pregnancy, and information about postnatal cannabis use was not available. Nevertheless, self-reports were in agreement with urinary THC levels assessed in a subsample,⁴ and the cannabis use prevalence in our sample was similar to national prevalence of cannabis use among young women.⁵² We could not investigate maternal cannabis use in the absence of paternal cannabis use, as approximately 80% of the mothers using cannabis during pregnancy had cannabis-using partners. Last, we were unable to study the directionality of the associations between parental cannabis use and parental psychopathology: whether cannabis use mediated the association between parental psychopathology and offspring emotional and behaviour problems, or contributed to the occurrence of parental psychiatric symptoms. Future studies with repeated assessments of parental psychopathology and substance use before, during and after pregnancy will be needed to address this.

Implications

The current study shows that maternal cannabis use was associated with offspring externalizing problems. Nonetheless, as offspring externalizing problems were also associated with maternal cannabis use before pregnancy, maternal tobacco smoking during pregnancy and paternal cannabis use, it is unlikely that intrauterine exposure to cannabis causally increases the risk for child externalizing problems.

Funding

This work was supported by the Erasmus University Rotterdam (EUR Fellow 2014), Stichting Volksbond Rotterdam, the Netherlands Organization for Health Research and Development (ZONMW Vici project 016.VICI.170.200), the European Union Seventh Framework Program (FP7/2007–2013), ACTION: Aggression in Children: Unravelling gene-environment interplay to inform Treatment and InterventiON strategies (grant number 602768), the European Union’s Horizon 2020 research and innovation programme (grant agreement No.633595 DynaHEALTH and No.733206 LifeCycle). The funding agencies had no role in the design or conduct of the study, collection, management, analyses or interpretation of the data, or preparation, review or approval of the manuscript or the decision to submit it for publication.

Acknowledgements

The Generation R Study is conducted by the Erasmus Medical Centre in close collaboration with the Municipal Health Service Rotterdam area, the Rotterdam Homecare Foundation and the Stichting Trombosedienst & Artsenlaboratorium Rijnmond (STAR), Rotterdam. We gratefully acknowledge the contribution of general practitioners, hospitals, midwives and pharmacies in Rotterdam. Supercomputing computations were supported by the NWO Physical Sciences Division: Exacte Wetenschappen, and SURFsara: Lisa computer cluster [www.surfsara.nl].

Conflict of interest: None declared.

References