Use of prescription opioids and other psychotropic drugs during pregnancy and their impact on the mother and developing child: protocol for a cohort study using linked administrative data from Manitoba and British Columbia, Canada

Abstract

Introduction

Opioids are prescribed to manage pain. Approximately 1 in 20 pregnant women in Canada are prescribed opioids during the prenatal period, which may occur concurrently with other psychotropic drug use. The health implications of the independent and concurrent prenatal use of these drugs are not fully understood; however, adverse neonatal and longer-term outcomes have been suggested. This protocol describes a study to update the epidemiology of prenatal exposure to opioid and other psychotropic drug use during pregnancy, providing an enhanced understanding of the potential impacts on the mother and child to help inform decisions regarding prescription and use.

Methods and analysis

The retrospective cohort study design uses population-based administrative data from Manitoba and British Columbia, Canada, to investigate the effect of prenatal opioid and concurrent psychotropic drug use on maternal and child outcomes. All mother–child dyads from 2000/2001 to 2019/2020 (approximately 1M pairs) will be identified and assigned to exposure groups based on the number of opioid and other psychotropic drug dispensations to the mother during the prenatal period. Maternal sociodemographic characteristics, prescribing patterns, short- and long-term child health and education outcomes and maternal outcomes will be examined.

Ethics and dissemination

The study was approved by the University of Manitoba Human Research Ethics Board (No. HS24397 – H2020:470) and the University of British Columbia Clinical Research Ethics Board (No. H21-02262). The study will generate findings that will add to the growing body of evidence of potential short- and long-term adverse effects on children exposed to these drugs prenatally and will help to inform safe prescribing guidelines during pregnancy. Results will be published in peer-reviewed journals.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• The study will use comprehensive whole-population administrative datasets to identify all mother-child dyads in two Canadian provinces during the 20-year study period.

• The inclusion of two provinces increases the sample size and statistical power, enhancing the reliability and generalisability of the study findings.

• We will use a robust high-dimensional propensity score approach to create study groups that are comparable on many sociodemographic, health status and health service use characteristics.

• The observational nature of the study limits our ability to account for all potential confounders, such as over-the-counter analgesics, hospital inpatient opioid use and unreported nicotine, alcohol and illicit drug use.

Introduction

Opioids are prescribed to patients experiencing acute or chronic pain.¹ Pregnant women may also be prescribed opioids for conditions such as low back and pelvic pain, myalgia, joint pain and migraines.² Opioids are known to cross the placenta and have been reported to be associated with adverse outcomes for the mother and developing fetus.3,7 In addition, people using opioids often have co-occurring mental health disorders and may be prescribed other psychotropic drugs to manage these concerns. Pregnant women may therefore be exposed to opioids and other psychotropic drugs concurrently; however, the risk of adverse outcomes associated with concurrent use of opioids and other psychotropic drugs is not well understood.^{8 9} This research will provide valuable insights into the clinical implications of opioid use during pregnancy, enabling healthcare providers to better understand the risks and mitigate adverse effects, and will help clinicians support their patients in informed decision-making to optimise their health.

This protocol describes a retrospective cohort study using population-based administrative data from two Canadian provinces. We will investigate the epidemiology of opioid prescribing for pain management during pregnancy with and without concurrent psychotropic drug use and then examine maternal, neonatal and childhood outcomes.

Opioid use during pregnancy

The prevalence of opioid use during pregnancy varies across jurisdictions, but overall, the number of pregnant women prescribed opioids appears to be declining over time. For example, among Medicaid-enrolled women in the USA, 21.6% were dispensed an opioid during pregnancy between 2000 and 2007, ranging from 9.5% to 41.6% across 46 states,¹⁰ whereas approximately 10% of privately insured pregnant US women were prescribed or dispensed an opioid between 2001 and 2015;^{8 11} these studies report ‘any’ opioid use during pregnancy. In Canada, studies show that 5.4% of pregnancies between 1998 and 2015 involved ‘any’ opioid use,¹² and 7.7% of pregnancies between 2001 and 2013 and 1.1% of pregnancies between 2014 and 2020 involved opioid prescription or dispensation for pain management.^{13 14} The overall downward trend in opioid prescribing/dispensing to pregnant women in more recent years may be the result of the Centers for Disease Control and Prevention guidelines for prescribing opioids for pain introduced in 2016.^{5 8 14}

The most common opioid prescribed during pregnancy is codeine, which is considered a relatively weak opioid.¹³ While the prevalence of pregnant women being prescribed codeine has decreased over time, prescribing of stronger opioids (eg, morphine, hydromorphone and oxycodone) has increased.¹² However, the amount of opioids prescribed, measured by the morphine equivalent of the drug, has remained stable over time. These trends demonstrate that research examining changes in prescribing practice must take into account not only the number of people being prescribed opioids or the number of prescriptions they receive but also opioid dose and potency.¹³

Risk of adverse fetal and neonatal outcomes

Opioids are known to cross the placenta, which may increase the risk of adverse fetal and neonatal outcomes.3,7 Large population studies investigating the impact of prenatal analgesic opioid exposure on the developing fetus indicate a higher risk of fetal growth restriction, preterm birth, small for gestational age birth, low birth weight, stillbirth, congenital anomalies and neonatal intensive care unit admissions and neonatal abstinence syndrome (NAS).3,515 Other research examining women taking methadone, which is used in maintenance therapy for opioid dependence or opioid use disorder, demonstrates a higher risk of preterm birth, small for gestational age birth, admission to a neonatal unit, diagnosis of a major congenital anomaly and infant mortality.^{18 19} While a few population-based studies have shown no evidence of increased risk, including a Swedish study of opioid-exposed births¹⁵ and an Australian study of oxycodone-exposed births,²⁰ a recent review and meta-analysis demonstrates a higher risk of several adverse birth outcomes, including preterm birth and neonatal death, following prenatal exposure to opioids.²¹ Opioid use during pregnancy remains a serious public health concern due to the potential for harm.

Risk of adverse childhood outcomes

There is still limited evidence of early and later childhood impacts of prenatal opioid exposure, likely because of the longer follow-up time from exposure to measurement and the methodological challenges in controlling for associated factors.²² However, a recent systematic review points to a higher risk of adverse neurodevelopmental outcomes among children exposed to opioids in utero, affecting their standardised test scores, attention capability and behaviour disorders, cognition and vision in early and later childhood.²³ A meta-analysis of five case-control studies showed a trend of poorer neurobehavioural outcomes measured in cognitive, psychomotor and behaviour domains among exposed children, but these differences did not reach statistical significance.²⁴ A series of small studies of children in Norway demonstrated lower cognitive function and IQ scores and more depression, anxiety, attention problems and antisocial behaviours in children exposed to opioids in utero.25,27 Another showed that prenatal exposure to opioids was associated with an increased risk of childhood immune-related conditions, such as asthma, eczema and dermatitis, but not allergies or anaphylaxis.²⁸ In terms of a mechanism for these findings, smaller neuroanatomical volumes, smaller cortical areas and a thinner cortex of the brain have been shown to persist from birth into youth and young adulthood among those who were exposed prenatally, which could explain some of the observed longer-term findings.²⁹

Whether prenatal opioid exposure is associated with school achievement among older children remains to be investigated. One small study of 9-year-old children born to opioid-dependent women shows that they performed more poorly in reading and math assessments and were at high risk of educational delay,³⁰ although the authors caution that both adverse prenatal exposures and postnatal social risk appeared to be at play. The extensive follow-up years available in our data repository will allow us to explore more distal outcomes.

Risk of adverse maternal outcomes

Research on longer-term maternal outcomes is also still limited. Cross-sectional studies show that women who use opioids during pregnancy are more likely to have depression, anxiety and other concurrent chronic medical conditions compared with women not using opioids.^{31 32} After controlling for sociodemographic characteristics and health conditions, opioid use of any type during pregnancy was associated with longer hospital stays, higher in-hospital mortality, preterm labour, early onset of delivery and higher rates of postpartum depression.³¹ To our knowledge, only one Canadian study has explored a long-term maternal outcome among women who used opioids during pregnancy;³³ however, the authors did not have access to direct measures of opioid use and instead used a diagnosis of NAS in the newborn as a proxy. The study reported that the 10-year maternal mortality rate was 11 times higher among mothers with an infant affected by NAS than among those without.³³

Concurrent use of opioids and other psychotropic drugs in pregnancy

Psychotropic drugs such as benzodiazepines and antidepressants are often concurrently prescribed with opioids during pregnancy.⁸ In a Canadian study, psychotropic drug prescriptions in pregnant women were seven times more common among those with an opioid prescription than those without.³⁴ Infants of women prescribed benzodiazepines or antidepressants during pregnancy may have a higher risk of adverse outcomes: the results of a recent umbrella review suggested evidence of a higher risk of preterm birth, small for gestational age birth and major congenital malformation in pregnancies exposed to psychotropic medications.³⁵ However, few studies have investigated how the interactions between opioids and other psychotropic drugs might affect the developing fetus, and those that have begun investigating this topic are so far inconsistent in their conclusions. For example, a study following Medicaid recipients in the USA reported a 34% increase in the risk of NAS associated with coexposure to opioids and antidepressants compared with opioids alone.³⁶ Meanwhile, a population-based cohort study from Norway did not find detrimental outcomes with respect to fine motor skills or attention deficit hyperactivity disorder (ADHD) symptoms after prenatal benzodiazepine exposure alone or in combination with opioids or antidepressants.³⁷

The concurrent use of prescription opioids and other psychotropic drugs among pregnant women and the potential for harm to the mother and child warrant further investigation. To date, most published studies on this topic use relatively small sample sizes and are unable to account for many confounding factors. There is a critical need for high-quality population-based, longitudinal research with the ability to control for important confounders to investigate the effects of prenatal opioid exposure, with and without concurrent psychotropic drug exposure, on maternal, neonatal and childhood outcomes.

In the proposed study, we will address these important knowledge gaps using population-based linked administrative data in two Canadian provinces. This work has important real-world clinical implications for women and their physicians in balancing the risks and benefits of treatment during pregnancy.

Methods and analysis

We will use a population-based retrospective cohort design to investigate the effect of prenatal opioid exposure alone and concurrently with other psychotropic drugs on fetal and neonatal health, childhood health and education, and maternal health and mental health outcomes. We will use deidentified linked administrative data covering a 20-year period (1 April 2000 to 31 March 2020) from the Canadian provinces of Manitoba (MB) and British Columbia (BC). The proposed design for this study is shown in figure 1.

Figure 1. Study design. BC, British Columbia; MB, Manitoba.

The study was funded in 2021, and data access approvals were obtained later that year; we expect to complete the analysis by the end of 2025.

The study aims to address the following three objectives:

1. Conduct a descriptive analysis of prescription opioid use during pregnancy, alone and with other psychotropic drugs.

2. Determine whether prenatal exposure to prescription opioids alone or in combination with other psychotropic drugs is associated with adverse fetal and neonatal health outcomes and short- and long-term childhood health and education outcomes compared with no opioid and psychotropic drug exposure.

3. Determine whether prenatal exposure to prescription opioids alone and in combination with other psychotropic drugs is associated with adverse maternal outcomes compared with no opioid and psychotropic drug exposure.

Setting

MB is a centrally located province in Canada with a population of approximately 1.4 million, while BC is the westernmost Canadian province with a population of approximately 5 million.³⁸ All residents of both provinces are insured for services provided by hospitals and physicians under the publicly funded healthcare system. Coverage for specific services and pharmaceuticals may vary somewhat across provinces and territories, but both MB and BC have pharmacare programmes that collect data on prescription medications dispensed regardless of whether the medications are covered by insurance or paid for by residents.³⁹

Data sources

Our study will leverage the administrative data housed at the Manitoba Centre for Health Policy (MCHP) and Population Data (PopData) BC. The Manitoba Repository at MCHP contains deidentified, individual-level information on >99% of the MB population in over 90 datasets spanning more than 40 years.^{40 41} PopData BC facilitates access to more than three decades of population-based health and social data for BC residents.⁴² The data sources that will be used in this study contain information on prescription drug dispensations, prescribing physician characteristics, hospitalisations, physician payments and long-term developmental and education outcomes. A complete list of data sources and years of data used in this study is provided in the online supplemental table 1.

Study cohorts

All hospital births from 2000/2001 to 2019/2020 in MB and BC will be identified and linked to their mothers to create mother–child dyads. To avoid double-counting pregnancies that resulted in multiples (twins, triplets, etc), only one randomly selected mother–child dyad will be retained. We will exclude dyads where the child is linked to multiple mothers or where the mother has an invalid personal health information number. If the mother is multiparous, we will randomly select one pregnancy to include in the study (for objective 1 only). The mother–child dyads will be linked to the hospital discharge abstracts and/or birth records and to the provincial health insurance registries to obtain the child’s gestational age at birth and the mother’s conception date. Dyads where the gestational age could not be found and mothers without continuous health coverage from their conception date until their delivery date (allowing for a maximum 30-day gap in coverage) will be excluded. To preserve our focus on prescription opioids for pain management, we will also exclude mothers with a known history of opioid use disorder or dependence based on having at least one maintenance therapy prescription opioid dispensation during pregnancy (methadone or buprenorphine) or having an International Classification of Diseases (ICD) code specific to opioid misuse and dependence in the physician claim or hospital discharge abstract database during the 6 months prior to conception or any time during pregnancy.

For objectives 2 and 3, we will exclude mother–child dyads where the birth resulted in a stillbirth, the birth was for multiples, the mother did not have continuous health coverage in the year prior to the conception date (allowing for maximum 30-day gaps in coverage) and the newborn did not have a health coverage registry record at birth. It is anticipated that approximately 5% of the mother–child dyads from objective 1 will be excluded to form the study cohort for objectives 2 and 3.

The number of pregnancies in each province is predetermined, so after applying exclusions, we used these sample sizes to calculate the minimum detectable effect sizes for one newborn outcome (intensive care unit admission), one early childhood outcome (autism spectrum disorder) and one school-entry age outcome (ADHD). The results of these analyses indicate sufficient sample size to detect clinically meaningful effects in our study population for all three outcomes (online supplemental table 2).

Exposure groups

For each study cohort, mothers will be observed during the prenatal period, defined as the conception date until (but not including) the delivery date for any study drugs dispensed. Study drugs will include all opioids, anxiolytics, antipsychotics and antidepressants that were approved for use in Canada as of 2020. The World Health Organisation’s Anatomical Therapeutic Classification (ATC) codes will be used to create a list of the relevant Drug Identification Numbers, which will be used to identify the dispensations in relevant categories (table 1).

Table 1. Drug categories and their Anatomical Therapeutic Chemical (ATC) codes/labels.

Category	ATC code	ATC label
Opioids	A07DA02	Opium
	A07AD52	Morphine, combinations
	M03BB53	Chlorzoxazone, combinations excluding psycholeptics (containing codeine)
	N01AH01	Fentanyl
	N02A	Opioids
	N02BA51	Acetylsalicylic acid, combinations excluding psycholeptics (containing codeine)
	N02BA71	Acetylsalicylic acid, combinations with psycholeptics (containing codeine)
	N02BE51	Acetaminophen, combinations excluding psycholeptics (containing codeine)
	N07BC01	Buprenorphine
	N07BC06	Diamorphine
	N07BC51	Buprenorphine, combinations
	R05DA04	Codeine
	R05DA20	Combinations of opium alkaloids and derivatives (containing codeine)
	R05FA02	Opium derivatives and expectorants (containing codeine)
	R05FB02	Cough suppressants and expectorants (containing codeine)
Other psychotropics	Anxiolytic
	A03CA02	Clidinium and psycholeptics
	N02BA71	Acetylsalicylic acid, combinations with psycholeptics (containing meprobamate)
	N03AE01	Clonazepam
	N05B	Anxiolytics
	Antipsychotic
	A03CA01	Isopropamide and psycholeptics (containing trifluoperazine)
	N05A	Antipsychotics
	N06CA01N07XX06	Amitriptyline and psycholeptics (containing perphenazine)tetrabenazine
	Antidepressant
	N06A	Antidepressants
	N06CA	Antidepressants in combination with psycholeptics

Each pregnancy will be placed in one of 12 exposure groups based on the number of dispensations from each drug category during the prenatal period (table 2). Exposure groups involving an opioid dispensation will be separated into those with only one dispensation and those with two or more dispensations. This will be done to identify mothers who we believe would be most likely to have taken the opioids (ie, those who had 1+ dispensation). Within each of the 1 and 2+ opioid dispensation groups, we will identify those that also concurrently used an anxiolytic, antipsychotic, antidepressant or two or more psychotropic drug types and those that used opioids only. Given the potential for small exposure group sizes and to assist in interpreting some of the results, we will collapse concurrent use groups 2–5 (those involving only one opioid dispensation) and concurrent use groups 7–10 (those involving 2+ opioid dispensations) (table 2). Pregnancies involving none of the study drugs will serve as the unexposed group.

Table 2. Study exposure group composition groups are based on the number of dispensations in each drug category during the prenatal period^* before and after collapsing concurrent use exposure groups involving anxiolytics, antipsychotics and antidepressants.

Group	Before collapsing				Group	After collapsing
	Number of dispensations					Number of dispensations
	Opioid	Anxiolytic	Antipsychotic	Antidepressant		Opioid	Psychotropic type
1	1	0	0	0	1	1	0
2	1	1+	0	0	2	1	1+
3	1	0	1+	0
4	1	0	0	1+
5	1	2+types
6	2+	0	0	0	3	2+	0
7	2+	1+	0	0	4	2+	1+
8	2+	0	1+	0
9	2+	0	0	1+
10	2+	2+ types
11	0	1+			5	0	1+
12	0	0	0	0	6	0	0

^*prenatal period=conception date until, but not including, the delivery date.

Analysis plan

Data analysis will be conducted separately in MB and BC using SAS statistical analysis software.⁴³ Common protocols for each objective will ensure comparability between sites.

Objective 1: descriptive analysis

The distribution of pregnancies by exposure group before and after collapsing concurrent groups along with the difference in the proportion of pregnancies between MB and BC within each exposure group will be calculated. We will examine annual trends in the proportion of mothers in each exposure group over the 20-year study period for the study cohort and separately for MB and BC using linear regression.

To assess the relationship between opioid exposure and concurrent use pregnancies, we will estimate the odds of a psychotropic drug dispensation in pregnancies with 2+ opioid dispensations and pregnancies with one opioid dispensation and report ORs with 95% CIs. We will assess the relationship between multiple psychotropic drug dispensations and the level of opioid exposure among only the concurrent use pregnancies and compare the odds of having 2+ psychotropic drug dispensations for pregnancies with 2+ opioid dispensations to those with one opioid dispensation.

To quantify the drugs dispensed during pregnancy, we will calculate the average number of opioid and psychotropic drug dispensations by exposure group, the strength of the opioids dispensed by exposure group and total days supplied. Morphine equivalent doses (MEDs), which account for different types, potencies and number of opioids, will provide a standardised measure of opioid strength. MEDs are calculated by multiplying the total number of milligrams of active drug in every opioid unit (eg, tablet, capsule, liquid or patch) by a conversion factor (online supplemental table 3). MEDs from all opioid dispensations within an exposure group will be summed, and the mean and SD for each pregnancy will be calculated. We will also divide the summed MED total by the number of opioid dispensations for each exposure group to provide an MED per dispensation value.

We will report the distribution of the 10 most common indications associated with each opioid dispensation by year by assessing the medical service claims and hospital discharge abstracts (from 7 days before to 7 days after dispensation) and assigning the ICD 9 or 10 code to the dispensation. For hospital discharge abstracts, only the primary (ie, most responsible) diagnosis will be assigned.

We will report descriptive statistics for maternal sociodemographic characteristics (age, rural/urban residency and income quintile) in each pregnancy by exposure group. To account for the difference in the proportion of women in rural and urban areas, we will stratify by residency and calculate the distribution of pregnancies in each exposure group. Similarly, we will stratify the exposure groups by income quintile and provide the exposure group distribution. Differences in the means, percentages and 95% CIs between MB and BC will be calculated.

We will estimate continuity of care during the prenatal period based on ambulatory visits to family physicians or nurse practitioners. The continuity of care index ranges from 0 to 1 based on how many visits are made to the same provider.⁴⁴ The distribution of mothers by continuity of care quintiles will be reported for each exposure group. For quality of prenatal care, we will use the Revised Graduated Prenatal Care Utilisation Index, where results of ‘Inadequate’ or ‘No Care’ are classified as inadequate prenatal care and ‘Adequate’, ‘Intermediate’ or ‘Intensive’ are classified as adequate prenatal care.⁴⁵

A complete list of the variables that will be examined and their definitions is provided in online supplemental table 3.

In objective 2, we will examine short- and longer-term health and education outcomes for the children in the study cohort (table 3). The short-term outcomes are related to fetal development, birth hospitalisation and infant morbidity. The longer-term outcomes include neurodevelopmental disorders, chronic childhood diseases and educational achievements. In objective 3, we will examine maternal outcomes related to prenatal health, labour characteristics, postpartum health and mental health (table 3). Detailed definitions of all outcome variables are presented in online supplemental table 3.

Table 3. Outcome categories for objectives 2 and 3.

Short-term child outcomes	Longer-term child outcomes	Maternal outcomes
Fetal development Preterm birth Post-term birth Small for gestational age Large for gestational age Low Apgar Score	Neurodevelopmental disorders ADHD Autism spectrum disorder Epilepsy	Prenatal health Quality of prenatal care Prenatal psychological distress Prenatal hospitalisation
Birth hospitalisation Length of stay NICU admission	Chronic diseases Diabetes Asthma Renal disease	Labour characteristics Induction of labour Epidural Anaesthesia Caesarean birth
Morbidity Postnatal adaptation syndrome Composite severe neonatal morbidity Composite congenital anomalies	Educational achievement Developmental vulnerability using the EDI domains High school completion MB Assessments: Grade 3 teacher-completed assessment (reading and numeracy) Grade 7 teacher-completed assessment (math and school engagement) Grade 8 teacher-completed assessment (reading and writing) Grade 12 standard tests (language arts and math) BC assessments: Grade 4 and 7 foundational skills assessments (student-written tests)	Postpartum health Postpartum readmission Postpartum psychological distress Postpartum depression
		Mental health (measured prenatally and postnatally) Mood and anxiety disorders Psychotic disorders Schizophrenia Personality disorders Suicide attempts Composite (‘any’) mental health disorder

ADHD, attention deficit hyperactivity disorder; BC, British Columbia; EDI, Early Development Instrument; MB, Manitoba; NICU, neonatal intensive care unit.

Primary analysis

To assess the risk of opioid and other psychotropic drug use during pregnancy, we will compare the unexposed group to exposure group 1 (one opioid dispensation) and groups 6–10 (2+ opioid dispensations) (figure 2). We are choosing to focus on the exposure groups with 2+ opioid dispensations because they are more likely to include mothers who actually took the opioids as opposed to mothers who received only one opioid dispensation.

Figure 2. Comparison groups for objectives 2 and 3.

For the short-term child outcomes, we will report the proportions of children affected and use logistic regression to estimate adjusted ORs with 95% CIs. For the longer-term child health-related outcomes, we will estimate incidence per 1000 person-years of follow-up and use Cox proportional hazards models to estimate adjusted HRs. Children will be followed from birth until 31 March 2020 or until death, a move out of province, or a diagnosis of the outcome under investigation, whichever occurs first. The same statistical tests will be used for the maternal outcomes.

To assess child education outcomes, we will examine teacher assessments and standards test results and use logistic regression to estimate adjusted ORs with 95% CIs. For the high school graduation outcome, the study groups will only include children who were in grade 9 in 2013/2014 or earlier to allow for 6 years to complete high school.

We will use a powerful high-dimensional propensity score (HDPS) algorithm to account for measured confounding variables present in the data, including indication, disease severity and maternal comorbidity. The HDPS algorithm identifies potential covariates by selecting variables correlated with the exposure and the outcome and prioritises these by their potential for causing bias. In our study, the propensity score will describe a mother’s probability of using an opioid during pregnancy while considering her measurable characteristics prior to conception, including physician tariff codes, physician diagnostic codes, hospital procedure codes, hospital diagnostic codes and prescription drug claims from the 365 days prior to conception. Examining healthcare use before the index pregnancy makes use of many proxy covariates and will improve control for confounding for morbidity and drug use compared with models with significantly fewer covariates.⁴⁶ We will match exposed pregnancies to unexposed pregnancies 1:1. Year of conception, age of mother at birth, region of residence, area-level socioeconomic status, inadequate prenatal care, mother’s history of the outcome and documented history of maternal substance use will also be included as covariates for all outcomes.

Additional analyses

To further control for potential confounding, we will conduct several additional analyses (detailed plans presented in online supplemental table 4).

1. Control for the effect of substance use (tobacco, alcohol and illicit drug use) by the mother.

2. Control for genetic risk and/or shared familial environment using discordant-matched sibling analyses.⁴⁷

3. Examine the dose-response relationship.

4. Compare women who were dispensed opioids in the period before conception (but not during pregnancy) to women who were not exposed to opioids before conception or during pregnancy (negative control group analysis).

5. Examine whether the timing of the opioid exposure is associated with preterm birth.

Secondary analyses

The primary analysis will also be conducted by the type of opioid dispensed (eg, codeine, morphine and hydromorphone) and by the gestational timing of exposure to opioids. However, assessing the latter may be challenging for exposure groups that include 2+ opioids as they may select a group of women who used opioids in more than one trimester, making it difficult to identify a critical exposure period. These analyses will also investigate sex-specific differences through stratification and modelling, including an interaction term for the childhood outcomes, to examine whether the impact of prenatal exposure to opioids with and without other psychotropic drugs is different for boys and girls.

Combining results across jurisdictions

Due to privacy legislation that prohibits individual-level data from leaving the province, data from MB and BC cannot be pooled. Instead, results for each objective will be produced separately and then summarised. For objectives 2 and 3, we will use meta-analysis techniques with a fixed effect model to generate inverse variance weighted pooled risk estimates and 95% CIs. This approach has been used by members of our team in pharmacoepidemiological studies across Canada, increasing generalisability, statistical power and timeliness of the study results.⁴⁸

Patient and public involvement

The research is overseen by an advisory group of key stakeholders and knowledge users in BC and MB who provide relevant context and contribute practical policy and clinical considerations. The team is also working closely with both provinces’ Canadian Institutes of Health Research (CIHR)-funded Strategy for Patient-Oriented Research (SPOR) Support for People and Patient-Oriented Research and Trials (SUPPORT) Units to develop a patient and public engagement strategy and receive guidance on knowledge mobilisation to key audiences, which include women of childbearing age and/or pregnant and the physicians who care for them.

Ethics and dissemination

This study involves human participants and was approved by the ethics boards at the University of Manitoba (No. HS24397 – H2020:470) and the University of British Columbia (No. H21-02262). The Manitoba Government’s Health Information Privacy Committee (No. 2020/2021-57) reviewed the proposal and waived the requirement for individual informed consent on the basis that the study uses deidentified administrative data, that none of the participants were directly involved in the study and that there was a low risk of any individual being personally identified.

The cross-provincial multidisciplinary research team includes leading experts in paediatrics, pharmacoepidemiology, psychiatry and data science, as well as senior scientists from the Canadian Network for Observation Drug Effect Studies and national leaders of the Health Data Research Network Canada. Findings will be shared within these academic and professional networks in the form of seminars, presentations at national and international conferences and manuscripts in peer-review journals. To ensure the widespread dissemination of our results outside the academic community, we will work with The Conversation Canada to make the latest evidence available to the media and readers from the general public.