527172 research-article2014
JIVXXX10.1177/0886260514527172Journal of Interpersonal ViolenceReingle et al.
Article Journal of Interpersonal Violence 2014, Vol. 29(16) 2951–2970 © The Author(s) 2014 Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/0886260514527172 jiv.sagepub.com
On the Pervasiveness of Event-Specific Alcohol Use, General Substance Use, and Mental Health Problems as Risk Factors for Intimate Partner Violence
Jennifer M. Reingle, PhD,1 Wesley G. Jennings, PhD,2 Nadine M. Connell, PhD,3 Michael S. Businelle, PhD,1 and Karen Chartier, PhD4
Abstract The aim of this study was to evaluate the role of demographic, mental health, and substance use as risk factors for intimate partner violence (IPV). Data were derived from Wave II of the National Epidemiological Survey on Alcohol and Related Conditions (2004-2005). Eligible participants (N = 25,778) reported having an intimate partner 1 year before the survey. Clustered survey multivariate multinomial regression methods were used to assess risk factors for episodes of IPV. IPV victimization, perpetration, and both victims/perpetrators were assessed. Bivariate analyses indicated that African Americans, Hispanics, and women were more likely to be victims, perpetrators, or victim/perpetrators as compared with men and Whites. Multivariate analyses suggested that having a marijuana use disorder was strongly associated with IPV victimization (odds ratio [OR] = 2.61) and victim/ 1University
of Texas, School of Public Health, Dallas Regional Campus, Dallas, USA of South Florida, Tampa, USA 3The University of Texas at Dallas, Richardson, USA 4Virginia Commonwealth University, Richmond, USA 2University
Corresponding Author: Jennifer M. Reingle, Assistant Professor, University of Texas School of Public Health, Dallas Regional Campus, 5323 Harry Hines Blvd., V8.112, Dallas, TX 75390, USA. Email: [email protected]
Downloaded from jiv.sagepub.com at UNIV ARIZONA LIBRARY on June 5, 2015
2952
Journal of Interpersonal Violence 29(16)
perpetration (OR = 2.65). Post-traumatic stress disorder was consistently associated with all IPV typologies. Depression was associated with victimization (OR = 2.00) and IPV victim/perpetration (OR = 1.74). Antisocial Personality Disorder and Mania were both related to IPV perpetration (ORs = 2.53 and 2.32) and victim/perpetration (ORs = 3.15 and 2.31). Results also indicated that alcohol use during episodes of IPV is common (i.e., 35% of those who reported IPV also reported that alcohol was involved). Results indicate several substance- and mental health–related correlates of IPV. In addition, findings indicate that alcohol use by the victim and/or perpetrator is common during IPV events. Policy implications and directions for future research are discussed. Keywords alcohol, marijuana, intimate partner, violence, mental health
Introduction In the United States, 26% of adults ages 18 to 65 reported experiencing intimate partner violence (IPV) during their lifetime (Centers for Disease Control and Prevention [CDC], 2011; Coker, Davis, et al., 2002). IPV has been associated with several negative health consequences, including unhealthy weight maintenance, social insecurity, unplanned pregnancy, suicidality, chronic disease, physical injury, and substance use (Ackard, Neumark-Sztainer, & Hannan, 2003; Campbell, 2002; Coker et al., 2000; Coker, Davis, et al., 2002; Kim-Godwin, Clements, McCuiston, & Fox, 2009; Pandey, Dutt, & Banerjee, 2009; Roberts, Auinger, & Klein, 2006; Silverman, Raj, Mucci, & Hathaway, 2001; C. A. Smith, Elwyn, Ireland, & Thornberry, 2010; Swahn, Bossarte, & Sullivent, 2008). Mental health, particularly depression, anxiety, and posttraumatic stress disorder, has been related to IPV (Coker, Smith, et al., 2002). Drug and alcohol use have also been associated with IPV, as marijuana users and binge drinkers are at elevated risk for being perpetrators and victims of IPV (Moore et al., 2008; Reingle, Staras, Jennings, Branchini, & MaldonadoMolina, 2012; P. H. Smith, Homish, Leonard, & Cornelius, 2012). Although there is evidence that alcohol and marijuana use generally precede IPV (Lipsky, Caetano, Field, & Larkin, 2005; Reingle et al., 2012), there is no literature evaluating the influence of event-specific alcohol use on IPV using a nationally representative sample. With the exception of one study by Caetano, Schafer, and Cunradi (2001) who conducted a national sample of couples, convenience samples have been more common in the literature assessing the effect of alcohol consumption on partner violence at the event level. Specifically, Caetano and colleagues
Downloaded from jiv.sagepub.com at UNIV ARIZONA LIBRARY on June 5, 2015
2953
Reingle et al.
(2001) surveyed 1,635 married couples throughout the United States, focusing on racial and ethnic differences on partner violence. The results from the Caetano et al. study suggested that 30% to 40% of the men and 27% to 34% of the women perpetrated partner violence while under the influence of alcohol. Unfortunately, alcohol impairment of the victim, or victim/perpetrator status, was not obtained among these respondents. In contrast, convenience samples have found that perpetrators are more likely than victims to use alcohol during IPV (Ernst, Weiss, Enright-Smith, Hilton, & Byrd, 2008), and that IPV is more common retrospectively on days that heavy drinking occurred (Fals-Stewart, 2003). Recent research has demonstrated the existence of a persistent victim– perpetrator overlap, whereby individuals are both the perpetrator and victim during IPV episodes (Reingle et al., 2012), and these findings have been documented cross-culturally and internationally (Cunradi, Ames, & Duke, 2011; Jennings, Tomsich, Gover, & Akers, 2011; McKinney, Caetano, Ramisetty-Mikler, & Nelson, 2009; Paterson, Feehan, Butler, Williams, & Cowley-Malcolm, 2007). Furthermore, research has suggested that when this overlap is apparent among individuals, the occurrence of IPV and its associated severity and related injuries is also amplified (Swahn, Alemdar, & Whitaker, 2010; Whitaker, Haileyesus, Swahn, & Saltzman, 2007). Substance use in general; particularly, marijuana, has also been identified as a predictor of IPV overlap (Jennings, Reingle et al., 2012; Reingle et al., 2012). Despite this evidence in support of the link between substance use in general and the IPV overlap specifically, there may be differences in the event-specific alcohol use by victim/perpetrator overlap typology (e.g., non-victims/non-perpetrators, victims only, perpetrators only, and/or victim/perpetrators). For instance, those who only perpetrate IPV may be more likely than those who are victims only, or who are victim/perpetrators to be using alcohol at the time of the event. The relationship between violence in general and alcohol use has been studied in detail, and four theories have been formulated to explain the relationship between alcohol use and violence, which directly apply to the role of alcohol during episodes of IPV: (a) psychopharmacological, meaning that the intoxicating effects of alcohol cause people to be violent (Goldstein, 1985); (b) the relationship is causal, in that alcohol use causes violence because aggressive people self-select into situations that encourage alcohol consumption (Johnston, O’Malley, & Eveland, 1978); (c) the relationship is reciprocal, and the arrow between alcohol use and violence may point in either or both directions (White, Loeber, Stouthamer-Loeber, & Farrington, 1999); or (d) the relationship is spurious, as problem behaviors cluster as part of a general problem behavior syndrome (Jessor, Donovan, & Costa, 1991).
Downloaded from jiv.sagepub.com at UNIV ARIZONA LIBRARY on June 5, 2015
2954
Journal of Interpersonal Violence 29(16)
Furthermore, mental health is strongly related to alcohol use, drug use, and violent behavior; particularly, partner violence (Lipskey, Caetano, Field, & Bazargan, 2005). In this regard, alcohol and drug use may be self-medicating for the effects of IPV or other stressors; alternatively, they may be reducing self-control inhibitions and assisting in the perpetration of violence. Therefore, the current study will help clarify the temporality of the relationship between event-level alcohol use and partner violence perpetration and victimization (and victim/perpetration), removing the effects of mental health confounders as risk factors for IPV. Recent evidence has also documented that specific ethnic minority groups such as Blacks and Hispanics evince a higher level of risk regarding their likelihood of being both victims and perpetrators of IPV (Caetano, Field, Ramisetty-Mikler, & McGrath, 2005; Caetano, Vaeth, & Ramisetty-Mikler, 2008; Cunradi, 2007; Field & Caetano, 2005). For instance, one such study recently noted that 16.4% of Blacks, 10.2% of Hispanics, and 6.0% of Whites were identifiable as IPV victim–perpetrators or representative of illustrating IPV overlap (Cunradi, 2007). Caetano, Ramisetty-Mikler, & Field, (2005) also observed a similar difference in the prevalence of IPV overlap between Whites and ethnic minorities with Blacks and Hispanics displaying greater risk. Overall, there appear to be substantial differences in the concentration of the victim/perpetrator overlap within racial/ethnic groups. In addition to racial/ethnic differences in IPV, the role of gender has been an area of debate in the literature on partner violence. Recent assessments of the literature have concluded that there are no gender differences in IPV perpetration (Archer, 2000; Gover, Park, Tomsich, & Jennings, 2011; Langhinrichsen-Rohling, 2005); however, other studies and reviews suggest that this literature ignores the severity of violent episodes, and the extant behavioral theories guiding domestic violence research (Reed, Raj, Miller, & Silverman, 2010). In fact, a recent report by the CDC (2011) reported that, according to the National Intimate Partner and Sexual Violence Survey, 25% of women (and 14% of men) reported some form of intimate partner victimization. Unfortunately, this study assessed victimization only; therefore, knowledge of any victim/perpetration that may have occurred can only be speculated. Within the literature that suggests no gender differences, the absence of differences in IPV perpetration between men and women could indicate that behavioral risk factors, such as alcohol and other substance use, may mediate demographic differences in offending (Capaldi, Knoble, Wu Shortt, & Kim, 2012). A consistent and somewhat controversial finding in the criminological literature is that the frequency of aggression in intimate partner relationships is similar for men and women, and that in some cases women are more likely
Downloaded from jiv.sagepub.com at UNIV ARIZONA LIBRARY on June 5, 2015
2955
Reingle et al.
to be the aggressors in the relationship compared with men (Archer, 2000; Langhinrichsen-Rohling, 2005). Yet, other studies have suggested that women were more likely to be victimized in intimate partner relationships compared with men, and more likely than men to experience injury (Stets & Pirog-Good, 1987). Recognizing these gaps and inconsistencies in the research, the current study builds on the extant public health literature by extending the recent work of P. H. Smith et al. (2012) using the National Epidemiologic Survey on Alcohol and Related Conditions (NESARC) dataset to examine prevalence rates by race and gender in both victimization, perpetration, and victim/perpetration, and substance-related risk factors for IPV. Although P. H. Smith et al. (2012) evaluated substance use as a risk factor for IPV perpetration and victimization independently, the results did not acknowledge the considerable overlap between victimization and perpetration (Jennings, Reingle et al., 2012). The current study extends the previous research by simultaneously evaluating mental health and substance use variables as correlates of IPV, with an additional focus on acute alcohol intoxication during the IPV event. This extension is a substantial contribution because these disorders are strongly related, and mental health conditions may the basis for the observed relationship between substance use and IPV (Burns & Teesson, 2002). Specifically, the current study will contribute to the literature by evaluating three specific aims: (a) to identify IPV differences by race/ethnicity and gender, (b) to examine the mental health and substance use–related correlates of IPV using a nationally representative sample of adults in the United States, and (c) to evaluate the nature of the event-specific relationship between alcohol consumption and IPV.
Method Data were obtained from Wave II of the NESARC (Chen, Yi, Dawson, Stinson, & Grant, 2010). Respondents (N = 34,653) comprised a representative sample of the U.S. adult population, including both citizens and noncitizens who were eligible if they participated in Wave I (2001-2002) of NESARC. The initial eligibility criteria for Wave I included non-institutionalized civilians, ages 18 years and older, residing in any of the 50 states or the District of Columbia. These criteria included persons living in households and non-institutional group quarters. Wave I used a multistage stratified design and oversampled young adults (ages 18-24) and non-Hispanic Black and Hispanic households. Data collection for Wave II occurred during 2004 and 2005. Sampling weights for Wave II consider the survey and sampling design characteristics of the NESARC survey as well as for non-response and
Downloaded from jiv.sagepub.com at UNIV ARIZONA LIBRARY on June 5, 2015
2956
Journal of Interpersonal Violence 29(16)
sample attrition, to maintain generalizability to the national population. All secondary analyses were approved by a university institutional review board. In-person respondent interviews were administered by trained and experienced lay interviewers from the U.S. Census Bureau. The survey instrument was computerized with software that included built-in skip, logic, and consistency checks. Surveys were conducted either in English or Spanish based on respondent’s preference in face-to-face household settings. Supervisors recontacted 10% of all respondents at random by telephone for quality control purposes.1 For the purposes of this study, respondents were included if they reported that they were “married, dating, or involved in a romantic relationship in the past year” (N = 25,778; 74.4% of the complete NESARC sample), as the outcome measures were IPV victimization and/or perpetration in the past year. Those who reported their relationship status within the prior year as “unknown” (n = 143) were excluded. This item was included as an eligibility criterion because only those respondents who were in a relationship in the past year can have the opportunity to experience dating violence as a perpetrator or victim with a romantic partner. As an acknowledged limitation of the survey, we are unaware of the duration of the relationship, or if the respondent was in a relationship at the time of the survey.
Measures Demographics. Respondents were asked to self-identify their age (retained as a continuous measure), sex (1 = male or 0 = female), and race/ethnicity at the time of the interview. Race/ethnicity was measured using the response options, “White, non-Hispanic,” “Black, non-Hispanic,” “American Indian/Alaska Native, non-Hispanic,” “Asian/Native Hawaiian/Other Pacific Islander, nonHispanic,” and “Hispanic, any race.” For analytical purposes, respondents were considered “White,” “Black,” “Other race,” or “Hispanic.” The decision to use the broad racial/ethnic categories of “White,” “Black,” “Other race,” and “Hispanic” was made by the authors due to the small numbers of Asians, Pacific Islanders, and Native Americans who experienced IPV. These categorizations would have led to cell sizes of smaller than 5, prohibiting quantitative analysis. Therefore, these “other” groups were lumped into “Other races” so that they were not excluded entirely from the current study. Mental health problems. All mental health problems (major depression, mania, dysthymia, hypomania, panic disorder, social phobia, specific phobia, general anxiety disorder, post-traumatic stress disorder, and antisocial personality disorder) were measured as mental health symptomology in the past
Downloaded from jiv.sagepub.com at UNIV ARIZONA LIBRARY on June 5, 2015
2957
Reingle et al.
year, in accordance with the Diagnostic and Statistical Manual of Mental Disorders (4th ed., DSM-IV; American Psychiatric Association, 1994) version Alcohol Use Disorder and Associated Disabilities Interview Schedule– IV (AUDADIS-IV), a fully structured diagnostic interview designed specifically for lay interviewers and self-administration. For each type of disorder, a respondent would need to respond affirmatively to a predetermined number of DSM-IV or AUDADIS-IV symptoms in the past year. These symptoms must be neither substance-related nor due to bereavement or a medical condition (Grant et al., 2005). Reliability and validity of the NESARC mental health measures have been demonstrated to be acceptable (see Grant et al., 2005) and they are commonly used in the epidemiological literature to assess mental health problems. Furthermore, they have been used thoroughly and pervasively in the epidemiological literature in measuring mental health disorders (Grant et al., 2008; Grant et al., 2005; Grant et al., 2004; Petry, Stinson, & Grant, 2005). Substance use disorders. Marijuana use disorder was coded as “disordered” or “non-disordered/non-user” in accordance with the DSM-IV. Alcohol use was assessed as quantity (“Number of drinks [of any alcohol] usually consumed on days when you drank in the last 12 months”) of alcohol use in the past year. IPV. Twelve items were used to create typologies of victimization, perpetration, and victim/perpetration of IPV. Specifically, (a) “How often did [you/ your spouse or partner] push, grab, or shove you in the past year?”; (b) “How often did [you/your spouse or partner] slap, kick, bite or hit [your spouse or partner/you] in the past year?”; (c) “How often did [you/your spouse/partner] threaten [your spouse/partner/you] with a weapon like a knife or gun in the past year?”; (d) “How often did [you/your spouse or partner] cut or bruise [your spouse or partner/you] in the past year?”; (e) “How often did [you/your spouse or partner] force [your spouse or partner/you] to have sex in the past year?”; and (f) “How often did [you/your spouse or partner] injure [your spouse or partner/you] enough that [they/you] had to get medical care in the past year?” Response options for each of these items included “Never,” “once,” “2-3 times,” “once a month,” and “more than once a month.” For the purposes of the current analyses, individuals were considered “exposed” to IPV if they reported that they were victims, perpetrators, or both, at least once within the past year. These items were derived from the Conflict Tactics Scale (CTS), Form R (Straus, 1990b). The CTS’s measure of IPV has good internal consistency reliability with alpha coefficients ranging from .69 to .88 (Straus, 1990a). A test−retest reliability study conducted as a part of the NESARC data collection process found these IPV items to be reliable over time
Downloaded from jiv.sagepub.com at UNIV ARIZONA LIBRARY on June 5, 2015
2958
Journal of Interpersonal Violence 29(16)
(Intraclass Correlation [ICC] for respondent IPV = 0.79; 95% confidence interval [CI] = [0.76, 0.82]; ICC for partner IPV = 0.76; 95% CI = [0.73, 0.80]; Ruan et al., 2008). Using these 12 items, four distinct typologies were created: (a) those who reported no IPV in the past year (e.g., non-victims, non-perpetrators); (b) those who reported at least one of the episode of victimization, but no perpetration behavior/s (e.g., victims only); (c) those who reported perpetration behavior/s only (e.g., perpetrators only); and (d) those who reported at least one perpetration behavior and at least one victimization experience (e.g., victim–perpetrators of IPV; Jennings, Reingle et al., 2012). Event-specific alcohol use. The following two questions were asked if respondents reported IPV over the past 12 months: “How often had you been drinking at the time this/these things(s) happened?” and “How often had your spouse or partner been drinking at that time?” Responses were dichotomized into “never” (referent) and “have been drinking during at least one IPV episode.”
Analytical Methods Consistent with the recommendations for NESARC data analysis, all analyses were conducted considering the clustered two-stage sampling design, and observations were weighted due to the unequal probability of selection of each primary sampling unit (Grant & Dawson, 2006). To account for the clustering of individuals within counties, survey multinomial logistic regression was used to provide weighted effect estimates and confidence intervals for aim 1. Respondents in the “non-victim, non-perpetrator” category served as the dependent variable reference group/typology for all models. Three models were used to assess the effects of event-specific alcohol use, mental health, and substance use disorders on IPV. First, we evaluated the bivariate relationship between substance use, mental health, and demographic variables described above as correlates of IPV. Only independent variables that met the inclusion criteria (bivariate relation p < .10) for at least one dependent variable were included in the multivariate models. In separate analyses, we examined the bivariate relation between IPV and event-specific alcohol use. All analyses were conducted using STATA 12 (StataCorp, 2012).
Results The sample was 46% male, 59.3% White or Caucasian, 16.7% Black or African American, and 19.4% Hispanic or Latino. The average age of
Downloaded from jiv.sagepub.com at UNIV ARIZONA LIBRARY on June 5, 2015
2959
Reingle et al.
respondents was 46.08 (SE = 0.10). Consistent with expectations, we found that most people who were in a relationship were neither victims nor perpetrators of IPV (91.2%). However, slightly more than 4% of the population reported being victim/perpetrators of IPV. A smaller proportion reported victimization only (2.3%) or perpetration only (2.5%). See Table 1 for demographic, mental health, and IPV-related participant characteristics. As detailed in Table 2, there were large gender and racial/ethnic differences in IPV. Women were more likely than men to be victim/perpetrators of IPV and perpetrators only, while men were more likely than women to be victims only. The largest proportion of victim/perpetration occurred among African Americans, followed by Hispanics. African Americans and Hispanics were also more likely than Whites to report victimization only and perpetration of IPV only. Table 3 depicts the relationship between respondents’ alcohol consumption (when any perpetration was reported by the respondent) and alcohol use by their partner (when the partner was also a perpetrator). Although 69% of respondents reported no alcohol consumption by either partner during IPV events, 15% reported that both partners were drinking at the time of the event. In 11% of IPV events, the partner had consumed alcohol while the respondent had not. In 4% of IPV cases, the respondent reported recent alcohol consumption while the partner had not consumed alcohol. Overall, one partner had consumed alcohol in 15% of IPV cases and both partners had consumed alcohol in an additional 15% of IPV cases. Bivariate tests were conducted to detect differences in the correlates of IPV victimization, perpetration, and victimization/perpetration (results not shown). Using the results from these bivariate tests, we built a multivariate model (Table 4) of the most theoretically relevant and potent behavioral disorders that we believe would differentiate the three typologies of IPV. The reference group for the dependent variable was “non-victims and non-perpetrators.” Consistently across the three typologies, Black respondents were more likely than Whites to report being IPV victims only (odds ratio [OR] = 2.13; 95% CI = [1.62, 2.80]), perpetrators only (OR = 2.36; 95% CI = [1.71, 3.26]) and victim/perpetrators during IPV events (OR = 2.61; 95% CI = [2.13, 3.19]). Hispanics were also at increased risk for IPV across all three typologies as compared with Whites: victims only (OR = 1.49; 95% CI = [1.14, 1.96]), perpetrators only (OR = 1.59; 95% CI = [1.22, 2.09]), and victim/perpetrators (OR = 1.76; 95% CI = [1.43, 2.16]). Younger age was consistently protective from all three IPV typologies. Quantity of alcohol use (in general, not event-specifically) was associated with victimization only (OR = 1.08; 95% CI = [1.01, 1.16]) and victim/perpetration (OR = 1.12; 95% CI = [1.05, 1.20]). Likewise, having a marijuana use disorder was strongly
Downloaded from jiv.sagepub.com at UNIV ARIZONA LIBRARY on June 5, 2015
2960
Journal of Interpersonal Violence 29(16)
Table 1. Sample Description of Those Who Reported Being in a Relationship in the Past Year, NESARC II, n = 25,778. %
n Intimate Partner Violence (IPV) No IPV Victimization only Perpetration only Both perpetration and victimization Alcohol use (among those who used alcohol during IPV) Alcohol use during IPV by respondent only Alcohol use during IPV by partner only Alcohol use by both partners during IPV Substance use Marijuana use disorder (past year) Alcohol use (Quantity typically used each occasion in the past year, range: 1-10)a Mental health Depression (past year) Mania (past year) Dysthymia Hypomania Panic disorder Social phobia Specific phobia General anxiety Post-traumatic stress disorder Antisocial personality disorder Demographic covariates Male Race and ethnicity White or Caucasian Black or African American Hispanic or Latino Other race Agea
23,370 580 647 1,028 43 146 153 399 4.72
91.2 2.26 2.52 4.01 n = 342 4.19 14.22 14.90 1.55 0.05
2,143 560 302 382 494 658 2,082 984 1,849 62 11,850
8.31 2.17 1.17 1.48 1.92 2.55 8.08 3.82 7.17 0.24 45.97
15,274 59.25 4,298 16.67 4,999 19.39 1,207 4.68 46.08 (0.10)
Note. NESARC = National Epidemiologic Survey on Alcohol and Related Conditions. aM and SE are reported.
associated with IPV victimization (OR = 2.61; 95% CI = [1.75, 3.89]) and victim/perpetration (OR = 2.65; 95% CI = [1.81, 3.89]), compared with nonvictims and non-perpetrators.
Downloaded from jiv.sagepub.com at UNIV ARIZONA LIBRARY on June 5, 2015
2961
Reingle et al. Table 2. Prevalence Rates of IPV by race, ethnicity, and gender, NESARC II, n = 25,625. IPV Victimization/Perpetration
Men
Women
χ2
Non-victim/non-perpetrator Victim only Perpetrator only Victim and perpetrator
.93 .03 .01 .03
.91 .02 .03 .04
28.59***
IPV Victimization/Perpetration White Non-victim/non-perpetrator Victim only Perpetrator only Victim and perpetrator
.94 .02 .02 .03
African American
Hispanic
χ2
.86 .04 .04 .07
.89 .03 .03 .05
10.18***
Note. IPV = intimate partner violence; NESARC = National Epidemiologic Survey on Alcohol and Related Conditions. ***p
JIVXXX10.1177/0886260514527172Journal of Interpersonal ViolenceReingle et al.
Article Journal of Interpersonal Violence 2014, Vol. 29(16) 2951–2970 © The Author(s) 2014 Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/0886260514527172 jiv.sagepub.com
On the Pervasiveness of Event-Specific Alcohol Use, General Substance Use, and Mental Health Problems as Risk Factors for Intimate Partner Violence
Jennifer M. Reingle, PhD,1 Wesley G. Jennings, PhD,2 Nadine M. Connell, PhD,3 Michael S. Businelle, PhD,1 and Karen Chartier, PhD4
Abstract The aim of this study was to evaluate the role of demographic, mental health, and substance use as risk factors for intimate partner violence (IPV). Data were derived from Wave II of the National Epidemiological Survey on Alcohol and Related Conditions (2004-2005). Eligible participants (N = 25,778) reported having an intimate partner 1 year before the survey. Clustered survey multivariate multinomial regression methods were used to assess risk factors for episodes of IPV. IPV victimization, perpetration, and both victims/perpetrators were assessed. Bivariate analyses indicated that African Americans, Hispanics, and women were more likely to be victims, perpetrators, or victim/perpetrators as compared with men and Whites. Multivariate analyses suggested that having a marijuana use disorder was strongly associated with IPV victimization (odds ratio [OR] = 2.61) and victim/ 1University
of Texas, School of Public Health, Dallas Regional Campus, Dallas, USA of South Florida, Tampa, USA 3The University of Texas at Dallas, Richardson, USA 4Virginia Commonwealth University, Richmond, USA 2University
Corresponding Author: Jennifer M. Reingle, Assistant Professor, University of Texas School of Public Health, Dallas Regional Campus, 5323 Harry Hines Blvd., V8.112, Dallas, TX 75390, USA. Email: [email protected]
Downloaded from jiv.sagepub.com at UNIV ARIZONA LIBRARY on June 5, 2015
2952
Journal of Interpersonal Violence 29(16)
perpetration (OR = 2.65). Post-traumatic stress disorder was consistently associated with all IPV typologies. Depression was associated with victimization (OR = 2.00) and IPV victim/perpetration (OR = 1.74). Antisocial Personality Disorder and Mania were both related to IPV perpetration (ORs = 2.53 and 2.32) and victim/perpetration (ORs = 3.15 and 2.31). Results also indicated that alcohol use during episodes of IPV is common (i.e., 35% of those who reported IPV also reported that alcohol was involved). Results indicate several substance- and mental health–related correlates of IPV. In addition, findings indicate that alcohol use by the victim and/or perpetrator is common during IPV events. Policy implications and directions for future research are discussed. Keywords alcohol, marijuana, intimate partner, violence, mental health
Introduction In the United States, 26% of adults ages 18 to 65 reported experiencing intimate partner violence (IPV) during their lifetime (Centers for Disease Control and Prevention [CDC], 2011; Coker, Davis, et al., 2002). IPV has been associated with several negative health consequences, including unhealthy weight maintenance, social insecurity, unplanned pregnancy, suicidality, chronic disease, physical injury, and substance use (Ackard, Neumark-Sztainer, & Hannan, 2003; Campbell, 2002; Coker et al., 2000; Coker, Davis, et al., 2002; Kim-Godwin, Clements, McCuiston, & Fox, 2009; Pandey, Dutt, & Banerjee, 2009; Roberts, Auinger, & Klein, 2006; Silverman, Raj, Mucci, & Hathaway, 2001; C. A. Smith, Elwyn, Ireland, & Thornberry, 2010; Swahn, Bossarte, & Sullivent, 2008). Mental health, particularly depression, anxiety, and posttraumatic stress disorder, has been related to IPV (Coker, Smith, et al., 2002). Drug and alcohol use have also been associated with IPV, as marijuana users and binge drinkers are at elevated risk for being perpetrators and victims of IPV (Moore et al., 2008; Reingle, Staras, Jennings, Branchini, & MaldonadoMolina, 2012; P. H. Smith, Homish, Leonard, & Cornelius, 2012). Although there is evidence that alcohol and marijuana use generally precede IPV (Lipsky, Caetano, Field, & Larkin, 2005; Reingle et al., 2012), there is no literature evaluating the influence of event-specific alcohol use on IPV using a nationally representative sample. With the exception of one study by Caetano, Schafer, and Cunradi (2001) who conducted a national sample of couples, convenience samples have been more common in the literature assessing the effect of alcohol consumption on partner violence at the event level. Specifically, Caetano and colleagues
Downloaded from jiv.sagepub.com at UNIV ARIZONA LIBRARY on June 5, 2015
2953
Reingle et al.
(2001) surveyed 1,635 married couples throughout the United States, focusing on racial and ethnic differences on partner violence. The results from the Caetano et al. study suggested that 30% to 40% of the men and 27% to 34% of the women perpetrated partner violence while under the influence of alcohol. Unfortunately, alcohol impairment of the victim, or victim/perpetrator status, was not obtained among these respondents. In contrast, convenience samples have found that perpetrators are more likely than victims to use alcohol during IPV (Ernst, Weiss, Enright-Smith, Hilton, & Byrd, 2008), and that IPV is more common retrospectively on days that heavy drinking occurred (Fals-Stewart, 2003). Recent research has demonstrated the existence of a persistent victim– perpetrator overlap, whereby individuals are both the perpetrator and victim during IPV episodes (Reingle et al., 2012), and these findings have been documented cross-culturally and internationally (Cunradi, Ames, & Duke, 2011; Jennings, Tomsich, Gover, & Akers, 2011; McKinney, Caetano, Ramisetty-Mikler, & Nelson, 2009; Paterson, Feehan, Butler, Williams, & Cowley-Malcolm, 2007). Furthermore, research has suggested that when this overlap is apparent among individuals, the occurrence of IPV and its associated severity and related injuries is also amplified (Swahn, Alemdar, & Whitaker, 2010; Whitaker, Haileyesus, Swahn, & Saltzman, 2007). Substance use in general; particularly, marijuana, has also been identified as a predictor of IPV overlap (Jennings, Reingle et al., 2012; Reingle et al., 2012). Despite this evidence in support of the link between substance use in general and the IPV overlap specifically, there may be differences in the event-specific alcohol use by victim/perpetrator overlap typology (e.g., non-victims/non-perpetrators, victims only, perpetrators only, and/or victim/perpetrators). For instance, those who only perpetrate IPV may be more likely than those who are victims only, or who are victim/perpetrators to be using alcohol at the time of the event. The relationship between violence in general and alcohol use has been studied in detail, and four theories have been formulated to explain the relationship between alcohol use and violence, which directly apply to the role of alcohol during episodes of IPV: (a) psychopharmacological, meaning that the intoxicating effects of alcohol cause people to be violent (Goldstein, 1985); (b) the relationship is causal, in that alcohol use causes violence because aggressive people self-select into situations that encourage alcohol consumption (Johnston, O’Malley, & Eveland, 1978); (c) the relationship is reciprocal, and the arrow between alcohol use and violence may point in either or both directions (White, Loeber, Stouthamer-Loeber, & Farrington, 1999); or (d) the relationship is spurious, as problem behaviors cluster as part of a general problem behavior syndrome (Jessor, Donovan, & Costa, 1991).
Downloaded from jiv.sagepub.com at UNIV ARIZONA LIBRARY on June 5, 2015
2954
Journal of Interpersonal Violence 29(16)
Furthermore, mental health is strongly related to alcohol use, drug use, and violent behavior; particularly, partner violence (Lipskey, Caetano, Field, & Bazargan, 2005). In this regard, alcohol and drug use may be self-medicating for the effects of IPV or other stressors; alternatively, they may be reducing self-control inhibitions and assisting in the perpetration of violence. Therefore, the current study will help clarify the temporality of the relationship between event-level alcohol use and partner violence perpetration and victimization (and victim/perpetration), removing the effects of mental health confounders as risk factors for IPV. Recent evidence has also documented that specific ethnic minority groups such as Blacks and Hispanics evince a higher level of risk regarding their likelihood of being both victims and perpetrators of IPV (Caetano, Field, Ramisetty-Mikler, & McGrath, 2005; Caetano, Vaeth, & Ramisetty-Mikler, 2008; Cunradi, 2007; Field & Caetano, 2005). For instance, one such study recently noted that 16.4% of Blacks, 10.2% of Hispanics, and 6.0% of Whites were identifiable as IPV victim–perpetrators or representative of illustrating IPV overlap (Cunradi, 2007). Caetano, Ramisetty-Mikler, & Field, (2005) also observed a similar difference in the prevalence of IPV overlap between Whites and ethnic minorities with Blacks and Hispanics displaying greater risk. Overall, there appear to be substantial differences in the concentration of the victim/perpetrator overlap within racial/ethnic groups. In addition to racial/ethnic differences in IPV, the role of gender has been an area of debate in the literature on partner violence. Recent assessments of the literature have concluded that there are no gender differences in IPV perpetration (Archer, 2000; Gover, Park, Tomsich, & Jennings, 2011; Langhinrichsen-Rohling, 2005); however, other studies and reviews suggest that this literature ignores the severity of violent episodes, and the extant behavioral theories guiding domestic violence research (Reed, Raj, Miller, & Silverman, 2010). In fact, a recent report by the CDC (2011) reported that, according to the National Intimate Partner and Sexual Violence Survey, 25% of women (and 14% of men) reported some form of intimate partner victimization. Unfortunately, this study assessed victimization only; therefore, knowledge of any victim/perpetration that may have occurred can only be speculated. Within the literature that suggests no gender differences, the absence of differences in IPV perpetration between men and women could indicate that behavioral risk factors, such as alcohol and other substance use, may mediate demographic differences in offending (Capaldi, Knoble, Wu Shortt, & Kim, 2012). A consistent and somewhat controversial finding in the criminological literature is that the frequency of aggression in intimate partner relationships is similar for men and women, and that in some cases women are more likely
Downloaded from jiv.sagepub.com at UNIV ARIZONA LIBRARY on June 5, 2015
2955
Reingle et al.
to be the aggressors in the relationship compared with men (Archer, 2000; Langhinrichsen-Rohling, 2005). Yet, other studies have suggested that women were more likely to be victimized in intimate partner relationships compared with men, and more likely than men to experience injury (Stets & Pirog-Good, 1987). Recognizing these gaps and inconsistencies in the research, the current study builds on the extant public health literature by extending the recent work of P. H. Smith et al. (2012) using the National Epidemiologic Survey on Alcohol and Related Conditions (NESARC) dataset to examine prevalence rates by race and gender in both victimization, perpetration, and victim/perpetration, and substance-related risk factors for IPV. Although P. H. Smith et al. (2012) evaluated substance use as a risk factor for IPV perpetration and victimization independently, the results did not acknowledge the considerable overlap between victimization and perpetration (Jennings, Reingle et al., 2012). The current study extends the previous research by simultaneously evaluating mental health and substance use variables as correlates of IPV, with an additional focus on acute alcohol intoxication during the IPV event. This extension is a substantial contribution because these disorders are strongly related, and mental health conditions may the basis for the observed relationship between substance use and IPV (Burns & Teesson, 2002). Specifically, the current study will contribute to the literature by evaluating three specific aims: (a) to identify IPV differences by race/ethnicity and gender, (b) to examine the mental health and substance use–related correlates of IPV using a nationally representative sample of adults in the United States, and (c) to evaluate the nature of the event-specific relationship between alcohol consumption and IPV.
Method Data were obtained from Wave II of the NESARC (Chen, Yi, Dawson, Stinson, & Grant, 2010). Respondents (N = 34,653) comprised a representative sample of the U.S. adult population, including both citizens and noncitizens who were eligible if they participated in Wave I (2001-2002) of NESARC. The initial eligibility criteria for Wave I included non-institutionalized civilians, ages 18 years and older, residing in any of the 50 states or the District of Columbia. These criteria included persons living in households and non-institutional group quarters. Wave I used a multistage stratified design and oversampled young adults (ages 18-24) and non-Hispanic Black and Hispanic households. Data collection for Wave II occurred during 2004 and 2005. Sampling weights for Wave II consider the survey and sampling design characteristics of the NESARC survey as well as for non-response and
Downloaded from jiv.sagepub.com at UNIV ARIZONA LIBRARY on June 5, 2015
2956
Journal of Interpersonal Violence 29(16)
sample attrition, to maintain generalizability to the national population. All secondary analyses were approved by a university institutional review board. In-person respondent interviews were administered by trained and experienced lay interviewers from the U.S. Census Bureau. The survey instrument was computerized with software that included built-in skip, logic, and consistency checks. Surveys were conducted either in English or Spanish based on respondent’s preference in face-to-face household settings. Supervisors recontacted 10% of all respondents at random by telephone for quality control purposes.1 For the purposes of this study, respondents were included if they reported that they were “married, dating, or involved in a romantic relationship in the past year” (N = 25,778; 74.4% of the complete NESARC sample), as the outcome measures were IPV victimization and/or perpetration in the past year. Those who reported their relationship status within the prior year as “unknown” (n = 143) were excluded. This item was included as an eligibility criterion because only those respondents who were in a relationship in the past year can have the opportunity to experience dating violence as a perpetrator or victim with a romantic partner. As an acknowledged limitation of the survey, we are unaware of the duration of the relationship, or if the respondent was in a relationship at the time of the survey.
Measures Demographics. Respondents were asked to self-identify their age (retained as a continuous measure), sex (1 = male or 0 = female), and race/ethnicity at the time of the interview. Race/ethnicity was measured using the response options, “White, non-Hispanic,” “Black, non-Hispanic,” “American Indian/Alaska Native, non-Hispanic,” “Asian/Native Hawaiian/Other Pacific Islander, nonHispanic,” and “Hispanic, any race.” For analytical purposes, respondents were considered “White,” “Black,” “Other race,” or “Hispanic.” The decision to use the broad racial/ethnic categories of “White,” “Black,” “Other race,” and “Hispanic” was made by the authors due to the small numbers of Asians, Pacific Islanders, and Native Americans who experienced IPV. These categorizations would have led to cell sizes of smaller than 5, prohibiting quantitative analysis. Therefore, these “other” groups were lumped into “Other races” so that they were not excluded entirely from the current study. Mental health problems. All mental health problems (major depression, mania, dysthymia, hypomania, panic disorder, social phobia, specific phobia, general anxiety disorder, post-traumatic stress disorder, and antisocial personality disorder) were measured as mental health symptomology in the past
Downloaded from jiv.sagepub.com at UNIV ARIZONA LIBRARY on June 5, 2015
2957
Reingle et al.
year, in accordance with the Diagnostic and Statistical Manual of Mental Disorders (4th ed., DSM-IV; American Psychiatric Association, 1994) version Alcohol Use Disorder and Associated Disabilities Interview Schedule– IV (AUDADIS-IV), a fully structured diagnostic interview designed specifically for lay interviewers and self-administration. For each type of disorder, a respondent would need to respond affirmatively to a predetermined number of DSM-IV or AUDADIS-IV symptoms in the past year. These symptoms must be neither substance-related nor due to bereavement or a medical condition (Grant et al., 2005). Reliability and validity of the NESARC mental health measures have been demonstrated to be acceptable (see Grant et al., 2005) and they are commonly used in the epidemiological literature to assess mental health problems. Furthermore, they have been used thoroughly and pervasively in the epidemiological literature in measuring mental health disorders (Grant et al., 2008; Grant et al., 2005; Grant et al., 2004; Petry, Stinson, & Grant, 2005). Substance use disorders. Marijuana use disorder was coded as “disordered” or “non-disordered/non-user” in accordance with the DSM-IV. Alcohol use was assessed as quantity (“Number of drinks [of any alcohol] usually consumed on days when you drank in the last 12 months”) of alcohol use in the past year. IPV. Twelve items were used to create typologies of victimization, perpetration, and victim/perpetration of IPV. Specifically, (a) “How often did [you/ your spouse or partner] push, grab, or shove you in the past year?”; (b) “How often did [you/your spouse or partner] slap, kick, bite or hit [your spouse or partner/you] in the past year?”; (c) “How often did [you/your spouse/partner] threaten [your spouse/partner/you] with a weapon like a knife or gun in the past year?”; (d) “How often did [you/your spouse or partner] cut or bruise [your spouse or partner/you] in the past year?”; (e) “How often did [you/your spouse or partner] force [your spouse or partner/you] to have sex in the past year?”; and (f) “How often did [you/your spouse or partner] injure [your spouse or partner/you] enough that [they/you] had to get medical care in the past year?” Response options for each of these items included “Never,” “once,” “2-3 times,” “once a month,” and “more than once a month.” For the purposes of the current analyses, individuals were considered “exposed” to IPV if they reported that they were victims, perpetrators, or both, at least once within the past year. These items were derived from the Conflict Tactics Scale (CTS), Form R (Straus, 1990b). The CTS’s measure of IPV has good internal consistency reliability with alpha coefficients ranging from .69 to .88 (Straus, 1990a). A test−retest reliability study conducted as a part of the NESARC data collection process found these IPV items to be reliable over time
Downloaded from jiv.sagepub.com at UNIV ARIZONA LIBRARY on June 5, 2015
2958
Journal of Interpersonal Violence 29(16)
(Intraclass Correlation [ICC] for respondent IPV = 0.79; 95% confidence interval [CI] = [0.76, 0.82]; ICC for partner IPV = 0.76; 95% CI = [0.73, 0.80]; Ruan et al., 2008). Using these 12 items, four distinct typologies were created: (a) those who reported no IPV in the past year (e.g., non-victims, non-perpetrators); (b) those who reported at least one of the episode of victimization, but no perpetration behavior/s (e.g., victims only); (c) those who reported perpetration behavior/s only (e.g., perpetrators only); and (d) those who reported at least one perpetration behavior and at least one victimization experience (e.g., victim–perpetrators of IPV; Jennings, Reingle et al., 2012). Event-specific alcohol use. The following two questions were asked if respondents reported IPV over the past 12 months: “How often had you been drinking at the time this/these things(s) happened?” and “How often had your spouse or partner been drinking at that time?” Responses were dichotomized into “never” (referent) and “have been drinking during at least one IPV episode.”
Analytical Methods Consistent with the recommendations for NESARC data analysis, all analyses were conducted considering the clustered two-stage sampling design, and observations were weighted due to the unequal probability of selection of each primary sampling unit (Grant & Dawson, 2006). To account for the clustering of individuals within counties, survey multinomial logistic regression was used to provide weighted effect estimates and confidence intervals for aim 1. Respondents in the “non-victim, non-perpetrator” category served as the dependent variable reference group/typology for all models. Three models were used to assess the effects of event-specific alcohol use, mental health, and substance use disorders on IPV. First, we evaluated the bivariate relationship between substance use, mental health, and demographic variables described above as correlates of IPV. Only independent variables that met the inclusion criteria (bivariate relation p < .10) for at least one dependent variable were included in the multivariate models. In separate analyses, we examined the bivariate relation between IPV and event-specific alcohol use. All analyses were conducted using STATA 12 (StataCorp, 2012).
Results The sample was 46% male, 59.3% White or Caucasian, 16.7% Black or African American, and 19.4% Hispanic or Latino. The average age of
Downloaded from jiv.sagepub.com at UNIV ARIZONA LIBRARY on June 5, 2015
2959
Reingle et al.
respondents was 46.08 (SE = 0.10). Consistent with expectations, we found that most people who were in a relationship were neither victims nor perpetrators of IPV (91.2%). However, slightly more than 4% of the population reported being victim/perpetrators of IPV. A smaller proportion reported victimization only (2.3%) or perpetration only (2.5%). See Table 1 for demographic, mental health, and IPV-related participant characteristics. As detailed in Table 2, there were large gender and racial/ethnic differences in IPV. Women were more likely than men to be victim/perpetrators of IPV and perpetrators only, while men were more likely than women to be victims only. The largest proportion of victim/perpetration occurred among African Americans, followed by Hispanics. African Americans and Hispanics were also more likely than Whites to report victimization only and perpetration of IPV only. Table 3 depicts the relationship between respondents’ alcohol consumption (when any perpetration was reported by the respondent) and alcohol use by their partner (when the partner was also a perpetrator). Although 69% of respondents reported no alcohol consumption by either partner during IPV events, 15% reported that both partners were drinking at the time of the event. In 11% of IPV events, the partner had consumed alcohol while the respondent had not. In 4% of IPV cases, the respondent reported recent alcohol consumption while the partner had not consumed alcohol. Overall, one partner had consumed alcohol in 15% of IPV cases and both partners had consumed alcohol in an additional 15% of IPV cases. Bivariate tests were conducted to detect differences in the correlates of IPV victimization, perpetration, and victimization/perpetration (results not shown). Using the results from these bivariate tests, we built a multivariate model (Table 4) of the most theoretically relevant and potent behavioral disorders that we believe would differentiate the three typologies of IPV. The reference group for the dependent variable was “non-victims and non-perpetrators.” Consistently across the three typologies, Black respondents were more likely than Whites to report being IPV victims only (odds ratio [OR] = 2.13; 95% CI = [1.62, 2.80]), perpetrators only (OR = 2.36; 95% CI = [1.71, 3.26]) and victim/perpetrators during IPV events (OR = 2.61; 95% CI = [2.13, 3.19]). Hispanics were also at increased risk for IPV across all three typologies as compared with Whites: victims only (OR = 1.49; 95% CI = [1.14, 1.96]), perpetrators only (OR = 1.59; 95% CI = [1.22, 2.09]), and victim/perpetrators (OR = 1.76; 95% CI = [1.43, 2.16]). Younger age was consistently protective from all three IPV typologies. Quantity of alcohol use (in general, not event-specifically) was associated with victimization only (OR = 1.08; 95% CI = [1.01, 1.16]) and victim/perpetration (OR = 1.12; 95% CI = [1.05, 1.20]). Likewise, having a marijuana use disorder was strongly
Downloaded from jiv.sagepub.com at UNIV ARIZONA LIBRARY on June 5, 2015
2960
Journal of Interpersonal Violence 29(16)
Table 1. Sample Description of Those Who Reported Being in a Relationship in the Past Year, NESARC II, n = 25,778. %
n Intimate Partner Violence (IPV) No IPV Victimization only Perpetration only Both perpetration and victimization Alcohol use (among those who used alcohol during IPV) Alcohol use during IPV by respondent only Alcohol use during IPV by partner only Alcohol use by both partners during IPV Substance use Marijuana use disorder (past year) Alcohol use (Quantity typically used each occasion in the past year, range: 1-10)a Mental health Depression (past year) Mania (past year) Dysthymia Hypomania Panic disorder Social phobia Specific phobia General anxiety Post-traumatic stress disorder Antisocial personality disorder Demographic covariates Male Race and ethnicity White or Caucasian Black or African American Hispanic or Latino Other race Agea
23,370 580 647 1,028 43 146 153 399 4.72
91.2 2.26 2.52 4.01 n = 342 4.19 14.22 14.90 1.55 0.05
2,143 560 302 382 494 658 2,082 984 1,849 62 11,850
8.31 2.17 1.17 1.48 1.92 2.55 8.08 3.82 7.17 0.24 45.97
15,274 59.25 4,298 16.67 4,999 19.39 1,207 4.68 46.08 (0.10)
Note. NESARC = National Epidemiologic Survey on Alcohol and Related Conditions. aM and SE are reported.
associated with IPV victimization (OR = 2.61; 95% CI = [1.75, 3.89]) and victim/perpetration (OR = 2.65; 95% CI = [1.81, 3.89]), compared with nonvictims and non-perpetrators.
Downloaded from jiv.sagepub.com at UNIV ARIZONA LIBRARY on June 5, 2015
2961
Reingle et al. Table 2. Prevalence Rates of IPV by race, ethnicity, and gender, NESARC II, n = 25,625. IPV Victimization/Perpetration
Men
Women
χ2
Non-victim/non-perpetrator Victim only Perpetrator only Victim and perpetrator
.93 .03 .01 .03
.91 .02 .03 .04
28.59***
IPV Victimization/Perpetration White Non-victim/non-perpetrator Victim only Perpetrator only Victim and perpetrator
.94 .02 .02 .03
African American
Hispanic
χ2
.86 .04 .04 .07
.89 .03 .03 .05
10.18***
Note. IPV = intimate partner violence; NESARC = National Epidemiologic Survey on Alcohol and Related Conditions. ***p
