Gender differences in circumstances associated with cannabis use

Abstract

Background and Objectives

Identifying circumstances associated with cannabis use is critical to the development of effective interventions for cannabis use disorder (CUD) and circumstances may differ by gender. The Inventory of Drug Taking Situations (IDTS) assesses the types of situations in which individuals most often use drugs through eight subscales: dealing with (1) Unpleasant Emotions; (2) Physical Discomfort; (3) Pleasant Emotions; (4) Testing Personal Control; (5) Urges/Temptations; (6) Conflict with Others; (7) Social Pressure; and (8) Pleasant Times with Others. The aims of this study were to determine if IDTS scores varied by gender in individuals with CUD, and to ascertain if behavioral factors such as cannabis use or stress were differentially associated with IDTS scores by gender.

Methods

Baseline data were obtained from a study of 148 non-treatment-seeking individuals with CUD. Data included demographic characteristics, IDTS scores, self-reported past 90-day substance use, and measures of past-month and childhood stress.

Results

Women reported using cannabis more than men in negative affective contexts, namely when experiencing conflict with others, testing personal control, physical discomfort, and unpleasant emotions. IDTS subscale scores associated with negative affect and pleasant emotions were positively associated with past 90-day cannabis use sessions across genders, use when experiencing negative affect was associated with childhood stress across genders, and use when experiencing unpleasant emotions was differentially associated with recent stress by gender.

Conclusion and Scientific Significance

Findings suggest gender differences in circumstances associated with cannabis use with implications for therapeutic development for CUD.

Introduction

In the United States, approximately 7% of individuals ages 12 and above met past-year DSM-5 criteria for cannabis use disorder (CUD) in 2022.¹ While treatment options for CUD are generally limited, psychotherapeutic approaches show modest evidence of efficacy.² One of the most commonly used psychotherapeutic approaches for CUD is cognitive behavioral therapy (CBT), which has been shown to reduce cannabis use frequency and quantity and is particularly efficacious when combined with other behavioral approaches, such as motivational enhancement therapy or contingency management.² CBT for CUD includes the identification of high-risk situations for cannabis use and subsequent development of relapse prevention and coping skills; as such, comprehensive understanding of the circumstances in which individuals are more likely to use cannabis is important for the improvement of behavioral interventions implementing CBT.

Importantly, gender significantly impacts both the effects of cannabis itself and the circumstances in which an individual is likely to use drugs more broadly. Women experience greater rewarding properties of cannabis³ and progress more quickly from first use to dependence⁴ relative to men. Women also report greater severity of cannabis withdrawal symptoms and withdrawal-related functional impairment relative to men,⁵ potentially putting them at higher risk for withdrawal-precipitated relapse. With respect to broader substance use behaviors, women are more likely to misuse opioids⁶ or benzodiazepines⁷ in order to relieve negative affect relative to men. Moreover, using alcohol to cope with negative feelings is associated with increased alcohol use in women, but not men.⁸ Though cannabis is most often used in the contexts of experiencing pleasant times with others or pleasant emotions⁹ and use is motivated by enhancement of positive feelings more often than tobacco or alcohol use,¹⁰ it is possible that there are gender differences in the contexts associated with cannabis use that may impact use behavior and optimal treatment approaches.

Gender has also been shown to interact with other life circumstances, including childhood or recent stress, to affect behavior. Both acute stress and stressful life events have been associated with problematic cannabis use¹¹ and show evidence of gender differences in their effects. Acute stress is associated with greater immediate cannabis craving and negative affect among women relative to men,^12,13 though men may self-administer more cannabis following stress exposure.¹⁴ Notably, use of cannabis to cope with stress has been associated with greater baseline negative affect,¹⁵ which may in turn promote greater stress-induced cannabis use. The interplay between acute stress, affect, and cannabis use may therefore have a differential salience in men. In contrast, adverse childhood experiences (ACEs) are stressful life events that are particularly prevalent among women.¹⁶ ACEs have been shown to influence the situations in which one is more likely to use drugs¹⁷ and are negatively associated with addiction treatment outcomes.¹⁸ Importantly, ACEs appear to be a greater risk factor for the development of substance use disorders in women relative to men.¹⁹ Taken together, while childhood and recent stress may independently predict cannabis use behavior, their interaction with gender is an important consideration when interpreting their effects on cannabis use.

The present study sought to evaluate gender differences in circumstances associated with cannabis use in non-treatment-seeking individuals with CUD, including the contexts that typically precede cannabis use in these individuals and the life circumstances that may interact with these contexts, such as lifetime and recent stress. We first hypothesized that women with CUD would be more likely to endorse use when experiencing negative affect relative to men, consistent with what has been shown for opioids and benzodiazepines.^6,7 We further hypothesized that gender would moderate the association between use to relieve negative affect and self-reported cannabis use, comparable to the relationship reported in the alcohol literature.⁸ Finally, with respect to relevant life circumstances that might interact with use contexts, we hypothesized that 1) individuals reporting high levels of lifetime and recent stress would be more likely to endorse use when experiencing negative affect and 2) that this association would also be moderated by gender, with stronger associations observed in women.

Material and Methods

Participants

This study used pre-intervention baseline data from 148 non-treatment-seeking individuals enrolled in a randomized controlled trial of progesterone in CUD (NCT03729869). Primary outcomes from this trial are not yet published. Participants included in the parent trial were: 1) age 18-45; 2) meeting DSM-5 criteria for CUD²⁰ in the past 3 months; and 3) using cannabis at least five times per week in the past month. Women needed to report regular menses (every 25-35 days), could not be pregnant or nursing, and had to be willing to use an effective, non-hormonal method of birth control for the study duration. Participant gender was obtained by self-report, though notably only binary options were provided. Participants were excluded if they showed evidence of major medical illness, psychotic or bipolar disorder, current suicidal or homicidal ideation, allergy to progesterone or peanuts, or met criteria for a moderate or severe substance use disorder other than nicotine or cannabis in the past 60 days. Participants were allowed to have a mild substance use disorder other than cannabis or tobacco if cannabis was still identified as the primary drug of choice; however, participants were asked to abstain from alcohol for at least 12 hours prior to study visits and from all other non-cannabis/nicotine substance use for the duration of the study.

Procedures

Participants were primarily recruited via media advertisements. If individuals appeared eligible for study participation based on a brief phone screening, written informed consent was obtained and an in-person baseline assessment was completed. At this baseline visit, the above inclusion/exclusion criteria were assessed and participants completed the Inventory of Drug-Taking Situations (IDTS).²¹ This 50-item measure asks participants to report on a scale of 1 (never) to 4 (almost always) situations in which they used cannabis in the past year. Measure items describe a variety of scenarios an individual might experience preceding cannabis use which can be grouped into 8 subscales: unpleasant emotions, physical discomfort, pleasant emotions, testing personal control, urges/temptations, conflict with others, social pressure to use, and pleasant times with others. Item responses within each subscale are summed to produce a relative estimate of frequency of use associated with each use context. Reliability of the IDTS reported elsewhere showed adequate consistency for the total scale score (Cronbach’s α=0.95) as well as for the eight individual subscale scores (α’s=0.70-0.92).²² Reliability in our sample similarly showed adequate consistency for the total scale score (Cronbach’s α=0.95) as well as for the eight individual subscale scores (α’s=0.60-0.90).

Participants also completed basic demographic surveys (including self-reported gender) and measures detailing their cannabis use (Timeline Follow-Back or TLFB; a self-reported use history),²³ past-month stress (10-item Perceived Stress Scale; PSS)²⁴, and self-reported adverse childhood experiences (ACEs).²⁵ Participants completed the 10-item version of the ACE questionnaire, which includes items related to neglect and parental divorce in addition to the 7 items found in the original questionnaire.²⁶ ACE outcomes were calculated as total number of unique ACEs endorsed and as prevalence across four categories of experiences: emotional/physical abuse, household dysfunction, sexual abuse, and/or neglect. These subtypes were derived from a three-factor model proposed by Ford and colleagues,²⁷ modified to include additional ACE items pertaining to neglect. Items on the PSS related to feelings of recent stress as well as perceived ability to cope with those feelings. Past 90-day cannabis use assessed via TLFB was quantified as number of use days, estimated grams used and number of use “sessions” per day (individual instances of use separated by at least 1 hour). The PSS and ACE questionnaires are reported to have adequate consistency (α=0.70 and 0.70, respectively) while our sample had slightly higher values (α=0.86 and 0.77, respectively).^24,26 All study procedures were approved by the Medical University of South Carolina IRB and were conducted in accordance with the Declaration of Helsinki.

Statistical Analysis

Demographic and clinical characteristics were collected at baseline and contrasted between men and women using standard non-parametric methods (e.g., Wilcoxon rank-sum tests for continuous variables and Pearson chi-squares test for categorical variables). Demographic and clinical characteristics are noted as observed means with standard deviations and percentages. IDTS subscale scores were tabulated for all participants and compared between men and women using linear regression models; results are presented as regression parameter estimates (β) and gender-specific model-based estimated means and standard errors (SE). Initial models were unadjusted for significant differences between men and women; following, models were adjusted for demographic and clinical characteristics that varied between genders. Associations between IDTS subscale scores with cannabis use sessions per day (TLFB), past-month stress (PSS), and adverse childhood experiences (ACE total and subscale scores), as well as the moderating effects of gender on the relationship with IDTS subscale scores, were examined through inclusion of main effects and interaction terms in models. Cannabis use sessions per day was selected as the most relevant metric of recent cannabis use for these analyses as it had the highest correlation with all IDTS subscale scores relative to past-90 day number of cannabis use days and grams used per day. Gender-stratified parameter estimates were calculated using model contrast statements. All statistical analyses were assessed at an uncorrected α=0.05.

Results

Sample Characteristics

Demographic and clinical characteristics are listed in Table 1. Although men and women were comparable across most domains, women were older (29.7 (SD=7.6) vs. 25.9 (SD=6.9) years old, p<.01) and had used cannabis for a longer duration prior to study entry (10.2 (SD=7.6) vs. 8.3 (SD=6.7) years, p=.07). Men reported marginally more frequent cannabis use in the past 90 days relative to women (85.4 (SD=8.8) vs. 81.1 (SD=13.5) use days, p=.01), though women endorsed a slightly higher quantity of use on cannabis use days (1.9 (SD=2.8) vs. 1.0 (SD=0.9) grams used per using day, p=.05). Women endorsed more ACEs (3.1 (SD=2.5) vs. 1.9 (SD=2.0), p=.02) relative to men and were more likely to report ACEs relating to emotional/physical abuse (50% vs 27%, p<.01), neglect (33% vs. 17%, p=.02) and sexual abuse (33% vs. 9%; p<.01).

Table 1.

Sample characteristics

	Women (n=66)	Men (n=82)	p-value
Age in years*	29.7 (7.6)	25.9 (6.9)	<.01
Race
Black/African American	26%	14%	.11
White	68%	83%	--
Native Hawaiian or Pacific Islander	0%	1%	--
Asian	1%	1%	--
Other	5%	1%	--
Currently married	20%	12%	.18
Cigarette smoker	26%	18%	.27
ACE Total Score*	3.1 (2.5)	1.9 (2.0)	.02
Any Emotional/Physical Abuse*	50%	27%	<.01
Any Neglect*	33%	17%	.02
Any Sexual Abuse*	33%	9%	<.01
Any Household Dysfunction	70%	63%	.38
PSS Total Score	16.5 (7.0)	15.4 (6.0)	.35
Cannabis Use
CUD severity
Mild	20%	29%	.49
Moderate	34%	32%	--
Severe	45%	40%	--
Age of first use*	19.1 (4.2)	18.0 (4.6)	.02
Years of use	10.2 (7.6)	8.3 (6.7)	.07
Use days*	81.1 (13.5)	85.4 (8.8)	.01
Use sessions/using day	3.8 (6.6)	2.9 (1.9)	.21
Grams used/using day	1.9 (2.8)	1.0 (0.9)	.05

Note: Values are presented as means (standard deviation) or as a proportion of participants of each gender. Proportions are rounded and may not total to 100%. Use days, use sessions/day, and grams used/day refer to cannabis use behavior in the 90 days prior to baseline, assessed via TLFB.

^*Indicates a significant gender difference, p<.05.

Gender differences in antecedents of cannabis use

IDTS subscale scores stratified by gender are shown in Figure 1. Men and women similarly endorsed past-year cannabis use when experiencing pleasant emotions or pleasant times with others at high rates, whereas testing personal control was endorsed the least across genders. Relative to men, however, women endorsed using cannabis more often when experiencing unpleasant emotions (50.6 (SE=2.7) vs. 33.8 (SE=2.4), p<.01), physical discomfort (41.6 (SE=2.4) vs. 30.7 (SE=1.6), p<.01), conflict with others (28.4 (SE=2.7) vs. 16.5 (SE=1.8), p<.01), or when testing personal control (20.9 (SE=2.1) vs. 13.3 (SE=1.6), p<.01). No significant gender differences were seen for cannabis use when experiencing pleasant emotions, urges/temptations, when feeling social pressure to use, or when having pleasant times with others (all p’s>.05). Following adjustment for covariates, no changes in overall significance were noted.

Figure 1. Contexts associated with cannabis use by gender.

Women were more likely to endorse cannabis use when experiencing unpleasant emotions, physical discomfort, conflict with others, or when testing personal control relative to men. Both genders endorsed high rates of cannabis use when experiencing pleasant emotions or pleasant times with others. Data are presented as mean subscale scores and associated standard errors. *Indicates a significant gender difference, p<.05.

Associations between use antecedents and cannabis use, adverse childhood events, and recent stress

Increased endorsement of use when experiencing unpleasant emotions (β=2.9, SE=0.9; p<.01), physical discomfort (β=2.4, SE=0.7; p<.01), pleasant emotions (β=2.6, SE=0.7; p<.01), conflict with others (β=3.2, SE=0.7; p<.01), and urges/temptations to use (β=1.9, SE=0.9; p=.04) was positively associated with number of cannabis use sessions per day in the past 90 days (Table 2). Gender interactions were nonsignificant (p’s>.25).

Table 2.

Associations between use contexts and past 90-day cannabis use

	Overall (n=148)	Women (n=66)	Men (n=82)	Interaction p
Unpleasant Emotions	2.93 (0.90); p=.001	2.55 (1.02); p=.015	3.24 (1.46); p=.029	0.7
Physical Discomfort	2.39 (0.67); p=.001	2.63 (0.90); p=.005	1.66 (0.96); p=.089	0.47
Pleasant Emotions	2.57 (0.74); p=.001	2.81 (0.87); p=.002	2.03 (1.31); p=.126	0.25
Testing Personal Control	0.51 (0.63); p=.421	1.09 (0.82); p=.191	−0.78 (0.97); p=.420	0.15
Urges/Temptations	1.92 (0.93); p=.041	1.49 (1.26); p=.243	2.61 (1.46); p=.077	0.18
Conflict with Others	3.24 (0.74); p=.001	3.19 (0.99); p=.002	3.05 (1.07); p=.006	0.92
Social Pressure to Use	0.74 (0.91); p=.421	−0.18 (1.18); p=.883	2.47 (1.50); p=.104	0.17
Pleasant Times with Others	0.95 (0.87); p=.274	0.68 (1.15); p=.556	1.39 (1.40); p=.323	0.7

Note: Data are presented as betas, associated standard errors, and p-values within each gender grouping. Values represent the association between cannabis use contexts and past 90-day cannabis use within each gender grouping. The interaction column refers to the interaction between 90-day cannabis use and gender in predicting use context. Bolded values indicate significant associations between a given context and cannabis use, p<.05.

Similarly, an increasing number of unique reported ACEs was positively associated with use when experiencing unpleasant emotions (β=3.2, SE=0.8; p<.01), physical discomfort (β=1.7, SE=0.6; p<.01), and conflict with others (β=2.3, SE=0.7; p<.01) (Table 3), again without significant gender interactions (p’s>.28). In addition to total ACEs reported, responses to the ACE questionnaire were categorized into factors relating to emotional/physical abuse, household dysfunction, sexual abuse, or neglect, and factor scores were assessed for associations with IDTS subscale scores. Like ACE total scores, endorsement of emotional/physical abuse, household dysfunction, and neglect were significantly and positively associated with use when experiencing unpleasant emotions (Emotional/Physical Abuse: β=11.4, SE=3.9, p<.01; Household Dysfunction: β=11.0; SE=4.0, p<.01; Neglect: β=15.4, SE=4.4, p<.01), physical discomfort (Emotional/Physical Abuse: β=7.4, SE=3.0, p=.01; Household Dysfunction: β=6.2, SE=3.0, p=.04; Neglect: β=9.2, SE=3.3, p<.01), and conflict with others (Emotional/Physical Abuse: β=7.3, SE=3.3, p=.03; Household Dysfunction: β=7.6, SE=3.4, p=.03; Neglect: β=10.4, SE=3.7, p<.01). No significant gender differences were seen in the relationship between ACE factors and IDTS scores (all ACE x gender interactions, p’s>.05).

Table 3.

Associations between use contexts and childhood and past-month stress

	Overall (n=148)	Women (n=66)	Men (n=82)	Interaction p
Childhood Stress (ACEs)
Unpleasant Emotions	3.22 (0.79); p=.001	2.13 (1.04); p=.044	2.79 (1.21); p=.023	0.678
Physical Discomfort	1.74 (0.61); p=.005	1.58 (0.94); p=.099	0.69 (0.81); p=.398	0.473
Pleasant Emotions	0.30 (0.69); p=.667	0.07 (0.95); p=.945	0.09 (1.11); p=.936	0.987
Testing Personal Control	0.74 (0.56); p=.193	0.96 (0.83); p=.251	−0.51 (0.80); p=.539	0.201
Urges/Temptations	0.00 (0.85); p=.998	0.16 (1.28); p=.905	−1.00 (1.23); p=.418	0.517
Conflict with Others	2.30 (0.68); p=.001	2.40 (1.03); p=.022	0.89 (0.93); p=.334	0.277
Social Pressure to Use	−0.38 (0.82); p=.643	−0.56 (1.18); p=.639	−0.22 (1.27); p=.865	0.854
Pleasant Times with Others	0.00 (0.78); p=.997	−0.17 (1.16); p=.876	−0.27 (1.17); p=.815	0.955
Past-Month Stress (PSS)
Unpleasant Emotions	1.39 (0.28); p=.001	0.65 (0.39); p=.101	1.96 (0.35); p=.001	0.012
Physical Discomfort	0.54 (0.22); p=.018	0.40 (0.35); p=.269	0.56 (0.26); p=.037	0.712
Pleasant Emotions	−0.01 (0.25); p=.959	0.26 (0.35); p=.461	−0.34 (0.36); p=.346	0.238
Testing Personal Control	0.00 (0.20); p=.994	−0.32 (0.31); p=.299	0.23 (0.26); p=.377	0.168
Urges/Temptations	0.25 (0.31); p=.411	−0.23 (0.47); p=.626	0.68 (0.40); p=.091	0.137
Conflict with Others	1.05 (0.24); p=.001	0.93 (0.38); p=.017	1.04 (0.28); p=.001	0.817
Social Pressure to Use	0.01 (0.30); p=.965	−0.26 (0.44); p=.552	0.25 (0.42); p=.556	0.399
Pleasant Times with Others	0.18 (0.28); p=.526	0.21 (0.43); p=3619	0.10 (0.39); p=.795	0.844

Note: Data are presented as betas, associated standard errors, and p-values within each gender grouping. Values represent the association between cannabis use contexts and ACE/PSS total scores within each gender grouping. The interaction column refers to the interaction between ACE/PSS total score and gender in predicting use context. Bolded values indicate significant associations between a given context and ACE/PSS score, p<.05.

Like ACE scores, increased past-month perceived stress was positively associated with use when experiencing unpleasant emotions (β=1.4, SE=0.3; p<.01), physical discomfort (β=0.5, SE=0.2; p=.018), and conflict with others (β=1.1, SE=0.2; p<.01) (Table 3). However, unlike ACEs, a significant gender interaction was observed for PSS and use when experiencing unpleasant emotions (p=.01); past-month stress was significantly associated with use when experiencing unpleasant emotions in men (β=2.0, SE=0.4, p<.01), but not women (β=0.7, SE=0.4, p=.10).

Discussion

In this study, women with CUD were more likely to endorse using cannabis when experiencing negative affective states—including unpleasant emotions, physical discomfort, conflict with others, or when testing personal control—relative to men, and both genders reported high incidence of use when experiencing positive affective states (pleasant emotions, pleasant times with others). Use when experiencing unpleasant emotions, physical discomfort, pleasant emotions, urges/temptations, and conflict with others were all positively associated with cannabis use sessions per day in the past 90 days, and use when experiencing unpleasant emotions, physical discomfort, or conflict with others were all positively associated with number of unique adverse childhood experiences (ACEs) and past-month perceived stress. Notably, gender differences were evident in the association between cannabis use when experiencing unpleasant emotions and past-month stress: in stratified analyses, cannabis use when experiencing unpleasant emotions was positively associated with past-month stress in men but not women. Taken together, results from this study show gender differences in circumstances associated with cannabis use, including the contexts in which individuals use (e.g. negative affective states) and other life circumstances (recent stress).

Consistent with previous work, both genders assessed reported high rates of cannabis use when experiencing pleasant emotions or pleasant times with others.⁹ The finding that women use cannabis more often than men when experiencing unpleasant emotions is also consistent with gender differences in use motives observed for opioids and benzodiazepines.^6,7 Importantly, use when experiencing negative affective states (unpleasant emotions, physical discomfort, urges/temptations, conflict with others), as well as pleasant emotions, were all positively associated with recent cannabis use in the whole sample. As such, behavioral interventions for CUD may need to treat both types of circumstances as high-risk situations for use and implement training for each circumstance simultaneously (e.g. coping skills training for experiencing negative affect vs. cost-benefit analysis for experiencing positive affect). Moreover, while the finding that recent cannabis use was positively associated with both negative and positive affective contexts may appear contradictory, the negative contexts positively associated with cannabis use in this sample map well with symptoms of cannabis withdrawal (depressed mood; physical symptoms such as stomach pains, hand tremors/shakes, sweating, hot flashes, chills, or headaches; irritability, anger, or aggression).²⁸ Taken together, this outcome may reflect an association between greater cannabis use and withdrawal symptoms, with withdrawal subsequently motivating further use. This is consistent with previous work indicating a role for withdrawal in perpetuating cannabis use.²⁸ Notably, women typically report more severe cannabis withdrawal symptoms and greater withdrawal-associated impairment relative to men.⁵ Withdrawal-perpetuated use might therefore be more prevalent among women with CUD, suggesting pharmacotherapeutics development targeting cannabis withdrawal symptoms may have increased efficacy in relapse prevention among women. Alternatively, simultaneous high endorsement of use when experiencing positive and negative affective states may simply be a consequence of these circumstances occurring more often in everyday life than the other circumstances measured (e.g. testing personal control). Future studies might look to disassociate cannabis use driven by certain circumstances with the gross incidence of those circumstances via use of daily diaries or ecological momentary assessment.

In addition to frequency of use, we found that a greater number of unique adverse childhood experiences (ACEs) and past-month perceived stress (PSS) were positively associated with use when experiencing negative affect (unpleasant emotions, physical discomfort, conflict with others). Our ACE findings are consistent with previous work showing that using cannabis to cope with distress mediates the relationship between childhood adversity and problematic cannabis use.²⁹ This mediation may be related to the impaired emotional regulation commonly observed in individuals with ACEs,³⁰ increasing reliance on maladaptive emotional regulation strategies like cannabis use. When considering gender, previous work indicates a positive association between cumulative ACEs and increased interpersonal difficulty in adulthood, with this association also being mediated by emotional dysregulation.³¹ While observed associations between ACEs and use when experiencing negative affect were directionally similar across genders in our sample, women reported more ACEs relative to men, consistent with what is seen in the general population.¹⁶ Women may therefore be at greater risk of using cannabis as a tool for emotional regulation, providing additional barriers to quitting during treatment for CUD.

Use when experiencing negative affective states was also associated with past-month perceived stress (PSS), though this relationship was driven by men with respect to use when experiencing unpleasant emotions. As with ACEs, the association between use when experiencing negative affect and recent perceived stress is likely rooted in the use of cannabis to regulate emotions, as several items on the PSS relate to feelings of inability to cope with distress. Importantly, Cavalli and Cservenka³² showed that emotion dysregulation had the greatest moderating effect on the association between perceived stress and problematic cannabis use when perceived stress was relatively low. Average PSS scores in the present study were 15.4 for men and 16.5 for women out of a total potential score of 40, only marginally higher than the 12.1 for men and 13.7 for women reported for the normative sample used for validating the PSS.²⁴ Thus, like in Cavalli and Cservenka,³² the cannabis use behavior reported in the current study may also have been impacted by poor emotion regulation. Regarding gender, men may be more susceptible to acute stress-induced cannabis use compared to women, whereas women might be more susceptible to chronic stress-induced use (e.g. when experiencing pathological anxiety^33–35). This is supported by human laboratory work in individuals that use cannabis regularly without CUD, which has shown that men self-administer more cannabis immediately post-stress compared to women.¹⁴ Further, as stress-associated cannabis use has been shown to increase baseline negative affect,¹⁵ men may be specifically at risk of entering a self-perpetuated cycle of negative affect-associated cannabis use. Thus, appropriate intervention strategies for stress-associated cannabis use may differ by gender. Men may benefit from acute stress management or relapse prevention techniques, as to disrupt the association between acute stress and cannabis use. In contrast, women may benefit from more generalized approaches to stress management.

Results should be interpreted in consideration of several limitations. First, this study relies on retrospective assessments of 1) cannabis use contexts, 2) lifetime and recent stress, and 3) recent cannabis use. Outcomes may have been impacted by recall bias, particularly negativity bias³⁶, leading to overrepresentation of negative affective states and stressful circumstances in our data. As there is evidence to suggest women are more susceptible to negativity bias compared to men³⁷, prospective studies are needed to confirm our gender-related findings reported herein. Relatedly, our analyses include associations between past-month (PSS) or lifetime (ACE) stresses and past-year cannabis use contexts. As the present study only collected outcomes at a single time point, the differing timelines of these assessments impact interpretability of our results. For example, past-month stress might not reflect past-year stress, and thus an association between past-month stress and past-year cannabis use contexts might be biased. Longitudinal studies examining whether associations between recent or lifetime stress and cannabis use contexts hold over time could provide stronger insights into the relationships between these outcomes. Next, this study only evaluated individuals that used cannabis heavily and met criteria for CUD but were not currently seeking treatment. Near-daily use can bias results toward greater endorsement of use across contexts, which may inflate endorsement for use in maladaptive circumstances. Further, cannabis use motivated by coping has been associated with cannabis-related problems and dependence symptoms in previous work.^38–40 Individuals that use cannabis heavily that do not meet criteria for CUD may be less likely to endorse use when experiencing negative affective states than individuals in our sample. Conversely, it is possible that individuals seeking treatment for CUD may be more likely to endorse use to cope with negative affective states than the individuals included in our sample. Future research should look to compare use motives across treatment- and non-treatment-seeking individuals with CUD to determine optimal strategies for treatment development. Participants were also only provided with binary options for self-reported gender; future studies should include more comprehensive response options (e.g. non-binary, transgender, two-spirit, etc.) to better understand associations between circumstances associated with cannabis use and gender. With respect to our analytic approach, outcomes reported herein were uncorrected for multiple comparisons, inflating chance of type I error, and this study sample included a marginally greater number of men relative to women. Replication studies including larger samples of women are necessary to validate our findings.

In conclusion, our study found that women were more likely to endorse cannabis use when experiencing negative affective states relative to men; frequency of use and childhood adversity were associated with use when experiencing negative affect; and past-month stress was positively associated with use when experiencing negative affect differentially across genders. These results suggest that cannabis use may be differentially associated with contextual and life circumstantial factors across genders, indicating a need for gender-specific behavioral therapeutic interventions for CUD.