The association between benzodiazepine proliferation in the unregulated drug supply and experiences of violence: A gender-based analysis

Abstract

Background:

Rising contamination of illicitly-manufactured benzodiazepines in unregulated opioids in Canada poses concern as their strong sedative effects may increase vulnerability to violence for people who use drugs (PWUD), particularly women. The current study longitudinally examined the relationship between suspected exposure to benzodiazepines (SEB) and violent victimization, assessing for gender differences.

Methods:

Data were drawn from 1049 participants (40.9% women) from three harmonized prospective cohorts of community-recruited PWUD in Vancouver, Canada between 2021-2023. We conducted gender-stratified analyses, with a three-level exposure variable 1) primary SEB: through using unregulated benzodiazepines, 2) secondary SEB: only through using other unregulated drugs, and 3) no exposure (reference category). Patterns of unregulated opioid and stimulant use were examined as effect modifiers.

Findings:

At baseline, 249 (58.0%) women and 349 (56.3%) men reported any SEB; 84 (19.6%) women and 126 (20.3%) men reported experiencing violence. For all participants, secondary SEB was significantly related to higher odds of violence (adjusted odds ratio [AOR]=1.48; 95% confidence interval [CI]:1.14-1.92), however this association did not persist for primary SEB (AOR=1.26; 94% CI:0.73-2.18). In gender-stratified analyses, the association between secondary SEB and violence was only significant for women (AOR=1.62; 95% CI:1.10-2.38). Secondary SEB was also associated with higher odds of violence when participants reported daily stimulant and no daily opioid use (AOR=2.76; 95% CI:1.58-4.82).

Conclusions:

Our findings suggest lack of agency over exposure to benzodiazepines, as created by drug supply unpredictability, may exacerbate risk of violence among women who use drugs and PWUD who do not frequently use opioids.

1. INTRODUCTION

Over the last decade, the unregulated drug supply in North America has become increasingly unpredictable and contaminated with illicitly-manufactured synthetic drugs resulting in more than 40,000 deaths in Canada between 2016 and 2023 (Special Advisory Committee on Toxic Drug Poisonings, 2024), and over 100,000 deaths in both 2021 and 2022 in the US (Spencer et al., 2023) from toxic drug poisoning. When the Canadian province of British Columbia (B.C.) declared a public health emergency in 2016 following a sudden rise in fatal drug poisonings, the primary concern was the emergence of fentanyl, a highly potent synthetic opioid (Ministry of Health, 2016). However, the unregulated drug supply is continuing to change rapidly, with additional contaminants emerging. While existing research has largely focused on how rising toxicity has impacted mortality rates, people who use drugs (PWUD) have raised concerns that the recent proliferation of benzodiazepines in the unregulated drug supply may be exacerbating their risk of violent victimization, particularly for women (Denis, 2021; Smyth, 2022; Speed et al., 2025).

Benzodiazepines, a class of drugs commonly prescribed for symptoms of anxiety and insomnia, have been increasingly detected in unregulated opioids in B.C. since around 2019 (Canadian Centre on Substance Use and Addiction, 2021; Laing et al., 2021; Russell et al., 2023; Scarfone et al., 2022). Many of the benzodiazepines detected are considered to be novel (i.e., illicitly manufactured, non-medically-approved sedatives in Canada or the US, such as etizolam or bromazolam, that can be several times stronger than their medically-approved counterparts; Russell et al., 2023). Post-mortem toxicology data from B.C.’s Coroners Service reports that detection of benzodiazepines in fatal overdose cases rose from 3% in 2018 to 28% in 2022 (B. C. Coroners Service, n.d.). Drug checking data in B.C. shows benzodiazepines were consistently found in about half of all expected-opioid samples by mid-2023 (BC Centre on Substance Use, 2023). When combined with opioids, benzodiazepines can increase risk of fatal and non-fatal drug poisoning (BC Centre on Substance Use, 2021). Symptoms of non-fatal drug poisoning from concurrent use of opioids and benzodiazepines include unconsciousness, extended periods of memory loss, and an inability to respond to external stimuli. The strong sedating effects of benzodiazepines may reduce the ability of people to respond if assaulted (BC Centre on Substance Use, 2021; Canadian Centre on Substance Use and Addiction, 2021). Notably, in some qualitative studies examining gendered experiences of the ongoing drug poisoning crisis, women reported that non-fatal overdose from fentanyl increases their risk of predatory or opportunistic physical and sexual violence (Boyd et al., 2018; Harris et al., 2021). In a recent qualitative study, women specifically described fear of unexpected benzodiazepine-related blackout or heightened intoxication exacerbating their vulnerability to violence (Speed et al., 2025).

Although the link between suspected exposure to benzodiazepines through contaminated drugs and interpersonal violence among PWUD remains underexplored, benzodiazepines have previously been implicated in producing vulnerability to violent victimization in research examining drug-facilitated sexual assaults (Morgillo et al., 2023; Navarro Escayola et al., 2023). Concerningly, PWUD, and disproportionately so women, already experience heightened rates of violence as a result of intersecting systems of structural violence, such as drug prohibition and related criminalization, capitalism-driven impoverishment, and patriarchy (Boyd et al., 2022; Fleming et al., 2023; McNeil et al., 2021; Perreault, 2020; Shannon et al., 2008). These oppressive systems force people who use unregulated drugs into over-policed and under-resourced environments that promote violence by perpetuating gender-based hierarchies and competition over limited resources. Although research shows PWUD use a variety of tactics to avoid violence, such as using alone (Collins et al., 2020; Ivsins et al., 2022), their ability to do so is also gendered (Marshall et al., 2008). For instance, women are more likely to be attacked by people they know and may not be able to avoid (Marshall et al., 2008). To our knowledge, there are no quantitative studies examining how increasing benzodiazepine contamination in the unregulated drug supply has impacted risk of violence among PWUD.

Therefore, we conducted a longitudinal, gender-based analysis of suspected exposure to benzodiazepines and experiences of physical and sexualized violence. Additionally, given that benzodiazepines are typically more common in the unregulated opioid supply and less so in unregulated stimulants (e.g., crystal methamphetamine, crack, and cocaine; Knill et al., 2023, 2022), the effect of benzodiazepines on risk for exposure to violence may differ depending upon whether people primarily use opioids or stimulants, or some combination of the two. As such, we analyzed whether associations between suspected exposure to benzodiazepines and experiences of violence were modified by different patterns of unregulated opioid and stimulant use.

2. METHODS

2.1. Study procedures

This study derived data from three ongoing, open, and prospective cohort studies of PWUD. These three cohorts are: 1) the Vancouver Injection Drug Users Study (VIDUS); 2) the AIDS Care Cohort to evaluate Exposure to Survival Services (ACCESS); and 3) the At-Risk Youth Study (ARYS). All three cohorts have been described in detail previously (Hayashi et al., 2018; Wood et al., 2006). For all three cohorts, participants are recruited using community-based methods (e.g., word of mouth, postering, community outreach), they must reside in the Greater Vancouver region at enrolment, and they must provide written informed consent. Eligibility criteria vary across the three cohorts: VIDUS participants must be aged 18 years or older, HIV-negative, and have injected unregulated drugs within the month prior to enrolment; ACCESS participants must be over the age of 18, living with HIV, and have used an unregulated drug (not including cannabis) within the month prior to enrolment (any consumption method); and ARYS participants must be between the ages of 14 and 26 at enrolment, street-involved, and have used an unregulated drug (not including cannabis) within the month prior to enrolment (any consumption method).

At baseline and every six months thereafter, participants complete an interviewer-administered questionnaire with items related to drug use patterns, income generating activities, drug-related harms and engagement with health and social services. Study protocols are harmonized to allow for pooled analyses. Participants receive $50 CAD for each study visit. ACCESS has received approval from the University of British Columbia/Providence Health Care Research Ethics Board and VIDUS and ARYS have approval from the Simon Fraser University Research Ethics Board.

The study period for the current analyses was from June 1, 2021 to May 31, 2023. Participants were eligible if they reported using unregulated drugs in the past six months—not including exclusive use of cannabis—and had valid responses for the primary exposure and outcome variables (described below). Participants were grouped according to their self-identified gender (men vs. women; trans-inclusive) to explore how gender expression or presentation that aligns with binary gender constructs—as created and sustained by patriarchy (Hoskin, 2020)—shapes experiences of violence. Two participants who identified as Two-Spirit or non-binary were excluded from the current study as their gender expression may not fit into binary gender categories and there were not enough counts to create a separate group. We also conducted sensitivity analyses to assess whether grouping transgender participants with cisgender participants was appropriate (described below).

2.2. Study measures

We used a binary gender variable (men vs. women; trans-inclusive) to conduct gender-based analysis. The primary outcome variable was having experienced physical or sexual violence (henceforth, violence) in the past six months (yes vs. no). Yes was defined as responding “yes” to one of the following survey questions: 1) “In the last six months, have you been physically attacked or suffered any kind of physical violence, including torture or punishment related to a drug debt? (Do not include sexual assault.)” or 2) “In the last six months, have you been forced to have sex or perform a sexual act against your will, or experienced any kind of sexual assault?”. No was defined as responding “no” to both aforementioned survey questions.

The main exposure variable was self-reported suspected exposure to benzodiazepines (primary exposure vs. secondary exposure vs. no exposure). Primary exposure was operationalized as reporting any unregulated benzodiazepine use in the past six months. Secondary exposure was operationalized as responding yes to the survey question “In the last six months, have you used any drugs that you knew, or now believe, contained benzodiazepines (benzos)?” AND not reporting any unregulated benzodiazepine use in the past six months. No exposure was defined as responding no to the described survey question, AND not reporting any unregulated benzodiazepine use in the past six months. Suspected exposure to benzodiazepines will be henceforth referred to as SEB, unless otherwise indicated.

A four-category variable assessing patterns of unregulated stimulant and opioid use (no daily use of either drug vs. daily stimulant use without daily opioid use vs. daily opioid use without daily stimulant use vs. daily use of both drugs) was incorporated as an explanatory variable in primary analyses, and an effect modifier in a secondary analysis.

Additional explanatory variables were determined based on previous research examining experiences of violence among PWUD (Boyd et al., 2018; Collins et al., 2020, 2018; Fairbairn et al., 2008; Harris et al., 2021; Hayashi et al., 2016; Ivsins et al., 2023, 2020; Jaffe et al., 2018; Kennedy et al., 2017; Marshall et al., 2008; Riley et al., 2020; Shannon et al., 2008). Demographic characteristics were age (per year older), and race/ethnicity/Indigenous ancestry (White vs. Black, Indigenous, and people of colour). In addition to the unregulated opioid and stimulant use patterns described above, drug-use patterns and practices include: daily alcohol use; supervised consumption services or overdose prevention site use; drug checking service use; used drugs alone (75% of the time or more vs. less than 75% of the time); accessed drug treatment (e.g., opioid agonist therapy, detox, residential or in community treatment, counselling); and accessed prescribed safer supply (i.e., prescribed alternatives to the unregulated supply; Ministry of Mental Health and Addictions and Ministry of Health, 2021). Social and structural factors include: resided in the Downtown Eastside of Vancouver (DTES); recent homelessness; worked in the unregulated drug market (e.g., drug selling, middling); engaged in sex work; street-based income generation (e.g., recycling, squeegeeing, or panhandling); other illegal income generation (e.g., theft, robbing, or stealing); and recent incarceration. Unless otherwise specified, the variables were dichotomized as yes vs. no, and aside from age and race/ethnicity/Indigenous ancestry, all variables in the current study refer to the past six months.

2.3. Statistical analysis

Baseline characteristics of participants stratified by SEB were examined using the Mann-Whitney Test for continuous variables and Chi-square test for categorical variables.

For the primary analysis, bivariate and multivariable generalized estimating equations (GEE) with a logit link and an exchangeable correlation structure were used to assess the association between SEB and experiencing violence. All exposures (except for race/ethnicity/ancestry, which was measured at baseline only) and the outcome were measured at the same study visits, and GEE’s exchangeable correlation structure addresses the correlation arising from repeated measures by assuming that all pairs of observations within a subject are equally correlated, regardless of the time difference between them. We fit three models, including one for the entire sample, one with only women (trans-inclusive), and one with only men (trans-inclusive). All multivariable models included the same variables used in the respective bivariate models (see Table 1). We used R (version 4.4.1, R Foundation for Statistical Computing, Vienna, Austria) for all analyses.

Table 1.

Baseline sample characteristics of participants stratified by suspected exposure to benzodiazepines (n=1049)

Characteristics	Total (n=1049) N (%)	Suspected exposure to benzodiazepinesa
		Yes (57%), (n=598) N (%)	No (43%), (n=451) N (%)	p-value
Age [median (IQR)]b	45 (34-56)	42 (33-53)	49 (36-58)	<0.001
Genderb
Men (trans-inclusive)	620 (59.1)	349 (58.3)	271 (60.1)	0.573
Women (trans-inclusive)	429 (40.9)	249 (41.6)	180 (39.9)
Race/Ethnicity/Ancestry
White	591 (56.3)	359 (60.0)	232 (51.4)	0.036
Indigenous	392 (37.4)	207 (34.6)	185 (41.0)
Black	22 (2.1)	9 (1.5)	13 (2.9)
Other people of colour	38 (3.6)	21 (3.5)	17 (3.8)
Missing	6 (0.6)	2 (0.3)	4 (0.9)
Experienced violencebc	210 (20.0)
Unregulated stimulant & opioid use patternsbd
No daily opioid or stimulant use	395 (37.7)	22 (45.8)	373 (37.3)	0.549
Daily opioid & no daily stimulant use	139 (13.3)	4 (8.3)	135 (13.5)
Daily stimulant & no daily opioid use	210 (20.0)	10 (20.8)	200 (20.0)
Daily opioid and stimulant use	305 (29.1)	12 (25.0)	293 (29.3)
Daily alcohol useb	130 (12.4)	67 (11.2)	63 (14.0)	0.176
Drug-use behaviours
Used drugs aloneb	515 (49.1)	286 (47.8)	229 (50.8)	0.38
Used a drug checking serviceb	206 (19.6)	146 (24.4)	60 (13.3)	<0.001
Used a supervised consumption siteb	308 (29.4)	244 (40.8)	64 (14.2)	<0.001
Accessed prescribed safer supplyb	291 (27.7)	227 (38.0)	64 (14.2)	<0.001
Accessed substance use treatmentb	665 (63.4)	437 (73.08)	228 (50.55)	<0.001
Social-structural factors
Engaged in sex workb	117 (11.2)	83 (13.9)	34 (7.5)	0.001
Worked in unregulated drug marketb	312 (29.7)	241 (40.3)	71 (15.7)	<0.001
Street-based income generationbe	337 (32.1)	226 (37.8)	111 (24.6)	<0.001
Illegal income generationbf	117 (11.2)	95 (15.9)	22 (4.9)	<0.001
Incarcerationb	66 (6.3)	45 (7.5)	21 (4.7)	0.061
Homelessnessb	191 (18.2)	147 (24.6)	44 (9.8)	<0.001
DTES residenceb	534 (50.9)	331 (55.4)	203 (45.0)	<0.001

Note: ^a defined as exposure through using unregulated benzodiazepines or through using other unregulated drugs

^bdenotes activities and events in the previous six months.

^crefers to physical and/or sexualized violence

^dstimulants include crack, cocaine, or crystal methamphetamine

^eincludes recycling/binning/panhandling

^fincludes theft/robbing/stealing/other criminal activity

IQR: interquartile range.

DTES: Downtown Eastside of Vancouver

For a secondary analysis, an additional model was fit using multivariable GEE with a logit link and an exchangeable correlation structure to test for modifying effects of patterns of unregulated stimulant and opioid drug use on the association between SEB and experiencing violence in the entire sample.

Finally, the gender stratified primary analyses were re-run as sensitivity analyses without transgender participants to determine whether the relationship between SEB and violence varied according to cis and transgender identity.

3. RESULTS

A total of 1049 participants provided 2598 observations. Median number of missing follow-up interviews per participant among all participants was 1 (interquartile range [IQR] = 1-3) out of maximun three follow-up interviews to be completed. Table 1 presents baseline sample characteristics. Median age of participants at baseline was 45 years (IQR = 34-56), 429 (40.9%) were women (including 9 trans women), and 620 (59.1%) were men (including 1 trans men). At baseline, 598 (57%) participants reported SEB in the previous six months. Among those who reported SEB, 48 (8.0%) reported primary exposure (i.e., using unregulated benzodiazepines in the past six months), and 591 (98.8%) reported secondary exposure (i.e., exposure to benzodiazepines through using other unregulated drugs). Further, 249 (58.0%) women and 349 (56.3%) men reported SEB, with no significant difference in reporting SEB between the two groups (p=0.573). Overall, 84 (19.6%) women and 126 (20.3%) men reported experiencing violence. There was no significant gender difference in the prevalence of experiencing violence overall (p=0.768) or in experiencing only physical violence (p=0.723), however, prevalence of only sexualized violence was significantly higher among women (p<0.002).

Results for primary analyses of the entire sample and gender-stratified samples examining the association between SEB and experiencing violence in the past six months are reported in Tables 2, 3, and 4. In the model with all participants, after adjusting for confounding variables, SEB – when limited to secondary exposure cases – was significantly associated with experiencing violence (adjusted odds ratio [AOR] = 1.48; 95% confidence interval [CI]: 1.14 – 1.92). However, when SEB was defined as primary exposure, it was not significantly associated with experiencing violence (AOR = 1.26; 95% CI: 0.73 – 2.18).

Table 2.

Bivariate and multivariable GEE analyses of the relationship between suspected exposure to benzodiazepines and experiencing physical and/or sexualized violence among people who use drugs in Vancouver, Canada, July 2021 – May 2023.

Characteristics	Odds Ratio (95% CI)	Adjusted Odds Ratio (95% CI)
Suspected exposure to benzodiazepinesa
(primary exposureb vs. no)	1.63 (0.97-2.72)	1.26 (0.73-2.18)
(secondary exposurec vs. no)	2.10 (1.69-2.60) **	1.48 (1.14-1.92) **
Agea (per year older)	0.96 (0.95-0.97) **	0.97 (0.96-0.98) **
Gendera (men vs. women; trans-inclusive)	1.11 (0.87-1.40)	1.32 (1.01-1.73) *
Race/Ethnicity/Ancestry (White vs. BIPOC)	1.14 (0.90-1.44)	1.17 (0.91-1.49)
Unregulated stimulant and opioid use patternsad
(daily stimulant & no daily opioid use vs. no daily use)	1.26 (0.89-1.79)	1.15 (0.79-1.86)
(daily opioid & no daily stimulant use vs. no daily use)	1.81 (1.37-2.37)	0.97 (0.71-1.34)
(daily opioid & daily stimulant use vs. no daily use)	1.45 (1.19 – 1.75)	0.94 (0.69-1.27)
Daily alcohol usea (yes vs. no)	1.11 (0.80-1.53)	1.51 (1.06-2.15) *
Used drugs alonea (75% of the time or more vs. <75% of the time)	1.07 (0.88-1.30)	1.19 (0.96-1.48)
Used a drug checking servicea (yes vs. no)	1.43 (1.15-1.79) **	1.14 (0.88-1.46)
Used a supervised consumption sitea (yes vs. no)	1.84 (1.47-2.30) **	1.23 (0.95-1.61)
Accessed prescribed safer supplya (yes vs. no)	1.66 (1.36-2.03) **	1.23 (0.96-1.57)
Accessed substance use treatmenta (yes vs. no)	1.39 (1.11-1.74) **	1.21 (0.91-1.60)
Engaged in sex worka (yes vs. no)	2.31 (1.73-3.08) **	1.89 (1.36-2.62) **
Worked in the unregulated drug marketa (yes vs. no)	2.45 (2.00-3.00) **	1.55 (1.23-1.98) **
Street-based income generationae (yes vs. no)	1.46 (1.20-1.78) **	1.18 (0.94-1.47)
Illegal income generationaf (yes vs. no)	2.89 (2.24-3.75) **	1.40 (1.04-1.88) *
Recent incarcerationa (yes vs. no)	2.78 (1.90-4.07) **	1.86 (1.19-2.91) **
Recent homelessnessa (yes vs. no)	2.41 (1.90-3.05) **	1.61 (1.24-2.11)
DTES residencea (yes vs. no)	1.27 (1.02-1.58) *	1.22 (0.94-1.58)

Note: ^a denotes activities and events in the past six months.

^bprimary exposure refers to exposure to benzodiazepines through using unregulated benzodiazepines

^csecondary exposure refers to exposure to benzodiazepines through using other unregulated drugs only, without the primary exposure

^dstimulants include cocaine, crack or crystal methamphetamine

^eincludes recycling/binning/panhandling

^fincludes theft/robbing/stealing/other criminal activity

^**p<0.01

^*p<0.05

CI: confidence interval.

BIPOC: Black, Indigenous, and people of colour

DTES: Downtown Eastside of Vancouver

Table 3.

Bivariate and multivariable GEE analyses of the relationship between suspected exposure to benzodiazepines and experiencing physical and/or sexualized violence among men (trans-inclusive)

Characteristics	Odds Ratio (95% CI)	Adjusted Odds Ratio (95% CI)
Suspected exposure to benzodiazepinesa
(primary exposureb vs. no)	1.77 (0.90-3.48)	1.23 (0.57-2.68)
(secondary exposurec vs. no)	2.31 (1.73-3.08) **	1.38 (0.96-1.98)
Agea (per year older)	0.97 (0.96-0.98) **	0.98 (0.97 - 0.99) **
Race/Ethnicity/Ancestry (White vs. BIPOC)	1.21 (0.88-1.67)	1.22 (0.88-1.70)
Unregulated stimulant and opioid use patternsad
(daily stimulant & no daily opioid use vs. no daily use)	0.88 (0.53-1.45)	0.79 (0.45-1.37)
(daily opioid & no daily stimulant use vs. no daily use)	2.19 (1.57-3.05) **	1.09 (0.72-1.64)
(daily opioid & stimulant use vs. no daily use)	2.41 (1.76-3.30) **	0.93 (0.64-1.36)
Daily alcohol usea (yes vs. no)	1.19 (0.76-1.85)	1.76 (1.07-2.90) *
Used drugs alonea (75% of the time or more vs. <75% of the time)	0.96 (0.74-1.23)	1.02 (0.76-1.36)
Used a drug checking servicea (yes vs. no)	1.32 (0.98-1.76)	0.89 (0.64-1.25)
Used a supervised consumption sitea (yes vs. no)	2.14 (1.57-2.92) **	1.38 (0.95-2.00)
Accessed prescribed safer supplya (yes vs. no)	2.11 (1.63-2.74) **	1.40 (1.00-1.97)
Accessed substance use treatmenta (yes vs. no)	1.69 (1.26-2.26) **	1.27 (0.86-1.86)
Engaged in sex worka (yes vs. no)	4.36 (2.41-7.91) **	3.51 (1.82-6.79) **
Worked in the unregulated drug marketa (yes vs. no)	3.09 (2.36-4.05) **	1.87 (1.37-2.59) **
Street-based income generationae (yes vs. no)	1.41 (1.07-1.84) *	1.08 (0.79-1.48)
Illegal income generationaf (yes vs. no)	3.58 (2.55-5.04) **	1.43 (0.96-2.14)
Recent incarcerationa (yes vs. no)	2.74 (1.78-4.21) **	1.66 (0.94-2.93)
Recent homelessnessa (yes vs. no)	2.41 (1.80-3.24) **	1.52 (1.09-2.14) *
DTES residencea (yes vs. no)	1.29 (0.97-1.70)	1.28 (0.92-1.80)

Note: ^a denotes activities and events in the past six months.

^bprimary exposure refers to exposure to benzodiazepines through using unregulated benzodiazepines

^csecondary exposure refers to exposure to benzodiazepines through using other unregulated drugs only, without the primary exposure

^dstimulants include cocaine, crack or crystal methamphetamine

^eincludes recycling/binning/panhandling

^fincludes theft/robbing/stealing/other criminal activity

^**p<0.01

^*p<0.05

CI: confidence interval.

BIPOC: Black, Indigenous, and people of colour

DTES: Downtown Eastside of Vancouver

Table 4.

Bivariate and multivariable GEE analyses of the relationship between suspected exposure to benzodiazepines and experiencing physical and/or sexualized violence among women (trans-inclusive)

Characteristics	Odds Ratio (95% CI)	Adjusted Odds Ratio (95% CI)
Suspected exposure to benzodiazepinesa
(primary exposureb vs. no)	1.47 (0.69 - 3.15)	1.25 (0.60-2.64)
(secondary exposurec vs. no)	1.85 (1.34 - 2.56) **	1.62 (1.10-2.38) *
Age (per year older)	0.95 (0.94-0.97) **	0.96 (0.94-0.98) **
Race/Ethnicity/Ancestry (White vs. BIPOC)	1.01 (0.70 - 1.46)	1.01 (0.68 – 1.51)
Unregulated stimulant and opioid use patternsad
(daily stimulant & no daily opioid use vs. no daily use)	1.89 (1.15-3.11) *	1.86 (1.08-3.20) *
(daily opioid & no daily stimulant use vs. no daily use)	1.30 (0.80-2.10)	0.85 (0.50-1.43)
(daily opioid & stimulant use vs. no daily use)	2.01 (1.33-3.02) **	0.92 (0.55-1.53)
Daily alcohol usea (yes vs. no)	1.02 (0.64-1.62)	1.26 (0.78-2.04)
Used drugs alonea (75% of the time or more vs. <75% of the time)	1.25 (0.92-1.72)	1.48 (1.05-2.07) *
Used a drug checking servicea (yes vs. no)	1.62 (1.15-2.29) **	1.54 (1.03-2.30) *
Used a supervised consumption sitea (yes vs. no)	1.47 (1.06-2.02) *	1.08 (0.73-1.60)
Accessed prescribed safer supplya (yes vs. no)	1.22 (0.90-1.65)	1.05 (0.72-1.52)
Accessed substance use treatmenta (yes vs. no)	1.06 (0.75-1.51)	1.14 (0.75-1.73)
Engaged in sex worka (yes vs. no)	2.29 (1.60-3.29) **	1.63 (1.11-2.40) *
Worked in the unregulated drug marketa (yes vs. no)	1.80 (1.32 -2.45) **	1.24 (0.86-1.79)
Street-based income generationae (yes vs. no)	1.56 (1.17-2.07) **	1.30 (0.94-1.79)
Illegal income generationaf (yes vs. no)	2.17 (1.45-3.25) **	1.39 (0.87 -2.21)
Recent incarcerationa (yes vs. no)	2.81 (1.26-6.25) *	2.47 (1.10-5.56) *
Recent homelessnessa (yes vs. no)	2.40 (1.59-3.62) **	1.75 (1.12-2.74) *
DTES residencea (yes vs. no)	1.29 (0.91-1.84)	1.18 (0.78-1.78)

Note: ^a denotes activities and events in the past six months.

^bprimary exposure refers to exposure to benzodiazepines through using unregulated benzodiazepines

^csecondary exposure refers to exposure to benzodiazepines through using other unregulated drugs only, without the primary exposure

^dstimulants include cocaine, crack or crystal methamphetamine

^eincludes recycling/binning/panhandling

^fincludes theft/robbing/stealing/other criminal activity

^**p<0.01

^*p<0.05

CI: confidence interval.

BIPOC: Black, Indigenous, and people of colour

DTES: Downtown Eastside of Vancouver

Among only men, SEB was significantly associated with experiencing violence in bivariate GEE analyses when SEB was through secondary exposure (OR = 2.31; 95% CI: 1.73 – 3.08), however, SEB through secondary exposure did not retain a significant association after adjusting for confounding variables (AOR = 1.38; 95% CI: 0.96 – 1.98). Among only women, after adjusting for confounding variables, SEB through secondary exposure remained significantly associated with experiencing violence (AOR = 1.62; 95% CI: 1.10 – 2.38). SEB through primary exposure was not significantly associated with violence in the bivariate or multivariable analyses for either men or women.

The results did not significantly change in sensitivity analyses of gender-stratified models where trans-women and trans-men were removed (results tables are in the supplementary material).

In the secondary analysis, only daily stimulant without daily opioid use significantly modified the association between SEB through secondary exposure and experiencing violence (p=0.04). Specifically, SEB through secondary exposure was significantly associated with higher odds of experiencing violence among participants who reported daily stimulant without daily opioid use (AOR = 2.76; 95% CI: 1.58 – 4.82). Table 5 shows all marginal effects from the secondary analysis.

Table 5.

Marginal effects of suspected exposure to benzodiazepines conditional on unregulated stimulant and opioid use patterns

Interaction model	Adjusted Odds Ratio (95% CI)
SEB through primary exposurea vs. no SEB
Among those with no daily stimulant or opioid use	2.06 (0.90-4.72)
Among those with daily stimulant use & no daily opioid use	0.55 (0.13-2.25)
Among those with daily opioid use & no daily stimulant use	0.87 (0.26-2.97)
Among those with daily opioid and stimulant use	1.00 (0.37-2.73)
SEB through secondary exposureb vs. no SEB
Among those with no daily stimulant or opioid use	1.34 (0.89-2.04)
Among those with daily stimulant use & no daily opioid use	2.76 (1.58-4.82) **
Among those with daily opioid use & no daily stimulant use	1.57 (0.86- 2.86)
Among those with daily opioid and stimulant use	1.08 (0.70-1.68)

Note: ^a primary exposure refers to exposure to benzodiazepines only through using unregulated benzodiazepines

^bsecondary exposure refers to exposure to benzodiazepines through using other unregulated drugs

^dstimulants include cocaine, crack or crystal methamphetamine

^**p<0.01

^*p<0.05

As described in the supplementary material, missing observations were <4% in each model.

4. DISCUSSION

This study highlights critical gender-based differences related to the impacts of benzodiazepine contamination of the unregulated drug supply on experiences of violent victimization. Most participants in this study (57%) reported SEB at the time of their first interview during the study period. Among the entire sample, odds of experiencing violence were 1.5 times higher for people who reported SEB through secondary exposure (i.e., only through using other unregulated drugs and not through using unregulated benzodiazepines) compared to those who did not. This association did not persist for people who reported SEB through primary exposure (i.e., using unregulated benzodiazepines). These findings suggest the unpredictability in exposure to benzodiazepines through other drugs is more relevant to risk of violence than simply using benzodiazepines. This is an important finding as it differentiates “drug use”-related harm from “drug prohibition”-related harm, as it is drug prohibition that creates conditions for an unregulated drug market to exist in which profits are prioritized, including by cutting drugs with contaminants and buffers, without regulatory oversight (Beletsky and Davis, 2017; Cano et al., 2024). Our findings contrast from common narratives blaming PWUD for their negative health outcomes (Rhodes, 2009), and instead suggests lack of agency over drug use shaped by drug prohibition is more relevant to health than individual choice.

Further, we found odds of experiencing violence were over 1.6 times higher among women who reported SEB through secondary exposure compared to women who did not, while there was no significant relationship between either type of SEB and violence among men, after adjusting for potential confounders in gender-stratified models. There was also no significant association between SEB through primary exposure and experiencing violence among women. This is noteworthy as our results lacked evidence of a gender difference in prevalence of experiencing violence in our study sample, although women reported more experiences of sexualized violence. These results align with Harris et al.’s (2021) study where women described physical and sexualized violence after non-fatal overdose as major concerns, while men expressed concern for police and street violence more generally, as well as with findings from Marshall and colleague’s (2015) analysis of gender differences in physical violence among PWUD. That said, the current findings highlight gendered harms of drug prohibition (e.g., increasing unpredictability in the unregulated drug supply), specifically that it may heighten odds of violence only for women, even after adjusting for previously identified risk factors.

The current study builds on previous studies that have emphasized how misogynistic and patriarchal norms promote violence against women who use drugs both interpersonally (e.g., within street-based environments; McNeil et al., 2021, 2014) and structurally (e.g., as a result of policies catered to men, including in supportive housing or harm reduction services; Boyd et al., 2018; Collins et al., 2020). Our findings suggest that interventions that reduce unpredictability in the drug supply—for example, regulated production/sale of currently unregulated drugs—might reduce the risk of violence against women. While these interventions have largely been called for in the context of overdose prevention (BC Death Review Panel, 2023; Ledlie et al., 2024; Office of the Provincial Health Officer, 2024; Tyndall, 2020), the current findings suggest additional benefits in terms of preventing violence. Recently published literature evaluating impacts of the Drug User Liberation Front’s community-led compassion club that provided members access to purchase a tested and dosed supply of heroin, cocaine, and crystal methamphetamine shows promising results (Bowles et al., 2025; Kalicum et al., 2025). In one study, 72% of participants reported club membership helped reduce their risk of physical violence, however, additional research is needed (Kalicum et al., 2025).

We also found that daily stimulant use with no daily opioid use modified the relationship between SEB (when defined as through using other drugs) and violence. This finding was expected as benzodiazepines are primarily found in unregulated opioids (Knill et al., 2023), and people who use opioids less frequently will have a lower tolerance to the sedative effects of benzodiazepines or may be less likely to be aware of effects and take precautions against potential violence. This result builds on prior research which has identified people who use stimulants are also impacted by the increased toxicity and contamination in unregulated drugs (Ivsins et al., 2024). For instance, people who use stimulants have cited cross-contamination (i.e., through sharing drug use equipment, or while weighing/packaging) or unintentionally using opioids as risk factors for drug poisoning and sedation (Ivsins et al., 2024). Taken together, our results and previous research suggest access to regulated drugs may reduce risk of violence by reducing risk of cross-contamination and providing occasional opioid users with agency over dosage and benzodiazepine use.

5. LIMITATIONS

An important limitation of the current study is that our study instruments do provide information on the temporal ordering of SEB and violence and thus, for this reason and others, we cannot conclude the associations are necessarily causal. A second limitation is that grouping cis and transgender participants may have obscured unique differences between groups. However, separate analyses would not have had adequate power due to low counts, and sensitivity analyses showed results did not significantly change when transgender participants were excluded. Further, gender grouping is complex, and while trans-men may experience gender discrimination that aligns more closely with experiences of women (compared to cis-men), previous research shows people who were assigned male at birth and have feminine gender presentation experience greater discrimination than people who were assigned female at birth and have masculine gender presentation (Hoskin, 2020). We also excluded non-binary and Two-Spirit participants because low counts would not allow for a separate group analysis. Given the challenges of gender grouping, future research should prioritize recruiting gender-diverse people and provide disaggregated gender data to address the underrepresentation of these populations in research. Other limitations include sampling bias and the use of self-reported data, which is subject to social desirability and recall bias. As over half of participants reported living in Downtown Eastside of Vancouver, a unique inner-city setting characterized by concentrated social services, poverty, and policing, the sample in this study may not be representative of all people who use unregulated drugs in Canada (McNeil et al., 2014). Further, experiences of violence are likely to have been underreported due to social desirability bias, particularly for sexualized violence, and also because benzodiazepines can result in memory loss and blackout periods (BC Centre on Substance Use, 2021). Underreporting of violence may have obscured a stronger relationship between SEB and violence, particularly among women. Another limitation could be that our results were impacted by an unmeasured confounder as this was an observational study and we were limited to specific variables. However, this concern may be mitigated by the comprehensive set of demographic, drug-use related, socioeconomic, health, and other factors confounders examined in this study based on previous research; and because the cohort data questionnaires that we drew from are reviewed and updated yearly, and are informed by community members, a team of frontline interviewer staff and researchers. Finally, the analyses in this study were not pre-registered on a publicly available platform and results should be considered exploratory.

6. CONCLUSION

In summary, SEB was associated with higher odds of experiencing violence for PWUD. This was particularly in cases when exposure was through using other unregulated drugs. SEB was also particularly related to higher odds of experiencing violence for women who use drugs. Further, people who use unregulated opioids more frequently may have a higher tolerance to the sedating effects of benzodiazepines or may be more likely to take precautions against potential effects, while people who only use opioids occasionally and have a lower tolerance to benzodiazepines or are exposed to benzodiazepines through contamination of stimulants, may be at higher risk of violence.

These findings suggest that the deleterious health impacts of drug prohibition extend beyond drug poisoning. Further, while men generally have higher rates of unregulated drug poisoning death (B. C. Coroners Service, n.d.), this study highlights limitations to common framing of the toxic drug crisis as primarily impacting men. Future research should explore whether increasing access to regulated drugs reduces risk of violence among women who use drugs.

Highlights:

• We surveyed 1049 people (40% women) who used unregulated drugs in Vancouver, Canada (2021-23).

• 57% reported suspected exposure to unregulated benzodiazepines

• Women who reported benzodiazepine exposure had higher odds of violent victimization

• This was only when benzodiazepine exposure was defined as exposure through using other unregulated drugs

• Unregulated benzodiazepine use alone was not related to violence