A causal relationship between controlled substances and firearm violence has been widely assumed in the United States, and federal law prohibits individuals who are “unlawful users of or addicted to any controlled substance” from purchasing or possessing firearms (68 FR 3750. 2003. Codified at 27 CFR §478.11). However, the law does a poor job of defining “unlawful users,” resulting in recent calls for a revised, actionable definition. Such a definition should be informed by research evidence, but to date the epidemiologic research on the relationship between controlled substances and violence has not been comprehensively reviewed. The initial goal of this review was to summarize the best available evidence on the relationship between controlled substances and firearm violence, but only 1 study specific to firearm violence was identified. We therefore reviewed studies of this relationship using broader measures of interpersonal violence and suicide, all of which included but were not limited to firearm violence, and measures of illicit firearm carrying. Prospective longitudinal studies (n = 22) from 1990 to 2014 were identified by using searches of online databases and citation tracking. Information was extracted from each study by using a standardized protocol. Quality of evidence was independently assessed by 2 reviewers. Aggregate measures of controlled substance use were associated with increased interpersonal violence and suicide, but evidence regarding the relationship between specific substances and violence was mixed. Involvement in illegal drug sales was consistently associated with interpersonal violence. To effectively revise extant federal law and delineate appropriate prohibiting criteria, more research is needed to understand the relationship between controlled substances and firearm violence.

INTRODUCTION

Since the late 1990s, the United States has directed billions of dollars toward efforts to prevent trafficking, use, and abuse of controlled substances, such as cocaine, heroin, methamphetamine, and marijuana. A key motivator of the “war on drugs” is concern that controlled substances are associated with heightened risk of violence, particularly firearm violence (1, 2). In the United States, the crack cocaine epidemic of the 1980s and 1990s was widely perceived as an important driver of elevated rates of firearm violence during that period (3) and, over the past 4 decades, drug-related violence perpetrated by gangs and cartels has been widely reported (46). Although public attention has focused predominantly on interpersonal violence, the majority of firearm deaths (60% in 2013) in the United States are suicides (7), and prior research suggests that controlled substance use may elevate risk of suicide as well as homicide (8, 9).

The perceived connection between controlled substances and firearm violence is evident in federal law, which prohibits individuals who are “unlawful users of or addicted to any controlled substance” from purchasing or possessing firearms (10). However, the law does not clearly define “unlawful users,” and in recent years stakeholders have called for revision and clarification of this policy (1113). The changing landscape of drug control policy in the United States, where recreational marijuana use is now legal in multiple states but still prohibited under federal law, suggests that a nuanced approach to revisiting this prohibition is needed. The ongoing policy debate surrounding this provision should be informed by the research evidence, but to date no comprehensive review of the epidemiologic relationship between controlled substances and firearm violence (and the implications of that relationship for policy) has been conducted.

The initial goal of this review was to summarize the best available evidence on the relationship between controlled substance and firearm violence. However, we identified only 1 study meeting our inclusion criteria that was specific to violence committed with firearms. We therefore reviewed studies of this relationship using broader measures of interpersonal violence and suicide, all of which included but were not limited to firearm violence. We also included studies measuring the relationship between controlled substances and illicit firearm carrying, a precursor to interpersonal violence (14).

The lack of firearm-specific studies is a key limitation of our review. Nonetheless, understanding the best available evidence on the relationship between controlled substances and measures of violence that include but are not limited to firearm violence can provide important insights into the role of illicit drugs in firearm violence and the implications for policy. The majority of risk factors for firearm and nonfirearm violence overlap (15, 16). There is therefore no strong theoretical reason to hypothesize that individuals' involvement with controlled substances would be associated with elevated risk for nonfirearm violence but not firearm violence, or vice versa, unless demand for and access to firearms were very low or nonexistent among controlled substance users. Prior literature suggests that this is not the case: Rather, studies show that rates of firearm possession and carrying among US drug users are similar to or higher than rates among non-drug users, depending upon the specific study population (4, 1721). Importantly, some measures of controlled substance involvement—for example, drug dealing or crack cocaine use—may be associated with higher demand for and access to firearms, and therefore higher risk of firearm violence, than others (e.g., recreational marijuana use) (4, 17, 18, 20, 22), and these relationships may differ for interpersonal violence versus suicide.

Prior research has shown that nonfirearm violent acts, such as assault, are predictors of future interpersonal firearm violence (23, 24). Thus, studies assessing the relationship between controlled substances and serious violence generally, while not firearm specific, are measuring an outcome known to be directly related to firearm violence. Lack of firearm-specific studies is not unique to the controlled substances and violence literature. The best available epidemiologic research on the relationship between mental illness and interpersonal violence also uses measures of violence that include but are not limited to firearm incidents (25). Nonetheless, this research has informed the development and implementation of recent evidence-based firearm policies (11, 12, 25, 26). We could usefully take a similar approach in the case of drugs and firearm violence, where the active policy discussion is largely uninformed by research evidence.

We begin by briefly summarizing rates of controlled substance use and firearm violence in the United States and discussing the pathways by which controlled substances might influence violence. We then describe the methods and results of our comprehensive review and conclude with discussion of research gaps and implications for firearm policy in the United States.

Prevalence of controlled substance use

Compared with the general population, both homicide offenders and suicide victims are more likely to use controlled substances (Figure 1). In the overall US population in 2013, an estimated 9% of Americans aged 12 years or older used any controlled substances in the past month (27). The most common illicit substance, marijuana, was used by 7.5% of Americans, followed by nonmedical use of prescription drugs such as opioid analgesics, tranquilizers, and sedatives (2.5%); cocaine (0.6%); hallucinogens such as lysergic acid diethylamide (LSD), phencyclidine, and ecstasy (0.5%); methamphetamine (0.2%); and heroin (0.1%) (27). By contrast, in a 2004 survey of US state and federal prisoners, 24% of federal inmates and 28% of state inmates incarcerated for violent crimes reported being under the influence of 1 or more controlled substances at the time of the crime, and about 50% of both groups reported using controlled substances in the month prior to committing the violent offense that led to their incarceration (28). In a 2010 toxicology study of suicide victims in 16 US states, 17% of suicide victims tested positive for amphetamines, cocaine, marijuana, and/or opiates. Within specific categories, 3% tested positive for amphetamines, 5% for cocaine, 10% for marijuana, and 20% for opiates (29).

Figure 1.

Controlled substance use showing overall population, perpetrators of violent crime, and suicide victims, United States, 2004–2014. Figure 1 illustrates percent of the population using controlled substances in the overall US population and among state prisoners convicted of violent crimes, federal prisoners convicted of violent crimes, and suicide victims. Controlled substance use in the overall US population was measured as self-reported past-month use; controlled substance use in state and federal prisoners convicted of violent crime was measured as self-reported use at the time of the crime; and controlled substance use in suicide victims was measured by using toxicity screens (2729). 

Figure 1.

Controlled substance use showing overall population, perpetrators of violent crime, and suicide victims, United States, 2004–2014. Figure 1 illustrates percent of the population using controlled substances in the overall US population and among state prisoners convicted of violent crimes, federal prisoners convicted of violent crimes, and suicide victims. Controlled substance use in the overall US population was measured as self-reported past-month use; controlled substance use in state and federal prisoners convicted of violent crime was measured as self-reported use at the time of the crime; and controlled substance use in suicide victims was measured by using toxicity screens (2729). 

Firearm violence in the United States

In 2013, there were 33,636 firearm deaths in the United States. Of these, 11,208 were homicides (33%) and 21,175 were suicides. The remaining 1,253 deaths were unintentional shootings, related to legal intervention, or of undetermined intent (7). Another 80,000 or more individuals are nonfatally wounded with firearms each year (30). High rates of firearm ownership in the United States are associated with rates of firearm morbidity and mortality that are considerably higher than in other high-income nations (31).

Relationships between controlled substances and interpersonal violence

In 1989, Goldstein et al. (32) first published their tripartite conceptual framework (Table 1), which theorized 3 pathways by which controlled substances lead to interpersonal violence: the psychopharmacological, economic compulsive, and systemic pathways. On the psychopharmacological pathway, the physical and psychological effects of controlled substances, such as agitation, aggression, and cognitive impairment, heighten risk for violent behavior and impair the decision-making and communication skills necessary to avoid violence. On the economic compulsive pathway, controlled substances are related to violence when addicts turn to armed robbery or other violent crimes in order to finance their drug use. On the systemic pathway, disputes within illegal drug markets (e.g., conflicts over turf) lead to violence.

Table 1.

Goldstein et al.'s Tripartite Framework for Pathways by Which Controlled Substances May Influence Perpetration of Violence Toward Others (32)

Pathways by Which Controlled Substances May Influence Violence Definition 
Pathway 1 (psychopharmacological) Physical and psychological effects of controlled substances on violence 
Pathway 2 (economic compulsive) Violence as the means for financing illicit drug use (e.g., assault in the course of robbery) 
Pathway 3 (systemic) Violence arising from disputes within illegal drug markets 
Pathways by Which Controlled Substances May Influence Violence Definition 
Pathway 1 (psychopharmacological) Physical and psychological effects of controlled substances on violence 
Pathway 2 (economic compulsive) Violence as the means for financing illicit drug use (e.g., assault in the course of robbery) 
Pathway 3 (systemic) Violence arising from disputes within illegal drug markets 

To begin to test his framework, Goldstein et al. (32) examined the causes of drug-related homicides committed in New York City in 1988 in the midst of that city's crack cocaine epidemic. They worked with law enforcement to identify drug-related homicides (68% of which involved the use of firearms) and, based on detailed information collected about each case, to classify the homicide as most likely related to the psychopharmacological, economic compulsive, or systemic pathway to violence. A total of 218 drug-related homicides were identified (32). Of these, 14% were classified as psychopharmacological, 4% as economic compulsive, 75% as systemic, and 7% as multidimensional (32). The most frequently involved drug in all cases was crack cocaine, followed by powder cocaine, marijuana, and heroin (32). Alcohol was also involved in a high proportion of the homicides determined to fall along the psychopharmacological pathway (32). Our review of studies of the relationship between controlled substances and interpersonal violence is guided by the paradigm of Goldstein et al. In recent years, this framework has been critiqued and expanded upon but nonetheless remains the predominant paradigm guiding research on the causal relationship between illicit drugs and interpersonal violence (33). Although other paradigms such as opportunity theory and strain theory have provided important insights into the broader causes of interpersonal violence (3436), to our knowledge the tripartite framework of Goldstein et al. is the only theory specific to controlled substances. Since the seminal work by Goldstein et al. in the late 1980s, multiple studies have attempted to examine the relationship between controlled substances and violence. Many of these studies are retrospective and/or cross-sectional (3745). Although such studies can yield useful descriptive information, they are open to multiple threats to validity that preclude assessment of a causal association between controlled substances and violence perpetration. For example, retrospective and cross-sectional studies make it difficult or impossible to establish temporality of involvement with controlled substances and violence, which is particularly problematic in the context of prior studies suggesting that antisocial and violent behavior can be risk factors for future drug use (46). To truly measure the association between controlled substances and violence toward others, it is critical to establish that individuals' involvement with illicit drugs occurred prior to violence perpetration. It is also important to measure and establish the temporality of other factors associated with controlled substance use and violence, including but not limited to low socioeconomic status, delinquent peer groups, alcohol use and abuse, and history of violent or aggressive behavior. Although many cross-sectional studies measure the presence of these potentially confounding or mediating factors, they cannot establish temporality. For these reasons, in our review we include only prospective longitudinal studies of the relationship between controlled substances and interpersonal violence.

Relationship between controlled substances and suicide

Unlike the relationship between controlled substances and interpersonal violence, which has received considerable attention in the past 4 decades from researchers, policymakers, and the public, the relationship between illicit drugs and suicide has received relatively little attention. As with perpetration of violence toward others, some evidence suggests that the physical and psychological effects of using some controlled substances, particularly increased impulsivity, may heighten risk of suicide (47, 48). To our knowledge, however, no theoretical frameworks comparable to the paradigm of Goldstein et al. have been developed to explain the relationship between controlled substances and suicidal behavior. Models of suicide risk often conceptualize 2 separate, although sometimes interrelated, causal pathways: the distal risk pathway and the proximal risk pathway (47). Proximal risk factors are those present in the hours and minutes leading up to suicidal behavior. For example, acute intoxication and access to a firearm during the short period when an individual is considering suicide are proximal risk factors. In contrast, distal risk factors are those that increase suicide risk over a longer period of time (47). For the purposes of this review, we examine the distal relationship between controlled substance use and suicide, for 3 reasons. First, it is difficult to disentangle the role of proximal substance use as a risk factor versus suicide attempt mechanism. Second, the distal pathway is conceptually similar to Goldstein et al.'s psychopharmacological pathway, allowing us to compare the evidence related to this pathway for interpersonal violence versus suicide. Third, because of the potential for early identification and amelioration, the distal pathway is more relevant for policy intervention.

Some suicide and interpersonal risk factors overlap, and others are distinct. Despite its apparent role in several recent mass shootings and large amounts of public attention to the issue, mental illness is only rarely the direct cause of interpersonal violence: The best available research suggests that 3%–5% of all interpersonal violence in the United States is directly attributable to mental illness (25). In contrast, mental illness is strongly associated with risk of suicide (25). Mood disorders such as depression place individuals at particularly high risk of suicide (49), and these disorders often co-occur with use of controlled substances. In the United States in a given year, about 8% of individuals with any mood disorder also have a drug use disorder, compared with about 2% of the overall US population (50). Importantly, alcohol abuse, another risk factor for suicide, often co-occurs with both controlled substance use and mood disorders (50). As with interpersonal violence, use of controlled substances and suicide share many other risk factors, including personality traits such as impulsivity and hopelessness and situational factors such as stressful life circumstances related to poverty, unemployment, or divorce (47, 51). These factors may confound, mediate, or moderate the relationship between controlled substances and suicide (47). For the reasons given above, we limit our review of studies of the relationship between controlled substance use and suicide to prospective longitudinal studies.

METHODS

Search strategy

We identified peer-reviewed original research studies published in English-language sources between January 1, 1980, and December 31, 2014, using the online databases PubMed, Embase, and Web of Science. The search terms used to identify relevant articles included keywords related to controlled substances, firearms, interpersonal violence, and suicide. The full set of Boolean search terms was as follows: “drug” [tiab] or “controlled substance” [tiab] or “illicit substance” [tiab] or “narcotic” [tiab] or “stimulant” [tiab] or “hallucinogen” [tiab] or “depressant” [tiab] or “anabolic steroid” [tiab] or “marijuana” [tiab] or “heroin” [tiab] or “opioid” [tiab] or “opiate” [tiab] or “cocaine” [tiab] or “amphetamine” [tiab] or “methamphetamine” [tiab] or “barbiturate” [tiab] or “benzodiazepine” [tiab] or “inhalant” [tiab] AND “violence” [ti] or “violent” [ti] or “homicide” [ti] or “assault” [ti] or “suicide” [ti] or “aggress*” [ti] or “gun” [ti] or “firearm” [ti], where [ti] and [tiab] are the search tags for title and title and abstract, respectively. To identify articles that may have been missed by our electronic search, we also hand-searched the reference lists of 14 relevant review articles published between 1980 and 2014 (8, 9, 3740, 4345, 47, 5255). To narrow the initial search return, the lead author (E.E.M.) screened the titles and abstracts returned by the initial search. The first and second authors (E.E.M. and S.C.) then screened the remaining full-text articles for eligibility.

Study selection

Inclusion criteria

We included studies meeting the following inclusion criteria: 1) prospective longitudinal design measuring the association between controlled substance use/involvement (e.g., selling) and violence toward others or suicide; 2) use of a non-drug user comparison group selected from the same cohort as the drug-user group; 3) publication in the peer-reviewed literature between 1980 and 2014; and 4) inclusion of at least 100 study participants. We defined prospective longitudinal studies as those that measured controlled substance use/involvement prior to measurement of violence. Studies that asked violent offenders about their drug use in the preceding days/weeks/months and attempted to estimate a longitudinal association in this manner were not included. We included studies measuring either aggregate use of controlled substances or use of 1 or more specific substances and placed no restrictions on the types of controlled substances measured, except for the exclusion related to infrequently misused prescription drugs described in the following paragraph. Although we used the tripartite framework of Goldstein et al. and the distal pathway model for suicide risk to guide the scope and organize the results of our review, our selection of studies was motivated primarily by the desire to capture the best available research designed to assess the causal relationship between controlled substances and violence. Thus, we included prospective longitudinal studies where clear temporality between controlled substance involvement and violence could be established.

Exclusion criteria

We excluded studies on the basis of the following criteria: 1) used a retrospective or cross-sectional design; 2) measured the association between prescription drugs that are not frequently misused, for example, antiepileptics, antidepressants, and antipsychotics, with violence perpetration or suicide; 3) used nonhuman subjects; 4) measured only the combined association between alcohol and drugs with outcomes of interest; 5) examined the relationship between controlled substances and violent victimization, rather than perpetration or suicide; and 6) solely measured drug overdose as the method of suicide (only studies that examined controlled substance use as a risk factor preceding suicide were included).

Data extraction

We used a structured instrument (Web Appendix available at http://aje.oxfordjournals.org/) to extract key information from articles. Development of the instrument was informed by the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist (56) and the Cochrane Handbook (57). Information extracted for each study included the following: title; authors; study objective; study design; study population, including inclusion and exclusion criteria; participant recruitment; study setting; data sources; method and timeframe of assessment; definitions of controlled substance and violence measures; covariates controlled for in final analytical model; and measure(s) of association between controlled substances and violence. For interpersonal violence, we classified measures of association as psychopharmacological, economic compulsive, or systemic. Measures testing the association between individual drug use and interpersonal violence outcomes except armed robbery were classified as psychopharmacological. Measures testing the association between index measures of economic compulsion and interpersonal violence and measures testing the association between individual drug use and robbery were classified as economic compulsive. Measures testing the association between involvement in drug sales and violence perpetration were classified as systemic. Measures of the relationship between controlled substances and suicide were not classified into separate subcategories. Initial data extraction was performed by S.C. and then verified by E.E.M.

Study quality assessment

We used a modified version of the Cochrane risk of bias assessment tool for nonrandomized studies (ACROBAT-NRSI) (57) to grade the quality of each study included in our review (Web Table 1) (57). Because the Cochrane tool is designed to assess the quality of observational intervention studies and the literature we reviewed is epidemiologic rather than interventional, we modified the tool by excluding the checklist items relevant only for intervention studies. Fourteen quality criteria were assessed across 5 domains (confounding, bias in selection of participants into the study, bias due to missing data, bias in measurement of outcomes, and bias in selection of the reported result). For each domain, studies received a rating of low, moderate, serious, or critical risk of bias. Studies were then assigned an overall bias rating based on the domain-specific scores. The tool is designed so that only randomized controlled trials can receive an overall “low” bias rating. As no randomized controlled trials on the topic of interest exist, the highest rating a rigorous prospective longitudinal study could achieve was “moderate.” Following Cochrane conventions, studies receiving a “critical” bias score are not reviewed. Quality assessments were conducted independently by 2 authors (E.E.M. and S.C.), and interrater reliability for each item was assessed by using κ statistics. The majority of individual items and all overall bias judgments met conventional standards for reliability of 0.69 or higher (58). Three individual items had κ statistics between 62 and 68. In the case of discrepancy in the final bias rating, the 2 raters discussed disagreements and came to consensus.

Data analysis

Because of the heterogeneity in study populations, measures, and analytical strategies used to assess the associations between controlled substances and violence perpetration and suicide, we did not attempt to conduct a meta-analysis and relied instead on a narrative synthesis. Results are summarized in Tables 2345, which are organized by alphabetical order of author surnames and the text of the Results.

Table 2.

Summary of Included Studies (n = 22), 1990–2014

First Author, Year (Reference No.) Years of Data Collection Frequency of
Data Collection 
Study Population Location Study No. Drugs/Violence Pathway Studied Susceptibility to Bias (Low, Moderate, or Serious Risk) 
Perpetration of Violence Toward Others 
Bellair, 2009 (751997–2001 5 waves of data collection in 1997, 1998, 1999, 2000, and 2001 A nationally representative sample of youth aged 12–16 years in 1997 United States 5,567 Systemic Moderate 
Brady, 2008 (711996–2000 3 waves of data collection in 1996, 1999, and 2000 Youth aged 12–15 years insured by a large health maintenance organization United States 302 Psychopharmacological Serious 
Brook, 2011 (591990–2004 4 years of data collection at 5-year intervals: 1990, 1994, 1999, and 2004 Students in grades 7–10 (in 1990) in 11 schools East Harlem,
New York, New York 
1,332 Psychopharmacological Serious 
Cerdá, 2010 (741990–1999 10 years of data collection at 1-year intervals, 1990–1999 Gun homicide victims from 1990 to 1999 New York, New York 8,820 Systemic Moderate 
Dembo, 1990 (691986–1987 2 measures 6 months apart, exact timing unspecified Youth aged 10–18 years admitted to a regional detention center Tampa, Florida 201 Psychopharmacological, systemic Serious 
Ellickson, 2000 (701985–1990 2 waves of data collection, 1985 and 1990 Seventh grade students (in 1985) from 30 schools California and Oregon 4,390 Psychopharmacological Moderate 
Friedman, 2001 (66Years not given; 2.5-year period 2 measures during 2.5-year period, exact timing unspecified Inner-city, low socioeconomic status African-American adults Philadelphia, Pennsylvania 612 Psychopharmacological, economic compulsive, systemic Moderate 
Green, 2010 (681966–2003 4 waves of data collection in 1966, 1975–1977, 1992–1993, and 2002–2003 Community cohort of urban African Americans followed from age 6 to 42 years Woodlawn,
Chicago, Illinois 
702 Psychopharmacological, systemic Moderate 
Kuhns, 2005 (601976–1977 3 waves of data collection in 1976, 1977, and 1978 Representative sample of US youth aged 11–17 years in 1976 United States 1,725 Psychopharmacological Moderate 
McKetin, 2014 (672006–2010 4 waves of data collection at baseline, 3 months after baseline, 1 year after baseline, and 3 years after baseline Methamphetamine-dependent individuals aged ≥16 years Sydney and Brisbane, Australia 278 Psychopharmacological Serious 
Menard, 2001 (611976–1992 9 waves of data collection in 1976, 1977, 1978, 1979, 1980, 1983, 1986, 1989, and 1992 A representative sample of adolescents aged 11–17 years in 1976 and 22–33 years in 1992 United States 1,725 Psychopharmacological, systemic Serious 
Mulvey, 2006 (62Year not reported 26 weekly interviews, exact timing unspecified Individuals identified as high risk for involvement in repeated violence in the emergency room of an urban psychiatric hospital Northeastern
United States 
132 Psychopharmacological Serious 
Pedersen, 2010 (731992–2005 4 waves of data collection at ages 13, 15, 20, and 27 years Population-based sample of adolescents between 12 and 16 years of age in 1992 Norway 1,353 Psychopharmacological Moderate 
Sussman, 2004 (631994–2000 2 waves of data collection in 1994–1995 and 1999–2000 A representative sample of high school students Five-county region in southern California 676 Psychopharmacological Serious 
Van Dorn, 2012 (642001–2005 Two waves of data collection in 2001–2002 and 2004–2005 Nationally representative sample of the civilian noninstitutionalized adult population United States 34,653 Psychopharmacological Moderate 
Wei, 2004 (721991–2000 10 waves of data collection at ages 11, 12, 13, 14, 15, 16, 17, 18, 19, and 20 years Males aged 11 (in 1991) with antisocial behavior participating in the Pittsburgh Youth Survey Pittsburgh, Pennsylvania 503 Psychopharmacological Serious 
Weiner, 2005 (65Years not reported; 5-year study period 2 waves of data collection at baseline and an average of 5 years after baseline Continuation (alternative) of high school students 21 districts in
southern California 
1,867 Psychopharmacological, economic compulsive Serious 
Suicide 
Allebeck, 1990 (771969–1983 Baseline data collected at conscription in 1969 were linked to inpatient registry data; all inpatient hospitalizations from 1969 to 1983 were measured Men conscripted for military training in 1969–1970 Sweden 50,465 Psychopharmacological Serious 
Allgulander, 1992 (801973–1987 Used inpatient and suicide registry data; all relevant records between 1973 and 1987 were measured Patients discharged with at least 1 psychiatric diagnosis from inpatient hospitals Stockholm County,
Sweden 
80,970 Psychopharmacological Moderate 
Nilsson, 2014 (781999–2008 Used homeless registry, civil registration system, psychiatric central registry, and cause of death registry data. All relevant records between 1999 and 2008 were measured Individuals aged ≥16 years with at least 1 contact with a homeless shelter during the study period Denmark 32,010 Psychopharmacological Moderate 
Petronis, 1990 (791984–1985 Two waves of data collection in 1984 and 1985 ECA participants Five US cities: New Haven, Connecticut; Baltimore, Maryland; St. Louis, Missouri; Durham, North Carolina; and Los Angeles, California 13,673 Psychopharmacological Moderate 
Yen, 2003 (76Years not reported; 2-year study period Four waves of data collection at baseline, 6 months after baseline, 1 year after baseline, and 2 years after baseline Individuals aged 18–45 years with diagnosed personality disorders recruited from treatment clinics Four CLPS sites 621 Psychopharmacological Serious 
First Author, Year (Reference No.) Years of Data Collection Frequency of
Data Collection 
Study Population Location Study No. Drugs/Violence Pathway Studied Susceptibility to Bias (Low, Moderate, or Serious Risk) 
Perpetration of Violence Toward Others 
Bellair, 2009 (751997–2001 5 waves of data collection in 1997, 1998, 1999, 2000, and 2001 A nationally representative sample of youth aged 12–16 years in 1997 United States 5,567 Systemic Moderate 
Brady, 2008 (711996–2000 3 waves of data collection in 1996, 1999, and 2000 Youth aged 12–15 years insured by a large health maintenance organization United States 302 Psychopharmacological Serious 
Brook, 2011 (591990–2004 4 years of data collection at 5-year intervals: 1990, 1994, 1999, and 2004 Students in grades 7–10 (in 1990) in 11 schools East Harlem,
New York, New York 
1,332 Psychopharmacological Serious 
Cerdá, 2010 (741990–1999 10 years of data collection at 1-year intervals, 1990–1999 Gun homicide victims from 1990 to 1999 New York, New York 8,820 Systemic Moderate 
Dembo, 1990 (691986–1987 2 measures 6 months apart, exact timing unspecified Youth aged 10–18 years admitted to a regional detention center Tampa, Florida 201 Psychopharmacological, systemic Serious 
Ellickson, 2000 (701985–1990 2 waves of data collection, 1985 and 1990 Seventh grade students (in 1985) from 30 schools California and Oregon 4,390 Psychopharmacological Moderate 
Friedman, 2001 (66Years not given; 2.5-year period 2 measures during 2.5-year period, exact timing unspecified Inner-city, low socioeconomic status African-American adults Philadelphia, Pennsylvania 612 Psychopharmacological, economic compulsive, systemic Moderate 
Green, 2010 (681966–2003 4 waves of data collection in 1966, 1975–1977, 1992–1993, and 2002–2003 Community cohort of urban African Americans followed from age 6 to 42 years Woodlawn,
Chicago, Illinois 
702 Psychopharmacological, systemic Moderate 
Kuhns, 2005 (601976–1977 3 waves of data collection in 1976, 1977, and 1978 Representative sample of US youth aged 11–17 years in 1976 United States 1,725 Psychopharmacological Moderate 
McKetin, 2014 (672006–2010 4 waves of data collection at baseline, 3 months after baseline, 1 year after baseline, and 3 years after baseline Methamphetamine-dependent individuals aged ≥16 years Sydney and Brisbane, Australia 278 Psychopharmacological Serious 
Menard, 2001 (611976–1992 9 waves of data collection in 1976, 1977, 1978, 1979, 1980, 1983, 1986, 1989, and 1992 A representative sample of adolescents aged 11–17 years in 1976 and 22–33 years in 1992 United States 1,725 Psychopharmacological, systemic Serious 
Mulvey, 2006 (62Year not reported 26 weekly interviews, exact timing unspecified Individuals identified as high risk for involvement in repeated violence in the emergency room of an urban psychiatric hospital Northeastern
United States 
132 Psychopharmacological Serious 
Pedersen, 2010 (731992–2005 4 waves of data collection at ages 13, 15, 20, and 27 years Population-based sample of adolescents between 12 and 16 years of age in 1992 Norway 1,353 Psychopharmacological Moderate 
Sussman, 2004 (631994–2000 2 waves of data collection in 1994–1995 and 1999–2000 A representative sample of high school students Five-county region in southern California 676 Psychopharmacological Serious 
Van Dorn, 2012 (642001–2005 Two waves of data collection in 2001–2002 and 2004–2005 Nationally representative sample of the civilian noninstitutionalized adult population United States 34,653 Psychopharmacological Moderate 
Wei, 2004 (721991–2000 10 waves of data collection at ages 11, 12, 13, 14, 15, 16, 17, 18, 19, and 20 years Males aged 11 (in 1991) with antisocial behavior participating in the Pittsburgh Youth Survey Pittsburgh, Pennsylvania 503 Psychopharmacological Serious 
Weiner, 2005 (65Years not reported; 5-year study period 2 waves of data collection at baseline and an average of 5 years after baseline Continuation (alternative) of high school students 21 districts in
southern California 
1,867 Psychopharmacological, economic compulsive Serious 
Suicide 
Allebeck, 1990 (771969–1983 Baseline data collected at conscription in 1969 were linked to inpatient registry data; all inpatient hospitalizations from 1969 to 1983 were measured Men conscripted for military training in 1969–1970 Sweden 50,465 Psychopharmacological Serious 
Allgulander, 1992 (801973–1987 Used inpatient and suicide registry data; all relevant records between 1973 and 1987 were measured Patients discharged with at least 1 psychiatric diagnosis from inpatient hospitals Stockholm County,
Sweden 
80,970 Psychopharmacological Moderate 
Nilsson, 2014 (781999–2008 Used homeless registry, civil registration system, psychiatric central registry, and cause of death registry data. All relevant records between 1999 and 2008 were measured Individuals aged ≥16 years with at least 1 contact with a homeless shelter during the study period Denmark 32,010 Psychopharmacological Moderate 
Petronis, 1990 (791984–1985 Two waves of data collection in 1984 and 1985 ECA participants Five US cities: New Haven, Connecticut; Baltimore, Maryland; St. Louis, Missouri; Durham, North Carolina; and Los Angeles, California 13,673 Psychopharmacological Moderate 
Yen, 2003 (76Years not reported; 2-year study period Four waves of data collection at baseline, 6 months after baseline, 1 year after baseline, and 2 years after baseline Individuals aged 18–45 years with diagnosed personality disorders recruited from treatment clinics Four CLPS sites 621 Psychopharmacological Serious 

Abbreviations: CLPS, Collaborative Longitudinal Personality Disorders Study; ECA, Epidemiologic Catchment Area referring to 5 cities in the United States: New Haven, Connecticut; Baltimore, Maryland; St. Louis, Missouri; Durham, North Carolina; and Los Angeles, California.

Table 3.

Summary of Results for Studies of the Psychopharmacological Relationship Between Controlled Substance Use and Violence Perpetration, 1990–2014

First Author, Year (Reference No.) Controlled Substance Measure Violence Measure Key Finding Covariates Accounted for
in Analysis 
Association Between Drugs and Violence 
Aggregate Measures of Controlled Substance Usea 
Brook, 2011 (59Self-reported past year illicit drug use Serious violence scale. Individual items comprising the scale assessed frequency threatening with a weapon; shooting at or hitting with a weapon; cutting with a knife; and beating up/throwing something at someone else Illegal drug use in 1994 was associated with violent behavior in 1999 (coefficient = 0.08, t statistic = 2.02) and 2004 (coefficient = 0.12, t statistic = 3.67) Ethnicity, sex Positive 
Kuhns, 2005 (60Frequency of drug use, including marijuana, cocaine, heroin, barbiturates, amphetamines, and hallucinogens Serious violence offending, defined by using the Crimes Against Persons Scale Drug use in wave 1 of the survey was not associated with increased odds of serious violence offending at wave 2 (odds ratio = 0.05; P > 0.05). Drug use in wave 2 of the survey was not associated with increased odds of serious violent offending at wave 3 (odds ratio = 4.92; P > 0.05) Attitudes toward violence, exposure to delinquent peers, neighborhood problems, family attachment, perceived family importance, alcohol use, minor delinquency No association 
Menard, 2001 (61Self-reported use of hallucinogens, amphetamines, heroin, cocaine, and barbiturates in the last calendar year “Index violence,” defined as felony assault or robbery Drug use had no association with
violence 
None No association 
Mulvey, 2006 (62Self-reported daily and weekly use of drugs excluding marijuana Self-reported daily and weekly frequency of 9 aggressive acts, including pushing, hitting, and using a weapon Drug use had no association with violence (odds ratio = 1.5, 95% CI: 0.8, 2.8) Marijuana use and sales, alcohol use and sales, hard drug sales No association 
Sussman, 2004 (63Self-reported current use and frequency of use of “hard drugs” including cocaine, hallucinogens, stimulants, inhalants, PCP, heroin, and steroids Violent behavior using a 4-item index adapted from the 1981 Monitoring the Future Survey Current hard drug use was associated with violence perpetration (F statistic = 8.7; P < 0.05) Baseline perpetration of violence, sex, beliefs about violence, and the acceptability and morality of drug use, self-identification with a high-risk group, and perceived stress Positive 
Van Dorn, 2012 (64Current (at baseline) drug abuse and/or dependence, measured using a structured interview schedule Any violence, defined as any of the following having occurred in the time since the prior interview (≈2 years): 1) using a weapon like a stick, knife, or gun in a fight; 2) hitting someone so hard you injured them or they had to see a doctor; 3) starting a fire on purpose to destroy someone's property or just to see it burn; 4) force someone to have sex against their will; 5) getting into a physical fight when or right after drinking; 6) getting into a fight when under the influence of a drug; 7) physically hurting another person in any way on purpose; 8) getting into a fight that came to swapping blows with someone like a husband, wife, boyfriend, or girlfriend; 9) getting into a lot of fights that you started Drug use disorder had no association with violence (odds ratio = 1.51, 95% CI: 0.59, 3.84; P > 0.05) Length of time between waves 1 and 2, age, sex, race, education, income, marital status, urban/rural, household size, history of abuse or neglect, mental illness and alcohol use disorders, stressful life circumstances No association 
Weiner, 2005 (65Self-reported monthly frequency of marijuana, cocaine, hallucinogen, stimulant, inhalant, and other drug use Violence perpetration scale comprising items assessing the annual frequency of incidents in which a weapon was used to injure or threaten someone; injury was perpetrated without a weapon; and property was damaged or stolen on purpose. Adapted from the 1981 Monitoring the Future Survey Illegal drug use was associated with increased likelihood of violence perpetration 5 years later (coefficient = 0.15, SE, 0.04; P < 0.05) Sex and ethnicity, baseline level of violence perpetration Positive 
Amphetamines and Methamphetamines 
Friedman, 2001 (66Self-reported frequency of use of amphetamines Self-reported conviction for assault Amphetamine use had no association with assault conviction (P > 0.05) Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use No association 
  Self-reported conviction for weapons offenses Amphetamine use had no association with conviction for weapons offenses (P > 0.05)  No association 
  Self-reported conviction for attempted homicide Amphetamine use had no association with attempted homicide conviction (P > 0.05)  No association 
  Self-reported conviction for homicide Amphetamine use had no association with homicide conviction (P > 0.05)  No association 
  Self-reported gang drug-war fighting Amphetamine use was associated with increased likelihood of gang drug-war fighting (partial correlation coefficient = 0.1; P < 0.05)  Positive 
McKetin, 2014 (67Days of methamphetamine use in past 4 weeks Violent behavior defined as history of assault with no physical harm, destruction of property, attack of others with intention to harm, or commission of actual physical harm (e.g., with a weapon) Compared with individuals with no methamphetamine use in the past month, individuals with 1–15 days of use (odds ratio = 2.8, 95% CI: 1.6, 4.9) and ≥16 days of use (odds ratio = 9.5, 95% CI: 4.80, 19.1) had increased odds of acting violently Psychotic symptoms, other substance use (including drugs and alcohol), and sociodemographic characteristics Positive 
Barbiturates 
Friedman, 2001 (66Self-reported frequency of use of barbiturates Self-reported conviction for assault Barbiturate use was associated with decreased likelihood of assault conviction (partial correlation coefficient = −0.1; P < 0.05) Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use Negative 
  Self-reported conviction for weapons offenses Barbiturate use had no association with conviction for weapons offenses (P > 0.05)  No association 
  Self-reported conviction for attempted homicide Barbiturate use was associated with decreased likelihood of attempted homicide conviction (partial correlation coefficient = −0.13; P < 0.05)  Negative 
  Self-reported conviction for homicide Barbiturate use had no association with homicide conviction (P > 0.05)  No association 
  Self-reported gang drug-war fighting Barbiturate use was associated with decreased likelihood of gang drug-war fighting (partial correlation coefficient = −0.33; P < 0.05)  Negative 
Cocaine 
Dembo, 1990 (69Cocaine (crack and powdered) use and frequency of use, measured by urinalysis and self-report Crimes against persons, defined as self-reported aggravated assault, gang fights, hit a teacher/student, committed sexual assault, strong-armed students/teachers/others There was no association between cocaine use and crimes against persons None, HIV risk No association 
Friedman, 2001 (66Self-reported frequency of use of cocaine/crack Self-reported conviction for assault Cocaine use had no association with assault conviction (P > 0.05) Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use No association 
  Self-reported conviction for weapons offenses Cocaine use had no association with conviction for weapons offenses (P > 0.05)  No association 
  Self-reported conviction for attempted homicide Cocaine use had no association with attempted homicide conviction (P > 0.05)  No association 
  Self-reported conviction for homicide Cocaine use was associated with increased likelihood of homicide conviction (partial correlation coefficient = 0.14; P < 0.05)  Positive 
  Self-reported gang drug-war fighting Cocaine use was associated with increased likelihood of gang drug-war fighting (partial correlation coefficient = 0.12; P < 0.05)  Positive 
Marijuana 
Brady, 2008 (71Self-reported use of marijuana in past year Self-reported minor or serious violence in the past year, ranging from starting a fight with other kids to firing a gun at another teenager Marijuana use at age 15 years was associated with increased risk of violent perpetration at age 19 years (odds ratio = 2.90, 95% CI: 1.08, 7.82; P < 0.05). Marijuana use at age 18 years was not associated with violent perpetration at age 19 years (odds ratio = 1.66, 95% CI: 0.56, 4.90; P > 0.05) Prior levels of violence involvement, age, sex, ethnicity, socioeconomic status Mixed 
Ellickson, 2000 (70Self-reported frequency of marijuana use in the past year Predatory violence, defined as use of force to obtain money or things from people, involvement in gang fights, attacking someone with intent to injure or kill, and carrying a concealed weapon Marijuana use was associated with increased odds of perpetration of predatory violence (odds ratio = 0.11; P < 0.05) Alcohol use, school bonds, family bonds, problem behavior, peer drug use, drug offers, self-esteem, rebelliousness, age, sex, race, parent education, neighborhood socioeconomic status, school drug-use prevalence Positive 
Friedman, 2001 (66Self-reported frequency of use of marijuana Self-reported conviction for assault Marijuana use had no association with assault conviction (P > 0.05) Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use No association 
  Self-reported conviction for weapons offenses Marijuana use was associated with increased likelihood of weapons offenses (partial correlation coefficient = 0.11; P < 0.05)  Positive 
  Self-reported conviction for attempted homicide Marijuana use was associated with increased likelihood of attempted homicide (partial correlation coefficient = 0.10; P < 0.05)  Positive 
  Self-reported conviction for homicide Marijuana use had no association with homicide conviction (P > 0.05)  No association 
  Self-reported gang drug-war fighting Marijuana use had no association with gang drug-war fighting (P > 0.05)  No association 
Green, 2010 (68Self-reported lifetime frequency of heavy marijuana use, defined as lifetime frequency of 20 or more times during adolescence Violent crime, defined as murder, assault, battery, and domestic violence and obtained from law enforcement records Heavy adolescent marijuana use had no association with violent crime (odds ratio = 1.09, 95% CI: 0.72, 1.65; P > 0.05) Sex, socioeconomic status, family background, school adaptation, school achievement, tobacco smoking, and delinquency No association 
Menard, 2001 (61Self-reported use of marijuana, in the last calendar year “Index violence,” defined as felony assault or robbery Marijuana use was associated with violent crime (risk ratio = 2.60; P < 0.05) None Positive 
Mulvey, 2006 (62Self-reported daily and weekly use of marijuana Self-reported daily and weekly frequency of 9 aggressive acts, including pushing, hitting, and using a weapon Prior-day marijuana use was associated with increased risk of violence (odds ratio = 1.6, 95% CI: 1.2, 2.0; P < 0.05) Hard drug (nonmarijuana) use and sales, alcohol use and sales Positive 
Pedersen, 2010 (73Self-reported frequency of cannabis use in past 12 months Serious crime including theft, robbery, and violence, as recorded by Statistics Norway Marijuana use 1–10 times in the past year was not associated with serious crime (odds ratio = 1.6, 95% CI: 0.6, 4.7; P > 0.05) or ≥11 times in the past year (odds ratio = 1.4, 95% CI: 0.4, 5.2; P > 0.05) Age, sex, parental cultural capital, parental monitoring, school grades, conduct problem, early adolescent marijuana use, cohabitation status, previous criminal charges, alcohol use, other illegal drug use No association 
Wei, 2004 (72Self-reported past-year frequency of use of marijuana, assessed by using the Substance Use Scale Violence, defined by using a delinquency scale including items assessing self-reported past-year frequency of gang fighting, strong-arming, attacking someone with a weapon or intent to seriously hurt or kill, and rape or forced sex Prior-year frequent marijuana use was associated with increased odds of violence among those aged 14 years (P < 0.05); 15 years (odds ratio = 3.07; P < 0.05); 16 years (odds ratio = 3.36; P < 0.05); 17 years (odds ratio = 2.67; P < 0.05); and 19 years (odds ratio = 3.83; P < 0.05). Prior-year frequent marijuana use was not associated with increased odds of violence among those aged 18 years or 20 years Academic achievement, depressed mood, hyperactivity/ impulsivity/inattention problems, poor communication with caretaker, poor supervision, caretaker perception of bad neighborhood, race/ethnicity Mixed 
Opiatesb 
Friedman, 2001 (66Self-reported frequency of use of opiates Self-reported conviction for assault Opiate use was associated with increased likelihood of assault conviction (partial correlation coefficient = 0.14; P < 0.05) Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use Positive 
  Self-reported conviction for weapons offenses Opiate use had no association with conviction for weapons offenses (P > 0.05)  No association 
  Self-reported conviction for attempted homicide Opiate use was associated with increased likelihood of attempted homicide conviction (partial correlation coefficient = 0.11; P < 0.05)  Positive 
  Self-reported conviction for homicide Opiate use had no association with homicide conviction (P > 0.05)  No association 
  Self-reported gang drug-war fighting Opiate use was associated with increased likelihood of gang drug-war fighting (partial correlation coefficient = 0.36; P < 0.05)  Positive 
PCP/Hallucinogens 
Friedman, 2001 (66Self-reported frequency of use of PCP/hallucinogens Self-reported conviction for assault PCP/hallucinogen use had no association with assault conviction (P > 0.05) Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use No association 
  Self-reported conviction for weapons offenses PCP/hallucinogen use had no association with conviction for weapons offenses (P > 0.05)  No association 
  Self-reported conviction for attempted homicide PCP/hallucinogen use had no association with attempted homicide conviction (P > 0.05)  No association 
  Self-reported conviction for homicide PCP/hallucinogen use had no association with attempted homicide conviction (P > 0.05)  No association 
  Self-reported gang drug-war fighting PCP/hallucinogen use had no association with gang drug-war fighting conviction (P > 0.05)  No association 
Tranquilizers 
Friedman, 2001 (66Self-reported frequency of use of tranquilizers Self-reported conviction for assault Tranquilizer use was associated with increased likelihood of assault conviction (partial correlation coefficient = 0.11; P < 0.05) Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use Positive 
  Self-reported conviction for weapons offenses Tranquilizer use had no association with conviction for weapons offenses (P > 0.05)  No association 
  Self-reported conviction for attempted homicide Tranquilizer use had no association with attempted homicide conviction (P > 0.05)  No association 
  Self-reported conviction for homicide Tranquilizer use had no association with homicide conviction (P > 0.05)  No association 
  Self-reported gang drug-war fighting Tranquilizer use was associated with decreased likelihood of gang drug-war fighting (partial correlation coefficient = −0.007; P < 0.0.05)  Negative 
First Author, Year (Reference No.) Controlled Substance Measure Violence Measure Key Finding Covariates Accounted for
in Analysis 
Association Between Drugs and Violence 
Aggregate Measures of Controlled Substance Usea 
Brook, 2011 (59Self-reported past year illicit drug use Serious violence scale. Individual items comprising the scale assessed frequency threatening with a weapon; shooting at or hitting with a weapon; cutting with a knife; and beating up/throwing something at someone else Illegal drug use in 1994 was associated with violent behavior in 1999 (coefficient = 0.08, t statistic = 2.02) and 2004 (coefficient = 0.12, t statistic = 3.67) Ethnicity, sex Positive 
Kuhns, 2005 (60Frequency of drug use, including marijuana, cocaine, heroin, barbiturates, amphetamines, and hallucinogens Serious violence offending, defined by using the Crimes Against Persons Scale Drug use in wave 1 of the survey was not associated with increased odds of serious violence offending at wave 2 (odds ratio = 0.05; P > 0.05). Drug use in wave 2 of the survey was not associated with increased odds of serious violent offending at wave 3 (odds ratio = 4.92; P > 0.05) Attitudes toward violence, exposure to delinquent peers, neighborhood problems, family attachment, perceived family importance, alcohol use, minor delinquency No association 
Menard, 2001 (61Self-reported use of hallucinogens, amphetamines, heroin, cocaine, and barbiturates in the last calendar year “Index violence,” defined as felony assault or robbery Drug use had no association with
violence 
None No association 
Mulvey, 2006 (62Self-reported daily and weekly use of drugs excluding marijuana Self-reported daily and weekly frequency of 9 aggressive acts, including pushing, hitting, and using a weapon Drug use had no association with violence (odds ratio = 1.5, 95% CI: 0.8, 2.8) Marijuana use and sales, alcohol use and sales, hard drug sales No association 
Sussman, 2004 (63Self-reported current use and frequency of use of “hard drugs” including cocaine, hallucinogens, stimulants, inhalants, PCP, heroin, and steroids Violent behavior using a 4-item index adapted from the 1981 Monitoring the Future Survey Current hard drug use was associated with violence perpetration (F statistic = 8.7; P < 0.05) Baseline perpetration of violence, sex, beliefs about violence, and the acceptability and morality of drug use, self-identification with a high-risk group, and perceived stress Positive 
Van Dorn, 2012 (64Current (at baseline) drug abuse and/or dependence, measured using a structured interview schedule Any violence, defined as any of the following having occurred in the time since the prior interview (≈2 years): 1) using a weapon like a stick, knife, or gun in a fight; 2) hitting someone so hard you injured them or they had to see a doctor; 3) starting a fire on purpose to destroy someone's property or just to see it burn; 4) force someone to have sex against their will; 5) getting into a physical fight when or right after drinking; 6) getting into a fight when under the influence of a drug; 7) physically hurting another person in any way on purpose; 8) getting into a fight that came to swapping blows with someone like a husband, wife, boyfriend, or girlfriend; 9) getting into a lot of fights that you started Drug use disorder had no association with violence (odds ratio = 1.51, 95% CI: 0.59, 3.84; P > 0.05) Length of time between waves 1 and 2, age, sex, race, education, income, marital status, urban/rural, household size, history of abuse or neglect, mental illness and alcohol use disorders, stressful life circumstances No association 
Weiner, 2005 (65Self-reported monthly frequency of marijuana, cocaine, hallucinogen, stimulant, inhalant, and other drug use Violence perpetration scale comprising items assessing the annual frequency of incidents in which a weapon was used to injure or threaten someone; injury was perpetrated without a weapon; and property was damaged or stolen on purpose. Adapted from the 1981 Monitoring the Future Survey Illegal drug use was associated with increased likelihood of violence perpetration 5 years later (coefficient = 0.15, SE, 0.04; P < 0.05) Sex and ethnicity, baseline level of violence perpetration Positive 
Amphetamines and Methamphetamines 
Friedman, 2001 (66Self-reported frequency of use of amphetamines Self-reported conviction for assault Amphetamine use had no association with assault conviction (P > 0.05) Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use No association 
  Self-reported conviction for weapons offenses Amphetamine use had no association with conviction for weapons offenses (P > 0.05)  No association 
  Self-reported conviction for attempted homicide Amphetamine use had no association with attempted homicide conviction (P > 0.05)  No association 
  Self-reported conviction for homicide Amphetamine use had no association with homicide conviction (P > 0.05)  No association 
  Self-reported gang drug-war fighting Amphetamine use was associated with increased likelihood of gang drug-war fighting (partial correlation coefficient = 0.1; P < 0.05)  Positive 
McKetin, 2014 (67Days of methamphetamine use in past 4 weeks Violent behavior defined as history of assault with no physical harm, destruction of property, attack of others with intention to harm, or commission of actual physical harm (e.g., with a weapon) Compared with individuals with no methamphetamine use in the past month, individuals with 1–15 days of use (odds ratio = 2.8, 95% CI: 1.6, 4.9) and ≥16 days of use (odds ratio = 9.5, 95% CI: 4.80, 19.1) had increased odds of acting violently Psychotic symptoms, other substance use (including drugs and alcohol), and sociodemographic characteristics Positive 
Barbiturates 
Friedman, 2001 (66Self-reported frequency of use of barbiturates Self-reported conviction for assault Barbiturate use was associated with decreased likelihood of assault conviction (partial correlation coefficient = −0.1; P < 0.05) Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use Negative 
  Self-reported conviction for weapons offenses Barbiturate use had no association with conviction for weapons offenses (P > 0.05)  No association 
  Self-reported conviction for attempted homicide Barbiturate use was associated with decreased likelihood of attempted homicide conviction (partial correlation coefficient = −0.13; P < 0.05)  Negative 
  Self-reported conviction for homicide Barbiturate use had no association with homicide conviction (P > 0.05)  No association 
  Self-reported gang drug-war fighting Barbiturate use was associated with decreased likelihood of gang drug-war fighting (partial correlation coefficient = −0.33; P < 0.05)  Negative 
Cocaine 
Dembo, 1990 (69Cocaine (crack and powdered) use and frequency of use, measured by urinalysis and self-report Crimes against persons, defined as self-reported aggravated assault, gang fights, hit a teacher/student, committed sexual assault, strong-armed students/teachers/others There was no association between cocaine use and crimes against persons None, HIV risk No association 
Friedman, 2001 (66Self-reported frequency of use of cocaine/crack Self-reported conviction for assault Cocaine use had no association with assault conviction (P > 0.05) Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use No association 
  Self-reported conviction for weapons offenses Cocaine use had no association with conviction for weapons offenses (P > 0.05)  No association 
  Self-reported conviction for attempted homicide Cocaine use had no association with attempted homicide conviction (P > 0.05)  No association 
  Self-reported conviction for homicide Cocaine use was associated with increased likelihood of homicide conviction (partial correlation coefficient = 0.14; P < 0.05)  Positive 
  Self-reported gang drug-war fighting Cocaine use was associated with increased likelihood of gang drug-war fighting (partial correlation coefficient = 0.12; P < 0.05)  Positive 
Marijuana 
Brady, 2008 (71Self-reported use of marijuana in past year Self-reported minor or serious violence in the past year, ranging from starting a fight with other kids to firing a gun at another teenager Marijuana use at age 15 years was associated with increased risk of violent perpetration at age 19 years (odds ratio = 2.90, 95% CI: 1.08, 7.82; P < 0.05). Marijuana use at age 18 years was not associated with violent perpetration at age 19 years (odds ratio = 1.66, 95% CI: 0.56, 4.90; P > 0.05) Prior levels of violence involvement, age, sex, ethnicity, socioeconomic status Mixed 
Ellickson, 2000 (70Self-reported frequency of marijuana use in the past year Predatory violence, defined as use of force to obtain money or things from people, involvement in gang fights, attacking someone with intent to injure or kill, and carrying a concealed weapon Marijuana use was associated with increased odds of perpetration of predatory violence (odds ratio = 0.11; P < 0.05) Alcohol use, school bonds, family bonds, problem behavior, peer drug use, drug offers, self-esteem, rebelliousness, age, sex, race, parent education, neighborhood socioeconomic status, school drug-use prevalence Positive 
Friedman, 2001 (66Self-reported frequency of use of marijuana Self-reported conviction for assault Marijuana use had no association with assault conviction (P > 0.05) Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use No association 
  Self-reported conviction for weapons offenses Marijuana use was associated with increased likelihood of weapons offenses (partial correlation coefficient = 0.11; P < 0.05)  Positive 
  Self-reported conviction for attempted homicide Marijuana use was associated with increased likelihood of attempted homicide (partial correlation coefficient = 0.10; P < 0.05)  Positive 
  Self-reported conviction for homicide Marijuana use had no association with homicide conviction (P > 0.05)  No association 
  Self-reported gang drug-war fighting Marijuana use had no association with gang drug-war fighting (P > 0.05)  No association 
Green, 2010 (68Self-reported lifetime frequency of heavy marijuana use, defined as lifetime frequency of 20 or more times during adolescence Violent crime, defined as murder, assault, battery, and domestic violence and obtained from law enforcement records Heavy adolescent marijuana use had no association with violent crime (odds ratio = 1.09, 95% CI: 0.72, 1.65; P > 0.05) Sex, socioeconomic status, family background, school adaptation, school achievement, tobacco smoking, and delinquency No association 
Menard, 2001 (61Self-reported use of marijuana, in the last calendar year “Index violence,” defined as felony assault or robbery Marijuana use was associated with violent crime (risk ratio = 2.60; P < 0.05) None Positive 
Mulvey, 2006 (62Self-reported daily and weekly use of marijuana Self-reported daily and weekly frequency of 9 aggressive acts, including pushing, hitting, and using a weapon Prior-day marijuana use was associated with increased risk of violence (odds ratio = 1.6, 95% CI: 1.2, 2.0; P < 0.05) Hard drug (nonmarijuana) use and sales, alcohol use and sales Positive 
Pedersen, 2010 (73Self-reported frequency of cannabis use in past 12 months Serious crime including theft, robbery, and violence, as recorded by Statistics Norway Marijuana use 1–10 times in the past year was not associated with serious crime (odds ratio = 1.6, 95% CI: 0.6, 4.7; P > 0.05) or ≥11 times in the past year (odds ratio = 1.4, 95% CI: 0.4, 5.2; P > 0.05) Age, sex, parental cultural capital, parental monitoring, school grades, conduct problem, early adolescent marijuana use, cohabitation status, previous criminal charges, alcohol use, other illegal drug use No association 
Wei, 2004 (72Self-reported past-year frequency of use of marijuana, assessed by using the Substance Use Scale Violence, defined by using a delinquency scale including items assessing self-reported past-year frequency of gang fighting, strong-arming, attacking someone with a weapon or intent to seriously hurt or kill, and rape or forced sex Prior-year frequent marijuana use was associated with increased odds of violence among those aged 14 years (P < 0.05); 15 years (odds ratio = 3.07; P < 0.05); 16 years (odds ratio = 3.36; P < 0.05); 17 years (odds ratio = 2.67; P < 0.05); and 19 years (odds ratio = 3.83; P < 0.05). Prior-year frequent marijuana use was not associated with increased odds of violence among those aged 18 years or 20 years Academic achievement, depressed mood, hyperactivity/ impulsivity/inattention problems, poor communication with caretaker, poor supervision, caretaker perception of bad neighborhood, race/ethnicity Mixed 
Opiatesb 
Friedman, 2001 (66Self-reported frequency of use of opiates Self-reported conviction for assault Opiate use was associated with increased likelihood of assault conviction (partial correlation coefficient = 0.14; P < 0.05) Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use Positive 
  Self-reported conviction for weapons offenses Opiate use had no association with conviction for weapons offenses (P > 0.05)  No association 
  Self-reported conviction for attempted homicide Opiate use was associated with increased likelihood of attempted homicide conviction (partial correlation coefficient = 0.11; P < 0.05)  Positive 
  Self-reported conviction for homicide Opiate use had no association with homicide conviction (P > 0.05)  No association 
  Self-reported gang drug-war fighting Opiate use was associated with increased likelihood of gang drug-war fighting (partial correlation coefficient = 0.36; P < 0.05)  Positive 
PCP/Hallucinogens 
Friedman, 2001 (66Self-reported frequency of use of PCP/hallucinogens Self-reported conviction for assault PCP/hallucinogen use had no association with assault conviction (P > 0.05) Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use No association 
  Self-reported conviction for weapons offenses PCP/hallucinogen use had no association with conviction for weapons offenses (P > 0.05)  No association 
  Self-reported conviction for attempted homicide PCP/hallucinogen use had no association with attempted homicide conviction (P > 0.05)  No association 
  Self-reported conviction for homicide PCP/hallucinogen use had no association with attempted homicide conviction (P > 0.05)  No association 
  Self-reported gang drug-war fighting PCP/hallucinogen use had no association with gang drug-war fighting conviction (P > 0.05)  No association 
Tranquilizers 
Friedman, 2001 (66Self-reported frequency of use of tranquilizers Self-reported conviction for assault Tranquilizer use was associated with increased likelihood of assault conviction (partial correlation coefficient = 0.11; P < 0.05) Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use Positive 
  Self-reported conviction for weapons offenses Tranquilizer use had no association with conviction for weapons offenses (P > 0.05)  No association 
  Self-reported conviction for attempted homicide Tranquilizer use had no association with attempted homicide conviction (P > 0.05)  No association 
  Self-reported conviction for homicide Tranquilizer use had no association with homicide conviction (P > 0.05)  No association 
  Self-reported gang drug-war fighting Tranquilizer use was associated with decreased likelihood of gang drug-war fighting (partial correlation coefficient = −0.007; P < 0.0.05)  Negative 

Abbreviations: CI, confidence interval; HIV, human immunodeficiency virus; PCP, phencyclidine; SE, standard error.

a Studies in this category examined broad categories of controlled substance use, abuse, and addiction. Independent controlled substance variables in these studies included multiple specific substances.

b Including heroin and other opiates.

Table 4.

Summary of Results for Studies of the Economic Compulsive and Systemic Relationships Between Controlled Substances and Violence Perpetration, 1990–2014

First Author, Year (Reference No.) Controlled Substance Controlled Substance Measure Violence Measure Key Finding P Value Covariates Accounted for in Analysis Association Between Drugs and Violence 
Pathway 2 (Economic Compulsion) 
Weiner, 2005 (65Aggregate measures of controlled substance usea Measure of economic compulsion, defined as mean responses of 4 items that assessed past-year frequency of taking/selling things to pay for alcohol or drugs; doing personal favors to get/pay for alcohol/drugs; sold personal belongings to pay for alcohol/drugs; done illegal things other than selling drugs to pay for alcohol/drugs Violence perpetration scale comprising items assessing the annual frequency of incidents in which a weapon was used to injure or threaten someone; injury was perpetrated without a weapon; and property was damaged or stolen on purpose Measures of economic compulsion were not associated with violence perpetration >0.05 Sex, ethnicity, baseline level of violence perpetration No association 
Friedman, 2001 (66Amphetamines Self-reported frequency of use of amphetamines Self-reported conviction for robbery Amphetamine use was associated with increased likelihood of robbery conviction (partial correlation coefficient = 0.21) <0.05 Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use Positive 
Friedman, 2001 (66Barbiturates Self-reported frequency of use of barbiturates Self-reported conviction for robbery Barbiturate use was associated with decreased likelihood of robbery conviction (partial correlation coefficient = −0.12) <0.05 Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use Negative 
Friedman, 2001 (66Cocaine Self-reported frequency of use of cocaine Self-reported conviction for robbery Cocaine use had no association with robbery conviction >0.05 Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use No association 
Friedman, 2001 (66Marijuana Self-reported frequency of use of marijuana Self-reported conviction for robbery Marijuana use had no association with robbery conviction >0.05 Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use No association 
Friedman, 2001 (66Opiatesb Self-reported frequency of use of opiates Self-reported conviction for robbery Opiate use was associated with increased likelihood of robbery conviction (partial correlation coefficient = 0.13) <0.05 Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use Positive 
Friedman, 2001 (66PCP/hallucinogens Self-reported frequency of use of PCP/hallucinogens Self-reported conviction for robbery PCP/hallucinogen use had no association with robbery conviction >0.05 Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use No association 
Friedman, 2001 (66Tranquilizers Self-reported frequency of use of tranquilizers Self-reported conviction for robbery Tranquilizer use had no association with robbery conviction >0.05 Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use No association 
Pathway 3 (Systemic Violence) 
Bellair, 2009 (75Aggregate measures of sale of controlled substances Self-reported sale of any illicit drug in the past 12 months Violent acts, defined as self-report of attacking someone with intent to hurt them in the past year Selling drugs was associated with increased risk of violence among non-gang members (incidence rate ratio = 1.1; P < 0.05) and gang members (incidence rate ratio = 2.0) <0.05 Age, residential mobility, urban/suburban locale, school dropout, marital status, employment status, race, peer gang membership, and drug involvement Positive 
Menard, 2001 (61Aggregate measures of sale of controlled substances Self-reported sale of hard drugs, including hallucinogens, amphetamines, heroin, cocaine, and barbiturates in the past calendar year “Index violence,” defined as felony assault or robbery Selling hard drugs was associated with increased risk of violence perpetration (risk ratio = 5.71, 95% CI: 5.06, 9.76) <0.05 None Positive 
Cerdá, 2010 (74Cocaine market penetration Percentage of accidental deaths that had positive toxicology results for cocaine, per police precinct per year Gun-related homicides identified from New York City Office of the Chief Medical Examiner files Declines in rates of accidental deaths involving cocaine were associated with declines in rates of gun-related homicides for youth aged 15–24 years (posterior median, 0.26, 95% Bayesian CI: 0.07, 0.45) and adults aged ≥35 years (posterior median, 0.07, 95% Bayesian CI: 0.02, 0.12) but not for young adults aged 25–34 years (posterior median, −0.02, 95% Bayesian CI: −0.18, 0.13) <0.05 for ages 15–24 and ≥35 years; >0.05 for ages 25–34 years Misdemeanor/ordinance arrest rate, firearm availability, alcohol consumption, incarceration rate; complaint rate, percent of felony arrests per complaints, policy manpower, sex, race, ethnicity, place of birth (United States vs. foreign), employment status, poverty status, residential stability Mixed 
Dembo, 1990 (69Cocaine sales Self-reported illicit drug sales in the prior year Crimes against persons, defined as self-reported aggravated assault, gang fights, hit a teacher/student, committed sexual assault, strong-armed students/teachers/others Selling drugs was associated with increased likelihood of violence perpetration  None Positive 
Menard, 2001 (61Marijuana sales Self-reported sale of marijuana in the past calendar year “Index violence,” defined as felony assault or robbery Selling marijuana was associated with increased risk of violence perpetration (risk ratio = 4.69, 95% CI: 3.83, 7.46) <0.05 None Positive 
First Author, Year (Reference No.) Controlled Substance Controlled Substance Measure Violence Measure Key Finding P Value Covariates Accounted for in Analysis Association Between Drugs and Violence 
Pathway 2 (Economic Compulsion) 
Weiner, 2005 (65Aggregate measures of controlled substance usea Measure of economic compulsion, defined as mean responses of 4 items that assessed past-year frequency of taking/selling things to pay for alcohol or drugs; doing personal favors to get/pay for alcohol/drugs; sold personal belongings to pay for alcohol/drugs; done illegal things other than selling drugs to pay for alcohol/drugs Violence perpetration scale comprising items assessing the annual frequency of incidents in which a weapon was used to injure or threaten someone; injury was perpetrated without a weapon; and property was damaged or stolen on purpose Measures of economic compulsion were not associated with violence perpetration >0.05 Sex, ethnicity, baseline level of violence perpetration No association 
Friedman, 2001 (66Amphetamines Self-reported frequency of use of amphetamines Self-reported conviction for robbery Amphetamine use was associated with increased likelihood of robbery conviction (partial correlation coefficient = 0.21) <0.05 Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use Positive 
Friedman, 2001 (66Barbiturates Self-reported frequency of use of barbiturates Self-reported conviction for robbery Barbiturate use was associated with decreased likelihood of robbery conviction (partial correlation coefficient = −0.12) <0.05 Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use Negative 
Friedman, 2001 (66Cocaine Self-reported frequency of use of cocaine Self-reported conviction for robbery Cocaine use had no association with robbery conviction >0.05 Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use No association 
Friedman, 2001 (66Marijuana Self-reported frequency of use of marijuana Self-reported conviction for robbery Marijuana use had no association with robbery conviction >0.05 Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use No association 
Friedman, 2001 (66Opiatesb Self-reported frequency of use of opiates Self-reported conviction for robbery Opiate use was associated with increased likelihood of robbery conviction (partial correlation coefficient = 0.13) <0.05 Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use Positive 
Friedman, 2001 (66PCP/hallucinogens Self-reported frequency of use of PCP/hallucinogens Self-reported conviction for robbery PCP/hallucinogen use had no association with robbery conviction >0.05 Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use No association 
Friedman, 2001 (66Tranquilizers Self-reported frequency of use of tranquilizers Self-reported conviction for robbery Tranquilizer use had no association with robbery conviction >0.05 Age, income, welfare status, head of household occupation, academic performance, school behavior, association with delinquent peers, adjustment, attitudes toward deviance, conduct disorder, antisocial personality, family problems and alcohol, marijuana, barbiturate, tranquilizer, cocaine/crack, heroin, opiate, PCP/hallucinogen use No association 
Pathway 3 (Systemic Violence) 
Bellair, 2009 (75Aggregate measures of sale of controlled substances Self-reported sale of any illicit drug in the past 12 months Violent acts, defined as self-report of attacking someone with intent to hurt them in the past year Selling drugs was associated with increased risk of violence among non-gang members (incidence rate ratio = 1.1; P < 0.05) and gang members (incidence rate ratio = 2.0) <0.05 Age, residential mobility, urban/suburban locale, school dropout, marital status, employment status, race, peer gang membership, and drug involvement Positive 
Menard, 2001 (61Aggregate measures of sale of controlled substances Self-reported sale of hard drugs, including hallucinogens, amphetamines, heroin, cocaine, and barbiturates in the past calendar year “Index violence,” defined as felony assault or robbery Selling hard drugs was associated with increased risk of violence perpetration (risk ratio = 5.71, 95% CI: 5.06, 9.76) <0.05 None Positive 
Cerdá, 2010 (74Cocaine market penetration Percentage of accidental deaths that had positive toxicology results for cocaine, per police precinct per year Gun-related homicides identified from New York City Office of the Chief Medical Examiner files Declines in rates of accidental deaths involving cocaine were associated with declines in rates of gun-related homicides for youth aged 15–24 years (posterior median, 0.26, 95% Bayesian CI: 0.07, 0.45) and adults aged ≥35 years (posterior median, 0.07, 95% Bayesian CI: 0.02, 0.12) but not for young adults aged 25–34 years (posterior median, −0.02, 95% Bayesian CI: −0.18, 0.13) <0.05 for ages 15–24 and ≥35 years; >0.05 for ages 25–34 years Misdemeanor/ordinance arrest rate, firearm availability, alcohol consumption, incarceration rate; complaint rate, percent of felony arrests per complaints, policy manpower, sex, race, ethnicity, place of birth (United States vs. foreign), employment status, poverty status, residential stability Mixed 
Dembo, 1990 (69Cocaine sales Self-reported illicit drug sales in the prior year Crimes against persons, defined as self-reported aggravated assault, gang fights, hit a teacher/student, committed sexual assault, strong-armed students/teachers/others Selling drugs was associated with increased likelihood of violence perpetration  None Positive 
Menard, 2001 (61Marijuana sales Self-reported sale of marijuana in the past calendar year “Index violence,” defined as felony assault or robbery Selling marijuana was associated with increased risk of violence perpetration (risk ratio = 4.69, 95% CI: 3.83, 7.46) <0.05 None Positive 

Abbreviations: CI, confidence interval; PCP, phencylidine.

a Studies in this category examined broad categories of controlled substance use, abuse, and addiction. Independent controlled substance variables in these studies included multiple specific substances.

b Including heroin and other opiates.

Table 5.

Summary of the Longitudinal Relationship Between Controlled Substance Use and Suicide, 1990–2014

First Author, Year (Reference No.) Controlled Substance Controlled Substance Measure Violence Measure Key Finding P Value Covariates Accounted for in Analysis Association Between Drugs and Suicide 
Allebeck, 1990 (77Aggregate measures of controlled substance usea Drug dependence, defined on the basis of structured diagnostic interview conducted at time of conscription, or any primary or secondary inpatient drug-related diagnosis Suicide, as recorded in national cause-of-death registry Drug dependence was associated with increased risk of suicide when using 2 different measures of drug dependence: diagnosis at time of military conscription (27.1 suicides per 1,000 among individuals with drug dependence) and inpatient diagnosis (46.2 suicides per 1,000) 0.001 for time of conscription;
<0.05 for inpatient diagnosis 
None Positive 
Nilsson, 2014 (78Aggregate measures of controlled substance usea ICD-9 diagnoses of drug use disorders from the Psychiatric Central Register Suicide, as recorded in the national cause-of-death registry Men with drug use disorders were at increased risk of suicide (hazard ratio = 2.7, 95% CI: 1.8, 4.1). Women with drug use disorders were at increased risk of suicide (hazard ratio = 8.3, 95% CI: 2.3, 29.4) <0.05 Sex, age at first homeless shelter contact, country of origin, main source of income, schizophrenia spectrum disorders, alcohol use disorders, affective disorders, personality disorders Positive (both men and women) 
Yen, 2003 (76Aggregate measures of controlled substance usea Drug use disorder at baseline, defined by using the SCID-1/P semistructured interview Self-reported number of suicidal attempts Drug use disorders were associated with increased risk of suicide attempts (risk ratio = 2.11, 95% CI: 1.43, 3.12) <0.05 Sex, borderline personality disorder, past self-injurious behavior Positive 
Nilsson, 2014 (78Cocaine ICD-9 diagnoses of cocaine use disorders from the Psychiatric Central Register Suicide, as recorded in the national cause-of-death registry Cocaine use disorders had no association with suicide among men (hazard ratio = 1.0, 95% CI: 0.3, 3.4). Cocaine use disorders had no association with suicide among women (no suicides among individuals with this condition) >0.05 Sex, age at first homeless shelter contact, country of origin, main source of income, schizophrenia spectrum disorders, alcohol use disorders, affective disorders, personality disorders No association (both men and women) 
Petronis, 1990 (79Cocaine Self-reported current use of cocaine in year prior to suicide attempt, obtained by using the Diagnostic Interview Schedule Self-reported suicide attempt, obtained using the Diagnostic Interview Schedule Past-year cocaine use was associated with increased odds of suicide attempt (relative odds = 61.9, 95% CI: 2.51, 1527.8) <0.05 Sex, race/ethnicity, education, employment, marital status, major depression, manic episode, alcohol use, other drug use Positive 
Nilsson, 2014 (78Marijuana ICD-9 diagnoses of cannabis use disorders from the Psychiatric Central Register Suicide, as recorded in the national cause-of-death registry Cannabis use disorders had no association with suicide among men (hazard ratio = 1.3, 95% CI: 0.8, 2.2). Cannabis use disorders had no association with suicide among women (hazard ratio = 0.9, 95% CI: 0.3, 2.6) >0.05 Sex, age at first homeless shelter contact, country of origin, main source of income, schizophrenia spectrum disorders, alcohol use disorders, affective disorders, personality disorders No association (both men and women) 
Petronis, 1990 (79Marijuana Self-reported current use of marijuana in year prior to suicide attempt, obtained by using the Diagnostic Interview Schedule Self-reported suicide attempt, obtained using the Diagnostic Interview Schedule Past-year marijuana use had no association with suicide attempts >0.05 Sex, race/ethnicity, education, employment, marital status, major depression, manic episode, alcohol use, other drug use No association 
Nilsson, 2014 (78Opioidsb ICD-9 diagnoses of opioid use disorders from the Psychiatric Central Register Suicide, as recorded in the national cause-of-death registry Men with opioid use disorders were at increased risk of suicide (hazard ratio = 2.0, 95% CI: 1.2, 3.4). Opioid use disorders had no association with suicide among women (hazard ratio = 0.8, 95% CI: 0.2, 2.4) <0.05 (men); >0.05 (women) Sex, age at first homeless shelter contact, country of origin, main source of income, schizophrenia spectrum disorders, alcohol use disorders, affective disorders, personality disorders Positive (men); no association (women) 
Allgulander, 1992 (80Prescription drugs Inpatient diagnosis “prescription drug addiction” (not defined) Suicide, as recorded in national cause-of-death registry Addiction to prescription drugs was associated with increased risk of suicide (relative risk = 2.38, 95% CI: 1.61, 3.51). <0.05 Sex and age Positive 
Nilsson, 2014 (78Sedatives ICD-9 diagnoses of sedative use disorders from the Psychiatric Central Register Suicide, as recorded in the national cause-of-death registry Sedative use disorders had no association with suicide among men (hazard ratio = 1.5, 95% CI: 0.8, 2.9). Sedative use disorders had no association with suicide among women (hazard ratio = 1.3, 95% CI: 0.5, 3.3) >0.05 Sex, age at first homeless shelter contact, country of origin, main source of income, schizophrenia spectrum disorders, alcohol use disorders, affective disorders, personality disorders No association (both men and women) 
Petronis, 1990 (79Sedatives Self-reported current use of sedative hypnotics in year prior to suicide attempt, obtained by using the Diagnostic Interview Schedule Self-reported suicide attempt, obtained by using the Diagnostic Interview Schedule Past-year use of sedatives/hypnotics had no association with suicide attempts >0.05 Sex, race/ethnicity, education, employment, marital status, major depression, manic episode, alcohol use, other drug use No association 
Allgulander, 1992 (80“Street drugs” (not defined) Inpatient diagnosis of “street drug addiction” (not defined) Suicide, as recorded in national cause-of-death registry Addiction to “street drugs” was not associated with increased risk of suicide (relative risk = 1.02, 95% CI: 0.72, 1.45) >0.05 Sex and age No association 
First Author, Year (Reference No.) Controlled Substance Controlled Substance Measure Violence Measure Key Finding P Value Covariates Accounted for in Analysis Association Between Drugs and Suicide 
Allebeck, 1990 (77Aggregate measures of controlled substance usea Drug dependence, defined on the basis of structured diagnostic interview conducted at time of conscription, or any primary or secondary inpatient drug-related diagnosis Suicide, as recorded in national cause-of-death registry Drug dependence was associated with increased risk of suicide when using 2 different measures of drug dependence: diagnosis at time of military conscription (27.1 suicides per 1,000 among individuals with drug dependence) and inpatient diagnosis (46.2 suicides per 1,000) 0.001 for time of conscription;
<0.05 for inpatient diagnosis 
None Positive 
Nilsson, 2014 (78Aggregate measures of controlled substance usea ICD-9 diagnoses of drug use disorders from the Psychiatric Central Register Suicide, as recorded in the national cause-of-death registry Men with drug use disorders were at increased risk of suicide (hazard ratio = 2.7, 95% CI: 1.8, 4.1). Women with drug use disorders were at increased risk of suicide (hazard ratio = 8.3, 95% CI: 2.3, 29.4) <0.05 Sex, age at first homeless shelter contact, country of origin, main source of income, schizophrenia spectrum disorders, alcohol use disorders, affective disorders, personality disorders Positive (both men and women) 
Yen, 2003 (76Aggregate measures of controlled substance usea Drug use disorder at baseline, defined by using the SCID-1/P semistructured interview Self-reported number of suicidal attempts Drug use disorders were associated with increased risk of suicide attempts (risk ratio = 2.11, 95% CI: 1.43, 3.12) <0.05 Sex, borderline personality disorder, past self-injurious behavior Positive 
Nilsson, 2014 (78Cocaine ICD-9 diagnoses of cocaine use disorders from the Psychiatric Central Register Suicide, as recorded in the national cause-of-death registry Cocaine use disorders had no association with suicide among men (hazard ratio = 1.0, 95% CI: 0.3, 3.4). Cocaine use disorders had no association with suicide among women (no suicides among individuals with this condition) >0.05 Sex, age at first homeless shelter contact, country of origin, main source of income, schizophrenia spectrum disorders, alcohol use disorders, affective disorders, personality disorders No association (both men and women) 
Petronis, 1990 (79Cocaine Self-reported current use of cocaine in year prior to suicide attempt, obtained by using the Diagnostic Interview Schedule Self-reported suicide attempt, obtained using the Diagnostic Interview Schedule Past-year cocaine use was associated with increased odds of suicide attempt (relative odds = 61.9, 95% CI: 2.51, 1527.8) <0.05 Sex, race/ethnicity, education, employment, marital status, major depression, manic episode, alcohol use, other drug use Positive 
Nilsson, 2014 (78Marijuana ICD-9 diagnoses of cannabis use disorders from the Psychiatric Central Register Suicide, as recorded in the national cause-of-death registry Cannabis use disorders had no association with suicide among men (hazard ratio = 1.3, 95% CI: 0.8, 2.2). Cannabis use disorders had no association with suicide among women (hazard ratio = 0.9, 95% CI: 0.3, 2.6) >0.05 Sex, age at first homeless shelter contact, country of origin, main source of income, schizophrenia spectrum disorders, alcohol use disorders, affective disorders, personality disorders No association (both men and women) 
Petronis, 1990 (79Marijuana Self-reported current use of marijuana in year prior to suicide attempt, obtained by using the Diagnostic Interview Schedule Self-reported suicide attempt, obtained using the Diagnostic Interview Schedule Past-year marijuana use had no association with suicide attempts >0.05 Sex, race/ethnicity, education, employment, marital status, major depression, manic episode, alcohol use, other drug use No association 
Nilsson, 2014 (78Opioidsb ICD-9 diagnoses of opioid use disorders from the Psychiatric Central Register Suicide, as recorded in the national cause-of-death registry Men with opioid use disorders were at increased risk of suicide (hazard ratio = 2.0, 95% CI: 1.2, 3.4). Opioid use disorders had no association with suicide among women (hazard ratio = 0.8, 95% CI: 0.2, 2.4) <0.05 (men); >0.05 (women) Sex, age at first homeless shelter contact, country of origin, main source of income, schizophrenia spectrum disorders, alcohol use disorders, affective disorders, personality disorders Positive (men); no association (women) 
Allgulander, 1992 (80Prescription drugs Inpatient diagnosis “prescription drug addiction” (not defined) Suicide, as recorded in national cause-of-death registry Addiction to prescription drugs was associated with increased risk of suicide (relative risk = 2.38, 95% CI: 1.61, 3.51). <0.05 Sex and age Positive 
Nilsson, 2014 (78Sedatives ICD-9 diagnoses of sedative use disorders from the Psychiatric Central Register Suicide, as recorded in the national cause-of-death registry Sedative use disorders had no association with suicide among men (hazard ratio = 1.5, 95% CI: 0.8, 2.9). Sedative use disorders had no association with suicide among women (hazard ratio = 1.3, 95% CI: 0.5, 3.3) >0.05 Sex, age at first homeless shelter contact, country of origin, main source of income, schizophrenia spectrum disorders, alcohol use disorders, affective disorders, personality disorders No association (both men and women) 
Petronis, 1990 (79Sedatives Self-reported current use of sedative hypnotics in year prior to suicide attempt, obtained by using the Diagnostic Interview Schedule Self-reported suicide attempt, obtained by using the Diagnostic Interview Schedule Past-year use of sedatives/hypnotics had no association with suicide attempts >0.05 Sex, race/ethnicity, education, employment, marital status, major depression, manic episode, alcohol use, other drug use No association 
Allgulander, 1992 (80“Street drugs” (not defined) Inpatient diagnosis of “street drug addiction” (not defined) Suicide, as recorded in national cause-of-death registry Addiction to “street drugs” was not associated with increased risk of suicide (relative risk = 1.02, 95% CI: 0.72, 1.45) >0.05 Sex and age No association 

Abbreviations: CI: confidence interval; ICD-9, International Classification of Diseases, Ninth Revision; SCID-I/P, Structured Clinical Interview for DSM-IV Axis I Disorders (Patient Edition).

a Studies in this category examined broad categories of controlled substance use, abuse, and addiction. Independent controlled substance variables in these studies included multiple specific substances.

b Including heroin and other opiates.

RESULTS

Study identification

The initial electronic search returned 2,921 unique articles. An additional 9 studies were identified through searches of the reference lists of review articles. Title and abstract screening identified 217 potentially relevant studies. After full-article review, 195 of these were excluded because of failure to meet the review's inclusion criteria, for a final sample of 22 articles (Figure 2). Of these, 17 measured violence perpetration outcomes and 5 measured suicide outcomes (Table 2). Across all studies, controlled substance measures included aggregate measures of any controlled substance use, use of individual drugs or drug classes (amphetamine/methamphetamine, barbiturates, cocaine, marijuana, opiates, phencyclidine/hallucinogens, tranquilizers, prescription drugs, sedatives), sale of any illicit drug, sale of “hard drugs” excluding marijuana, and sale of marijuana. No studies meeting our inclusion criteria examined heroin use. Studies measured a range of violence outcomes, including index measures of violent behavior, assault, weapons offenses, homicide, and suicide.

Figure 2.

Flowchart explaining the identification of studies included in this review.

Figure 2.

Flowchart explaining the identification of studies included in this review.

Study quality

No studies were determined to have low risk of bias, a classification Cochrane defines as comparable to a well-conducted randomized controlled trial (57). Eleven of the 22 studies had moderate risk of bias, and 11 studies had serious risk of bias (Table 2). No studies were scored as having critical risk. The most frequent threats to validity were failure to measure key confounders and selection into the study population based on drug use or violence.

Interpersonal violence

The evidence on the relationship between measures of controlled substance use and interpersonal violence was mixed, with the highest quality studies that controlled for concurrent alcohol use tending to show no association. The evidence on the economic compulsive pathway was too limited to draw conclusions, and involvement with illegal drug sales (Goldstein et al.'s systemic pathway) was consistently associated with interpersonal violence.

The psychopharmacological pathway between controlled substances and violence perpetration

Aggregate measures

We extracted 7 such measures from 7 studies (5965) (Table 3). Three studies showed a positive relationship between controlled substance use and interpersonal violence (59, 63, 65) and 4 showed no association (6062, 64). Only 1 study measured a close temporal relationship between drug use and interpersonal violence; controlling for concurrent alcohol use, Mulvey et al. (62) found no association between prior-day use of controlled substances and subsequent commission of violent acts. All other studies in this category measured controlled substance use 1 year or more prior to measuring violence.

The 3 studies showing a positive association between aggregate measures of controlled substance use and interpersonal violence had study periods of 5 years or longer. Brook et al. (59) measured controlled substance use and violent behavior in a cohort of New York City youth in 1990, 1994, 1999, and 2004. Self-reported past-year illicit drug use in 1994 was associated with increased likelihood of self-reported violent behavior in 1999 and 2004. In 2 studies of California continuation high school students, Sussman et al. (63) and Weiner et al. (65) found a positive relationship between drug use and subsequent violence perpetration. The 3 studies showing a positive association all had serious threat of bias and did not control for concurrent alcohol use.

Three of the 4 studies showing no association used nationally representative US samples. Using the National Epidemiologic Survey of Alcohol and Related Conditions, Van Dorn et al. (64) found no difference in risk of interpersonal violence in 2004–2005 among persons with versus without a drug use disorder in 2001–2002. In a study by Kuhns (60) that used the first 2 waves of the National Youth Survey, a representative sample of US youth aged 11–17 years, controlled substance use was not associated with serious violence in 1976–1977 or 1977–1978. This study controlled for multiple factors, including alcohol use, delinquent peer group, and neighborhood problems (60). A study by Menard and Mihalic (61) using the first 9 waves of the same survey also showed no association between controlled substance use and interpersonal violence. A 26-week study of individuals identified as at high risk of violent offending in an urban US psychiatric hospital also found no association between controlled substance use and interpersonal violence (62). This null finding may be due to the fact that everyone in the study population, including controlled substance users and nonusers, was at high risk of violence. Of the 4 studies that showed no association, 2 had moderate threat of bias (60, 64), 2 had serious threat (61, 62), and 3 controlled for alcohol use (60, 62, 64).

Amphetamines and methamphetamines

Two studies examined the relationships between amphetamine or methamphetamine use and interpersonal violence. In a medium-bias study of inner-city African-American adults living in Philadelphia, Friedman et al. (66) found that amphetamine use was associated with self-reported gang drug-war fighting but not self-reported conviction for assault, weapons offenses (i.e., illicit carrying), attempted homicide, or homicide, potentially because of the fact that fighting is much more common than these other offenses. Although the exact timing of the drug use and violence outcomes was unspecified, the 2 measures occurred at some point over a 2.5-year period and thus appear to assess a distal relationship between substance use and the outcomes of interest. This analysis controlled for a range of key covariates, including alcohol and other drug use, conduct and antisocial personality disorder, and socioeconomic status (66). In a serious-bias study, McKetin et al. (67) examined the relationship between past-month methamphetamine use and violence perpetration in an Australian cohort from 2006 to 2010 and found that methamphetamine use was associated with increased odds of violent behavior, controlling for psychotic symptoms, use of other drugs and alcohol, and sociodemographic factors.

Barbiturates

Only 1 study measured the association between barbiturate use and interpersonal violence. In the study described above, Friedman et al. found that barbiturate use was associated with decreased likelihood of self-reported conviction for assault and attempted homicide, as well as self-reported gang drug-war fighting, and had no association with conviction for weapons offenses or homicide (66, 68). This study had a moderate risk of bias.

Cocaine

Two studies examined the relationship between cocaine use and interpersonal violence, with mixed results. In the moderate-bias study of inner-city African-American adults living in Philadelphia by Friedman et al., use of powdered or crack cocaine was associated with increased likelihood of conviction for homicide and increased self-reported gang drug-war fighting (66). In a 2-year study of youth aged 10–18 years admitted to a regional detention center in Florida, Dembo et al. (69) found that cocaine use was not associated with crimes against persons. Their study was rated as having a serious risk of bias.

Marijuana

Marijuana was examined most frequently, and study results were mixed. We extracted 12 measures of association from 8 studies. Five measures showed a positive relationship between marijuana use and interpersonal violence (61, 62, 66, 70), 2 showed mixed results (71, 72), and 5 showed no association (66, 68, 73). Three of the 5 measures showing a positive relationship (66, 70) and 4 of the 5 measures showing no relationship came from studies with moderate risk of bias (66, 68, 73). All remaining measures came from studies with serious risk of bias. Only 1 study, the paper by Mulvey et al. (62) described previously, measured the proximal relationship between prior-day controlled substance use and violence. All the other studies measured distal relationships between drug use in a given year and interpersonal violence in subsequent years.

Using 9 waves of the National Youth Survey, Menard and Mihalic (61) found that self-reported past-year use of marijuana was associated with felony-level violence, but this analysis did not control for any potentially confounding factors. In the Philadelphia, Pennsylvania, study by Friedman et al. (66), marijuana use was associated with increased likelihood of self-reported convictions for weapons offenses and attempted homicide. In a 5-year study of California middle and high school students, Ellickson and McGuigan (70) found that self-reported past-year marijuana use was associated with increased odds of perpetration of predatory violence. In a 26-week study of individuals identified as being at high risk for involvement in violence in an urban psychiatric hospital, Mulvey et al. (62) found that prior-day marijuana use was associated with increased odds of violence perpetration.

Two studies showed mixed results: In a study of insured youth aged 12–15 years in 1996 who were followed up in 1999 and 2000, Brady et al. (71) concluded that past-year marijuana use at age 15 years, but not at age 18 years, was associated with increased odds of violence perpetration at age 19 years. In a 10-year cohort study of Pittsburgh, Pennsylvania, youth with antisocial behavior, Wei et al. (72) found that past-year frequent marijuana use was associated with increased violence among youth aged 14–17 years and 19 years but not among those aged 18 or 20 years. Critically, this study also found that past-year violence predicted subsequent marijuana use, suggesting possible issues with reverse causation.

In a study of the Woodlawn cohort, a group of urban African Americans from Chicago, Illinois, followed from ages 6 to 42 years, no relationship between heavy adolescent marijuana use and violent crime was observed (68). In the inner-city Philadelphia study by Friedman et al. (66), the authors found no association between marijuana use and conviction for assault, homicide, or self-reported gang drug-war fighting. In a cohort study of Norwegian adolescents, neither marijuana use 1–10 times nor ≥11 times in the past year was associated with perpetration of serious crime (73).

Opiates

One moderate-bias study examined the relationship between opiate use (including heroin and other opiates, such as opioid analgesics) and interpersonal violence. In Philadelphia, Friedman et al. (66) found that opiate use was positively associated with self-reported conviction for assault and homicide and involvement in gang drug-war fighting, but not with conviction for weapons offenses or attempted homicide.

Phencyclidine/hallucinogens

The study by Friedman et al. (66) was the only one to examine the relationship between phencyclidine or other hallucinogen use and interpersonal violence. It found no association with any of the violence outcomes examined in the study.

Tranquilizers

The Friedman et al. study (66) was again the only source of relevant results. Self-reported tranquilizer use was associated with increased likelihood of assault conviction and decreased likelihood of self-reported gang drug-war fighting.

The economic compulsive pathway between controlled substances and violence perpetration

Prospective longitudinal studies of the economic compulsive pathway were limited and their results mixed (Table 4). In a 5-year study of students from alternative high schools in California, Weiner et al. (65) found no association between economic compulsion and violence perpetration, controlling for sex, ethnicity, and baseline violence levels. This was the only study to use an index measure of economic compulsion with face validity (65), but it was rated as having serious risk of bias. In their moderate-bias study of inner-city African-American males, Friedman et al. (66) examined the relationship between use of 7 controlled substances and self-reported conviction for robbery. They found a positive association between robbery and use of amphetamines or opiates, a negative association with barbiturate use, and no association with use of cocaine, marijuana, phencyclidine/hallucinogens, or tranquilizers. However, it was unknown whether the robbery was motivated by a need for money to obtain controlled substances.

The systemic pathway between controlled substances and violence perpetration

Four studies examined the systemic pathway between controlled substances and violence toward others. Study results were suggestive of a positive relationship. In a study of firearm homicides in New York City from 1990 to 1999, Cerda et al. (74) examined the relationship between the rate of accidental deaths with positive toxicology results for cocaine (which they conceptualize as a proxy measure for crack cocaine market penetration) and firearm homicide. The authors assessed this relationship using a 1-year lag and found that decreases in cocaine-related accidental deaths were associated with fewer firearm homicides among youth aged 15–24 and adults aged ≥35 years, but not among young adults aged 25–34 years. Using the first 5 waves of the National Longitudinal Survey of Youth (1997–2001), Bellair and McNulty (75) found that self-reported sale of any controlled substance in the past 12-months was associated with increased risk of violence among both gang members and non-gang members, controlling for socioeconomic status, race, peer gang membership, and peer drug involvement. In Menard and Mihalic's National Youth Survey study described previously (61), they found that selling either hard drugs or marijuana in the past year was associated with increased risk of committing felony-level violence. In a study of Florida youth admitted to a regional detention center in 1986–1987, Dembo et al. (69) found that past-year illicit drug sales were significantly associated with subsequent commission of violent crimes against persons. Only the study by Bellair and McNulty (75) controlled for any confounding factors. The studies by Cerda et al. (74) and Bellair and McNulty (75) were rated as having moderate risk of bias, and those by Menard and Mihalic (61) and Dembo et al. (69) had serious risk.

Controlled substance use and suicide

We identified 5 prospective longitudinal studies measuring the distal relationship between controlled substance use and suicide. Studies examining aggregate measures of controlled substance use showed a positive association, while studies examining the relationship between use of specific substances and suicide showed mixed results (Table 5).

Aggregate measures

Three studies examined the relationship between aggregate measures of controlled substance use and suicide. Only 1 study was conducted in a US population: In a 2-year study of a cohort of adults with personality disorders, Yen et al. (76) found that drug use disorders were associated with increased risk of suicide attempts, controlling for sex, borderline personality disorder, and past self-injurious behavior. In a study of Swedish men conscripted for military training in 1969–1970 and subsequently followed through 1983, Allebeck and Allgulander (77) found a positive association between drug dependence and suicide but did not control for any other risk factors. In a study of Danish individuals with at least 1 contact with a homeless shelter between 1999 and 2008, Nilsson et al. (78) found that any drug use disorder diagnosis was positively associated with risk of suicide among both men and women. Their analysis controlled for sex, age, source of income, country of origin, and mental health and alcohol use disorders (78). Risk of bias was moderate for the study by Nillson et al. (78) and serious for the studies by Yen et al. (76) and Allebeck and Allgulander (77).

Cocaine

Two studies assessed the relationship between cocaine use and suicide. Using 1984 and 1985 data from the Epidemiologic Catchment Area study, which collected data from individuals in 5 US cities, Petronis et al. (79) found no association between self-reported past-year use of cocaine and suicide attempts, controlling for socioeconomic factors, mental illness, and alcohol and other drug use. This study had moderate risk of bias. In the Danish homeless shelter study by Nilsson et al. (78) described above, cocaine use disorders had no association with suicide among either men or women.

Marijuana

The studies by Nilsson et al. (78) and Petronis et al. (79) also measured the association between marijuana use and suicide. Nilsson et al. found no relationship between cannabis use disorder diagnoses and suicide among either men or women, and Petronis et al. found no association between self-reported past-year marijuana use and suicide attempts.

Opioids

No US studies assessed the relationship between opioid use and suicide. In their Danish study, Nilsson et al. (78) found a positive association between opioid use disorder and suicide among men and no relationship among women.

Prescription drugs

In a study of Swedish individuals with at least 1 inpatient psychiatric diagnosis between 1973 and 1987, Allgulander et al. (80) found a positive association between prescription drug addiction and suicide, controlling for sex and age. They did not define the prescription drug addiction measure. This study had moderate risk of bias.

Sedatives

The studies by Nilsson et al. (78) and Petronis et al. (79) described previously examined the relationship between sedatives and suicide. Nilsson et al. found no association in their Danish study, and Petronis et al. found no association between self-reported past-year use of sedatives and suicide attempts in their Epidemiologic Catchment Area study sample

Street drugs

In the study of the Swedish cohort with inpatient psychiatric diagnoses, Allgulander et al. (80) found no association between street drug addiction and suicide. The street drug addiction measure was not defined.

DISCUSSION

Overview of findings

The results of our review do not support a straightforward positive relationship between controlled substance use and interpersonal violence. The best available, moderate-bias studies that controlled for key risk factors such as alcohol use and low socioeconomic status found no association between aggregate measures of controlled substance use and perpetration of violence toward others (60, 64). Studies suggested a positive association between distal use of any controlled substance and suicide, though results were mixed for specific substances and the number and quality of studies were low. The limited longitudinal prospective studies examining the systemic pathway of Goldstein et al. suggested that selling illicit drugs was consistently associated with perpetration of violence toward others. The evidence on the economic compulsive pathway between controlled substances and violence perpetration was too limited to suggest an overall trend in results.

Demand for and access to firearms, as well as base rates of violence overall and firearm violence specifically, likely varied across the study populations reviewed. The majority of studies did not report the proportion of total violent incidents that involved firearms or measure firearm access/demand. In studies of interpersonal violence that used nationally or regionally representative samples, rates of firearm violence were likely low as these studies exclude many of the individuals at highest risk for firearm violence, for example, those who are incarcerated, homeless, or (in the case of youth studies) not in school. Because of low base rates of violence in these groups, such studies tended to use broad definitions of interpersonal violence that included acts such as pushing/shoving and threatening someone without a weapon in addition to more serious firearm-related offenses (60, 61, 63, 64, 73). In contrast, other study samples were composed of individuals from groups, such as juvenile offenders (69), gang members (75), and disadvantaged inner-city populations (59, 66, 68), shown by prior research to have elevated rates of illicit firearm behaviors (4, 17, 18, 20, 22). Likely due to the higher base rate of seriously violent acts, studies of these groups tended to use narrower definitions of serious violence that included acts such as threatening or injuring with a weapon, aggravated assault, and homicide. These outcome metrics are relatively likely to include firearm-related offenses. In the United States, an estimated 68% of homicides, 41% of robberies, and 21% of assaults involve firearms (81).

For interpersonal violence, 5 of the best available, moderate-bias studies stand out as being most directly relevant to firearm violence. Three of these 5 studies examine the psychopharmacological pathway and 2 assess the systemic pathway. On the psychopharmacological pathway, Friedman et al. (66) examined the association between use of 7 individual types of drugs and 5 violence outcomes, including assault, weapons offenses, attempted homicide, homicide, and gang drug-war fighting in a sample of low-income African- American adults living in inner-city Philadelphia in the 1990s. They found that cocaine use was associated with increased risk of homicide, and use of marijuana and opiates was positively associated with attempted homicide. Opiate and tranquilizer use increased the risk of assault, and amphetamine/methamphetamine, cocaine, and opiate use were positively associated with gang drug-war fighting. Ellickson and McGuigan (70) found a positive association between middle-school marijuana use and predatory violence, which included attacking someone with intent to injure or kill, carrying a concealed weapon, and involvement in gang fights, in a large longitudinal sample of high school students and dropouts. In the Woodlawn cohort of urban African Americans from Chicago, Green et al. (68) found no association between heavy adolescent marijuana use and future violent crime, defined as murder, assault, battery, and domestic violence.

On the systemic pathway, Bellair and McNulty (75) found a positive relationship between sale of any illicit drug in the past 12 months and attacking someone with the intent to hurt them among the subset of youth from a nationally representative survey who self-identified as gang members. In the only firearm-specific study meeting our review's inclusion criteria, Cerda et al. (74) concluded that the increases in cocaine market penetration in New York City in the 1990s were positively associated with firearm homicides among youth aged 15–24 years and adults aged ≥35 years. With the exception of the Woodlawn study, this subset of the best-available and most firearm-relevant studies is suggestive of a positive relationship between drug use/sales and interpersonal violence, although this pattern should be interpreted cautiously in light of the limitations discussed in more detail below. The Woodlawn study measured the association between marijuana use and interpersonal violence over an extended period of time (15 years), which may account at least in part for that study's discrepant finding. The other studies in this group measured associations over periods of 5 years or less.

Although toxicology studies show that a high proportion of suicides involve controlled substances (29), the prospective longitudinal literature on the distal relationship between illicit drugs and suicide was limited. The 3 studies examining the association between aggregate measures of controlled substance use and suicide all showed a positive association, but 2 had serious risk of bias (76, 77), and the highest quality study in this group was conducted by using a Danish cohort (78). The other 2 moderate-bias studies suggested a positive association between use of cocaine, but not marijuana or sedatives, and attempted suicide in the 5-city Epidemiologic Catchment Area study sample (79) and a positive association between prescription drugs, but not “street drugs,” and suicide in a cohort of Swedish psychiatric patients (80). Three of the 5 total studies used cohorts from Scandinavian countries (77, 78, 80), where rates of firearm suicide are low (82), and the 2 US studies examined nonfatal suicide attempts (76, 79). Because of the extreme lethality of firearms, only 15% of nonfatal suicide attempts involve firearms (83). Although the studies reviewed likely did not include high rates of firearm suicide, about 50% of suicides in the United States each year involve firearms, and a robust body of literature shows that the contributor to differential rates of firearm versus nonfirearm suicide in the United States and other developed nations is firearm access (84). The results of this review suggest that distal use of controlled substances may contribute to increased risk of future suicide. Additional US-based, firearm-specific epidemiologic studies to further test and explain this relationship are needed.

Limitations of the evidence

Across all the categories of studies reviewed, the prospective longitudinal literature on the relationship between controlled substances and violence faces 4 key limitations that warrant consideration when interpreting the results of the review. The first limitation is lack of firearm-specific studies, as discussed in detail above. The second, temporality issues related to long periods, often a year or more, between independent measures of controlled substance involvement and dependent violence outcomes violate a key criterion for the establishment of a causal relationship. This extended duration between the exposure and outcome of interest provides ample opportunity for changes in study participants' involvement with controlled substances and the introduction of other factors that may confound the measured association between drug use and violence. This temporality issue is present in the majority of studies of all 3 interpersonal violence pathways, as well as the 5 studies of the relationship between distal controlled substance use and suicide. The one notable exception was the study by Mulvey et al. (62) that found a heightened risk of interpersonal violence associated with prior-day marijuana use but not prior-day use of hard (nonmarijuana) drugs.

Importantly, this temporality issue suggests that the majority of studies examining the association between controlled substance use and interpersonal violence did not measure Goldstein et al.'s psychopharmacological pathway per se. Although it is possible that the acute or chronic physical and psychological effects of drugs led directly to violence in some instances, we cannot reasonably conclude that this was always the case. An alternate explanation is that shared risk factors (e.g., alcohol use, delinquent peer group, history of aggressive behavior, low socioeconomic status) account for much of the observed relationship between drug use and interpersonal violence. The same issue exists in the studies of the distal relationship between controlled substance use and suicide, where factors such as alcohol use, mental illness, and stressful life circumstances are likely involved. The third key limitation of the existing evidence is that few studies were able to measure and control for these risk factors, and the fourth and related limitation is that few studies were able to accurately measure and account for the time-varying nature of the controlled substance, violence, and covariate variables measured. Many of the studies reviewed used measures of any self-reported substance use in the past year as the independent variable of interest. This type of measure does not capture frequency, duration, or volume of use and, as a result, prevents accurate assessment of a dose-response relationship.

The idea that unmeasured or inaccurately measured variables account for some of the observed positive relationships between drug use and violence in our review is supported by the physiological research (38). Although prior research suggests that direct pharmacological links exist between amphetamine/methamphetamine (38, 52, 85, 86) or cocaine use (38, 44) and perpetration of violence (and limited evidence suggests a similar relationship for the United States (38)), other substances including hallucinogens, opiates, and marijuana do not appear likely to trigger violent behavior directly (38, 39). In the face of this physiological evidence, it seems unlikely that the reviewed studies showing a positive relationship between marijuana use and interpersonal violence, for example, are measuring the direct physiological effects of the drug on violent behavior.

Implications for firearm policy

The current federal firearm prohibition related to controlled substances is inadequate. The law does not specify the criteria by which individuals who are “unlawful users of or addicted to any controlled substance” should be identified (10). The Code of Federal Regulations includes a 273-word definition, which includes time-sensitive elements that must be reassessed each time eligibility is determined: “[a]n inference of current use may be drawn from evidence of a recent use or possession of a controlled substance or a pattern of use or possession that reasonably covers the present time” (10). This complexity makes it difficult to enforce, at best. As a result, in practice few individuals are subject to this prohibition. As of January 31, 2015, only 24,104 persons were recorded as unlawful users of or addicted to any controlled substance in the NICS Index, the database accessed by the National Instant Criminal Background Check System for background checks related to firearms purchases (87).

It is clear from the results of our review that further research is needed to inform the development of criteria that accurately identify individuals involved with controlled substances who are at elevated risk of future firearm violence. This best available existing literature on the relationship between illicit drugs and violence is limited, but it does provide several insights. Despite public and policymaker perceptions of a clear and direct link between illicit drugs and firearm violence, the actual relationship is complex. Although some violent acts occur as a direct result of the physical and/or psychological effects of consumed substances, many firearm incidents involving controlled substances are likely caused by multiple and interacting factors. As a result, it is unclear whether and how firearm prohibitions targeting controlled substance users would influence rates of firearm violence. It is also important to consider how such policy might influence treatment seeking among users of controlled substances, who may avoid treatment if receiving care could in some way lead to a firearm disqualification. Review results also suggest that is also important to consider nonfirearm interventions to address the subset of violence stemming directly from use of controlled substances. Substance abuse is treatable, and early identification and treatment could reduce risk for subsequent firearm and nonfirearm interpersonal violence and suicide.

The existing literature suggests that drug dealing or other involvement in illicit drug markets (e.g., as a byproduct of gang membership) is a stronger and more consistent pathway to firearm violence than is drug use. Studies of the systemic pathway of Goldstein et al. showed a consistently positive relationship between drug sales and violence, and prior research suggests that firearm demand, access, and use are elevated in this context. However, distribution of controlled substances is typically charged as felony, and felony convictions are already tied to firearm prohibitions under federal law (88).

Importantly, our review suggests the need for a nuanced approach to controlled substance-based firearm prohibitions in the context of marijuana legalization by several states in the United States. Marijuana is the most frequently used controlled substance in the United States (27), and tying a firearm restriction to the use or sale of this drug has the potential to affect a significant number of marijuana users. This is problematic given physiological studies showing no direct link between marijuana use and violent behavior. However, many of the studies reviewed did show a positive link between marijuana use and violence perpetration. The unresolved questions may be less about the existence of a link and more about its nature. Although it seems based on evidence to date that this relationship, when present, can be attributed to shared risk factors for illicit drug use and violence rather than the physical and psychological effect of marijuana use, there is evidence that illicit marijuana use can be correlated with heightened risk of violent behavior. In this case, a related firearm restriction may have the potential to save lives.

More generally, our review points very clearly to the important role of shared risk factors for controlled substance use and violence, including but not limited to alcohol use and abuse, association with delinquent peer groups, socioeconomic risk factors such as poverty and low educational attainment, and, in the case of suicide, treatable mental illnesses such as depression. Policies and interventions to address these modifiable factors, a detailed discussion of which is outside the scope of this review, should be considered as potential upstream approaches to prevent controlled substance-related firearm violence.

Priorities for future research

The results of our review suggest several priorities for future research. First, studies designed to assess the relationship between controlled substances and firearm violence specifically, as opposed to violent behavior generally, are needed. Future studies on this topic should be explicitly designed to assess the complex association between controlled substance use and firearm violence and to account for as many of the key risk factors discussed above as possible. In particular, such studies should consider the joint effects of alcohol and controlled substance use. In controlled studies, alcohol is a consistent and direct risk factor for both violence perpetration and suicide (47, 8992), and prior studies suggest that co-occurring use of alcohol and controlled substances, which occurs frequently, may increase risk for violence above and beyond the effects of either behavior alone (64, 93).

Second, future research should assess the relationship between specific criteria that could be used to prohibit firearm ownership, such as conviction for multiple controlled substance-related misdemeanor crimes in a given time period or multiple drug-impaired driving offenses, and risk of firearm violence. Third, future research should continue to evaluate policies and practices designed to curb the systemic violence related to illicit drug sales, which the studies reviewed suggest is a consistent predictor of violence perpetration. Althogh multiple law enforcement initiatives targeting the illegal drug trade purport to reduce violence by taking drug dealers off the street, some research suggests that such practices may actually increase firearm violence by disrupting the illegal market (94).

Finally, it is critical to evaluate the relationship between marijuana use and violence perpetration in context of the changing landscape surrounding marijuana legalization in the United States. It is possible that legalization will uncouple marijuana use from some of the risk factors for violence, such as association with violent drug dealers. On the other hand, legalization is unlikely to completely eliminate the underground marijuana market (95), and involvement in this market may still increase risk of violence perpetration. Also, the underlying factors associated with both controlled substance use and a propensity for violence may not be affected by legalization of the drug.

ACKNOWLEDGMENTS

Author affiliations: Center for Gun Policy and Research, Department of Health Policy and Management, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland (Emma E. McGinty, Seema Choksy); and Violence Prevention Research Program, Department of Emergency Medicine, University of California Davis, Sacramento, California (Garen J. Wintemute).

This work was supported by a gift from Bloomberg Philanthropies (E.E.M.) and a grant (no. 2013-159) from the California Wellness Foundation (G.J.W.).

Conflict of interest: none declared.

REFERENCES

1
Blumstein
A
.
Youth violence, guns, and the illicit-drug industry
.
J Crim Law Criminol
 .
1995
;
86
(1)
:
10
36
.
2
Mosher
JF
,
Yanagisako
KL
.
Public health, not social warfare: a public health approach to illegal drug policy
.
J Public Health Policy
 .
1991
;
12
(3)
:
278
323
.
3
Johnson
BD
,
Golub
A
,
Dunlap
E
.
The rise and decline of hard drugs, drug markets, and violence in inner-city New York
. In:
Blumstein
A
,
Wallman
J
, eds.
The Crime Drop in America (Revised Edition)
 .
New York, NY
:
Cambridge University Press
;
2006
:
164
206
.
4
Howell
JC
,
Decker
SH
.
The Youth Gangs, Drugs, and Violence Connection
 .
Washington, DC
:
US Department of Justice
;
1999
.
5
Rodgers
D
.
Living in the shadow of death: gangs, violence and social order in urban Nicaragua, 1996–2002
.
J Lat Am Stud
 .
2006
;
38
(2)
:
267
292
.
6
Jütersonke
O
,
Muggah
R
,
Rodgers
D
.
Gangs, urban violence, and security interventions in Central America
.
Secur Dialogue
 .
2009
;
40
(4-5)
:
373
397
.
7
Centers for Disease Control and Prevention, US Department of Health and Human Services
.
Injury prevention & control: fatal injury data & statistics (WISQARS)
. .
Updated January 22, 2015
.
Accessed February 23, 2015
.
8
Vijayakumar
L
,
Kumar
MS
,
Vijayakumar
V
.
Substance use and suicide
.
Curr Opin Psychiatry
 .
2011
;
24
(3)
:
197
202
.
9
Wilcox
HC
,
Conner
KR
,
Caine
ED
.
Association of alcohol and drug use disorders and completed suicide: an empirical review of cohort studies
.
Drug Alcohol Depend
 .
2004
;
76
(suppl)
:
S11
S19
.
10
Commerce in Firearms and Ammunition
.
68 FR 3750. 2003. Codified at 27 CFR §478.11
.
11
McGinty
EE
,
Frattaroli
S
,
Appelbaum
PS
et al
.
Using research evidence to reframe the policy debate around mental illness and guns: process and recommendations
.
Am J Public Health
 .
2014
;
104
(11)
:
e22
e26
.
12
Consortium for Risk-Based Firearm Policy. Guns, public health, and mental illness: an evidence-based approach for state policy
. .
Published December 2, 2013
.
Accessed February 23, 2015
.
13
Government Accountability Office
.
Gun Control: Sharing Promising Practices and Assessing Incentives Could Better Position Justice to Assist States in Providing Records for Background Checks
 .
Washington, DC
:
US Government Printing Office
;
2012
.
(GAO publication no. 12-684)
.
14
Wintemute
GJ
,
Drake
CM
,
Beaumont
JJ
et al
.
Prior misdemeanor convictions as a risk factor for later violent and firearm-related criminal activity among authorized purchasers of handguns
.
JAMA
 .
1998
;
280
(24)
:
2083
2087
.
15
Hawkins
JD
,
Herrenkohl
T
,
Farrington
DP
et al
.
A Review of Predictors of Youth Violence
 .
Washington, DC
:
Office of Juvenile Justice and Delinquency Prevention
;
1998
.
16
Liu
J
.
Early health risk factors for violence: conceptualization, review of the evidence, and implications
.
Aggress Violent Behav
 .
2011
;
16
(1)
:
63
73
.
17
Sheley
JF
.
Drugs and guns among inner-city high school students
.
J Drug Educ
 .
1994
;
24
(4)
:
303
321
.
18
Cork
D
.
Examining space–time interaction in city-level homicide data: crack markets and the diffusion of guns among youth
.
J Quant Criminol
 .
1999
;
15
(4)
:
379
406
.
19
Presley
CA
,
Meilman
PW
,
Cashin
JR
.
Weapon carrying and substance abuse among college students
.
J Am Coll Health
 .
1997
;
46
(1)
:
3
8
.
20
Wilkinson
DL
,
Fagan
J
.
What we know about gun use among adolescents
.
Clin Child Fam Psychol Rev
 .
2001
;
4
(2)
:
109
132
.
21
Webster
DW
,
Vernick
JS
.
Keeping firearms from drug and alcohol abusers
.
Inj Prev
 .
2009
;
15
(6)
:
425
427
.
22
Lizotte
AJ
,
Krohn
MD
,
Howell
JC
et al
.
Factors influencing gun carrying among young urban males over the adolescent-young adult life course
.
Criminology
 .
2000
;
38
(3)
:
811
834
.
23
Wright
MA
,
Wintemute
GJ
.
Firearm prohibition for persons convicted of violent crimes: a potential non-legislative approach
.
Am J Prev Med
 .
2014
;
47
(2)
:
e3
e5
.
24
Wintemute
GJ
,
Wright
MA
,
Drake
CM
et al
.
Subsequent criminal activity among violent misdemeanants who seek to purchase handguns: risk factors and effectiveness of denying handgun purchase
.
JAMA
 .
2001
;
285
(8)
:
1019
1026
.
25
Swanson
JW
,
McGinty
EE
,
Fazel
S
et al
.
Mental illness and reduction of gun violence and suicide: bringing epidemiologic research to policy
.
Ann Epidemiol
 .
2015
;
25
(5)
:
366
376
.
26
Swanson
J
,
Robertson
AG
,
Frisman
LK
et al
.
Preventing gun violence involving people with serious mental illness
. In:
Webster
DW
,
Vernick
JS
, eds.
Reducing Gun Violence in America: Informing Policy with Evidence and Analysis
 .
Baltimore, MD
:
Johns Hopkins University Press
;
2013
:
33
51
.
27
Substance Abuse and Mental Health Services Administration
.
Results from the 2013 National Survey on Drug Use and Health: Summary of National Findings
 .
Rockville, MD
:
Substance Abuse and Mental Health Services Administration
;
2014
.
(NSDUH series H-48; DHHS publication no. (SMA) 14-4863)
.
28
Mumola
CJ
,
Karberg
JC
.
Drug use and dependence, state and federal prisoners, 2004
. .
Published October 2006
.
Updated January 19, 2007
.
Accessed February 23, 2015
.
29
Parks
SE
,
Johnson
LL
,
McDaniel
DD
et al
.
Surveillance for violent deaths—National Violent Death Reporting System, 16 states, 2010
.
MMWR Surveill Summ
 .
2014
;
63
(1)
:
1
33
.
30
National Center for Injury Prevention and Control, Centers for Disease Control and Prevention
.
WISQARS, nonfatal Injury reports, 2001–2013
. .
31
Hemenway
D
,
Miller
M
.
Firearm availability and homicide rates across 26 high-income countries
.
J Trauma
 .
2000
;
49
(6)
:
985
988
.
32
Goldstein
PJ
,
Brownstein
HH
,
Ryan
PJ
et al
.
Crack and homicide in New York City, 1988: a conceptually based event analysis
.
Contemp Drug Probl
 .
1989
;
16
(4)
:
651
687
.
33
Bennett
T
,
Holloway
K
.
The causal connection between drug misuse and crime
.
Br J Criminol
 .
2009
;
49
(4)
:
513
531
.
34
Warner
BD
,
Fowler
SK
.
Strain and violence: testing a general strain theory model of community violence
.
J Crim Justice
 .
2003
;
31
(6)
:
511
521
.
35
Collins
R
.
Violence: a Micro-sociological Theory
 .
Santa Barbara, CA
:
Greenwood Publishing Group
;
2009
.
36
Augustine
MC
,
Wilcox
P
,
Ousey
GC
et al
.
Opportunity theory and adolescent school-based victimization
.
Violence Vict
 .
2002
;
17
(2)
:
233
253
.
37
Bhatt
R
.
Domestic violence and substance abuse
.
Int J Gynecol Obstet
 .
1998
;
63
:
S25
S31
.
38
Boles
SM
,
Miotto
K
.
Substance abuse and violence: a review of the literature
.
Aggress Violent Behav
 .
2003
;
8
(2)
:
155
174
.
39
Moore
TM
,
Stuart
GL
.
A review of the literature on marijuana and interpersonal violence
.
Aggress Violent Behav
 .
2005
;
10
(2)
:
171
192
.
40
Kuhns
JB
,
Clodfelter
TA
.
Illicit drug-related psychopharmacological violence: the current understanding within a causal context
.
Aggress Violent Behav
 .
2009
;
14
(1)
:
69
78
.
41
Kuhns
JB
,
Wilson
DB
,
Maguire
ER
et al
.
A meta-analysis of marijuana, cocaine and opiate toxicology study findings among homicide victims
.
Addiction
 .
2009
;
104
(7)
:
1122
1131
.
42
Darke
S
.
The toxicology of homicide offenders and victims: a review
.
Drug Alcohol Rev
 .
2010
;
29
(2)
:
202
215
.
43
Ostrowsky
MK
.
Does marijuana use lead to aggression and violent behavior?
J Drug Educ
 .
2011
;
41
(4)
:
369
389
.
44
Fagan
J
.
Interactions among drugs, alcohol, and violence
.
Health Aff (Millwood)
 .
1993
;
12
(4)
:
65
79
.
45
Parker
RN
,
Auerhahn
K
.
Alcohol, drugs, and violence
.
Annu Rev Sociol
 .
1998
;
24
:
291
311
.
46
Friedman
AS
.
Substance use/abuse as a predictor to illegal and violent behavior: a review of the relevant literature
.
Aggress Violent Behav
 .
1998
;
3
(4)
:
339
355
.
47
Borges
G
,
Loera
CR
.
Alcohol and drug use in suicidal behaviour
.
Curr Opin Psychiatry
 .
2010
;
23
(3)
:
195
204
.
48
Goodman
A
.
Neurobiology of addiction: an integrative review
.
Biochem Pharmacol
 .
2008
;
75
(1)
:
266
322
.
49
Li
Z
,
Page
A
,
Martin
G
et al
.
Attributable risk of psychiatric and socio-economic factors for suicide from individual-level, population-based studies: a systematic review
.
Soc Sci Med
 .
2011
;
72
(4)
:
608
616
.
50
Grant
BF
,
Stinson
FS
,
Dawson
DA
et al
.
Prevalence and co-occurrence of substance use disorders and independent mood and anxiety disorders: results from the National Epidemiologic Survey on Alcohol and Related Conditions
.
Arch Gen Psychiatry
 .
2004
;
61
(8)
:
807
816
.
51
Conner
KR
,
Duberstein
PR
.
Predisposing and precipitating factors for suicide among alcoholics: empirical review and conceptual integration
.
Alcohol Clin Exp Res
 .
2004
;
28
(5
suppl)
:
6S
17S
.
52
Miczek
KA
,
Tidey
JW
.
Amphetamines: aggressive and social behavior
.
NIDA Res Monogr
 .
1989
;
94
:
68
100
.
53
Mino
A
,
Bousquet
A
,
Broers
B
.
Substance abuse and drug-related death, suicidal ideation, and suicide: a review
.
Crisis
 .
1999
;
20
(1)
:
28
35
.
54
Darke
S
,
Ross
J
.
Suicide among heroin users: rates, risk factors and methods
.
Addiction
 .
2002
;
97
(11)
:
1383
1394
.
55
Harris
EC
,
Barraclough
B
.
Suicide as an outcome for mental disorders. A meta-analysis
.
Br J Psychiatry
 .
1997
;
170
(3)
:
205
228
.
56
Vandenbroucke
JP
,
Von Elm
E
,
Altman
DG
et al
.
Strengthening the Reporting of Observational Studies in Epidemiology (STROBE): explanation and elaboration
.
Ann Intern Med
 .
2007
;
147
(8)
:
W163
W194
.
57
Higgins
JPT
,
Green
S
eds.
Cochrane Handbook for Systematic Reviews of Interventions
 .
Version 5.1.0.
London, England
:
The Cochrane Collaboration
;
2011
.
58
Landis
JR
,
Koch
GG
.
The measurement of observer agreement for categorical data
.
Biometrics
 .
1977
;
33
(1)
:
159
174
.
59
Brook
DW
,
Brook
JS
,
Rubenstone
E
et al
.
Developmental associations between externalizing behaviors, peer delinquency, drug use, perceived neighborhood crime, and violent behavior in urban communities
.
Aggress Behav
 .
2011
;
37
(4)
:
349
361
.
60
Kuhns
JB
3rd
.
The dynamic nature of the drug use/serious violence relationship: a multi-causal approach
.
Violence Vict
 .
2005
;
20
(4)
:
433
454
.
61
Menard
S
,
Mihalic
S
.
The tripartite conceptual framework in adolescence and adulthood: evidence from a national sample
.
J Drug Issues
 .
2001
;
31
(4)
:
905
939
.
62
Mulvey
EP
,
Odgers
C
,
Skeem
J
et al
.
Substance use and community violence: a test of the relation at the daily level
.
J Consult Clin Psychol
 .
2006
;
74
(4)
:
743
754
.
63
Sussman
S
,
Skara
S
,
Weiner
MD
et al
.
Prediction of violence perpetration among high-risk youth
.
Am J Health Behav
 .
2004
;
28
(2)
:
134
144
.
64
Van Dorn
R
,
Volavka
J
,
Johnson
N
.
Mental disorder and violence: Is there a relationship beyond substance use?
Soc Psychiatry Psychiatr Epidemiol
 .
2012
;
47
(3)
:
487
503
.
65
Weiner
MD
,
Sussman
S
,
Sun
P
et al
.
Explaining the link between violence perpetration, victimization and drug use
.
Addict Behav
 .
2005
;
30
(6)
:
1261
1266
.
66
Friedman
AS
,
Glassman
K
,
Terras
BA
.
Violent behavior as related to use of marijuana and other drugs
.
J Addict Dis
 .
2001
;
20
(1)
:
49
72
.
67
McKetin
R
,
Lubman
DI
,
Najman
JM
et al
.
Does methamphetamine use increase violent behaviour? Evidence from a prospective longitudinal study
.
Addiction
 .
2014
;
109
(5)
:
798
806
.
68
Green
KM
,
Doherty
EE
,
Stuart
EA
et al
.
Does heavy adolescent marijuana use lead to criminal involvement in adulthood? Evidence from a multiwave longitudinal study of urban African Americans
.
Drug Alcohol Depend
 .
2010
;
112
(1-2)
:
117
125
.
69
Dembo
R
,
Williams
L
,
Wothke
W
et al
.
The relationship between cocaine use, drug sales, and other delinquency among a cohort of high-risk youths over time
.
NIDA Res Monogr
 .
1990
;
103
:
112
135
.
70
Ellickson
PL
,
McGuigan
KA
.
Early predictors of adolescent violence
.
Am J Public Health
 .
2000
;
90
(4)
:
566
572
.
71
Brady
SS
,
Tschann
JM
,
Pasch
LA
et al
.
Violence involvement, substance use, and sexual activity among Mexican-American and European-American adolescents
.
J Adolesc Health
 .
2008
;
43
(3)
:
285
295
.
72
Wei
EH
,
Loeber
R
,
White
HR
.
Teasing apart the developmental associations between alcohol and marijuana use and violence
.
J Contemp Crim Justice
 .
2004
;
20
(2)
:
166
183
.
73
Pedersen
W
,
Skardhamar
T
.
Cannabis and crime: findings from a longitudinal study
.
Addiction
 .
2010
;
105
(1)
:
109
118
.
74
Cerdá
M
,
Messner
SF
,
Tracy
M
et al
.
Investigating the effect of social changes on age-specific gun-related homicide rates in New York City during the 1990s
.
Am J Public Health
 .
2010
;
100
(6)
:
1107
1115
.
75
Bellair
PE
,
McNulty
TL
.
Gang membership, drug selling, and violence in neighborhood context
.
Justice Q
 .
2009
;
26
(4)
:
644
669
.
76
Yen
S
,
Shea
MT
,
Pagano
M
et al
.
Axis I and axis II disorders as predictors of prospective suicide attempts: findings from the Collaborative Longitudinal Personality Disorders Study
.
J Abnorm Psychol
 .
2003
;
112
(3)
:
375
381
.
77
Allebeck
P
,
Allgulander
C
.
Psychiatric diagnoses as predictors of suicide. A comparison of diagnoses at conscription and in psychiatric care in a cohort of 50,465 young men
.
Br J Psychiatry
 .
1990
;
157
(3)
:
339
344
.
78
Nilsson
SF
,
Hjorthøj
CR
,
Erlangsen
A
et al
.
Suicide and unintentional injury mortality among homeless people: a Danish nationwide register-based cohort study
.
Eur J Public Health
 .
2014
;
24
(1)
:
50
56
.
79
Petronis
KR
,
Samuels
JF
,
Moscicki
EK
et al
.
An epidemiologic investigation of potential risk factors for suicide attempts
.
Soc Psychiatry Psychiatr Epidemiol
 .
1990
;
25
(4)
:
193
199
.
80
Allgulander
C
,
Allebeck
P
,
Przybeck
TR
et al
.
Risk of suicide by psychiatric diagnosis in Stockholm County. A longitudinal study of 80,970 psychiatric inpatients
.
Eur Arch Psychiatry Clin Neurosci
 .
1992
;
241
(5)
:
323
326
.
81
National Institute of Justice
.
Gun violence
. .
Updated April 4, 2013
.
Accessed June 23, 2015
.
82
Richardson
EG
,
Hemenway
D
.
Homicide, suicide, and unintentional firearm fatality: comparing the United States with other high-income countries, 2003
.
J Trauma
 .
2011
;
70
(1)
:
238
243
.
83
Harvard University
.
Guns & suicide: the hidden toll
. .
Accessed June 23, 2015
.
84
Miller
M
,
Hemenway
D
.
Guns and suicide in the United States
.
N Engl J Med
 .
2008
;
359
(10)
:
989
991
.
85
Fischman
M
,
Haney
M
.
Neurobiology of stimulants
. In:
Ganalter
M
,
Kleber
H
eds.
Textbook of Substance Abuse Treatment
 .
2nd ed
.
Washington, DC
:
American Psychiatric Press
;
1999
:
21
31
.
86
Kosten
T
,
Singha
A
.
Stimulants
. In:
Ganalter
M
,
Kleber
H
eds.
Textbook of Substance Abuse Treatment
 .
2nd ed
.
Washington, DC
,
American Psychiatric Press
;
1999
:
183
193
.
87
Federal Bureau of Investigation
.
Active records in the NICS index
. .
Published 2014
.
Accessed February 23, 2015
.
88
The Brady Handgun Violence Prevention Control Act
.
18 USC §921–922 (1993)
.
89
Afifi
TO
,
Henriksen
CA
,
Asmundson
GJ
et al
.
Victimization and perpetration of intimate partner violence and substance use disorders in a nationally representative sample
.
J Nerv Ment Dis
 .
2012
;
200
(8)
:
684
691
.
90
Harford
TC
,
Yi
HY
,
Grant
BF
.
Other- and self-directed forms of violence and their relationships to DSM-IV substance use and other psychiatric disorders in a national survey of adults
.
Compr Psychiatry
 .
2013
;
54
(7)
:
731
739
.
91
Bagge
CL
,
Lee
H-J
,
Schumacher
JA
et al
.
Alcohol as an acute risk factor for recent suicide attempts: a case-crossover analysis
.
J Stud Alcohol Drugs
 .
2013
;
74
(4)
:
552
558
.
92
Kaplan
MS
,
Huguet
N
,
McFarland
BH
et al
.
Use of alcohol before suicide in the United States
.
Ann Epidemiol
 .
2014
;
24
(8)
:
588
592.e1-2
.
93
Elbogen
EB
,
Johnson
SC
.
The intricate link between violence and mental disorder: results from the National Epidemiologic Survey on Alcohol and Related Conditions
.
Arch Gen Psychiatry
 .
2009
;
66
(2)
:
152
161
.
94
Werb
D
,
Rowell
G
,
Guyatt
G
et al
.
Effect of drug law enforcement on drug market violence: a systematic review
.
Int J Drug Policy
 .
2011
;
22
(2)
:
87
94
.
95
Pacula
RL
,
Kilmer
B
,
Wagenaar
AC
et al
.
Developing public health regulations for marijuana: lessons from alcohol and tobacco
.
Am J Public Health
 .
2014
;
104
(6)
:
1021
1028
.