Does cannabis use predict aggressive or violent behavior in psychiatric populations? A systematic review

ABSTRACT

Background: Despite an increase in information evaluating the therapeutic and adverse effects of cannabinoids, many potentially important clinical correlates, including violence or aggression, have not been adequately investigated.

Objectives: In this systematic review, we examine the published evidence for the relationship between cannabis and aggression or violence in individuals with psychiatric disorders.

Methods: Following PRISMA guidelines, articles in English were searched on PubMed, Google Scholar, MEDLINE, and PsycINFO from database inception to January 2022. Data for aggression and violence in people with psychiatric diagnoses were identified during the searches.

Results: Of 391 papers identified within the initial search, 15 studies met inclusion criteria. Cross-sectional associations between cannabis use and aggression or violence in samples with post-traumatic stress disorder (PTSD) were found. Moreover, a longitudinal association between cannabis use and violence and aggression was observed in psychotic-spectrum disorders. However, the presence of uncontrolled confounding factors in the majority of included studies precludes any causal conclusions.

Conclusion: Although cannabis use is associated with aggression or violence in individuals with PTSD or psychotic-spectrum disorders, causal conclusions cannot be drawn due to methodological limitations observed in the current literature. Well-controlled, longitudinal studies are needed to ascertain whether cannabis plays a causal role on subsequent violence or aggression in mental health disorders.

KEYWORDS:

• Cannabis
• aggression
• violence
• mental illness
• schizophrenia
• mood disorders

Introduction

Cannabis use is highly prevalent among individuals with psychiatric disorders, in particular schizophrenia (1,2). Increasing evidence suggests adverse effects of cannabis use in psychiatric disorders, including poorer treatment outcomes, more severe symptoms, and greater cognitive impairments (3–5). In psychotic disorders, cannabis consumption has been shown to worsen positive symptoms, such as paranoia and delusions (1,6). Furthermore, cannabis use is associated with less favorable treatment outcomes (e.g., relapse), homelessness, and risk of further criminal charges (4). Simultaneously, there is some evidence implicating a relationship between substance use and violence or aggression in mental health disorders (7,8), with the strongest effects found in individuals with schizophrenia (9,10). A systematic review and meta-regression evaluating risk factors for violence in schizophrenia obtained significant associations between violence and history of polysubstance abuse (OR = 10.3), recent alcohol misuse (OR = 2.2), and recent drug misuse (OR = 2.2) (11). However, no relationship between cannabis use and violence emerged. In contrast, a recent meta-analysis observed a moderate relationship between cannabis misuse and violence in serious mental illness (12). One hypothesis underlying these conflicting findings may comprise that cannabis use is not a precipitating factor per se, but rather interacts with a multitude of risk factors associated with violence or aggression, including high levels of impulsivity, greater symptom severity, and comorbid substance use (13,14). Indeed, confounding variables were not adequately addressed in these reviews (11,12), and it remains unknown whether cannabis use predicts subsequent violence or aggression in psychiatric disorders. Accordingly, the purpose of this systematic review is to critically evaluate the literature concerning cannabis use and risk of aggression (i.e., hostile language or behavior) or violence (i.e., physically harmful behavior) in individuals with mental health disorders, beginning with an overview of potential explanations for this relationship.

Proposed biological explanations for cannabis use and aggression

Associations between cannabis use and aggression or violence have been investigated in individuals with first-episode psychosis, schizophrenia-spectrum disorders, mood, and anxiety (including posttraumatic stress) disorders. The primary psychoactive constituent in cannabis, Δ9-tetrahydrocannabinol (THC), has a high affinity for endogenous cannabinoid receptors that are densely located in limbic structures and the prefrontal cortex (15). Acute THC intoxication is commonly associated with relaxation and euphoria and has been found to reduce aggressive responses to aggression-inducing stimuli in non-psychiatric, heavy cannabis users (16). However, with higher doses, THC can produce anxiogenic responses, in addition to agitation, hallucinations, paranoia, panic attacks, memory impairments, and psychosis (17–19). Direct effects of high doses of THC on aggression and violence in humans are unknown; however, pre-clinical models show mixed results of THC on aggressive behaviors that are also affected by chronicity of cannabis use, early life experiences, and current environment (20,21).

Psychotic disorders

Although the majority of patients with schizophrenia are nonviolent, in comparison with the general population, schizophrenia patients are more likely to commit violent crimes and exhibit aggressive behaviors, such as verbal and physical threats (22–24). Violence is a complex social behavior, and to date, no consistent neurobiological theory has adequately explained the increased risk of violence or aggression in schizophrenia. Relatedly, there are no established biomarkers that predict risk of violence or aggression in schizophrenia and other mental health disorders. Nonetheless, there is evidence that pathophysiology in mesocorticolimbic structures contributes both to symptoms of schizophrenia and vulnerability to addictive behaviors, including substance use (25), which is a well-known risk factor for violence and aggression in schizophrenia (11). Similarly, advances in psychiatric genetics suggest that the co-occurrence of substance use disorders (SUDs) and schizophrenia is partially attributed to a shared genetic etiology (26–28). In the largest genome-wide association study (GWAS) for lifetime cannabis use, Pasman et al. (26) identified 14 significant genes associated with cannabis use and schizophrenia. Furthermore, the authors observed that those 14 genes are also associated with other psychiatric disorders and behaviors, including mood disorders, smoking, alcohol use, and impulsivity (29,30). To date, there is no biological mechanism linking cannabis use with aggression or violence in schizophrenia.

Mood and anxiety disorders

While the prevalence of mood and anxiety disorders is relatively high in individuals with a history of violent behaviors (31), comparisons between forensic-involved persons and the general population are limited, with one study finding no significant differences between groups (32). In this vein, biological explanations for why mood or anxiety disorders may contribute to violence are not clear, with some evidence suggesting that serotonergic hypofunction, which is present in mood disorders, may influence aggression or violence, given its role in emotion regulation and impulse control (33). However, other studies suggest that alone, low serotonin transmission is not enough to influence aggression (34). Rather, a combination of low serotonin, low cortisol, and high testosterone is more predictive of aggression, along with numerous moderators, including sex, psychopathic traits, and environment (35,36). Moreover, biological explanations for why cannabis use might induce violence in individuals with mood or anxiety disorders are limited. One study suggests that individuals with mood disorders may be at a greater risk of experiencing cannabis-induced paranoia than those without mood disorders (37), but there is no direct evidence implicating a biological mechanism in terms of a cannabis–aggression relationship. Notably, the majority of individuals with a mood or anxiety disorder who use cannabis do not exhibit violent or aggressive behaviors, suggesting any association is likely a result of interactions between multiple risk factors.

Proposed psychosocial explanations for cannabis use and aggression

Indeed, alternative explanations exist as to why there may be an indirect association between cannabis use and a greater risk of aggression, violence, or criminal behaviors in individuals with comorbid psychiatric diagnoses. These factors include peer relationships and social interactions (38,39), age of onset of cannabis and other substance use affecting development (21), family violence (40), and associations between mental health disorders and substance use with the experience of intimate-partner (41) or gang-related (42,43) victimization, which may increase the risk of future violence. Moreover, individuals who experience higher levels of aggression may use cannabis to self-medicate aggressive impulses, as low doses of cannabis typically produce an anxiolytic effect (16) and relaxation is a commonly endorsed motive for cannabis use (44). Hence, a relationship between cannabis use and aggression may exist but might not be indicative of cannabis-induced aggression per se. Accordingly, studies investigating the relationship between cannabis use and aggression that do not use temporal analysis to assess directionality and/or control for these potential confounding variables to avoid spurious correlations, are methodologically limited. Finally, several studies suggest that individuals with mood disorders experience higher rates of cannabis withdrawal symptoms, including increased hostility, anger, and irritability (45,46). While these symptoms may precede aggressive behaviors, no study to date has explored the relationship between cannabis withdrawal symptoms and aggression cross-sectionally or longitudinally in psychiatric populations.

In light of these controversies, we systematically reviewed the literature that cannabis use may predict risk of violence and aggression in patients with comorbid psychiatric disorders. We aimed to investigate whether cannabis use predicts subsequent violence and aggression in psychiatric populations and to evaluate the methodological quality of the studies identified in our search and discuss why this raises doubt concerning a causative relationship.

Methods

Search strategy and study eligibility

Following PRISMA guidelines, original, peer-reviewed articles published until January 2022 were searched for using PubMed, Google Scholar, MEDLINE, and PsycINFO. The following search terms were used: (“cannabis” OR “marijuana” OR “weed” OR “cannabinoid”) AND (“violence” OR “aggression” OR “impulsivity”) AND (“psychosis” OR “severe mental disorder” OR “cannabis use disorder” OR “posttraumatic stress disorder” OR “schizophrenia” OR “major depressive disorder” OR “depression” OR “bipolar” OR “psychiatric patient” OR “inpatient”). Titles and abstracts were screened for relevance by one of the authors (AK) and relevant articles were subjected to full-text reviews for eligibility by AK and MS. Disagreements were resolved through consensus between the two authors or through consultation with the senior author (TPG).

Studies were included if they were: (1) longitudinal, cross-sectional, experimental, or retrospective in design; (2) included a sample demonstrating a psychiatric disorder diagnosis at baseline; (3) utilized a validated or objective measure to ascertain cannabis use (e.g., urine toxicology screens, previous drug reports, scores from The Cannabis Use Disorders Identification Test (CUDIT)); and (4) utilized a validated or objective measure to ascertain violence and/or aggression (e.g., Overt Aggression Scale [OAS], criminal records, hospital incident reports). Exclusion criteria were: (1) reviews, meta-analyses, and case studies; (2) studies solely assessing victimization; and (3) studies assessing violent acts committed prior to cannabis use.

Risk of bias

The quality of longitudinal and cross-sectional studies was assessed using the Newcastle Ottawa Scale (NOS). The rating system has a 9-point grading scale that measures the following domains: (1) the selection process; (2) comparability of patients based on design; (3) outcome assessment. Two authors (AK and MS) conducted independent grading of all articles that were processed at the full-text review stage (AK and MS) to determine the methodological quality of potential studies. To ensure that only methodologically sound studies were synthesized into this review, we included only publications that received a score of 5 or higher on the NOS (47).

Results

A total of 391 articles were identified from the initial search, and a subsequent 335 papers were excluded based on initial inspection of the titles and abstracts. After title and abstract screening, 56 papers were read in full and assessed for eligibility. A further 49 papers were excluded, leaving 15 studies included for qualitative synthesis (see Figure 1).

Figure 1. PRISM-A diagram.

Study characteristics

Characteristics of the 15 included studies for our systematic review are presented in Table 1. Although randomized controlled trials (RCTs) were considered throughout our search, none of the included studies employed this design. Two of the studies were a retrospective chart review, five of the publications employed prospective cohort designs, two utilized a retrospective longitudinal design, and six were cross-sectional. These included a total of 11,318 participants, with sample sizes ranging from 69 to 3,028. Diagnostic groups included schizophrenia, first-episode psychosis, schizoaffective disorder, posttraumatic stress disorder (PTSD), and bipolar disorder, and mixed inpatient psychiatric samples.

Table 1. Review of evidence for cannabis on aggression or violence in psychiatric populations.

Citation (N = 15)	Sample Size (N = 11,318)	Substances Evaluated	Outcome Measure	Study Design	Key Results	NOS SCORE
Psychotic-Spectrum Disorders & First-Episode Psychosis (N = 10 studies; N = 4,083)
Arsenault et al., 2000	N = 961; NZ young adults (21 yrs) followed from a birth cohort	Cannabis or alcohol	Court conviction records	Retrospective, longitudinal	Although cannabis dependence was associated with violence in patients with schizophrenia, the association become non-significant after controlling for previous substance use, excessive threat perception at age 18, and presence or history of a conduct disorder.	8
Beaudoin et al., 2020	N = 965; US adult (18–55) outpatients with schizophrenia	Cannabis, alcohol, stimulants	MacArthur Abbreviated Community Violence Instrument, General Clinical Status Interview	Prospective, cohort	After controlling for relevant covariates, including alcohol and stimulant use, cannabis use in patients with schizophrenia predicted future violent behaviors. However, past violent acts did not predict subsequent cannabis use.	6
Dharmawardene & Menkes, 2017	N = 141; Inpatient NZ adults (18–65) with schizophrenia or psychotic-spectrum disorders	Cannabis and alcohol	Clinical records, DSM-IV criteria, CUDIT-R, AUDIT	Retrospective, chart review	Severity of cannabis dependence strongly correlated with lifetime violence. Both cannabis and alcohol dependence predicted engagement in self-harm.	7
Hachtel et al., 2020	N = 69; Australian male inpatients with at least one episode of psychosis	Alcohol, cannabis, cocaine, amphetamines	ICD-10, ASSIST, BPRS, Scale for Assessment of Negative Symptoms, clinical records, Lifetime History of Aggression Scale, Lancashire Quality of Life Profile	Cross-sectional, retrospective	Individuals who reported a history of violent offenses were significantly more likely to meet criteria for cannabis, cocaine, or amphetamine use disorder, report depressive symptoms, and report a history of victimization. The strongest predictor of violent offenses comprised of scores on the BPRS.	7
Kalk et al., 2020	N = 1089; UK individuals in emergency psychiatric care with acute psychosis and non-psychiatric controls	Alcohol, cannabis, stimulants	Objective measures of substance use (drug screens, breathalyzer, and urine tests), police and clinical records	Retrospective, longitudinal	Psychotic symptoms increased the odds of a violent event. Combined, recent cannabis use with acute psychotic symptoms increased the odds of a physically violent event five-fold. While stimulant use alone did not increase the likelihood of a violent event, stimulant and cannabis use in tandem significantly increased the likelihood of a violent event in individuals with psychotic symptoms. Recent alcohol use was not associated with subsequent physical violence in patients with psychotic symptoms.	7
Lejoyeux et al., 2013	N = 100; French, adult (18–65) inpatients with schizophrenia	Alcohol, nicotine, cannabis	OAS, Cannabis Abuse Screening Test	Cross-sectional, retrospective	Cannabis dependence did not increase the odds of engaging in an aggressive behavior. Nicotine dependence identifying as male sex, and history of sexual violence significantly increased the odds of engaging in an aggressive behavior.	6
Maremmani et al., 2004	N = 111; Italian male inpatients with at least one psychotic episode	Cannabis and cannabis-induced psychosis	BPRS, OAS, Global Assessment of Functioning Scale	Cross-sectional, retrospective	In comparison to nonusers with psychosis, both current and past cannabis users with a history of psychosis demonstrated significantly elevated levels of aggression and violence.	6
Moulin et al., 2018	N = 265; Swiss inpatients (18–35) with early psychosis	Cannabis, alcohol, and other substances	DSM-IV criteria for psychopathology and substance use, Staff OAS, Case Manager Rating Scale	Prospective, cohort	Cannabis use, but not alcohol use significantly increased the odds of engaging in a violent behavior. Cannabis use was also the strongest risk factor for subsequent violent behavior. The effect remained after adjusting for relevant baseline covariates such as comorbid psychiatric diagnoses and demographic variables.	7
Moulin et al., 2020	N = 250; Swiss inpatients (18–35) with early psychosis	Cannabis, alcohol, and other substances	DSM-IV criteria for psychopathology and other substance use, case records, Staff OAS, Case Manager Rating Scale	Prospective, cohort	Violent behaviors were predicted by cannabis use in adolescence. After adjusting for relative confounds such as demographic information, and comorbid psychiatric disorders, cannabis use beginning at age 15 or younger significantly predicted subsequent violent behaviors. Other substance use did not predict violent behaviors.	7
Park et al., 2019	N = 132; patients from numerous East, South, and Southeast countries with schizophrenia	Cannabis	BPRS, Drug-induced Extrapyramidal Symptoms Scale, Charlson Comorbidity Index	Cross-sectional, retrospective	In comparison to patients with schizophrenia who do not use cannabis, patients who used cannabis and had a concomitant diagnosis of schizophrenia reported greater levels of aggressive behaviors and speech	6
Mixed Psychiatric Populations (N = 3 studies, N = 1,931 participants)
Bassir Nia et al., 2016	N = 594; American psychiatric inpatients (15.5% schizophrenia; 16% schizoaffective; 7.9% bipolar; 1.7% major depressive disorder; 13% unspecified psychotic; 42.4% unspecified depressive; 3.5% other)	Synthetic cannabinoids and cannabis	SCID, Patient medical records	Retrospective, chart review	In comparison to non-users, psychotic patients who used synthetic cannabinoids were more likely to demonstrate aggressive behaviors and agitation over their course of hospitalization. Psychotic patients who used cannabis only were more likely to demonstrate agitation over their course of hospitalization in comparison to non-users.	7
Dugre et al., 2017	N = 1136; American psychiatric inpatients (52.6% affective disorder; 21.6% schizophrenia spectrum; 23.9% substance use disorder; 1.8% personality disorder)	Cannabis and alcohol	SCID, Psychopathy Checklist Screening, Barratt Impulsiveness Scale, MacArthur Community Violence Instrument	Prospective, cohort	After adjusting for relevant covariates including other substance use, continuous use of cannabis significantly predicted violence in future time points. Violence did not predict subsequent cannabis use.	7
Johnson et al., 2016	N = 201; American inpatients (47.3& bipolar disorder, 29.4% schizophrenia or psychosis; 23.4% schizoaffective disorder, and CUD	Cannabis	Excited Component of the Positive and Negative Syndrome Scale, Inpatient medical records, Urinalysis	Cross-sectional	Positive urine screens for cannabis was associated with agitation at the time of admission, but was not associated with a history of violence or aggression.	7
Post-Traumatic Stress Disorder (N = 2 studies, N = 5,304 participants)
Dillon et al., 2021	N = 3028; US veterans with PTSD	Cannabis, alcohol, and other substances	DSM-IV criteria, Symptom Checklist-90-R, Beck Depression Inventory-II, Davidson Trauma Scale, Combat Exposure Scale	Cross-sectional	After adjusting for demographic and psychiatric covariates, in addition to other substance use, previous violence, current CUD was associated with difficulty controlling violence, increased odds of reporting aggressive impulses and urges, and difficulty controlling anger in the past 30 days	7
Wilkinson et al., 2015	N = 2276; US veterans with PTSD	Cannabis, alcohol	DSM-III or DSM-IV criteria, Mississippi PTSD scale, Clinical records, Addiction Severity Index, National Vietnam Veterans Readjustment Survey	Prospective, cohort	Veterans who began cannabis use at treatment admission showed significantly higher measures of violent behavior at follow-up in comparison to non-cannabis users, previous cannabis users, and continuing cannabis users. Findings remained significant after controlling for alcohol and substance use.	6

Psychotic-spectrum disorders

Five publications investigated the temporal associations between cannabis and violent behaviors in people with a psychotic spectrum disorder (48–52), with four studies concluding that cannabis use is a significant predictor for subsequent violence (49–52). Using the MacArthur Abbreviated Community Violence Instrument and an interview with family/caregivers, a recent longitudinal study found cannabis as a significant risk factor for subsequent violence in schizophrenia (49). This relationship held after controlling for baseline and demographic variables, in addition to alcohol and other substance use (49). However, other substance use, and violence were assessed as dichotomous variables (yes/no) and due to the low base rate of violence (8%), severity of violence was also not considered. Moreover, the temporal relationship between cannabis use and violent behaviors was not taken into account. In two studies assessing patients entering an early psychosis program, one demonstrated that cannabis use before the age of 15 predicted subsequent violent behaviors (51), while the other identified cannabis use, in tandem with either reduced treatment adherence, lack of insight, and/or elevated impulsivity, as predictors of violent behaviors (50). Notably, while both studies controlled for baseline and demographic variables, other substance use and previous violence/aggression were not included as covariates. Kalk et al. (52) examined whether recent cannabis or stimulant use was associated with higher risk of subsequent violent behaviors in individuals experiencing a psychotic episode and non-psychotic controls. Overall, the presence of a psychotic episode increased the odds ratio of a violent behavior by 4.0 (95% CI 2.2–7.4), and the presence of a psychotic episode in addition to cannabis use had increased the odds of a subsequent violent episode to 7.1 (95% CI 3.7–13.6). Stimulant use alone did not significantly increase the odds of subsequent violence in patients with a psychotic episode (OR = 1.7; CI 0.5–5.9). However, a history of aggressive and violent behaviors was not controlled for, and substance use and violent episodes were assessed as binary variables (yes/no) that included damage to property. In contrast, a prospective study following a birth cohort of 961 individuals did not find a significant relationship between comorbid cannabis use disorder and presence of a violent offense in patients with schizophrenia at age 21, after controlling for substance use at earlier time points, presence of a conduct disorder during adolescence, and excessive threat perception at age 18 (48).

Four cross-sectional studies concluded that cannabis use is associated with violent behaviors in people with a psychotic disorder. None of these studies included a history of aggression/violence or other substance or alcohol use as potential covariates (53–56). Two of the studies found a significant relationship between cannabis use and a history of violence in inpatients with psychotic-spectrum disorders (54,56). Notably, Hatchel et al. (56) found that alcohol use yielded a stronger correlation with a history of violent offenses, in addition to other psychiatric variables, including psychopathic personality traits and a lifetime history of aggression. Surprisingly, these variables were not controlled for in the analyses with cannabis use, limiting conclusions. Park et al. (55) compared scores on the Brief Psychiatric Rating Scale in Asian patients with schizophrenia who reported lifetime cannabis use, with those of their non-cannabis using counterparts. In comparison to non-users, patients with lifetime cannabis use indicated greater levels of aggressive behaviors and speech. Similarly, Maremmani and colleagues (53) reported that relative to non-users, male inpatients with psychotic symptoms who reported either past or current cannabis use showed elevated levels of aggression and violence against others as measured by the OAS. Importantly, no potential confounds were controlled for in this study. In contrast, a separate study that also utilized the OAS did not find that cannabis significantly predicted aggressive behaviors among a sample of N = 100 inpatients with schizophrenia (57). Rather, the authors found that nicotine dependence, male sex, and a history of sexual violence as positive correlates with aggressive behaviors (57).

Mixed inpatient psychiatric populations

Two cross-sectional designs explored the relationship between cannabis use and violence and aggression in mixed inpatient populations (58,59). Bassir Nia et al. (58) compared the prevalence of agitation and aggression over the course of hospitalization in four groups of psychotic patients: individuals who use synthetic cannabinoids only, individuals who use cannabis only, dual synthetic cannabinoids and cannabis users, and non-users. In comparison to non-users, patients reporting cannabis use demonstrated greater agitation but not aggression throughout their inpatient stay. Synthetic cannabinoid users demonstrated higher agitation and aggression than nonusers and cannabis users. However, there was no significant difference between dual users and nonusers in levels of aggression and agitation. Similarly, a separate group of investigators observed a significant relationship between recent/current cannabis use measured through drug urine screens at the time of hospitalization and greater agitation during inpatient stay among a mixed psychiatric population (59), which may have been indicative of acute withdrawal (60). However, cannabis use was not associated with aggressive behaviors or a history of violence in these patients.

One longitudinal study following 1,136 psychiatric patients over five time periods found that persistence of cannabis use was associated with a greater likelihood of reporting a violent act at each subsequent timepoint (61). At the final time-point, in comparison to non-users, the odds ratio of reporting an act of violence increased significantly to 2.44 (CI 1.06–5.63) if cannabis was reported at the prior four time-points, even after controlling for baseline variables and alcohol or other substance use. However, reports of violent acts at the previous time points were not included in the analyses.

Post-traumatic stress disorder (PTSD)

One recent cross-sectional study found a significant, positive association between current cannabis use disorder (but not lifetime cannabis use) with difficulty controlling anger, controlling aggressive impulses and urges, and problems controlling violent behaviors in the previous 30 days among a sample of 3,028 American veterans with PTSD (44). The relationship held after accounting for demographic variables, combat exposure, history of incarceration, alcohol and other substance use, and comorbid psychopathology.

Wilkinson et al. (62) longitudinally evaluated the impacts of cannabis use upon symptom severity and subsequent violent behaviors in 2,276 American veterans entering PTSD treatment. Participants were categorized into four classes according to cannabis use: non-users (i.e., never used cannabis), recent users (used cannabis until admission of treatment program), continuing users (used cannabis prior and during treatment), and new users (started cannabis upon treatment admission). At follow-up, in comparison to the other groups, new users demonstrated higher PTSD symptom severity and were significantly more likely to report violent behaviors, even after controlling for baseline and demographic characteristics, in addition to drug and alcohol use (62). However, no predictive relationships between cannabis and aggression were found.

Discussion

This systematic review examined whether cannabis use predicts subsequent violence or aggressive behaviors in psychiatric populations. The findings of our review suggest a cross-sectional relationship between problematic cannabis use and violence in patients with a psychotic-spectrum disorder or post-traumatic stress disorder (PTSD). Moreover, we observed a longitudinal association between cannabis use and subsequent aggression or violence in patients with psychosis. However, it is crucial to note that confounding variables were inadequately controlled for among these studies, where three of the four prospective studies observing a significant temporal association between cannabis use and subsequent violence or aggression, did not adjust for other substance use, previous history of violence, symptom severity, or conduct problems (50–52). Thus, the influence of potential confounding variables significantly limits the ability to draw causal conclusions between cannabis use and subsequent violence or aggression in psychotic-spectrum disorders.

With respect to the positive relationship between cannabis use and aggression in people with psychotic disorders, it is noted that cannabis use is associated with greater psychotic relapse rates, a higher rate of hospitalizations, and more frequent or distressing positive symptoms (63). Poorer illness prognosis and symptom exacerbation may cause psychological distress along with difficulties in interpersonal relationships and emotion dysregulation, contributing to risk of aggressive behaviors. Furthermore, lack of insight and lower treatment adherence associated with cannabis use may moderate risk of violent behaviors in individuals with psychotic-spectrum disorders (50). Additionally, acute cannabis withdrawal may also influence risk of aggressive or violent behaviors. Boggs et al. found that 28% of individuals with schizophrenia and concurrent cannabis use disorder reported feeling angry and 18% reported feeling aggressive during a “serious” quit attempt for cannabis (64). Moreover, studies of extended cannabis abstinence for 28 days in people with schizophrenia found reduced depression and anxiety (65) and improved verbal learning and memory (66).

With respect to associations among PTSD, cannabis use, and aggression, combat-exposure induced PTSD is related to a greater risk of future violence (67), including domestic violence in the context of alcohol use (68). As above, cannabis use disorder is associated with less success in PTSD treatment and greater PTSD symptom severity, particularly in avoidance-numbing and hyperarousal (69). Furthermore, symptoms of cannabis withdrawal are more pronounced in people with PTSD relative to those without PTSD (70). However, other studies have found that many veterans report using cannabis to self-medicate PTSD symptoms, including irritability (71), which may point to a reverse relationship where greater irritability predicts a tendency to seek relief through cannabis use.

Conversely, several contextual factors may also explain the observed positive associations. For example, until 2018, possession of cannabis was illegal in Canada and remains so in many American states. Individuals who use cannabis and other illicit substances may be more likely to engage in antisocial behaviors when interacting with criminal networks or antisocial peers when purchasing or using drugs (72). Such proximity to criminal networks is not mirrored in the use of illicit substances (e.g., alcohol and nicotine) that can be easily obtained. Other indirect positive relationships between cannabis and aggression or violence are also plausible, as the majority of studies included (11/15) did not control for other substance use, antisocial personality traits, or childhood/adolescent conduct problems (Table 2), and only two studies controlling for relevant confounders obtained a significant relationship between cannabis and violence or aggression (49,61). Co-occurring alcohol or stimulant use, for example, have been shown to directly elicit overt aggression (13), while early antisocial behaviors are the most significant predictor of future antisocial behaviors, thus confounding the observed findings.

Table 2. Summary of key findings and whether relevant covariates are controlled for.

Study (Author, Year)	Was Cannabis Use Associated with Aggressive Behaviors?	Was Cannabis Predictive of Subsequent Aggressive Behaviors?	Was Other Substance Use Controlled for?	Were Antisocial Traits or Conduct Behaviors Controlled for?
Psychotic-Spectrum Disorders & First-Episode Psychosis
Arsenault et al., 2000	-	-	+	+
Beaudoin et al., 2020	+	+	+	+
Dharmawardene & Menkes, 2017	+	-	-	-
Hachtel et al., 2020	+	-	-	-
Kalk et al., 2020	+	+	+	-
Lejoyeux et al., 2013	-	-	-	+
Maremmani et al., 2004	+	-	-	-
Moulin et al., 2018	+	+	-	-
Moulin et al., 2020	+	+	-	-
Park et al., 2019	+	-	-	-
Mixed Psychiatric Populations
Bassir Nia et al., 2016	+	-	+	-
Dugre et al., 2017	+	+	+	+
Johnson et al., 2016	-	-	-	-
Post-Traumatic Stress Disorder
Dillon et al., 2021	+	-	+	+
Wilkinson et al., 2015	+	-	+	-

Alternatively, there may exist a reverse relationship between cannabis use and violence in psychiatric populations, where individuals use cannabis to control aggressive impulses and urges. In one study investigating motivations for cannabis use in comorbid major depressive disorder and cannabis use disorder, participants who reported problems controlling violent behaviors were more likely to use cannabis to decrease aggression and increase relaxation in comparison to individuals without problems controlling violent behaviors (37). Similar findings have been obtained in PTSD, where cannabis use is commonly reported to manage anger and aggressive impulses (44,73). Few longitudinal investigations have been conducted to clarify the directionality of this relationship, and further studies utilizing temporal analysis of the relationship between cannabis and aggression or violence are needed to clarify the effects of cannabis use.

Study limitations

While we found a cross-sectional relationship between cannabis and aggression in individuals with PTSD and longitudinal relationships between cannabis and aggression in psychotic-spectrum disorders, the results need to be interpreted with caution due to certain methodological limitations of the included studies. Of ten publications assessing patients experiencing psychosis, only two had controlled for other substance use and previous violence (48,49). Moreover, only two studies assessing cannabis use and PTSD were included (44,62), and while both reported a bivariate relationship, only one (44) controlled for other substance use and previous violence. Previous research involving the general population has demonstrated that relationships between cannabis use and violence significantly weaken after accounting for alcohol use, poly-substance use, previous antisocial behaviors (the most significant predictor of future violence (74)), or antisocial personality traits (75,76). It should also be noted that both PTSD studies assessed combat-exposure PTSD, which may not generalize to PTSD from other forms of trauma.

Furthermore, over half of the included studies were cross-sectional, which significantly limits our understanding concerning the directionality between these variables. Moreover, there was substantial heterogeneity across the studies, such that metrics of cannabis use and the operationalization of violence or aggression greatly differed across studies. While some of the included studies employed self-report measures such as the MacArthur Community Violence Instrument to measure violence, other studies utilized observer rater scales or relied upon court conviction records. Notably, one (57) of the two studies (53,57) utilizing the OAS did not distinguish verbal aggression and self-aggression scales from physical aggression against others/objects. There was also significant heterogeneity across the studies in terms of methodology, duration of follow-up, confounders controlled for, and outcome measures assessed. Many of the studies had inconsistent reporting measures insofar as age of onset of cannabis use, lifetime versus recent substance use, methods of assessing cannabis use (e.g., self-report versus urine toxicology) and the potency and frequency of cannabis used. Relatedly, significant heterogeneity was observed among the reporting of aggression and violent outcomes for the included studies. Subsequently, we were unable to determine whether cannabis use is associated with specific forms of violence, in addition to severity of violence. Additionally, it should be noted that any relationship between cannabis use and violence or aggression outcomes in people with psychotic-spectrum disorders or PTSD remains associational, and causality cannot be assumed; we are unaware of any human experimental studies that have demonstrated violence and aggression as an acute effect of cannabis (THC) in a controlled setting.

Finally, we found no studies comparing individuals with a psychiatric diagnosis to individuals without a psychiatric diagnosis, despite similar correlational evidence in the general population (77). Thus, conclusions are limited on whether a cannabis and aggression relationship exists, and if it exists, whether the effects are specific to individuals with mental health disorders. Approaching this research question must be done with caution to avoid further stigmatizing psychiatric disorders. For example, official crime sources may be biased as individuals with SUDs and psychiatric diagnoses may be at a greater risk of identification by law enforcement (78,79) than those without such diagnoses or those who do not use substances.

Clinical implications

It is important to note that among the included studies in this systematic review, rates of violence were relatively low in psychiatric populations (50,52,59) irrespective of cannabis use. The results of this review underscore the need for accessible treatment for problematic cannabis use among individuals with mental health disorders, with or without histories of aggression or violence. Moreover, clinicians should take the time to warn individuals who are at greater risk for experiencing paranoia or hallucinations from cannabis use about potential consequences, so that they can make informed decisions about their use, including THC dose, potency, and frequency choices, and encourage use of cannabis strains with cannabidiol (CBD).

Conclusions and future directions

Cannabis use rates are increasing, coinciding with cannabis legalization and decreased perceptions of risk (2). Our findings suggest there may be an association between cannabis use and violent or aggressive behaviors in people with psychotic-spectrum disorders and PTSD. However, methodological limitations, including the use of retrospective or cross-sectional data and heterogeneity across controlled confounders, preclude causal conclusions between cannabis use and subsequent aggression or violence. Nonetheless, it is important to note that there have been reports of numerous unfavorable outcomes related to cannabis use in mental illness, including worsened symptomology, poorer treatment adherence, and lower life satisfaction (1,17).

To ascertain whether the relationship between cannabis and aggression or violence is direct or indirect, research on neurobiological underpinnings is required, in tandem with rigorous, well-controlled, longitudinal study designs. Randomized experimental studies in rodents, non-human primates, or human participants could further illuminate causal mechanisms. Moreover, studies should incorporate measures assessing relevant cannabis-related variables, including THC potency, frequency and amount of use, and age of onset (19).

While research is preliminary, our conclusions have major implications for healthcare and public policy. Robust prevention and treatment interventions for cannabis use among psychiatric populations are warranted (2).

Disclosure statement

Dr George has received consulting fees from the Canadian Center for Substance Use and Addiction (Chair, Scientific Advisory Committee), the American College of Neuropsychopharmacology (ACNP; Deputy Editor, Neuropsychopharmacology), Frutarom, Aelis, and Sanford Burnham Prebys Neurosciences.

Additional information

Funding

The preparation of this article was supported in part by NIDA grant R21-DA-043949 and the Astrid H. Flaska Funds from the CAMH Foundation (to Dr George), an award from the Soroptimist Society of Canada, a graduate student award from the TC3 Cannabis and Cannabinoid Consortium of the Temerty Faculty of Medicine, University of Toronto, and an Ontario Graduate Scholarship (to Ms. Sorkhou); Centre for Addiction and Mental Health [Astrid H. Flaska Funds]; National Institute on Drug Abuse [R21-DA-043949]