How sex and gender relate to problematic cannabis use and cannabis use disorder

According to the United Nations Office on Drugs and Crime (UNODC), 228 million adults worldwide used cannabis in 2022 [1]. There is a historical sex/gender gap in the prevalence of cannabis use, where men and boys are consistently more likely to use cannabis than women and girls across different metrics of use (e.g., lifetime use, past-month use, daily use) and in many countries globally [2–4]. We refer to this as a “sex/gender” gap, as differences in cannabis use are likely driven by both biological and sociocultural factors [2]. Importantly, the sex/gender gap seems to be narrowing. For example, the gap in past-month prevalence of cannabis use in the United States (US) declined from a high of 2.125 men to women using cannabis in 2002 to a low of 1.25 men to women in 2020 – which seems to be largely driven by an increase in women’s cannabis use [5]. Youth data from the Canadian province of Ontario show that, as of 2023, girls actually have a higher prevalence of cannabis use than boys do [6]. Changes in the sex/gender gap in prevalence of use may be driven in part by cannabis legalization and will likely lead to changes in cannabis-related harms, such as prevalence of cannabis use disorder (CUD) [2]. CUD is characterized by an inability to cease or reduce cannabis use despite persistent physical and/or psychological harm [7], as defined in the most recent edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5). While men are more likely to meet criteria for CUD and have greater symptom severity [8–11], women may escalate their use of cannabis faster than men (e.g., fewer years between age of first use and age of first CUD symptom) [8,10,12,13]. Recent data-driven work has suggested that there are sex/gender differences in the factors underlying CUD. In a recent study, individual risk factors (e.g., mental health or cognitive factors) had stronger associations with CUD in men, while social-structural risk factors (e.g., low education level) had stronger associations in women [14]. More work is needed to better understand how sex and gender impact CUD trajectories, correlates, and treatment and how to integrate this knowledge clinically to improve care for individuals with CUD.

Sex is an umbrella term that refers to complex biological traits and processes necessary for sexual reproduction, including genes and gene expression, hormones, body parts, and all the downstream biological impacts of sex-related gene expression and hormone production. Sex is thus a property of bodies and affects the way that drugs interact with bodies [3]. A growing literature has found that gonadal hormones like estrogens, progestogens, and androgens (which are essential for sex-related biological processes) can influence drug-seeking behaviours in rodents and the experience of subjective drug states in humans [15–18]. For example, female rodents show greater self-administration of cannabinoids than male rodents, yet this sex difference is reduced by ovariectomy (removal of the ovaries, and thus dramatic reduction in ovarian hormone production) [19]. The modulation of drug experiences by ovarian hormones is thought to contribute to variability in the risk of progression of drug use in humans [20]. Moreover, human laboratory experiments have found that adult female participants experience greater subjective effects of cannabis than male participants under certain conditions [21–23].

Gender is a term that describes the sociocultural context and correlates of sexed bodies – it is a social construct that describes the social meaning of sexed bodies and the organization of that meaning into socially relevant categories. Gender is typically used as an umbrella term to refer to multiple concepts, including gender identity (individual experiences of gender), gender roles and norms (expectations placed on individuals based on their presumed gender), gender relations (how individuals of different genders interact, especially as it relates to power), and institutional gender (the translation of gendered meaning into social institutions, e.g., laws and regulations dictating the distribution of power and resources) [3]. Gender roles and norms can influence drug use and harms, e.g., stigmatization and devaluation of women’s drug use partially explains lower drug use and more experiences of sociocultural harms associated with persistent drug use among women [2]. The historical association between substance use and masculinity [24] has likely had an important role in shaping sex/gender differences in cannabis use, where men and boys have a higher prevalence of use than women and girls across cultures [2]. Performative aspects of gender (i.e., engaging in behaviours to shape or reinforce one’s gender identity or expression) can influence how and why drugs are used and in which contexts [3]. In quantitative studies, greater adherence to traditional masculine norms is associated with increased use of cannabis, while adherence to feminine norms is associated with decreased use of cannabis [2,25]. Several studies have focused on initial opportunities to use cannabis, finding that boys can more easily access cannabis than girls [2].

Some nuances about sex and gender are worth noting. First, the scientific literature is increasingly acknowledging that neither sex nor gender are binary [26,27]. Further, there is a distinction between an individual’s gender identity (e.g., woman, man, or non-binary) and their expression or performance of gender (e.g., femininity and/or masculinity) [26,27]. Second, there is an increasing focus on using an intersectional framework in the health sciences to understand how multiple social categories (e.g., gender, race, sexuality) intersect at the individual level to influence health [28]. For example, while sexual and gender minorities have a higher prevalence of cannabis use than cisgender, heterosexual people, bisexual women appear to have the highest prevalence of use when examining intersections of gender and sexual identity [29]. Complex three-way intersections between gender, sexual, and racial identities have been documented for lifetime substance use [30]. To the extent possible, our review will keep these nuances in mind when synthesizing and interpreting the results of the identified literature.

There are no currently approved pharmacological treatments for CUD, so CUD treatment primarily involves psychological and psychosocial interventions [31]. The current evidence favours both motivational enhancement therapy (MET) and cognitive behavioural therapy (CBT) for treatment of CUD, while contingency management has shown promise in some studies as well [32]. Gender likely influences psychosocial interventions. For example, previous studies have found that women with anxiety disorders attend more psychotherapy sessions, are more engaged in therapy, and view psychotherapy as more helpful when compared to men [33]. Meta-analysis has identified that women with psychiatric disorders are more likely to prefer psychotherapy over pharmacotherapy when compared to men [34]. Pharmacological interventions for CUD have been tested in clinical trials, with the strongest evidence in favour of drugs that target the endocannabinoid system, though the evidence is still weak and more randomized controlled trials are needed to establish efficacy [31]. Importantly, there are sex differences in the endocannabinoid system. For example, circulating levels of endocannabinoid ligands fluctuate across the rodent estrous cycle (a hormonal cycle analogous to the human menstrual cycle) [35] and ovariectomy causes a change in density of the cannabinoid type-1 receptor (CB1R) that is reversed by administration of exogenous estradiol [36]. Thus, both sex and gender have the potential to influence various CUD treatment approaches. Furthermore, sex/gender differences in comorbidities have the potential to influence CUD treatment. For example, studies have consistently documented a stronger association between CUD and poorer mental health among women than among men, particularly mood and anxiety symptoms [37–41].

Previous narrative reviews have highlighted broad relationships between sex, gender, and cannabis (e.g., [2,15]). Several systematic reviews have each provided a specific piece of the puzzle – how sex interacts with cannabis use to impact the brain [42], changes in the sex/gender gap in the prevalence of cannabis use [43], and sex differences in the comorbidity of CUD with mental illness [44]. Finally, a large scoping review was undertaken pre-2020 that broadly covered the influence of sex and gender on cannabis use and related outcomes [3,25], including one paper that specifically focused on sex- and gender-based analysis in trials of pharmacological interventions for CUD [45]. However, what is currently missing is a structured overview of all clinically relevant literature relating sex, gender, and intersecting social categories to problematic cannabis use and CUD to inform prevention and treatment of CUD. We use the term “problematic cannabis use” as an umbrella term to refer to three groups: 1) people screening above threshold on a validated scale of harmful or hazardous cannabis use (e.g., the Cannabis Use Disorder Identification Test – Revised (CUDIT-R) [46]), 2) people meeting DSM or ICD criteria for relevant cannabis diagnoses (e.g., DSM-5 cannabis use disorder, DSM-IV cannabis dependence, or ICD-10 F12.1/12.2 codes), or 3) people seeking treatment to stop or reduce their cannabis use. We adapted this terminology in part from a systematic review of definitions of risky and problematic cannabis use [47]. We chose a scoping review to account for the breadth of evidence that is necessary to be clinically meaningful, and to help us organize the existing evidence to identify significant knowledge gaps necessary to advance care. We chose to use the Arksey and O’Malley (2005) framework for scoping reviews [48] as this is the most commonly used scoping review methodology [49].

The overarching objective of the present scoping review is to provide a synthesis of all published evidence (from 2010 to present) implicating sex and gender in problematic cannabis use and CUD treatment, trajectories, and correlates. Furthermore, across research questions, we will consider the scope of evidence for intersections of sex and/or gender with other social categories such as race, ethnicity, sexuality, and ability.

Primary research question: What is known from existing published evidence about the influence of sex, gender, and intersecting social categories on correlates and outcomes of treatment among people with problematic cannabis use?

We further break this research question down into three secondary questions:

Inclusion criteria:

In the event of any ambiguous situations that arise during title/abstract screening (e.g., an article missing an abstract that is not clearly ineligible based on the title), articles will automatically be moved on to full-text screening. This is particularly important for our review, as sex and gender terminology is often not included in the title or abstract of articles that would meet our eligibility criteria. This means that we will not apply criterion 3 at the title/abstract screening stage. At the full-text screening phase, any ambiguous situations will be brought to a study team meeting (including one of the first and last author and at least two other authors) to discuss, and decisions will be documented in a table, which can be published as supplementary material with the final review article.

The draft of the Medline search strategy was developed by a research librarian at a large tertiary care mental health and addictions teaching hospital. The strategy was developed in collaboration with the study team and facilitated through email communication and regular virtual meetings. Starting with text mining previously identified on topic articles, a multiphase search approach was utilized to validate the search; numerous preliminary searches were performed to determine the suitability of the search strategy and the article eligibility criteria, in order to optimize how to identify research focusing on sex and/or gender. The final step of validating the search was to ensure that all on topic articles identified by the text mining and preliminary searches were captured by the expanded final Medline search. For the final searches, the following databases will be searched: MEDLINE, Embase, CINAHL, PsycINFO and Web of Science for articles published between 2010 and to the present (inclusive). There will not be any language or study type limits applied to the search results, in order to not filter out relevant research unintentionally. Additionally, a focused grey literature search will also be completed, restricted to key sources of completed but unpublished primary research. Specifically, the grey literature search will be limited to dissertations, preprints and conference papers/proceedings. Resources for searching the grey literature will include ProceedingsFirst, PapersFirst, ProQuest Dissertations and Theses Globe and medRxiv. We will also manually search reference lists from related review papers and from articles included in our review. We will perform the search just once. If sufficient time (i.e., 6 months) has elapsed, we conduct an updated search and update the registration as needed. We will use EndNote 21 (Clarivate, Philadelphia, PA, United States) for reference management. The Peer Review of Electronic Search Strategies (PRESS) checklist [52] will be used to document peer review of the search strategy.

The full search strategy for MEDLINE is provided here as an example:

Database: Ovid MEDLINE(R) ALL < 1946 to November 19, 2024>

Duplicate records will be removed using Endnote or Covidence software before beginning the screening process. Eligibility criteria will be applied at two stages to determine article inclusion: (i) title/abstract screening; and (ii) full-text screening. Note that, as described earlier, not all eligibility criteria will be applied at the title/abstract screening stage in order to include as many relevant articles as possible (e.g., articles that do not mention sex or gender in the title or abstract but are otherwise eligible will move to full-text screening). Covidence systematic review software (Veritas Health Innovation, Melbourne, Australia; available at www.covidence.org↗) will be used for article screening and data charting. Two independent reviewers will screen each record at both stages. In the event of voting discrepancies between reviewers, we will first try to reach consensus between the original two reviewers; if this is not possible, we will discuss any discrepancies with the study team, including at least one of the first author and senior author, to reach a final decision.

Both stages of article screening will begin with a pilot process (sometimes called a calibration exercise [51]). The research librarian will generate a random sample of records (5% of total articles for both stages) and all reviewers will screen these records in Covidence. Once the pilot process is completed, the team (i.e., all reviewers and any additional authors who wish to be present) will discuss the review process and any discrepancies between reviewers. If there is lower than 80% proportionate agreement between any two reviewers (calculated in Covidence), the screening process will be reviewed, any necessary adjustments will be made, and the reviewers will perform a second pilot process. The pilot process will end once there are no pairs of reviewers achieving less than 80% proportionate agreement. Screening at both stages will proceed only once the pilot process is completed and all reviewers are in agreement with the screening process and feel they have adequate training to begin. We will use the pilot process as an opportunity to determine an appropriate pace for screening (i.e., we will decide on a weekly goal for number of articles reviewed per reviewer for each stage of screening).

Given the broad nature of the evidence we plan to synthesize in this scoping review, it is likely that we may be required to make changes to the study selection process (e.g., updating or revising eligibility criteria). Any changes made will be documented in a log and will be published along with the final review article (likely as supplementary material).

The data charting form will be created with input from the entire study team. Two independent reviewers will extract data using Covidence software. If there are any discrepancies in data extraction, we will follow the same process as described for article screening. Missing, incomplete, or unclear data will be coded accordingly (e.g., we will code “Not Reported” if the data are not included in the article or “Unclear” if the study team cannot reach a consensus on how to interpret the data including in the article). The reviewers will make attempts to contact the corresponding author of articles containing any missing or unclear data. If the corresponding author replies with the requested information, this will be updated in the charting form for that article. We will keep a log of author contacts to be transparent about data that was not included in the published article. If the corresponding author has not replied to the study team after one initial email and two follow-up emails, or if the contact information for the corresponding author is no longer valid, we will leave the data coded as “Not Reported” or “Unclear”. If we encounter any “friend studies” (i.e., multiple articles published from the same study), we will include them all if they contain separate relevant endpoints, or else we will use the article with the largest sample size and/or longest follow-up period for data extraction. We will not be conducting a critical appraisal or risk of bias analysis, given the anticipated heterogeneity in study design.

We will begin the data charting with a pilot process. The research librarian will randomly select 10 articles, and all reviewers will participate in pilot data charting. Once this is complete, the study team will meet to discuss the data charting process and any discrepancies. Data charting will proceed only once the pilot process is completed and all reviewers are in agreement with the data charting form and feel they have adequate training to begin. Regular study team meetings will be held with at least one of the first or senior author present to discuss any issues that arise during data charting. Any changes made to the data charting form will be documented in a log and will be published along with the final review article (likely as supplementary material).

The following data elements will be extracted from each article:

All extracted data will be reviewed by a third reviewer not originally involved in data extraction. In order to summarize data and present basic descriptive statistics, data will be exported into a spreadsheet in Microsoft Excel (Microsoft Corporation, Redmond, WA, United States). A PRISMA flow diagram will be used to graphically depict the flow of article capture, screening, and extraction [53].

Articles will first be grouped together with regards to the specific research question they best inform (1, 2, or 3, as described earlier). Within each group, subgroups will be created based on study design and whether the analysis or theoretical framework considers biological (i.e., sex-related) or sociocultural (i.e., gender-related) factors. Depending on the results we identify, we may further group by age (e.g., results for youth separate from adults). We will likely conduct a consultation exercise at this phase, in order to solicit feedback and perspectives from relevant stakeholders (e.g., clinicians) related to the study results. We will use tables to present relevant data for each article included in each of these subgroups. Descriptive analysis will be used to complement these tables and communicate the frequencies of each type of study.

This scoping review is exempt from ethics review as we will not be directly accessing any data.

Multiple members of the study team are clinician-researchers, who are well situated to both inform the review process and guide dissemination of the results to have maximum capacity to influence policy and care. The study team has extensive experience engaging with knowledge translation (KT) services within our primary institution (CAMH). Our standard practice is to engage KT efforts on two fronts. First, we engage in traditional KT, such as publication in peer-reviewed journals and presentation at both medical and scientific conferences. Second, we will work with internal services (such as media affairs and patient and family engagement services) to develop broader KT efforts that will reach a wider (non-academic) audience. This can include KT activities such as presentations to broad audiences through our Patient and Family Learning Space, dissemination of findings through social media networks, and development of brochures, infographics, and short videos to summarize our findings for a lay audience. Our specific KT plan will depend on the scope of the evidence we generate in this review. We may consider bringing on additional collaborators (e.g., clinicians or other knowledge users) to assist in dissemination of findings to additional networks not covered by the current study team.