Correlates of reproductive coercion among college women in abusive relationships: baseline data from the college safety study

Abstract

Objective:

The purpose of this study was to examine correlates of reproductive coercion (RC) among a sample of college women in abusive relationships.

Participants:

354 college students reporting a recent history of intimate partner violence (IPV).

Methods:

This study examines baseline data from a randomized controlled trial testing effectiveness of an interactive safety decision aid (myPlan).

Results:

Almost a quarter (24.3%) of the sample reported RC. Associated factors included races other than White (p=0.019), relationship instability (p=0.022), missing class due to relationship problems (p=0.001), IPV severity (p<0.001), technology abuse (p<0.001), traumatic brain injury-associated events (p<0.001), and depression (p=0.024).

Conclusions:

RC was a significant predictor of depression, with implications for providers working with abused college women regarding the need for mental health services concurrent with IPV/RC services. A larger proportion of women who experienced RC sought help from a healthcare provider for contraception, which suggests intervention opportunities for college health providers.

Introduction

Reproductive coercion (RC) involves coercive and controlling behaviors that interfere with a woman’s autonomous reproductive health decision-making. Examples may include preventing a woman from attending contraceptive appointments, throwing away birth control pills or refusing to use condoms, and pressure to get pregnant or to end a pregnancy. In community studies, prevalence of RC behaviors ranges from 5%¹ (past three months) to 25.9%² (lifetime). RC is strongly correlated with intimate partner violence (IPV) and unintended pregnancy;^3–6 however, studies of RC experiences among populations of IPV survivors are limited. IPV survivors in community samples have reported prevalence of RC behaviors ranging from 16.7%⁷ to 35%⁶ (both lifetime). An open question in this relatively new field of study is whether RC is a type of IPV or an independent construct with unique associated outcomes.

College-aged (18–24 years) women experience the highest prevalence of IPV/sexual violence,⁸ as well as unique forms of IPV such as technology abuse (abusive behaviors utilizing technology such as cellphones and computers);⁹ RC may be an important facet of such violence exposure, with significant impact on reproductive health and well-being. Studies of RC in college-aged women are limited. Two studies of college women aged 17–25 and 18–22 adapted existing RC measures to examine RC behaviors (such as condom refusal) without asking about intent to impregnate, and found a lifetime prevalence of 25.3¹⁰ to 29.6% for RC.¹¹ Another study of college women aged 18–25 used the standard 10 screening RC questions⁶ as well as an additional investigator-developed measure, and found a prevalence of 8% for past three months RC.¹² And one additional study of college students (male and female) used four investigator-developed screening questions and found a prevalence of 10.7% for past 12 months RC (13.9% among female students).¹³ This wide variation may be due to the difference in time frame examined or to the changes in sensitivity with removing intent from screening questions as well as different settings, definitions and measures used by investigators.

Studies of broad populations of women identify certain risk factors for RC, which include younger age,¹⁴ lower socioeconomic status (SES),¹⁵ single relationship status,^1,14,15 and African American race, Latina ethnicity or multiracial.^1,6,15 In addition to IPV and unintended pregnancy, RC is associated with health risks and outcomes such as early sexual initiation,¹⁶ multiple sex partners,¹⁶ decreased contraceptive use,¹⁶ increased depression¹⁶ and post-traumatic stress disorder (PTSD),^16,17 sexually transmitted infections (STIs)^16,18,19 and transactional sex.²⁰ IPV is associated with a range of health consequences; specifically among college-aged women, IPV is shown to be associated with mental health concerns such as PTSD²¹ and poor self-reported emotional health,²² traumatic brain injury (TBI),²³ substance use,²² increased STIs²² and decreased contraceptive use.²²

Studies of college women suggest several risk factors for RC specific to this population. Cohabitation with partners and living off-campus increased the odds of experiencing RC;¹² RC was also more likely to be perpetrated by older partners.¹⁰ Hispanic women had higher risk for RC in one study.¹² RC was also strongly associated with IPV for college women.^10–12. Studies identified negative health outcomes for college women associated with RC, such as decreased contraceptive use,¹¹ higher number of dating and sex partners,¹¹ and a higher risk of pregnancy, unintended pregnancy, and abortion.¹² College students also experience higher rates of mental health concerns such as stress, anxiety and depression relative to the general population.^24,25

From this existing literature it is clear that college women are extremely vulnerable to relationship violence which is strongly associated with RC, mental health concerns are important to address for this population, and other studies have not specifically looked at the relationship between RC and mental health outcomes in this population, nor established clear risk factors and outcomes related to RC. Therefore, the purpose of this study was to examine possible risk factors and outcomes of RC among a sample of college women in abusive relationships, with particular attention to the areas of mental health outcomes and risk factors specific to this age group, in order to inform screening and practice for providers of college health.

Materials and methods

This study examines baseline data from the myPlan College Safety Study conducted by Johns Hopkins University School of Nursing, a randomized controlled trial to test the effectiveness of an interactive safety decision aid app called myPlan (www.myPlanApp.org) designed to assist IPV survivors with difficult safety decisions and to develop a plan for safety.²⁶ Students were eligible to participate who were age 18–24, enrolled in college in Oregon or Maryland, had access to email and a device with Internet access that were safe from partner surveillance, identified as a woman, and self-reported experience of partner violence in the past 6 months. The study was approved by the Johns Hopkins University Medical Institutions (JHMI) institutional review board (Institutional Review Board (IRB) 00054334) for oversight of the research in Maryland and the Kaiser Center for Health Research IRB (Pro00004875) for oversight of the research in Oregon. Colleges and universities approached for partnership to recruit students individually determined if additional institutional approvals to advertise to their students would be required. IRB approval was required at three additional institutions (names of participating schools are confidential). Research assistants (RAs) received standardized human subjects research ethics training as well as extensive IPV advocacy training including safety assessment, technology safety, IPV resource referrals, and suicidality protocols.

In partnership with selected campuses in Maryland and Oregon that represented a diversity in campus environment (e.g. urban, rural, public, private, community colleges, historically black colleges and universities), the study was advertised on campuses via traditional methods that included flyers and referrals from advocates/counseling staff, and via online methods such as online campus bulletins, social media posts in campus groups, and student list serves. In addition, the study team advertised the study outside of specific campuses using online forums such as Craigslist and Facebook.

Advertisements describing a study on “unsafe relationships” directed interested students to the myPlan study registration website, where they received a brief overview of the study and then answered eligibility questions and received an automated determination of their eligibility. Eligible students were given the option to provide a safe email address or phone number and instructions on how to safely be contacted (e.g. only call in the afternoon, do not leave a message) by a trained RA to enroll in the study. Potential participants were also given the option to contact the study team directly by email or phone themselves rather than enter in their contact information into the website, to give multiple options to enroll using the method safest for them.

When the RA reached a potential participant using their preferred method of contact, the RA conducted verbal informed consent by phone or sent the consent script to the participant via email. Consent information included the purpose of the study, study procedures, confidentiality, and the risks and benefits of participation. Enrolled participants were directed to complete measures online via a study website or downloadable study app. Participants were provided a retail gift card to compensate for their time and expertise ($20 for completing the baseline survey).

Measures included demographics such as age, race, ethnicity, year in college/university, living situation, and employment, and demographics of and characteristics of the relationship with the abusive partner such as partner gender, relationship status, and length of relationship. Help-seeking by seeing a healthcare provider for contraception was also measured with one question asking “Have you talked to a healthcare provider about safer sex/birth control options?” in the prior year. Additional measures included:

IPV exposure:

1. The composite abuse scale (CAS) is a 30-item validated IPV measure with strong psychometric properties.^27,28 The CAS measures four independent types of abuse: Severe combined abuse (defined as experiencing one or more episodes of severe combined abuse, for example being kept from accessing medical care, use of a weapon, and rape or attempted rape, with or without other dimensions of abuse), physical abuse with emotional abuse and/or harassment, physical abuse only, and emotional abuse and/or harassment only.²⁷ We used an adapted response scale: “never, only once, several times or many times” in the last 6 months. Example items include: “Told me I was crazy,” “Pushed, grabbed or shoved me,” “Made me have sex when I didn’t want to.” Each of the 30 items is scored on a scale of 0–3. We then calculated a mean score and also calculated how many participants experienced any episode of each independent type of abuse.

2. The misuse of technology to perpetrate abuse (tech abuse) in the past 6 months was measured with our own scale compiled of items adapted from existing studies.^9,29,30 This consisted of 7 “yes or no” items, for example “Shared (or threatened to share) content that was supposed to be private (e.g. a nude picture)?” This yielded a score of 0–7 with higher values indicating more forms of technology abuse.

3. An item was included to measure the experience of severe abuse likely to cause traumatic brain injury (TBI) in the past 6 months (Has your partner done any of the following: drown you, suffocate or smother you, beating to the head, choke you?). Answering yes to any of these items was scored as experiencing 1 or more TBI events.³¹

4. As an additional measure of relationship safety, one investigator-developed item was included asking whether the participant had missed class due to relationship problems in the past 6 months (e.g. for getting help from campus health or a counselor, finding a new place to live, feeling too depressed to go to class, etc.).

Reproductive coercion was measured with two items. In the last 6 months: (1) Has your partner tried to get you pregnant when you didn’t want to be (e.g. by telling you not to use birth control, messing with your birth control, not using condoms, breaking condoms)? and (2) Have you prevented a pregnancy by using emergency contraception (Plan B, ella, Take Action, My Way, etc.) or ended a pregnancy using other methods, and did not tell your partner about it because you were afraid of your partner? RC was defined as a positive response to either or both questions.

Mental health:

Depression was measured with the Center for Epidemiologic Studies Depression Scale, Revised (CESD-R).^32,33 A 20-item self-report measure assesses for depressive symptoms in community samples. Items are rated based on frequency in the “past week or so” from 0 (Rarely or none of the time-less than one day) to 4 (Nearly every day for two weeks).

Drug and alcohol abuse:

The monitoring the future drug and alcohol questionnaire³⁴ with minor wording modifications was used to measure alcohol and drug use. Participants are asked to self-report on how many occasions they have used alcohol in the last 30 days and in the last 6 months, as well as binge drinking behavior. Participants are also asked on how many occasions they have used drugs (marijuana, club drugs/hallucinogens, stimulants/narcotics, prescription drugs, other) in the last 30 days and last 6 months. Alcohol abuse was defined based on current literature^35,36 as alcohol use on greater than 19 occasions in the past 30 days, or 40 or more occasions in the past 6 months, or any binge drinking (defined as drinking to the point of feeling “pretty drunk” on any occasions, or any occasions of four or more drinks in a row). Drug abuse was defined as any drug use in the past 6 months.

Descriptive statistics (means, standard deviations and frequencies) were used to describe the characteristics of the sample. Chi square and t-tests were used to examine differences between those who had and had not experienced RC on risk factors and covariates. Multivariable linear regression was used to examine the relationship between RC and depression controlling for IPV, age, race and ethnicity. All analyses were conducted in SPSS Statistics 25.0.³⁷

Results

Three hundred fifty-four college women were enrolled in the study. The mean age of the sample was 20.9 years (SD = 2.31) (Table 1). The sample was predominantly White (67.2%), and 13.6% of participants identified as Spanish/Hispanic/Latina. Thirty-seven percent described their relationship with their abuser as “on again/off again.” Having a partner who tried to get you pregnant when you didn’t want to be in the previous 6 months was reported by 14.7% of the total sample and preventing or ending a pregnancy without telling your partner due to fear of the partner was reported by 16.7% of participants. 24.3% of the sample reported any RC (answered yes to either or both of the RC questions).

Table 1.

Demographic characteristics of the total sample.

Characteristic	N (%)
Total sample	354
Age (mean, SD)	20.88 (2.31)
Age discrepancy with partner (mean, SD)	2.10 (3.43)
Race
White	238 (67.2)
Black or African American	69 (19.5)
Asian	40 (11.3)
American Indian or Alaskan Native	9 (2.5)
Native Hawaiian or Other Pacific Islander	7 (2.0)
Other	19 (5.4)
Ethnicity
Spanish/Hispanic/Latina	48 (13.6)
Not Spanish/Hispanic/ Latina	303 (85.6)
School enrollment
Full time	275 (77.7)
Part time	41 (11.6)
Did not answer	38 (10.7)
Type of school
Community college	60 (17.0)
Four-year state college or university	158 (44.6)
4-year private college or university	113 (31.9)
Other	23 (6.5)
Current student status
Undergraduate	332 (93.8)
Graduate student	20 (5.7)
Did not answer	2 (0.6)
Partner’s identified gender
Male	324 (91.5)
Female	29 (8.2)
Did not answer	1 (0.3)
Relationship status
Boyfriend/girlfriend	59 (16.7)
Ex-boyfriend/ex-girlfriend	241 (68.1)
Casual/hookup/friends with benefits	17 (4.8)
Married or married like relationship	9 (2.5)
Were in a married or married like relationship	7 (2.0)
Don’t know	9 (2.5)
Other	12 (3.4)
Living situation
Live off campus	209 (63.5)
Live on campus	120 (36.5)
Living with or effectively living with partner
Yes	63 (18.6)
No	277 (81.4)
Partner enrolled in any school (HS or college)
Yes	176 (51.6)
No	165 (48.4)
On again/off again relationship?
Yes	130 (37.6)
No	216 (62.4)
Reproductive coercion
Partner tried to get you pregnant when you didn’t want to be (e.g. by telling you not to use birth control, messing with your birth control, not using condoms, breaking condoms)	52 (14.7)
Prevented a pregnancy by using emergency contraception (Plan B, ella, Take Action, My Way, etc.) or ended a pregnancy using other methods, and did not tell partner about it because you were afraid of your partner	59 (16.7)
Experienced any form of RC in past 6 months	86 (24.3)
Severity of IPV
Missed class due to relationship problems
Yes	167 (49.4)
No	171 (50.6)
CAS score - mean (SD)	0.95 (0.6)
Tech abuse count - mean (SD)	3.16 (2.1)
1 or more TBI events
Yes	98 (27.8)
No	254 (72.2)
Substance abuse
Alcohol abuse
Yes	179 (50.9)
No	173 (49.2)
Drug abuse
Yes	162 (46.0)
No	190 (54.0)

Risk factors for RC

Table 2 summarizes the differences between women who had and had not experienced RC. Women who experienced RC were more likely to be races other than White than women who did not experience RC (43.0% vs. 29.3%, p = 0.019). They were also more likely to be in an “on again/off again” relationship (48.2% vs. 34.2%, p = 0.022). Women who experienced RC were more likely to report missing class due to relationship problems (66.3% vs. 44.2%, p = 0.001). Age, ethnicity, living with partner, alcohol abuse, drug abuse and age difference with partner were not associated with RC. Women who experienced RC had significantly higher scores on the CAS Scale (p < 0.001), indicating more types of IPV experienced, as well as the technology abuse scale (p < 0.001), indicating more forms of technology abuse. Women who experienced RC were nearly twice as likely to have also experienced an event that may cause TBI compared to those who had not experienced RC (43.0% vs. 22.9%, p < 0.001). RC was not significantly associated with seeking help from a provider for contraception, though a greater percent of those who experienced RC sought help from a provider compared to those who did not experience RC (65.9% vs. 55.3%, p = 0.087). Women who experienced RC were significantly more likely to also experience severe combined abuse (SCA) compared to other types of abuse as measured on the CAS scale than women who did not experience RC (89.5% vs. 70.3%, p < 0.001).

Table 2.

Risk factors for RC.

Characteristic	Experienced RC N (%)	Did not experience RC N (%)	p-valuea
Total sample	86 (24.3)	266 (75.1)
Age - mean (SD)	21.0 (1.88)	20.85 (2.43)	0.232
Age discrepancy with partner - mean (SD)	2.55 (3.34)	1.98 (3.45)	0.254
Race
White	49 (57.0)	188 (70.7)	0.019
All other races	37 (43.0)	78 (29.3)
Ethnicity
Spanish/ Hispanic/Latina	10 (11.8)	37 (14.0)	0.605
Not Spanish/ Hispanic/Latina	75 (88.2)	228 (86.0)
Partner school enrollment
Enrolled in any school (HS or college)
Yes	37 (45.1)	139 (53.7)	0.177
No	45 (54.9)	120 (46.3)
Alcohol abuse
Yes	42 (48.4)	137 (51.5)	0.667
No	44 (51.2)	129 (48.5)
Drug abuse
Yes	38 (44.2)	124 (46.6)	0.694
No	48 (55.8)	142 (53.4)
Living situation
Live off campus	54 (67.5)	155 (62.3)	0.396
Live on campus	26 (32.5)	94 (37.8)
Living with or effectively living with partner
Yes	16 (19.8)	47 (18.2)	0.757
No	65 (80.3)	211 (81.8)
On again/off again relationship?
Yes	40 (48.2)	90 (34.2)	0.022
No	43 (51.8)	173 (65.8)
Missed class due to relationship problems
Yes	53 (66.3)	114 (44.2)	0.001
No	27 (33.8)	144 (55.8)
CAS score - mean (SD)	1.26 (0.65)	0.85 (0.50)	<0.001
Tech abuse count - mean (SD)	4.01 (2.11)	2.91 (1.98)	<0.001
1 or more TBI events
Yes	37 (43.0)	61 (22.9)	<0.001
No	49 (57.0)	205 (77.1)
Type of abuse experienced
Severe combined abuse (SCA)	77 (89.5)	187 (70.3)	<0.001
Other types	9 (10.5)	79 (29.7)
Sought help from a provider for birth control
Yes	56 (65.9)	145 (55.3)	0.087
No	29 (34.1)	117 (44.7)

^aBased on t-test or chi-square test.

RC and depression

IPV, age, race and ethnicity are known to have independent associations with depression,^38–41 so when examining the relationship between RC and depression we controlled for these variables. This multivariable regression analysis revealed that RC was significantly associated with increased depression (p = 0.024). However, IPV had a stronger association with depression than RC (p < 0.001;Table 3).

Table 3.

Effect of RC and IPV on depression score - multivariable regression results (N = 257).

	Standardized regression coefficient (β)	95% Confidence interval	p- value
RC (yes/no)	0.141	0.019 to 0.263	0.024
IPV (CAS score)	0.302	0.180 to 0.424	<0.001
Age	−0.052	−0.174 to 0.070	0.380
Race (White/not White)	0.021	−0.101 to 0.143	0.725
Ethnicity (Hispanic/not Hispanic)	−0.010	−0.132 to 0.112	0.871

Note: R-squared = 0.139, adjusted R-squared = 0.122.

Discussion

Similar to other studies,^15,42 these findings demonstrate that being a race other than White is a risk factor for RC; however, ethnicity was not found to be associated with RC, though small numbers of Latina/Hispanic participants may account for this lack of association. Indicators of severity of IPV-missing class due to relationship problems, experiencing more forms of technology abuse, and experiencing 1 or more TBI events-were also strongly associated with RC, as was experiencing Severe Combined Abuse (in contrast to experiencing other categories of abuse). Previous studies of the association of IPV and RC have not examined types of IPV or severity of abuse. The associations suggest that RC is a marker for more severe forms of IPV and also could indicate that RC itself is a severe form of IPV. As the nature of RC with and without accompanying IPV continues to be explored, it is critical to understand how RC associates with specific types of IPV, in order to identify the independent and synergistic effects of RC on related health outcomes. Having an “on again/off again” relationship also emerged as associated with RC. It may be that describing one’s relationship this way is a marker for relationship instability in which a partner may seek to increase power and control, and RC is one tactic for achieving this. Other studies identify decreased relationship trust⁴³ and relationship inequality⁴⁴ as associated with RC, but further exploration is warranted.

It is challenging to directly compare the findings from this study with existing research on RC because of the variation in RC measures used and timeframes varying from past 3 months to lifetime. However, the prevalence of RC in this sample of college women who were in abusive relationships was not dissimilar from other studies of IPV survivors^6,7 and college women^10,11.

In contrast to existing studies,^44,45 living with a partner and larger age discrepancy with partner were not significantly associated with experiencing RC among this sample of abused college women. And contrary to what was expected, having a partner who was not enrolled in the same school or in any school was also not associated with experiencing RC. It may be that these factors do not have an effect independent of IPV, which would not be detectable in a population of abused women, or that motivation for RC behaviors in a college population is less about power and control. Further work on motivations for RC behaviors is needed to elucidate this.

RC may be an indicator that a woman is experiencing more severe violence and is at risk for severe sequelae such as TBI. Counselors, faculty, and college staff who may become alerted to a student with class attendance issues, should be aware that this may be associated with experiencing IPV and other forms of coercive behavior such as RC. Conversely, experiencing severe abuse may be an indicator of the need to assess specifically for RC; when one form of abuse is detected, providers should also assess for other forms to be able to offer appropriate harm reduction strategies and connection to resources for safety planning.

Both RC and IPV had significant effects on depression score, but the effect of IPV was stronger. That RC remained a significant predictor of depression after adjusting for IPV, indicates that RC’s influence is additive; RC has an independent association with depression, even when holding IPV constant. This finding has important implications for providers working with abused college women, regarding the need for depression screening and strong mental health services concurrent with IPV and RC services. As cross-sectional data, it is unclear whether RC contributes to depressive symptoms or if depressive symptoms increase vulnerability for RC exposure. RC was not significantly associated with abusing alcohol or drugs. Studies on IPV have found consistent associations between IPV and substance use,^46–48 so it may be that the independent association of RC with substance abuse was not able to be detected in this sample of abuse survivors. While this study focused on depression as a primary mental health outcome, PTSD is also known to be linked to IPV,⁴⁹ and would be useful to study in future research. PTSD is also associated with alcohol and substance use in college students,⁵⁰ which, while not significantly associated with RC, did appear at high rates in this study.

Clinical implications

In previous studies, women who experience RC had a higher odds of seeking certain reproductive health services, such as pregnancy and STI tests and emergency contraception.⁵¹ However, seeking help from a provider for contraceptive options has not been examined. In our study, women who experienced RC were not significantly more likely to seek help from a healthcare provider for safer sex or birth control options. Nonetheless, a larger proportion of women who experienced RC did seek help from a provider, and the majority of the sample (57.9%) also sought help from a provider, which indicates an important opportunity for women’s health and college health providers to intervene when women are experiencing RC. This also has implications for the importance of screening for both IPV and RC when college women present for routine reproductive health services, in accordance with clinical practice guidelines.^52,53 Applying a nonjudgmental and trauma-informed care approach, all women should be screened when they are alone for IPV and RC and regardless of response, should be provided with resources and support.^54,55 Likewise, campus services for dating violence should be vigilant in assessing for RC when working with survivors.

Though there is a national focus on sexual assault prevention programing on college campuses, these efforts do not include RC prevention. There is limited research on RC-specific campus interventions and we know that screening for any type of IPV in college health centers is limited.⁵⁶ Futures Without Violence, a nonprofit leading national efforts on quality health care for IPV survivors, developed an RC-specific brief intervention for integration into health centers,^57,58 however, there are currently no published studies on how widely this intervention is being implemented on college campuses.

Limitations

Instruments used to measure RC in survey research vary widely. A RC Scale has been developed and refined over time,⁶ along with a shortened RC scale,⁵⁹ but many studies use modified versions of these scales, or may use investigator-developed items in place of these scales. This study used two RC assessment questions, which may have lowered sensitivity; the actual prevalence of RC may be higher than reported. Additionally, these questions have not been assessed psychometrically. One of the questions (Have you prevented a pregnancy by using emergency contraception or ended a pregnancy using other methods, and did not tell your partner about it because you were afraid of your partner?) likely overlaps with IPV, and may also capture fear of partner violence not specifically related to RC. Data in this study are cross-sectional, limiting inferences that can be drawn about causality. Caution in interpretation of depression associations is indicated given the low r-squared statistic and relatively small sample size. Lastly, the sample was limited to college-enrolled women in abusive relationships, and as such, findings may not be generalizable to other populations.

Conclusions

This study adds to emerging literature regarding factors associated with RC among IPV survivor college women, by supporting some established associations and also revealing new factors that are important to assess including TBI, technology abuse, class attendance, and relationship instability. Future research should investigate these factors, as well as the mental health outcomes identified, in a general population of college women to clearly identify the independent impact of RC separate from IPV. RC should also be assessed using a more comprehensive measure to capture the varied dimensions of RC that women may be experiencing. College-age women are at a vulnerable age for experiencing coercive behavior from partners, and healthcare providers who work with this population have an opportunity to intervene to provide support and reduce negative outcomes.