Trends and correlates of cannabis use in pregnancy: a population-based study in Ontario, Canada from 2012 to 2017

Objective Forthcoming legislative changes will legalize and make cannabis widely available in Canada. We conducted an analysis of Ontario’s birth registry to determine recent trends and correlates of cannabis use in pregnancy. Methods We conducted a population-based retrospective cohort study assembled from the Better Outcomes Registry & Network (BORN) Ontario database, covering live births and stillbirths in Ontario between April 2012 and December 2017. Trends in self-reported cannabis use in pregnancy were analyzed according to maternal age and area-level socio-economic status (SES) using log binomial regression analysis. Results A total of 10,731 women reported cannabis use in pregnancy. Prevalence increased from 1.2% in 2012 to 1.8% in 2017 (p-trend, < 0.001), equivalent to a relative increase of 61% (relative risk [RR] 1.61, 95% confidence interval [CI] 1.51 to 1.72). The crude prevalence of cannabis use in pregnancy among women aged 15 to 24 years and in the lowest two area-level income quintiles was 6.7%, compared to 0.3% among women aged 35 years and over in the highest three income quintiles (RR 24.59, 95% CI 21.98 to 27.52). A majority (52.0%) of cannabis users were aged 15–24 years and 54.7% of users were in the lowest two income quintiles. Conclusion Cannabis use in pregnancy has increased since 2012 in Ontario and was reported in about 2% of pregnancies in 2017. Increases were predominately among women of younger ages and those of lower SES, and these groups account for half of users. Promoting cannabis cessation in pregnancy could lead to improved perinatal and later childhood outcomes and reduce health inequalities. Electronic supplementary material The online version of this article (10.17269/s41997-018-0148-0) contains supplementary material, which is available to authorized users.


Introduction
In 2017, the Government of Canada introduced legislation to allow the consumption, sale, and distribution of cannabis (marijuana) for non-medical uses by October 2018 (Government of Canada 2016). Cannabis is the most widely used illicit drug in Canada, and the prevalence has been increasing in recent years following a period of decline from 2004 to 2011 (Government of Canada 2015). The 2015 Canadian Tobacco, Alcohol and Drugs Survey indicated that the prevalence of past-year cannabis use was 12% of the overall population (3.6 million individuals), but was 21% among young women and men 15 to 19 years of age (Government of Canada 2015). The prevalence among women increased from 6% in 2011 to 10% in 2015, although only 3% of women reported a weekly or greater frequency of use. Legalization of recreational cannabis in the American states of Washington and Colorado led to reductions in perceived harmfulness of cannabis and increases in use among 8th and 10th graders (Cerda et al. 2017). In Canada, legalization may likely increase the availability of cannabis (Pacula et al. 2013), decrease the perceptions of harm (Ko et al. 2015), and lead to further increases in use, including among pregnant women.
A 2017 review concluded that although robust evidence exists for certain therapeutic effects of cannabis, including treatment of chronic pain and as an antiemetic for chemotherapy-induced nausea, there exists only limited or moderate evidence for other purported therapeutic uses, including treatment of appetite and weight loss associated with HIV/AIDS, improving anxiety symptoms, and improving PTSD (National Academies of Sciences Engineering and Medicine-Health and Medicine Division; Board on Population Health and Public Health-Practice Committee on the Health Effects of Marijuana 2017). Acute effects of cannabis use include disruption of psychomotor function, global cognitive impairments, and impairments in abstract thinking and executive function (Crean et al. 2011;Hoch et al. 2015).
Epidemiological research suggests that chronic cannabis use may lead to substance dependence in up to one in ten users (Hoch et al. 2015), and may have impacts on psychosocial development, educational attainment, and mental health in adolescents (Hall and Degenhardt 2009). Other effects on respiratory function and cardiovascular events have been suggested, although evidence remains limited (National Academies of Sciences Engineering and Medicine-Health and Medicine Division; Board on Population Health and Public Health-Practice Committee on the Health Effects of Marijuana 2017; Goyal et al. 2017). Cannabis intoxication, however, has been associated with a moderate and statistically significant increase in motor vehicle crash risk in an analysis of 28 observational studies (pooled odds ratio [OR] 1.36, 95% CI 1.15 to 1.61) (Rogeberg and Elvik 2016).
Potential health effects of perinatal cannabis exposure on pregnancy and neonatal outcomes have been investigated using various observational designs with inconsistent findings (Varner et al. 2014;Saurel-Cubizolles et al. 2014;Leemaqz et al. 2016;Hayatbakhsh et al. 2012;Fergusson et al. 2002;Conner et al. 2016;Gunn et al. 2016;El Marroun et al. 2009). Although statistically significant associations have been reported between prenatal cannabis use and low birth weight (LBW, < 2500 g) and admission to a neonatal intensive care unit (Gunn et al. 2016), it has not been robustly associated with fetal development, LBW, small for gestational age, or preterm birth after adjusting for tobacco use and other confounding factors (Fergusson et al. 2002;Conner et al. 2016;Witter and Niebyl 1990). A large study of 26,654 singleton births between 2009 and 2014 in Southwestern Ontario reported an adjusted OR of 2.7 (95% CI 1.7 to 4.4) for the association between history of cannabis use and LBW and an adjusted OR of 1.7 (95% CI 1.1 to 2.5) for < 10th percentile birth weight adjusted for history of tobacco smoking (Campbell et al. 2018). Inconsistencies in findings and methodological differences across epidemiological studies of cannabis exposure underscore the urgency for additional high-quality population-based data on this topic.
The Better Outcomes Registry and Network (BORN) is a comprehensive birth registry in the province of Ontario and encompasses nearly 40% of all births in Canada (Dunn et al. 2011). It contains robust data on prenatal screening, pregnancy complications, intrapartum events, admission to neonatal intensive care, newborn screening, and data on maternal exposures, including substance use. In this analysis, we evaluated recent trends and correlates of cannabis use in pregnancy.

Methods
Research ethics board approval for this study was obtained from the Ottawa Health Science Network Research Ethics Board and the Children's Hospital of Eastern Ontario.

Study population and data source
The BORN Ontario birth registry captures all births occurring in the province. The routine data collection includes information on maternal demographics and health behaviours, preexisting health problems, obstetric complications, and birth outcomes. Data are collected from medical records, clinical forms, and patient interviews when a woman is admitted to hospital to give birth. An ongoing program of data quality checks and formal training sessions assures a high level of data quality (Dunn et al. 2011).
For this study, we conducted a retrospective cohort analysis of the most recent available data from the BORN information system comprising women who delivered a singleton live birth or stillbirth in an Ontario hospital between April 1, 2012 and December 31, 2017.

Primary outcome
Maternal self-report of cannabis use in pregnancy was determined from the BORN database using a multi-select variable which collected information on substance use during pregnancy, including cannabis, cocaine, hallucinogens, opioids, and other substances.

Covariates
Maternal age was derived from maternal birth date and date of delivery. Area-level income quintiles and residential area population classification were assigned from the Canadian census based on data aggregated to a women's local dissemination area. To assign these data, maternal postal codes in the database were linked using the Postal Code Conversion File Plus (PCCF+) version 6, developed by Statistics Canada (Statistics Canada 2016). This program assigns women to their dissemination area using six-digit postal codes and uses probabilistic assignment in cases where areas overlap more than one postal code. Smoking, alcohol exposure, and pre-existing diabetes and hypertension were included as covariates.

Statistical analyses
We conducted descriptive analyses, including the Cochran-Armitage test to evaluate trends in cannabis use over time in the overall sample and by age groups and area-level income quintiles. Log binomial regression was used to assess the effect of year, maternal age, and income on rates of cannabis use, and to test for interaction between year and maternal age, and year and income. Multivariable models included year, age, income, and population size. A chi-square test was used for assessing differences in age and income distribution separately in cannabis users and in non-users.

Results
The initial study sample comprised 783,419 pregnancy records. In total, 50,601 (6.5%) records were excluded due to missing information on cannabis use, maternal age, or arealevel income. Records with maternal age less than 15 or greater than 50 were excluded (n = 219), yielding a final sample size of 732,818 for the analysis. Excluded participants were older, more likely to report alcohol use in pregnancy, to have diabetes, and higher income (Supplemental Table).
Among women with singleton live births in the analysis, 64.1% (n = 469,725) were aged 25 through 34 years, 13.4% were less than 25 years, and 22.5% were greater than 35 years of age. A majority (69.7%) were from urban areas with greater than 100,000 inhabitants and 24.8% (n = 182,009) fell into the highest income quintile based on residential postal code. 1.5% (n = 10,731) of women used cannabis in pregnancy and 9.1% used cigarettes at the first prenatal visit (Table 1). The lowest two income quintiles accounted for more than half (55%) of the population of cannabis users compared to about 30% of the population of non-users and these proportions were similar throughout the study period (p = 0.90). Cannabis use was substantially higher among women who reported alcohol and tobacco use in pregnancy (12.0% and 10.3%, respectively).
In Ontario, the overall prevalence of cannabis use in pregnancy increased from 1.2% in 2012 to 1.8% in 2017 (p-trend, < 0.001). This equates to a relative increase of 61% between 2012 and 2017 (relative risk [RR] 1.61, 95% confidence interval [CI] 1.51 to 1.72) adjusting for maternal age, income, and population size ( Table 2). Prevalence of prenatal cannabis use was highest among women aged 15-24 years, and increased from 4.9% in 2012 to 6.5% in 2017 (Fig. 1a). A test of interaction indicated that increases over time differed by age group (p < 0.001). There was an 18% relative decrease in prevalence with each additional year of maternal age (RR 0.82, 95% CI 0.819-0.825). Cannabis use varied by income, with women in the lowest two quintiles reporting the highest usage (2.2% in 2012, increasing to 3.1% in 2017) as compared to women in the highest three quintiles (0.8% in 2012, and increasing to 1.2% in 2017) (Fig. 1b). A test of time by area-  (Table 3). Across all ages, the relative risk of cannabis use in pregnancy was higher among women in the bottom two quartiles of area-level income.
Among those who reported cannabis use in pregnancy, the majority (52.0%) were aged 15-25 years, although due to increases in overall usage, this proportion declined with time from 60.1% in 2012 to 42.7% in 2017 (Fig. S1). There was a corresponding increase in the proportion of users at older ages over time between 2012 and 2017 (p < 0.001).

Discussion
There has been an increase in cannabis use among pregnant women in Ontario since 2012, and the increase occurred predominately among younger women and those of lower socioeconomic status (SES). In these groups, the prevalence of selfreported cannabis use during pregnancy was 5.5% and 3.1%, respectively, compared with 1.8% in the overall obstetrical population. Young women under 25 years of age and women from lower area-level SES represented more than half of all women who reported using cannabis during pregnancy, suggesting that the burden of potential health implications of cannabis use in pregnancy may be greater among these women.
Previous studies have indicated an association between prenatal cannabis use and adverse pregnancy outcomes such as preterm birth, small for gestational age, and admission to neonatal intensive care (Leemaqz et al. 2016;Hayatbakhsh et al. 2012;Campbell et al. 2018). In addition, there is evidence to suggest an increased risk of adverse developmental consequences in children with intrauterine exposure to cannabis (Fried and Watkinson 1990), although few studies exist  (Hayes et al. 1991;McLemore and Richardson 2016). With forthcoming legalization leading to increased availability of cannabis in Canada, it is likely that the prevalence of prenatal exposure will continue to increase. Identification of women who use cannabis in pregnancy may facilitate interventions to improve obstetrical and neonatal outcomes (Fried and Makin 1987). The potential for underreporting of cannabis exposure in pregnancy is a challenge for understanding prevalence and identifying opportunities for intervention (Cook et al. 2017). BORN data on the use of cannabis in pregnancy are from self-reports, which are likely to be influenced by stigma, social desirability bias, and fear of intervention by child protection services (Greaves and Poole 2004;Stone 2015;Yonkers et al. 2011;Jacobson et al. 1991;Johnson and Fendrich 2005). Previous studies have attempted to validate self-reports of cannabis with urinary biomarkers of exposures and have found moderate correlation (Markovic et al. 2000;El Marroun et al. 2011). The authors of the Generation R study in Rotterdam suggested that despite the challenges in obtaining accurate information on cannabis use in pregnancy, self-report does seem to be a reliable method for determining use during pregnancy in epidemiological studies (El Marroun et al. 2009). Although underreporting may affect prevalence estimates for cannabis exposure, our associations between maternal age and SES and cannabis use may also be influenced if underlying biases in selfreports are patterned by age and/or SES. In addition, women with missing data on cannabis exposure were more likely to be older and in higher quintiles of area-level income. If this patterning of missing data was correlated with cannabis use, our associations between these characteristics and cannabis use may be overestimated. However, only 5% of the cohort was missing cannabis exposure information and it is unlikely that a potential correlation would substantially alter any of the reported effects.
The BORN data closely align with data from Canada and the United States. A recent study from Ontario indicated that pregnant women were more likely to report cannabis use if they lived below the low-income cutoff and were single mothers . Data from the US indicate that cannabis Fig. 1 Self-reported cannabis use in pregnancy in Ontario, by (a) age (p-trend < 0.001 in four age groups) and (b) area-level income (p-trend < 0.001 in two income groups), 2012-2017 use among both pregnant and non-pregnant women of reproductive age has increased between 2002 and 2015 (Brown et al. 2017;Volkow et al. 2017). Among 200,510 women aged 18-44 who responded to the US National Survey on Drug Use and Health from 2002 to 2014, the adjusted prevalence of lastmonth cannabis use for pregnant women increased from 2.4% in 2002 to 3.9% in 2014, with an adjusted prevalence of 7.5% among those aged 18-25 (Brown et al. 2017). For non-pregnant women, a similar increasing trend was observed, from 6.3% last-month use in 2002, to 9.3% in 2014 (Brown et al. 2017). Across the literature, the prevalence of self-reported cannabis use during pregnancy varies between 1% and 6%, and may vary by trimester (Finnegan 2013).
Previous research has suggested that cannabis use in pregnancy is associated with a low perception of risk among mothers and this may be reflective of societal beliefs, including those of healthcare providers (Ko et al. 2015). Qualitative data have indicated that in prenatal visits among women who disclosed cannabis use to their obstetrical provider, nearly half of providers did not respond to the disclosure or offer counseling (Holland et al. 2016). Proposed changes to legislation and other societal trends which normalize the usage of cannabis could lead to further increases in the prevalence of prenatal usage in Canada and this may cause an increase in the burden of adverse pregnancy outcomes, especially among younger women or those of lower SES.
With 10,731 women who self-reported cannabis use in pregnancy, this is one of the largest studies of this topic, due to its being based on one of the largest perinatal databases globally. This study has identified that cannabis use disproportionately affects women from more vulnerable segments of society. From these data and given forthcoming legal changes to cannabis use, increases in prenatal screening for cannabis use may be warranted (El Marroun et al. 2011), in order to accurately identify women and neonates for intervention. Open communication between health practitioners and pregnant women to promote cannabis cessation in pregnancy may improve perinatal and potentially later childhood outcomes and reduce health inequalities.
Funding This study was funded by the Canadian Institutes of Health Research.

Compliance with ethical standards
Research ethics board approval for this study was obtained from the Ottawa Health Science Network Research Ethics Board and the Children's Hospital of Eastern Ontario.

Conflict of interest The authors declare no conflict of interest.
Open Access This article is distributed under the terms of the Creative Comm ons Attribution 4.0 International License (http:// creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. Table 3 Association of categories of age and area-level income on cannabis use in pregnancy, Ontario, Canada, 2012 Bottom two income quintiles, n = 225,116 Upper three income quintiles, n = 507,702 Cannabis use 15 to 24 y, n = 46,905 25 to 29 y, n = 68,881 30 to 34 y, n = 67,849 35 y and over, n = 41,481 15 to 24 y, n = 51,532 25 to 29 y, n = 134,477 30 to 34 y, n = 198,518 35 y and over, n = 123,175 Reported use,