Drinking and Intimate Partner Violence Severity Levels Among U.S. Ethnic Groups in an Urban Emergency Department

Abstract

Background:

Emergency departments (EDs) provide care to ethnically diverse populations with multiple health-related risk factors, many of which are associated with intimate partner violence (IPV). This paper examines ethnic-specific12-month rates of physical IPV by severity and their association with drinking and other sociodemographic and personality correlates in an urban ED sample.

Methods:

Research assistants surveyed patients at an urban ED regarding IPV exposure, as well as patterns of alcohol and drug use, psychological distress, adverse childhood experiences and other sociodemographic features.

Results:

The survey (N=1037) achieved an 87.5% participation rate. About 23% of the sample reported an IPV event in the past 12 months. Rates were higher (p<.001) among Blacks (34%), Whites (31%) and multiethnic (46%) respondents than among Asians (13%) and Hispanics (15%). Modeled results showed that Black respondents were more likely than Hispanics (reference) to report IPV (AOR=1.69; 95%CI=1.98–2.66; p<.05), and that respondents’ partner drinking was associated with IPV (AOR=1.85; 95%CI=1.25–2.73; p<.01), but respondents drinking was not. Use of illicit drugs, younger age, impulsivity, depression, partner problem drinking, adverse childhood experiences, and food insufficiency were all positively associated with IPV.

Conclusions:

There was considerable variation in IPV rates across ethnic groups in the sample. The null results for the association between respondents’ drinking and IPV was surprising and may stem from the relatively moderate levels of drinking in the sample. Results for ethnicity, showing Blacks as more likely than Hispanics to report IPV, support prior literature.

Introduction

Intimate partner violence (IPV) is the physical violence, sexual violence, stalking and psychological aggression (including coercive tactics) that is perpetrated against a partner in a romantic relationship.¹ This paper focuses on physical IPV, which as other types of IPV, remains a major public health problem in the United States (U.S.). Surveys with face-to-face interviews of heterosexual couples in the community over the last 30 years have showed that about 1 in 5 couples in the U.S. reports at least one episode of physical IPV in the past 12 months. ^2–4 These rates also vary considerably by ethnicity. For instance, in one survey of couples 17% of Hispanic couples, 23% of Black couples, and 11% of White couples reported an incident of male-to-female physical IPV in the past 12 months.⁴ Rates of female-to-male physical IPV were 21% among Hispanics, 30% among Blacks, and 15% among Whites. More recent cross-sectional data from the 2010–2012 National Intimate Partner and Sexual Violence Survey (NISVS) show 12-month rates for physical IPV of 3.9% among women and 4.7% among men. These rates are lower than those above likely due to differences in survey methods, especially telephone interviewing versus face-to-face, and interviews with one person only and not with both persons in the couple.

EDs remain the entry point, and sometimes the only setting for clinical care for physical and behavioral problems for a large part of the U.S. population, especially the 8.8% (28.3 million) without health insurance.⁵ EDs have higher prevalence rates of IPV than other health care settings, second only to substance use treatment units. ⁶ IPV screening of patients presenting at EDs show rates ranging from 9% to 37% for a 12-month timeframe, and as high as 46% for lifetime.^7–11 EDs are therefore an excellent setting for screening and identification of IPV and subsequent referral for treatment. Identification of ED patients involved in IPV also helps ED personnel to arrive at a better understanding of patients’ reasons for seeking care given that these patients may present with problems associated with IPV (e.g., bone fractures, PTSD, depression).

Unfortunately, the association between IPV and ethnicity has not been well examined in ED samples, even though EDs serve a high proportion of U.S. ethnic minority patients (e.g., ^12–14). Also, when ED data on ethnicity are analyzed, this is done in a limited way by using the variable as a correlate in multivariable analyses. The estimated rates usually show that Black women compared to Whites, Hispanics or all others combined, have higher rates of IPV. ^8,9 Black women reporting IPV are more likely to use tobacco, to be depressed, and abuse alcohol and drugs. ^9,15,16 Both Black and Hispanic women reporting IPV are also more likely to binge drink, use drugs and have a higher level of impulsivity personality traits than women who do not report IPV. ^12,17 Men who perpetrate IPV are more likely than non-perpetrators to be young, not married, Black or Hispanic, with lower education and income, unemployed, with heavier alcohol use, and higher rates of both alcohol and illicit drug abuse/dependence.¹⁷

Drinking has been consistently associated with IPV both in general population and in ED samples. Among U.S. couples in the community, about a third of those who report an IPV event in the past 12 months also report that one or both partners were drinking during the event.¹⁸ Alcohol has also been associated with more severe violence, more severe injuries from violence and with greater chronicity of violence.^19–21 In ED samples, binge drinking, heavier drinking, drinking problems, hazardous drinking as measured by higher Alcohol Use Disorder Identification Test (AUDIT) scores, mental health problems, a positive diagnosis of alcohol and or illicit drug abuse and dependence have all been associated with IPV. ^9,15,22

This paper examines ethnic-specific12-month rates of victimization, perpetration and mutual physical IPV and their association with drinking and other sociodemographic and personality correlates in an urban ED sample. It provides two distinct contributions to the existing literature: a) a detailed focus on the association between drinking, ethnicity and IPV; and b) analyses that recognize two levels of IPV severity, moderate and severe. Following results in the existing literature reviewed above, the analyses will: a) describe ethnic specific sociodemographic characteristics, drinking, drug use, psychological risk factors and IPV rates (perpetration, victimization, mutual violence, moderate and severe IPV) in the sample; b) test the expectation that Blacks and those who are younger will have a higher rate of IPV in general and also a higher rate of severe IPV; c) test the expectation that respondents weekly mean drinking volume, the mean monthly frequency of alcohol intoxication, and illicit drug use will be independently and positively associated with IPV, as will their partners’ problem drinking; and d) test the expectation that adverse childhood experiences, post-traumatic stress disorder (PTSD), depression, impulsivity, and anxiety will also be positively associated with IPV, with severity of IPV increasing as these problems become more severe.

Methods

Sample and data collection

Trained, bilingual (English and Spanish) research assistants (RAs) recruited non-emergent patients in the ED of a hospital Level I trauma center and county safety-net hospital. After receiving training about the study’s conceptual framework, data collection techniques, and protection of human subjects, the RAs pilot tested the survey with 41 patients (39% African American; 41% Hispanic; 44% male) who met the study’s eligibility criteria. The pilot testing process helped the research team identify obstacles to study recruitment and refine data collection procedures and provided the team with baseline information as to average survey interview length. After the RAs were debriefed with project staff and making minor adjustments to the questionnaire, the finalized survey data collection effort was launched.

The initial sample size estimate called for the enrollment of 800 married, cohabiting, or dating adults aged 18–50. This was based on calculations that using linear regression analyses, power would be 80% to detect a small overall effect (R² = .02) with 20 predictors, α = .05, and n=800. Power would be 85% to detect small incremental changes of adding single variables to the regression equations (ΔR² = .01) with 19 prior predictors, a prior R² of .10, and α = .05. The analytical objectives that guided these power analyses were: a) identify aspects of alcohol consumption and drinking context (e.g., type and frequency of venue utilization; amount of alcohol consumed per venue) that are associated with the occurrence and frequency of IPV; b) determine the extent that known IPV risk factors (e.g., adverse childhood experiences; impulsivity; depression) are associated with drinking contexts. Once the survey was being conducted, costs were lower than anticipated, and the sample N was increased to increase power for planned analyses and additional analyses.

Participants eligibility criteria included: 18–50 years old; English or Spanish speaker, residence in the county where the study was conducted, and married, cohabiting, or in a romantic (dating) relationship for the past 12 months. Patients who were intoxicated, experiencing acute psychosis or suicidal or homicidal ideation, were cognitively and/or psychologically impaired and unable to provide informed consent, in custody by law enforcement, or in need of immediate medical attention were excluded. Thirty-four patients who could not speak English or Spanish were excluded. Two interviewers per shift staffed the ED during weekday peak volume hours (9am – 9pm) to recruit eligible patients to the study.

Data were collected from February through December 2017. Patients could opt to be interviewed in English or Spanish. In this latter case, a Spanish version of the questionnaire, which had been validated through translation into Spanish and re-translation into English followed by verification, was then used. Figure 1 shows in sequence from top to bottom the number of patients identified in electronic health records, screened, found eligible, and interviewed. The side arrows and boxes show the number of patients from the preceding inclusion step that were not considered further. Initially, 1,066 patients (90% of those eligible) agreed to participate. Of these, 1037 were interviewed (87.5% participation rate, 53% female), of which 376 were interviewed in Spanish (36% of the full sample, and 72% of respondents who self-identified as Hispanic/Latino).

Figure 1.

Study sample recruitment

The RAs obtained informed consent in a private area adjacent to the ED waiting room, or in the patient’s room without others present. Twenty-nine patients terminated the survey interview before completion, mostly due to interruption for medical services. Ten interviewed patients identified themselves as American Indian/Alaskan Native, 7 as Native Hawaiian/Pacific islanders, and 26 did not report their race/ethnicity. These respondents are not part of the analytical sample herein (N=994). Patient survey data were collected by the RAs using computer assisted personal interview (CAPI) with computer tablets running the Qualtrics platform. Questionnaire development was guided by the study’s conceptual model as described in Cunradi et al. ²³ and formative qualitative research conducted at the outset of the study. Regarding the latter, project ethnographers conducted semi-structured interviews with 30 non-acute ED patients in which they asked open-ended questions as to respondents’ use of alcohol and the venues where they drank. Missing data were negligible; none of the variables analyzed in this paper had more than 4% information missing. These data therefore were left as missing. The project was approved by the Committee for the Protection of Human Subjects of the hospital where the study was conducted.

Measurements

Ethnicity:

This was based on self-identification. Respondents were asked: What racial or ethnic group(s) best describes you? (More than one category may be checked.): Asian; Black, African American; Latino, Hispanic; White, Caucasian; Native American Indian/Alaskan Native; Native Hawaiian/Other Pacific Islander; Some Other Race (specify). Respondents who selected more than one category were identified as multiethnic.

Intimate partner violence:

Physical IPV was measured with the 12 items on physical assault in the Revised Conflict Tactics Scale (CTS2), ²⁴ which have been used in prior ED-based IPV studies.^13,25,26 Two levels of IPV severity, moderate and severe, were operationalized based on previously published reports.²⁷ Moderate violence consisted of at least one of the following acts: threw something at partner that could hurt; pushed or shoved; grabbed; slapped; twisted partner’s arm or hair. Severe violence consisted of: kicked; punched or hit with something that could hurt; beat up; choked; burned or scalded on purpose; slammed against a wall; used a knife or gun. Cronbach’s alpha for the scale in the dataset under analysis was .85.

Quantity and frequency of drinking:

Respondents were asked the frequencies with which they had 1 or more, 2 or more, 3 or more, 6 or more, and 9 or more drinks in the past 4 weeks. A “drink” was defined as a 12-ounce can of beer, a 5-ounce glass of wine, or a 1-ounce shot of liquor. Respondents who did not use alcohol in the past 4 weeks were asked the same questions over the past year. Using a mathematical model described in Gruenewald et al.^28,29 a measure of the weekly mean drinking volume was calculated. This measure was log transformed due to skewness for inclusion in the ordered logistic regression. Test-retest reliabilities of these measures vary from r=0.65 for drinking quantities to r=0.85 for drinking frequencies.³⁰

Alcohol intoxication:

This was assessed with the following question: “During the past 12 months, about how many times did you drink enough to feel intoxicated or drunk, that is, when your speech was slurred, you felt unsteady on your feet, or you had blurred vision?”. This measure was log transformed due to skewness for inclusion in the ordered logistic regression.

Illicit Drug Use:

This measure covered drug use in the 12 months preceding the interview. Respondents were asked how many days they used the following drugs: marijuana or hashish (without a doctor’s prescription), amphetamines, cocaine, heroin, and pain relievers not prescribed for you. In the current study, illicit drug used was operationalized as a dichotomous variable of either any or no drug use.

Partner problem drinking:

The 3-item AUDIT-C was used to measure the respondent’s assessment of his/her spouse/partner’s drinking.^31,32 Male partners with a score above 4, and female partners with a score above 3 in the test 0–12 scale were considered hazardous drinkers. Internal consistency reliability for this scale in the dataset under analysis as measured by Cronbach’s alpha was .81.

Adverse Childhood Experiences (ACE):

The modified ACE ³³ measures exposure to six adverse experiences the respondent may have had “during their first 18 years of life:” (1) exposure to a mentally ill person in the home; (2) parent/caregiver alcoholism; (3) sexual abuse; (4) physical abuse; (5) psychological abuse; and (6) violence directed against the respondent’s mother. These six exposures are summed to create the ACE variable (range=0–6). Internal consistency reliability (Cronbach’s alpha) in the dataset under analysis was .74.

Impulsivity:

This was measured with 3 items that assessed respondents’ agreement with the following statements: I often act on the spur-of-the-moment without stopping to think; You might say I act impulsively; Many of my actions seem to be hasty. Four response categories could be chosen ranging from “not at all” to “quite a lot.” Alpha reliability for the scale in the dataset under analysis was .79.

Anxiety and Depression:

This was measured with Hospital Anxiety and Depression Scale (HADS) ³⁴, which has been successfully used in previous ED studies (see ^35,36). Both anxiety and depression were measured with 7 items each on a 4-point Likert-type scale (e.g., 1=not at all; 4=very often). Alpha reliability in the dataset under analysis was .69 for the depression scale and .81 for the anxiety scale. Following Brennan et al.³⁷ a cutoff point equal to or higher than 8 identified positives in both scales. This cut off gives sensitivities of .82 and .78, and specificity of .74 and .78 for depression and anxiety, respectively.

Post-Traumatic Stress Disorder (PC-PTSD):

This measure is from the Primary Care Screener for PTSD³⁸, and it too has been successfully used in ED studies (see ^39,40). It asks subjects about past-month symptoms they may have felt in response to a “frightening, horrible or upsetting” experience. Answers were coded yes or no, and a score of 3 or more is considered positive. Internal consistency of this scale in the data set under analysis was Alpha=.83.

Perceived Neighborhood Disorder (PND):

This was measured with Hill and Angel’s 10-item measure of neighborhood disorder.⁴¹ Items cover the extent to which assaults, muggings, drug dealing, gangs, unsafe streets, thefts, teenage pregnancy, abandoned houses, police not available, unsupervised children, and high unemployment, are neighborhood problems. Respondents could select one of the following 3 categories to answer each item: Not a problem, somewhat of a problem, a big problem. Cronbach’s alpha was .88 in the data set under analysis.

Other Sociodemographic variables:

Gender. A dichotomous variable coded as male and female (reference). Age. The age of respondents was used as a categorical variable: 18–29, 30–39, and 40 years and older (reference). Level of education. Respondents were categorized into four education categories: a) less than high school (reference); b) completed high school or GED; c) some college or technical or vocational school; d) completed 4-year college or higher. Importance of Religion. This variable had 4 categories: very important (reference), somewhat important, not very important, no important at all. Marital status: This is a 4-category variable: a) married (reference); living with partner; b) separated or divorced, c) never married. Widowers (n=33) were dropped from the analyses because 23 had no alcohol use disorder, which created estimation problems in the multivariable analysis. Food insufficiency: Respondents were asked their level of agreement with the statement, “In the past 12 months, the food we bought ran out and we didn’t have money to get more.” Response categories were: never (reference), sometimes true, often true.

Statistical Analyses

All analyses were conducted with Stata 15.0. ⁴² Associations in bivariate analyses (Tables 1, 2, and 3) were tested with chi-square tests with level of statistical significance adjusted using a Bonferroni correction when necessary. This indicated that significance levels for Tables 1 and 2 should be .05/10 = p<.005. Denominators for estimating drinking indicators in Table 2 were drinkers only. All other rates use the full sample as the denominator.

Table 1:

Sociodemographic indicators by ethnicity in an urban emergency room sample.

	Hispanics	Asians	Blacks	Whites	Multiethnic
Sociodemographic Characteristics (Ns)	(520)	(51)	(299)	(68)	(56)
Mean age (SE) a NS	35.3 (.35)	36.9 (1.11)	34.9 (.53)	35.7 (.99)	34.0 (1.22)
% Women NS	50	51	58	53	63
% High School Graduate ***	48	90	86	85	93
% Annual Income up to $40,000***	61	67	62	53	68
% Married***	51	69	24	22	14
% Unemployed***	22	26	39	34	31
% “Cannot make ends meet” NS	22	12	17	25	23
% Reports food Insufficiency “at least sometimes” NS	49	37	49	47	59

^aSignificance level for differences between age means after Bonferoni correction:.05/10, all pairwise t tests= NS Chi² NS: not significant

^***p<.001

Table 2:

Drinking, drug use, and psychological indicators by ethnicity in an emergency room sample.

	Hispanics	Asian Americans	Blacks	Whites	Multi-ethnic
Drinking Indicators (Ns)a	(270)	(28)	(221)	(50)	(45)
Weekly Mean N. of Drinks (SE)b	3.5 (.44)	3.3 (1.42)	6.3 (.85)	7.5 (1.82)	6.8 (1.59)
Monthly Mean N. of Intoxication Events (SE)c	1.0 (.22)	1.6 (1.08)	.7 (.26)	1.4 (.70)	3.4 (1.35)
Respondents Drug Use and Partner Drinking Indicator (Ns)	(520)	(51)	(299)	(68)	(56)
% Used Any Illicit Drug Past 12 Months***	17	18	53	57	57
% Partner AUDIT-C Positive***	17	14	28	29	29
Psychological Indicators (Ns)	(520)	(51)	(299)	(68)	(56)
% Two or More Adverse Childhood Experiences***	30	14	35	65	66
% PTSD Positive***	19	16	27	48	55
% Anxiety Positive***	28	20	42	63	52
% Depression Positive NS	16	12	18	15	18

^aDenominators for drinking indicators are drinkers only

^bpairwise t test Hispanic x Black p<0.01; Hispanic x White p<0.01; Hispanic x multiethnic p<01; all others p=NS.

^call pairwise t tests NS; Chi² NS: not significant

^***p<.001

Table 3:

IPV rates (proportions) by ethnicity in an emergency room sample.

	Hispanics (520)	Asian Americans (51)	Blacks (299)	Whites (68)	Multi-ethnic (56)
% Any IPV***	15	16	34	31	46
% Perpetration***	2	0	6	7	7
% Victimization	5	4	8	10	9
% Mutual Violence	8	9	20	13	30
IPV Severity***
% Moderate IPV	8	8	16	15	23
% Severe IPV	6	6	17	16	23

Chi²

^***p<.001

Multivariable analysis of IPV severity in Table 4 was conducted with Stata’s “ologit” procedure, which implements an ordered logistic regression under a proportional odds assumption. Test results indicated that the model tested fit the proportional odds assumption: chi²=36.01 with df=27 and p=0.11. Therefore, AORs in Table 4 represent both the odds of Moderate plus Severe IPV contrasted with No IPV, and the odds of Severe IPV contrasted with No IPV plus Moderate IPV. Independent variables were entered in the model in one step. These variables were selected for inclusion in the model based on previous results in the literature indicating they had a statistically significant association with IPV in community samples (e.g., neighborhood disorder), ED samples, or both (e.g., gender, age, marital status, ethnicity, religion, drinking volume and intoxication, illicit drug use, impulsivity, childhood adverse experiences, depression).^{4,15,9,25,43–45}

Table 4:

Multivariate Ordered Logistic Regression of Intimate Partner Violence (IPV) Severity on Sociodemographic, Psychological and Drinking-Related Variables (N=917).

	AOR	95% CI
Ethnicity (Ref: Hispanics)
Asian American	1.63	0.61–4.35
Black*	1.69	1.98–2.66
White	1.17	.55–2.45
Multiethnic	1.61	0.80–3.23
Any illicit Drug Past 12 Months***	2.44	1.65–3.62
Age (Ref: 40+)
18–29**	2.08	1.31–3.33
30–39	1.24	0.79–1.94
Impulsivity Scale Score***	1.15	1.07–1.24
Partner AUDITC Positive**	1.85	1.25–2.73
Post-Traumatic Stress Disorder*	1.57	1.06–2.34
Childhood Adverse Experiences*	1.14	1.02–1.29
Food Insufficiency (Ref: Never True)
Sometimes True**	1.76	1.19–2.61
Often True*	1.95	1.14–3.35
Neighborhood Social Disorder*	1.04	1.004–1.07
Depression Scale**	1.07	1.01–1.13
Male (Ref: Female)	1.04	0.72–1.50
Anxiety Scale	0.96	0.91–1.01
Log Weekly Mean Drinking Volume	1.15	0.93–1.42
Log Monthly Mean Frequency of Intoxication	1.11	0.80–1.55
Marital Status (Ref: Married)
Living with Partner	1.14	0.72–1.81
Separated/divorced	1.62	0.62–4.20
Never Married	1.02	0.63–1.66
Education (Ref: No High School)
High School Completed	1.36	0.86–2.14
Some College/Technical	1.36	0.80–2.30
College Degree or More	0.68	0.29–1.60
Importance of Religion (Ref: Very Important)
Somewhat Important	0.67	0.43–1.04
Not Very Important	0.78	0.39–1.55
Not Important at All	0.79	0.32–1.96

Note: *p<05

^**p<.01

^***p<.001

Results

Sociodemographic indicators

There was no variation in the mean age of participants nor in the proportion of women across the ethnic groups (Table 1). However, the proportion of high school graduates, of participants with an annual income up to $40,000/year, of married participants and unemployed participants varied significantly across ethnic groups. The proportion of respondents reporting not being able to “make ends meet” and the proportion reporting food insufficiency “at least sometimes” did not vary significantly across ethnic groups.

Drinking, Drug Use, and Psychological Status Indicators

Pairwise comparisons indicated that there were significant differences in the monthly mean number of alcohol intoxication events between Hispanics and Blacks, Hispanics and Whites, and Hispanic and multiethnic participants (Table 2). The proportion of participants that used drugs in the past 12 months was higher among Blacks, Whites and multiethnic respondents, with over half of these respondents reporting such use. About a third of Blacks, Whites and multiethnic respondents had a partner with a positive AUDIT-C, while among Hispanics and Asian Americans this was true of less than a fifth of respondents.

The analyses of drinking indicators in Table 2 was repeated with each ethnic group split between those who reported and those who did not report IPV (data not shown). There were significant differences in the weekly mean number of drinks between Hispanics and the multiethnic group. There also were statistically significant differences for the proportion of Hispanics and Blacks with an AUDIT-C positive partner.

The proportion of respondents reporting 2 or more ACEs, PTSD positive, and Anxiety positive was significantly different across ethnic groups. The proportion with 2 or more ACEs and PTSD positive was higher among Whites and those with a multiethnic background. Blacks, Whites and multiethnic respondents were more likely to score positively in the anxiety scale.

Intimate Partner Violence

About 23% of the sample reported at least one incident of IPV in the past 12 months (data not shown). Any IPV as well as IPV perpetration, victimization and mutual violence were significantly higher among multiethnic respondents, followed by Blacks and Whites (Table 3). Mutual violence was significantly higher among Blacks and multiethnic respondents. Multiethnic respondents, Blacks, and Whites reported higher levels of both moderate and severe IPV than Hispanics and Asian Americans.

Correlates of Intimate Partner Violence

Correlates with statistically significant associations with IPV severity are shown first in Table 4. Blacks were a little over 1.5 times more likely than Hispanics to report IPV. Respondents who reported illegal drug use in the past 12 months were over two times more likely than those who did not report drug use to report IPV. Younger respondents, those who have higher scores in the impulsivity scale, those who had a partner with a positive AUDIT-C score, those with a positive PTSD screener, and those with a higher ACE were also more likely to report IPV. In addition, respondents who reported that they experienced food insufficiency “sometimes” or “often” were about 1.76 and 1.95 times more likely, respectively, than those who never experienced food insufficiency to report IPV. Finally, respondents who scored higher in the neighborhood social disorder scale as well as those with higher scores in the depression scale were also more likely to report IPV.

Discussion

The only hypothesis put forward in the Introduction that was not confirmed was that about a positive association between anxiety and IPV. Results in Tables 1, 2 and 3 and the multivariable model in Table 4 show that the sample analyzed herein has a profile of IPV and its correlates that closely resembles that of other ED samples. This includes an overall rate of IPV (23%) that is in the mid-range of the rates of 9% to 37% for a 12-month timeframe that have been described in ED samples.^7–9,11

The model in Table 4 indicates that, with controls for several confounders, Black respondents compared to Hispanics, those who used illicit drug use, and those ages 18–29 compared to age 40+ have increased odds of IPV. Regarding ethnicity, it is possible that the relatively small number of Asian, Whites and multiethnic respondents lead to lack of power to assess some of the associations in these three groups. As in other community and ED samples, Blacks were significantly more likely to report IPV.^8,9,27,43,46

Other variables that increased significantly the odds of IPV were impulsivity, partner with a positive score in the AUDIT-C, PTSD, ACEs, food insufficiency, neighborhood social disorder, and depression. Impulsive behavior is associated with lower behavior control, and thus a higher chance of acting out a violent behavior.^12,47 A partner with a positive score in the AUDIT-C underscores that although drinking by one partner in the dyad may be enough to increase the odds of IPV, frequently (in a third of the cases), IPV occurs when both partners are drinking.^4,43

PTSD is associated with increased levels of anxiety and depression, which can be a result of victimization from IPV ⁸ or increase the odds of aggressive behavior and thus IPV. ACEs, which often include neglect, maltreatment, sexual, emotional and physical abuse, as well as observing IPV between parents or other household members, has been consistently identified as a factor of risk for involvement in IPV during adult life.^45,48 Food insufficiency is an indicator of socioeconomic difficulties for a couple and other household members, but it perhaps triggers more anxiety and a stronger sense of insecurity than, say, less education or lower income, which then increases the odds of IPV. Finally, depression and perceived neighborhood social disorder are also positively associated with IPV. Recent studies have reported a relatively important irritability component in depression, which when present could lead to increased conflicts in a couple and then to IPV.^49–51 Depression can also be a consequence of perpetration or victimization by IPV.^43,47,52 Perceived neighborhood social disorder can lead to conditions where behavioral norms are more lax and more accepting of violence, and informal social controls that minimize violence (e.g., neighbors who call the police or intervene) are not present.⁴⁴

Altogether, the results show that IPV is associated with an array of risk factors, and that it is comorbid with other behavioral and psychological disorders. Health professionals working in ED settings must be aware of this comorbidity and look for common factors of risk across behavioral problems, apply screens to detect these problems and be ready to intervene and refer to specialized services if necessary. Given that IPV is often associated with physical trauma (e.g., lacerations, fractures), the presence of IPV factors of risk identified in this paper in patients with such presentations, especially if the patient is a woman, should guide ED personnel to investigate history of IPV. ED personnel can also use the knowledge about IPV correlates and risk factors in this paper to develop more personalized and culturally appropriate assessments of patients who have screened positive for IPV, building rapport, and thus maximizing the chances that these patients will follow up with suggested referrals.

Limitations

The major limitation in the analyses is the small number of respondents in some of the ethnic groups interviewed. This reflects the ethnic composition of ED samples in urban areas and in the region the study took place. This limitation leads to lack of power for some of the analyses presented herein. Because the study was conducted in a single ED, results may not generalize to other EDs and other health settings. Exclusion of subjects who were alcohol intoxicated from the study may have led to an underestimation of the association between drinking and IPV. In addition, recall bias may have affected subjects’ information about events that reached back over 12 months. Finally, if under-reporting of IPV varies across ethnic groups, those groups that are more willing to report will appear to have higher prevalence rates.

Conclusions

First, as seen here and in other ED studies reviewed above, the association between IPV and ethnicity matters; Blacks and perhaps multiethnic groups are more at risk for involvement in this type of problem than other groups. In the long term, this identification of population groups with higher prevalence of IPV will further understanding of potential social and cultural correlates of this public health problem, which may help develop more effective approaches to address IPV in and outside ED settings. More immediately, this informs decisions about what groups should be the focus of specialized personnel and specific IPV-related actions such as brief intervention or referral to treatment.