Predictors of expressed partner and non-partner violence among patients in substance abuse treatment

Abstract

This study examined reports of expressed partner and non-partner violence among men (n=126) and women (n=126) in the 12 months prior to substance abuse treatment. Rates of violence were 57% for partner, 53% for non-partner, and 75% collapsing across partner and non-partner relationships. Factors associated with partner and non-partner violence severity differed substantially. Partner violence was predicted by age, marital status, and drug problem severity. Non-partner violence was predicted by gender, income, alcohol and drug problem severity. The results highlight that individuals in substance abuse treatment are at high risk for violence, and targeted screening and intervention approaches should be routine in addictions treatment.

Introduction

There is a wealth of evidence suggesting an association between alcohol-related problems and violent behavior. For example, studies have found rates of past year partner violence among individuals with alcohol-related problems to range from 39 to 68% (Powers et al., 1983, Gondolf and Foster, 1991, Stith et al., 1991), compared to 16% found in epidemiological surveys of community samples (Straus and Gelles, 1990). Conversely, there is evidence of high rates of alcoholism (30–64%) among batterers (Fitch and Papantonio, 1983, Van Hasselt et al., 1985, Stith et al., 1991) and individuals incarcerated for violent offenses (Weiczorek et al., 1990). With regard to treatment of alcoholism, a number of studies have provided evidence that violent acting-out may be a marker for overall alcohol problem severity and poor treatment response (Schuckit and Russell, 1984, Murphy and O’Farrell, 1994). Given the obvious costs of violence with regard to physical and emotional injuries, relationship problems, legal problems, and high levels of health care services utilization, as well as the evidence that violence and poor treatment outcome appear to covary, additional research in this area is critical. Increased understanding of the extent of violent acting-out, and the impact of various risk factors will provide critical information relevant to the assessment and treatment of individuals with alcohol and violence problems.

A number of factors potentially related to violence risk for both partner relationships and more general violence have been identified. For example, along with high rates of alcohol-related problems among samples of partner-abusing men (Fitch and Papantonio, 1983, Van Hasselt et al., 1985, Straus, 1990, Stith et al., 1991) and women victims of domestic abuse (Miller and Downs, 1995), there is evidence that variables such as age (Straus and Gelles, 1990), gender (Cantos et al., 1994; Pernanen, 1991), ethnicity (Hampton and Gelles, 1994, Kantor, 1997), socioeconomic status (Centerwall, 1984, Lockhart, 1987, Pan et al., 1994), alcohol consumption patterns (Leonard et al., 1985) and drug-related problems (Fagan and Brown, 1994, Pan et al., 1994) are related to risk or severity of perpetrating domestic or more general violence. Some researchers have suggested that the factors [e.g. socioeconomic status (SES), gender, ethnicity, age, alcohol and drug-related problems] potentially related to violence risk appear similar for partner and non-partner violence (Fagan and Brown, 1994). However, there also is evidence that predictors of aggression differ depending on the nature of the relationship between individuals involved in violent encounters (Pernanen, 1991).

In spite of the evidence that violence is common among alcohol and drug abusers, and that the presence of violent acting-out or violent victimization has clinical implications, there have been relatively few studies with samples of individuals seeking substance abuse treatment. Furthermore, the research that has been conducted with clinical samples has had important methodological limitations. First, a number of studies have focused on only violence perpetrated by males (Schuckit and Russell, 1984, Jaffe et al., 1988). Such designs obviously do not allow for the examination of gender related influences. Second, the majority of studies focus exclusively on violence within the context of a marital or conjoint relationship (Powers et al., 1983, Miller et al., 1989, Gondolf and Foster, 1991, Stith et al., 1991, O’Farrell and Murphy, 1995), and do not attempt to assess violent behavior in other relationship types (e.g. friends, co-workers, strangers, etc.). The majority of studies focusing on violence in treatment populations have recruited participants exclusively from inpatient settings (Schuckit and Russell, 1984, Jaffe et al., 1988, Gondolf and Foster, 1991, Murphy and O’Farrell, 1994), and some studies have used relatively small sample sizes of less than 80 alcoholics (Powers et al., 1983, Stith et al., 1991).

Research on violence with other samples, including individuals enrolled in treatment for domestic violence, incarcerated individuals and community samples, also has been criticized on several grounds (Chermack and Giancola, 1997). For example, some studies have relied on samples consisting of only male participants (Valdez et al., 1997), have failed to measure a number of factors that may influence risk of violence (e.g. alcohol/drug use and related consequences) (Cantos et al., 1994, Doumas et al., 1994, Vivian and Malone, 1997), and also have focused exclusively on violence in partner relationships (Kalmuss, 1984, Riggs et al., 1990). The studies that have addressed non-partner aggression have typically measured violence with only one question, or have asked about history of arrests for violent offenses (which would produce under-reports of actual violence given that most acts of violence do not result in arrest). Further, such research has been very nonspecific regarding the type of relationship between respondents and victims. For example, studies have measured aggression by using items that ask whether respondents had ever hit someone in a fight, or whether they have committed a number of aggressive acts ‘toward anyone’ outside of their family (Gully et al., 1981, Pollock et al., 1990). Thus, given methodological limitations of prior research, more information is needed about the extent of violence across relationships types among men and women with substance abuse problems, factors related to violence risk, and whether such risk factors are different for partner and non-partner violence.

The present study addressed a number of limitations by including both men and women participants, individuals recruited from both inpatient and outpatient treatment settings, measures of a number of patient presenting characteristics, and measures of expressed violence both within partner and non-partner relationships. Analyses examined the relationship between patient presenting characteristics (income, age, education, marital status, gender, minority status, alcohol and drug problem severity) and violence severity in both partner and non-partner relationships. The patient presenting characteristics were selected based on prior research on violence and because such information is routinely collected as part of intake assessment procedures for substance abuse treatment. Partner violence consisted of aggressive behaviors expressed towards partners and non-partner violence consisted of aggressive behaviors expressed towards individuals across a variety of relationship types (strangers, friends, co-workers, bosses, people in bars). Finally, the present study used an approach to data analysis (Structural Equation Modeling, SEM) that allows for more flexibility to model complex mediational relationships among variables.

The purpose of the study was to: (1) provide information regarding the extent and nature of expressed violence in both partner and non-partner relationships; (2) present descriptive information regarding factors associated with violence severity; and (3) examine with SEM the inter-relationships among potential predictor variables (age, gender, minority status, SES and alcohol and drug problem severity) in order to identify how such factors are related to both partner and non-partner violence.