The Prevalence of Intimate Partner Violence and Association with Depression in University Students: Results of a Cross-sectional Study

Abstract

The purpose of this cross-sectional study was to determine the prevalence of IPV among university students, investigate potential predictors of IPV in this population, and study the link between IPV and depression. The survey included socio-demographic, relationship quality, and depression related questions. From 498 respondents, prevalence of IPV was 4.8%, depression was 30.9%, and suicidal ideation was 20.3%. After adjusting for co-variates and confounders, relationship satisfaction (OR=0.201, 95% CI: 0.101 to 0.401, p<0.001) and jealousy (OR=0.270, 95% CI: 0.094 to 0.776, p=0.015) were significant predictors of IPV. Relationship satisfaction predicted depressive disorders (OR=0.504, 95% CI: 0.365 to 0.698, p<0.001). IPV trended towards predicting presence of a depressive disorder (OR=0.436, 95% CI: 0.170 to 1.113, p=0.083). Relationship satisfaction and jealousy predicted IPV. Although IPV did not predict depression, poor relationship satisfaction increased odds of depression, implicating the influence of relationship satisfaction on both IPV and depression.

Introduction

Intimate partner violence (IPV) is of considerable concern in the university setting. However, previous reports to document rates of physical abuse among intimate partners have been limited by small sample sizes and conflicting results. Furthermore, several studies lack data from male students or restricted data collection to undergraduate students. For example, Lewis and Fremouw described intimate partner abuse in one of three college couples (Lewis and Fremouw, 2001). By comparison, Kaukinen, Gover, & Hartman reported that 27% of female college students experienced some form of dating violence (Kaukinen et al., 2012). Studying rates of IPV among males and females, and among undergraduate and graduate students, is necessary to lead grounded efforts to prevent and address IPV in the university setting.

More than physical injury, IPV can also cause deterioration in mental health and even lead to the development of major psychopathology (Sugg et al., 2015). In fact, Lagdon, Armour, & Stringer reported a significant association between IPV and both anxiety and depression in a systematic review of the literature (Lagdon et al., 2014). Although, the relationship between IPV and poor mental health outcomes has been previously documented, research in college students has produced conflicting findings, been limited by small sample sizes, or lacked data from male students (Sugg et al., 2015; Kamimura et al., 2016; Kamimura et al., 2016; Garner and Sheridan, 2017; Christopher and Kisler, 2012). College students have high rates of anxiety and depressive disorders as well as other mental disorders but often do not seek treatment (Pendi et al., 2015; Eisenberg et al., 2007; Zivin et al., 2009). Therefore, investigating the potential link between IPV and mental disorders among college/university students is paramount in order to better understand the etiological factors associated with psychopathology and spur efforts to screen for these problems on-campus.

In order to address these issues, the study was designed with the following objectives: (1) to establish the prevalence of IPV, depression and suicidal ideation in university students, (2) identify potential predictors of IPV, and (3) study the link between IPV and depression in this population.

Method

Institutional Review Board approval was obtained prior to conducting this study. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines were used in the reporting of this study (Vandenbroucke et al., 2007).

The survey tool consisted of a socio-demographic and relationship quality questionnaires, Relationship Assessment Scale (RAS), Hurt-Insult-Threaten-Scream (HITS) Domestic Violence Screening Tool, and Patient Health Questionnaire-9 (PHQ-9). Socio-demographic items included questions about sex, age, sexual/affectional orientation, institution, student type, student class, religious affiliation, ethnicity and/or race, parental household income, relationship type, relationship status, and substance use. Relationship quality items included questions about jealousy, flirting, cohabitation, sexual activity, and sexually explicit electronic messaging.

The Relationship Assessment Scale (RAS) was included as a global measure of relationship quality after previously demonstrating test-retest reliability and validity in the university setting (Renshaw et al., 2010). HITS Domestic Violence Screening Tool, a highly-studied questionnaire to measure IPV, was selected as a concise tool to measure intimate partner violence (Shakil et al., 2005; Rabin et al., 2009). The scale was modified to include additional items to capture past instances of IPV up to one year, such that individuals who were not in a relationship at the time of participation could report previous experiences of IPV. A total score ≥10 was considered positive for experience of IPV (Sherin et al., 1998). Finally, the PHQ-9 scales have each demonstrated reliability and validity in numerous settings and are widely accepted as screening tools of depression severity (Manea et al., 2015). The PHQ-9 has been previously used to identify depression in undergraduate and graduate students (Eisenberg et al., 2007). A cutoff PHQ-9 score ≥10 was used to detect moderate to severe depression (Manea et al., 2015). Suicidal thoughts were identified by a score ≥1 on question 9 of the PHQ-9. The rationale for incorporating these standardized scales into one questionnaire was based in part on the recommendation of the campus sexual assault prevention and resource services team that was consulted for this study.

The survey was administered online to students at a large, public university in California from September 30^th, 2016 to January 30^th, 2017. Recruitment began on June 1^st 2016. Participants were offered an opportunity to be entered into a lottery for an e-gift card. Only full-time graduate or undergraduate students aged 18 years or older were eligible to participate. Participants constituted a voluntary sample recruited indirectly via emails sent by university faculty listed in the directory (n=1997). Data were not collected on whether faculty chose to send electronic communications. Respondents were given an option for participating in a lottery for an e-gift card.

The survey was made available online via secure and encrypted online tool, Research Electronic Data Capture (REDCap) (Harris et al., 2009). To calculate a suitable sample size to assess IPV as a risk factor and predictor for depression, a power analysis was performed with α set at 0.05 and power of 80%. Based on a conservative estimate of 20% for moderate to severe depressive disorders in those who experienced IPV, a sample size of n=342 was required to detect odds ratio (OR) of 2.0 predicted by IPV as modeled by logistic regression.

Responses with missing data were excluded. Potential respondents were given the opportunity to end the survey at any time with the understanding that their partial data would be excluded from analyses. Categorical socio-demographic and relationship quality items with more than two response categories were dichotomized prior to further analysis and sensitivity analyses were performed for all variables dichotomized this way. Age (18–20 years vs 21 or older), sexual orientation (heterosexual vs non-heterosexual), class (lowerclassmen including freshman and sophomore students vs upperclassmen including junior, senior and students in their 5^th year or later), religion (affiliated vs non-affiliated), ethnicity (Asian, Caucasian), race (Hispanic vs non-Hispanic), parental income ($34,000 and below vs $35,000 and above), alcohol (use vs no use), tobacco (use vs no use), marijuana (use vs no use), other drugs (use vs no use), relationship type (exclusive vs non-exclusive or single), and relationship status (committed vs non-committed). Group comparisons were made using Student’s t-test for continuous variables and chi-square for categorical variables. Variables that achieved statistical significance according to t-test were also incorporated as potential confounders into a binary logistic regression model. Logistic regression models were used to identify potential predictors.

Analyses were carried out using Microsoft^® Excel Version 15.11.2 for Mac (Microsoft, Redmond, WA, USA) and IBM^® SPSS^® Version 22.0 for Mac (IBM, Chicago, IL, USA).

Results

Responses were collected from 687 participants. The university enrollment during the time period of data collection was 33,467. There were 687 entries with 498 responses incorporated into analyses, constituting 2.1% and 1.5% of total enrollment, respectively. Of those, 498 (72.5%) had complete responses and were incorporated into further analyses. Due to the recruitment strategy used for this study, the total number of students reached could not be determined, and a response rate could not be calculated. Descriptive information regarding socio-demographic and relationship quality data are provided in Table 1 and Table 2, respectively.

Table 1.

Participant Socio-demographic Characteristics

Category	n (%)	Category	n (%)
Sex		Ethnicity or race
Female	388 (77.9)	Asian	188 (37.8)
Male	110 (22.1)	Caucasian	115 (23.1)
Age (years)		Hispanic	99 (19.9)
18–20	269 (54.0)	Other	96 (19.3)
21 and above	229 (45.0)	Household income
Sexual orientation		Less than $34,999	148 (29.7)
Heterosexual	400 (80.3)	More than $35,000	350 (70.3)
Non-heterosexual	98 (19.7)	Alcohol use
Student type		No	139 (27.9)
Undergraduate	438 (88.0)	Yes	359 (72.1)
Graduate	60 (12.0)	Tobacco use
Student class		No	431 (86.5)
Lowerclassmen	219 (44.0)	Yes	67 (13.5)
Upperclassmen	279 (56.0)	Marijuana use
Religion affiliation		No	348 (69.9)
Affiliated	292 (58.6)	Yes	150 (30.1)
Unaffiliated	206 (41.4)	Other drug use
		No	454 (91.2)
		Yes	44 (8.8)

Table 2.

Prevalence of Intimate Partner Violence, Depressive Disorders, and Suicidal Ideation

Category	n (%)	Category	n (%)
Intimate Partner Violence		Suicidal ideation
No	24 (4.8)	Not at all	397 (79.7)
Yes	474 (95.2)	Several days	72 (14.5)
Depressive Disorders		More than half the days	14 (2.8)
Minimal	206 (41.4)	Nearly every day	15 (3.0)
Mild	138 (27.7)
Moderate	80 (16.1)
Moderately severe	46 (9.2)
Severe	28 (5.6)

Intimate partner violence was experienced by 5% of respondents (n=24), a depressive disorder was experienced by 31% of respondents (n=154), and suicidal thoughts were experienced by 20% of respondents (Table 3).

Table 3.

Participant Relationship Quality Characteristics

Category	n (%)	Category	n (%)
Relationship type		Cohabitation
Single	198 (39.8)	No	425 (85.3)
Exclusive	285 (57.2)	Yes	73 (14.7)
Non-exclusive	15 (3.0)	Sexual activity
Relationship status		No	198 (39.8)
Single	185 (37.1)	Yes	300 (60.2)
Dating	243 (48.8)	Explicit messaging
Non-exclusive	15 (3.0)	No	347 (69.7)
Jealousy		Yes	151 (30.3)
No	385 (77.3)
Yes	113 (22.7)
Worry of flirting
No	381 (76.5)
Yes	117 (23.5)

None of the socio-demographic variables were significantly associated with IPV. Among relationship measures, only relationship satisfaction (OR=0.201, 95% CI: 0.101 to 0.401, p<0.001) and jealousy (OR=0.270, 95% CI: 0.094 to 0.776, p=0.015) were significant predictors of IPV. Relationship satisfaction was the sole significant predictor of depression (OR=0.504, 95% CI: 0.365 to 0.698, p<0.001).

Discussion

This study found the prevalence of IPV among university students to be 5%, a low percentage when compared to previous studies. This discrepancy may be attributable to differences in study settings, eras, and demographic characteristics. For example, rates of physical violence have been reported to be lower in a four-year university compared to a community college setting (Daley and Noland, 2001). Also, although Rouse, Breen, & Howell found that physical abuse was experienced by 11% of college students, that study was conducted nearly three decades ago (Rouse et al., 1998). More recently, Kaukinen, Gover, & Hartman described victimization rates as high as 27%, but among female college students only (Kaukinen et al., 2012). By comparison, the sample in this study consisted of both male and female students enrolled full-time at a four-year university. Further, the study institution has an active office that provides campus-wide support, education, and training to students to prevent or respond to acts of violence.

Intimate partner violence is likely underreported in this sample. There is a discrepancy between reporting IPV and prevalence data from large-scale studies, which suggest that a large percentage of cases of domestic violence go underreported (Garcia, 2004). In fact, victims often underreport domestic abuse due to privacy concerns, fear of violence from the offender, and even a desire to protect the offender (Felson et al., 2002). In addition, because the HITS contains items addressing primarily physical and verbal abuse, the prevalence of sexual violence, a component of IPV that may involve stalking, dating violence, or forced sex, was not captured by this study. Furthermore, the students may have had concerns regarding privacy as the survey link was forwarded to them by their professors. This is less likely in light of the high rates of self-reported depressive disorders and suicidal ideation. In addition, an incentive lottery was offered to increase response rate but may have degraded data quality because respondents that receive incentive may be more likely to skip items, rush, or answer items inaccurately (Laguilles et al., 2011; Barge and Gehlbach, 2011). Kaukinen, Gover, & Hartman suggested that since many students engage in their first romantic relationships during college, they may not perceive some forms of violence as abnormal and/or may not have developed the appropriate communication skills to resolve disputes encountered in their romantic relationships (Kaukinen et al., 2012; Kaukinen, 2014). As a result, students may not have thought to report certain violent behaviors in this survey.

Another major finding of this study was that nearly one-third of students (31%) endorsed moderate-to-severe depression and one-fifth (20%) reported suicidal ideation during several days over the previous two weeks. These results indicate an increasing burden of depression compared to previous studies such as that of Eisenberg, Gollust, Golberstein, & Hefner, which found the prevalence of depression to be 13.8% in undergraduate students and 11.3% in graduate students (Eisenberg et al., 2007). It is likely that the reasons for this increase are multi-factorial but may be in-part due to factors not measured in the study, such as increased use of internet and television, sedentary lifestyle, and lack of sleep, which appear to be pertinent to the current student population (Carli et al., 2014). It is also notable that the increasing prevalence of depression in this population mirrors global trends (Pendi et al., 2016; Lépine and Briley, 2011). These findings underscore the need for effective depression and suicide screening efforts in universities at the undergraduate and graduate levels.

In this study, severity of IPV did not differ significantly between males and females. Although previous studies found that female students experienced more severe forms of abuse, a recent study found that women experienced more minor injuries and men experienced more severe injuries (Rouse et al., 1988; Amanor-Boadu et al., 2011; Cercone, Beach, & Arias, 2005). Moreover, it has been suggested that relationship violence is often perpetrated mutually (Kaukinen et al., 2012). Neither age nor sexual orientation was a predictor of IPV according to this study. However, Porter & Williams described a significant association between gay, lesbian, and bisexual students with physical and sexual abuse in intimate relationships (Porter and Williams, 2011). The reason for this discrepancy may be due to differences in the sample. Porter & Williams studied college students from a private university in the northeastern United States while this study was a survey of undergraduate and graduate students from a large, public university in the southwest (Porter and Williams, 2011).

This study was the first to evaluate potential differences in IPV rates according to student type (undergraduate or graduate) and class (lowerclassmen and upperclassmen), finding that neither was a significant predictor of intimate partner abuse. Similarly, religious affiliation was not noted to affect experience of IPV in this sample of university students. Although it has been suggested that spirituality may reduce risk of victimization, there remains a paucity of research identifying protective factors such as religiosity in terms of preventing domestic violence (Kaukinen, 2014). Ethnicity or race was also not associated with increased experience of physical violence in intimate relationships, a finding supported in-part by the previous study by Porter & Williams, which did not identify a significant link between racial or ethnic minorities and physical abuse (Porter and Williams, 2011). Parental household income approximating the poverty level or below was the final socio-demographic predictor not significantly associated with IPV among university students.

Neither relationship type nor relationship status was a significant predictor for experience of IPV according to this study. While Kaukinen, Gover, & Hartman similarly reported few differences in physical violence between casual and exclusive relationships among female college students, this study supported those findings among a heterogeneous sample of both male and female students (Kaukinen et al., 2012). The majority of undergraduate and graduate students in the study sample reported alcohol consumption and nearly 1/3 endorsed marijuana use, but few indicated use of tobacco or other drugs. Alcohol, in particular, is a classic predictor of relationship violence, but there are no longitudinal studies that describe the effect of drug use on victimization in college students (Kaukinen, 2014). As a result, this study suggests that neither use of alcohol, marijuana, nor other drugs was a predictor of IPV in university students.

Notably, relationship satisfaction was linked to lower odds of IPV in this sample. Whether addressing these relationship attributes through counseling may help to prevent IPV constitutes one area for further study. Interestingly, jealousy was linked to lower odds of IPV. While previous reports have linked jealousy to IPV (Hilberman and Munson, 1977; Rounsaville, 1978), the findings of the current study suggest that the inverse is true in the university setting and that jealousy may be a protective factor. Why relationship jealousy negatively predicted IPV in this sample remains an avenue for future research.

This study did not find a significant correlation between IPV and depression in university students. Although the relationship between IPV and poor mental health outcomes has been previously documented, previous studies in the college setting have indicated conflicting findings (Sugg, 2015; Kamimura et al., 2016; Kamimura et al., 2016; Garner and Sheridan, 2017; Christopher and Kisler, 2012). For example, Kamimura, Nourian, Assasnik, & Franchek-Roa found that Iranian college students that experienced IPV reported greater depression severity (Kamimura et al., 2016). However, IPV was not significantly associated with depression in Japanese, Singaporean, South Korean, and Taiwanese college students [6]. But American college women that suffered from physical abuse from an intimate partner, experienced symptoms of both anxiety and depression (Christopher and Kisler, 2012). Also, among female college students with pre-existing disorders or disabilities, victimization by an intimate partner was linked to exacerbation of depression (Bonomi et al., 2017). In younger female students, depressive symptoms have been associated with an increased risk of moderate to severe IPV, suggesting that addressing depressive symptoms in adolescent women may help limit their experience of IPV (Lehrer et al., 2006). It has also been suggested that cultural factors may mediate the relationship between IPV and depression, although the present study did not find the related factors of ethnicity or race to significantly affect rates of IPV or depression (White and Satyen, 2015).

Relationship satisfaction predicted lower odds of depression in this sample in line with previous studies that have documented significant links between relationship satisfaction and depression (Whitton and Whisman, 2010; Whitton and Kuryluk, 2012; Whitton and Kuryluk, 2014). Moreover, Stavianopoulos noted that couples therapy to enhance relationship satisfaction produced a statistically significant reduction in symptoms of depression (Stavrianopoulos, 2015). These findings suggest utility in asking university students about their relationships in the context of mental health screening. A post-hoc analysis to determine the effects of relationship quality measures on suicidal ideation did not demonstrate similar predictive value.

The findings of this study are moderated by certain limitations in the design and methodology. First, this study and the majority of studies performed to measure domestic violence have utilized a cross-sectional study design, but more longitudinal studies are necessary to accurately address the questions posed. Second, although this study obtained a large sample size, convenience sampling may lead to a response bias toward students in a relationship at the time of participation or experiencing hardship in their relationships. Given that participants were asked to recall cases of abuse over a 12-month period, there is also potential for recall bias. Third, although the HITS domestic violence screening tool has been noted for its reliability and validity, it was not designed specifically to measure violence among university students and did not include items to address sexual abuse or financial abuse behaviors. Finally, given that the study sample may not completely reflect the total student enrollment (e.g. nearly 78% of respondents were female), the generalizability of these findings are limited.

Conclusions

This study found the prevalence IPV to be 5% among university students, although this is likely to be an underestimation. Prevalence of moderate to severe depressive disorders and suicidal thoughts was 30% and 20%, respectively. Both relationship satisfaction and jealousy were significant predictors of IPV, while only relationship satisfaction was a predictor of a depressive disorder. Further research on interventions for IPV, depressive disorders, and suicidal ideation in the university setting are needed.