A Cluster-Randomized Trial of a College Health Center-based Alcohol and Sexual Violence Intervention (GIFTSS): Design, Rationale, and Baseline Sample

Abstract

Introduction

Sexual violence (SV) on college campuses is common, especially alcohol-related SV. This is a 2-arm cluster randomized controlled trial to test a brief intervention to reduce risk for alcohol-related sexual violence (SV) among students receiving care from college health centers (CHCs). Intervention CHC staff are trained to deliver universal SV education to all students seeking care, to facilitate patient and provider comfort in discussing SV and related abusive experiences (including the role of alcohol). Control sites provide participants with information about drinking responsibly.

Methods

Across 28 participating campuses (12 randomized to intervention and 16 to control), 2293 students seeking care at CHCs complete surveys prior to their appointment (baseline), immediately after (exit), 4 months later (T2) and one year later (T3). The primary outcome is change in recognition of SV and sexual risk. Among those reporting SV exposure at baseline, changes in SV victimization, disclosure, and use of SV services are additional outcomes. Intervention effects will be assessed using generalized linear mixed models that account for clustering of repeated observations both within CHCs and within students.

Results

Slightly more than half of the participating colleges have undergraduate enrollment of ≥3000 students; two-thirds are public and almost half are urban. Among participants there were relatively more Asian (10 v 1%) and Black/African American (13 v 7%) and fewer White (58 v 74%) participants in the intervention compared to control.

Conclusions

This study will offer the first formal assessment for SV prevention in the CHC setting.

INTRODUCTION

Background and rationale

Sexual violence (SV) (including sexual coercion, non-consensual sexual contact, and rape) on college campuses is common, with 20–25% of women and approximately 5–7% of men reporting sexual contact involving force or incapacitation during college.^1–5 Alcohol use plays a significant role in the occurrence of SV on college campuses, with about half of sexual assault cases occurring in the context of alcohol intoxication by the victim, perpetrator, or both.^6–10 The negative consequences of SV, which include physical, sexual, and mental health sequelae, are well documented.^11–19

Among modifiable factors for alcohol-related SV are social norms regarding hazardous drinking as part of college campus culture and expectations for SV occurring in the context of such heavy drinking. Such norms facilitate both the occurrence and underreporting of such violence.^9,10,20–22 Attitudes and behaviors related to SV in the context of alcohol consumption may be improved by increasing awareness of sexual consent and strategies to increase safety for oneself and peers (to decrease occurrence) and reducing stigma and self-blame associated with alcohol-related SV (to increase reporting).

Despite the high rates of SV, the majority of students do not report these experiences to university, health care or law enforcement officials.^1,5,23–29 Limited reporting of such victimization experiences among adolescents and young adults is related to stigma, self-blame, fear of consequences from illegal alcohol consumption, and belief that nothing will be done.^27,30–32 Community-based violence victim services provide support for survivors of SV^33–35 and can reduce post-traumatic stress symptoms,³⁵ self-blame,³⁶ and re-victimization.^37–39 Low levels of awareness^40,41 and use^23–29 of professional support services for SV highlight the need to better connect students who have experienced SV with these resources.

College health centers providing health and/or counseling services for students are ideal sites for introducing brief interventions and facilitating connection to services. College health centers provide care to a substantial proportion of students (estimates 43–48%),⁴² and thus represent an ideal setting to interact with a large number of students on an individual basis, to provide clear SV prevention messages, and to serve as a critical resource for SV victims. As experiences of SV contribute to poor health, including repeat violence victimization and sexual risk behaviors,^6,8,43 health centers present an opportunity to reach students who have experienced or are at high risk for SV. The limited research on college health center-based brief interventions have focused on high risk drinking,^44,45 without addressing its intersection with SV. Therefore, the purpose of the described study protocol is to test a brief alcohol and SV universal prevention education and harm reduction counseling intervention, titled Giving Information for Trauma Support and Safety (GIFTSS), to increase awareness of alcohol-related SV risk among college students.

METHODS

Objectives

Interventions effective in reaching college-age youth who either witness or experience alcohol-related SV are needed. This is a stakeholder-engaged longitudinal study to test, via a 2-arm cluster randomized controlled trial, a brief intervention to reduce risk for alcohol-related SV among 2293 male and female youth ages 18–24 receiving care from 28 college student health centers. This study draws on several research-informed intervention strategies: 1) the bystander intervention approach in which individuals are taught skills as active interveners in SV prevention rather than responding with apathy or tolerance is an effective strategy for promoting change within social contexts; 2) universal education about SV regardless of disclosure in the clinical setting can result in increased recognition of SV and use of relevant services; and 3) harm reduction strategies introduced by clinicians to increase intentions to use strategies to help oneself and friends that increase safety. The brief counseling intervention used a palm-size safety card with information about SV and harm reduction given to all patients during clinical encounters. GIFTSS provides (a) patient education and assessment regarding SV; (b) discussion of harm reduction behaviors to reduce risk of alcohol-related SV for self and peers (including bystander intervention); and (c) supported referrals to victim services. The control intervention used National Institute of Alcohol Abuse and Alcoholism (NIAAA) tools for brief alcohol risk reduction counseling.

We hypothesize that, compared to controls, male and female students seeking care at college health centers that offer the GIFTSS intervention will have greater recognition of what constitutes alcohol-related SV (primary outcome), knowledge of and self-efficacy to enact harm reduction strategies, intentions to intervene, and knowledge of and self-efficacy to use SV-related services, in both the short term (4 months) and longer term (12 months). Our second hypothesis is among students who witness peer SV during the study period, those who received GIFTSS will be more likely to report interventions to interrupt peer’s harmful behaviors than students not receiving GIFTSS. As an exploratory third hypothesis, students with SV victimization history who receive GIFTSS are expected to be more likely to disclose SV during their clinic visit, report greater use of SV-related services, and report less recent SV victimization at follow up compared to students on control campuses. This is the first study in a college health center setting that integrates prevention (using a bystander approach) and intervention (counseling and connection to victim services as needed) to address alcohol-related SV. The intervention model is integrated into standard clinical practice.

Trial Design

The experimental design for this evaluation involves a 2-armed cluster randomized controlled trial conducted with college students (male and female) ages 18–24 recruited from 28 college health centers that were randomly assigned to the intervention or control condition. Study participants (N=2293) in both conditions completed online surveys at baseline (prior to clinical visit and intervention) and immediately after the clinical visit (exit survey). Ongoing collection of follow up surveys at 4 months (T2), and 12 months (T3) occurs primarily via emailing or texting participants a link to the online survey; participants also have the option of completing the surveys in-person at the college health centers or over the phone (with a research assistant reading the questions aloud). Retention of participants is facilitated via tracking of participants using extensive contact information and an escalating incentive system ($15 for baseline and exit survey, $25 for T2, $40 for T3).

Participants, Interventions, and Outcomes

Study Setting and Eligibility Criteria

The sample includes 28 Pennsylvania and West Virginia college student health, wellness, and counseling centers, all located on co-ed campuses within a 3–4 hour radius of Pittsburgh. Recruitment of participating colleges was facilitated by contacting the colleges’ respective Dean’s offices, women’s programs, and counseling centers, in addition to the college health centers. At each college, a key stakeholder was identified to assist with initial training as well as set-up of student recruitment. Eligibility for a college campus was based on having a student health or counseling center that is accessible and staffed by a health professional (i.e., counselor, nurse, nurse practitioner, physician’s assistant, or physician) who provides one-on-one services to students at least five days a week. Each site recruited students seeking services within their health and wellness centers; for some campuses, this also included students seeking services with one-on-one providers such as alcohol and drug counseling and students seeking disability services. The Human Research Protections Office at the University of Pittsburgh approved all study procedures. Most of the 28 schools chose to use the University of Pittsburgh’s IRB approval; additional IRB approval was received from three schools that required their own IRB committees to review the study protocol.

Most of the college health centers had some alcohol and SV screening already in place (some in intake assessments, others in the electronic health record) and were variably involved in alcohol misuse and SV prevention efforts on campus, which were reviewed in detail at the time of recruiting college health centers into the study. Assessment for alcohol use was a routine practice in each of these college health centers prior to study initiation, and specific screening tools and interventions being used were assessed as part of a college health center practice assessment. However, such alcohol assessment had not been connected to discussions of SV prevention, specifically. While some of the college health centers routinely assessed for SV on intake forms, the connection of SV to alcohol consumption (by patient or peers) had also not been integrated into routine practice and no sites had protocol specifications for educating patients about alcohol-related SV. In addition to the health center assessments, an environmental scan of the alcohol and SV prevention programs on campus was conducted for each site.

Eligibility for this study includes students at the participating universities who meet the following criteria: 1) age 18–24 years, 2) literate in English, 3) seeking care at the college health center for any reason, and 4) sufficient time prior to seeing their provider to complete a 20-minute survey.

Recruitment of student participants

Students were recruited one of two ways, depending on whether the student was visiting the college health center for a walk-in or scheduled visit. Participating schools used a combination of one or more options, depending upon their scheduling system (6 =walk in only, 13 = scheduled visits only, 9 = both).

Students visiting the wellness center on a walk-in basis (no scheduled appointment) were identified by clinic staff at registration and informed about the study via printed flyers containing basic study information (purpose, time required, incentives, and confidentiality protections). Registration staff referred age eligible clients to research assistants staffing the clinic. Research assistants assessed students who were interested in participating for eligibility and reviewed an information sheet about the study with them; a waiver of signed consent was approved by the IRB that covered 27 colleges, with the remaining college requiring in-person signed consent obtained by an on-site research assistant. At some sites (n=22), flyers also included a number for Textmarks (www.textmarks.com), a text-messaging platform; interested students texted a designated code to the Textmarks number that sent the students directly to study information, an eligibility screen, and baseline survey.

Students with scheduled appointments were informed of the study by clinical staff; depending on school preference, potential participants were recruited by a) clinical staff providing the study team contact information directly to the interested student, b) clinical staff providing the interested students’ contact information (with the students’ permission) to the study team, or c) students could contact the study team via Textmarks, as described above. With all of these approaches, interested students were then directed to an eligibility survey via electronic link. Eligible students provided information regarding the time and day of their scheduled health center appointment and a phone number for a research assistant to follow-up with them. Research assistants recorded data regarding eligibility and reasons for non-participation where applicable.

Intervention descriptions and health center staff training

Giving Information for Trauma Support and Safety (GIFTSS)

The brief counseling intervention (GIFTSS) used a palm-sized educational card with information about alcohol-related SV and harm reduction given to patients during clinical encounters. After receiving training on how to deliver the intervention, college health center staff or clinicians (depending on reason for visit and clinic flow) were instructed to offer the educational card to all students in private, after a student completed intake assessment procedures for their college health center visit. The suggested introductory script was: “We are giving this information to all of our students because alcohol and sexual violence on our campus are far too common. We want to make sure students know how to help a friend and that we are a resource for them.” During the clinic visit, the clinician or health center staff were instructed to review specific panels on the card depending on the type of visit and reinforced universal education messages. Should a patient disclose problem drinking or SV victimization on a college health center screening tool or during the clinical encounter, the clinician was to conduct a more in-depth assessment for safety and make a warm referral to appropriate counseling support, including SV advocacy services. Each of the college health centers had a designated sexual assault advocate to whom they refer students (either to a local SV agency or an on-campus SV advocate - appropriate referral locations for each site were clarified during the GIFTSS training).

A local SV victim service agency, together with the PI, led the GIFTSS implementation training at each site. This three-hour training in use of the card involved all college health center staff, from registration and administration to medical assistants, counselors, health educators, and clinicians. A few of the campuses asked for this three hour training to be divided into two 1.5 hour sessions. Staff unable to attend training in-person were offered the opportunity to review the training materials online. The victim service advocates who led the training also connected with these individuals by phone, in-person, or by video conference to review the materials and ensure they were comfortable with the process of counseling and the core intervention components including harm reduction strategies, bystander behaviors, and making warm referrals to SV advocates. A recorded webinar using many of the original training slides was made available (https://www.futureswithoutviolence.org/addressing-and-preventing-intimate-partner-and-sexual-violence-in-campus-health-centers/). Providers were asked to notify their clinic supervisor when they had completed the webinar and meeting with one of the victim service advocate trainers to review the materials. Additionally, for sites where recruitment extended longer than an academic semester, a booster training was offered to staff at the start of the new academic semester.

Control Condition

The control sites received a three-hour training led by a local alcohol abuse prevention expert on screening and brief intervention using recommendations from the NIAAA to assess for alcohol misuse and abuse as well as a short introduction to motivational interviewing. These sites were instructed to give all patients a palm-sized card about drinking responsibly titled “Read Before Drinking.”⁴⁶ Similar to the intervention sites, the card is introduced in private with a healthcare or behavioral health professional using the script “We’ve started giving this card about alcohol use to all our patients so they can be informed, know how to get help for themselves, or help others.” Providers and staff not able to attend the training in person were offered an opportunity to review the materials on their own, and to meet with the clinic supervisor to answer questions about how to integrate this counseling intervention into their visits. A recorded version of the training was also available as an alternative for providers and staff unable to attend the in-person training, or for new staff who joined a site after the training. Any questions about the training materials were directed to the lead trainer for the brief alcohol prevention intervention. As with the GIFTSS intervention, booster trainings were offered to health centers where recruitment extended beyond an academic semester.

Additional Intervention Monitoring – Process Evaluation

All sites were assessed at baseline with a clinic quality assessment tool developed by Futures Without Violence for this intervention to determine current SV and alcohol screening policies and protocols and available resources including referral protocols within the health center. Additionally, all sites also completed an environmental scan of their campus’ existing SV and alcohol prevention programs and services. These scans included information such as whether students are required to complete a primary prevention program on SV or on alcohol abuse, if there are clear reporting options for experiences of SV, and if there are support services available for students who have experienced SV. These data will be used qualitatively to explore whether implementation quality varies by the extent of existing alcohol and SV related services and prevention programs. For example, we will descriptively assess whether schools that have high levels of existing alcohol and SV services (both within the CHC and campus-wide) are also those with high intervention fidelity (described below). We will also use these data to assess for potential threats to validity (e.g., control sites with robust campus prevention activities that may influence our ability to identify intervention effects).

Multiple strategies are being instituted in the study protocol to assess quality of intervention implementation. Fidelity of intervention implementation is checked via: 1) standardization of training for college health centers, 2) client exit surveys (completed after the clinic visit) that ask about the clinical encounter and content of the discussion; 3) biweekly phone meetings between the site lead and research team to review and document difficulties in implementation and consider possible solutions; 4) baseline and post training surveys with clinic staff; and 5) convenience sample of client interviews about receiving the intervention.

Chart extraction will provide additional information regarding clinic utilization, clinician assessment for SV, disclosures, and referrals made. Participant clinical records will be confidentially reviewed 12 months post-baseline as a supplemental data collection measure to assess a) changes in clinical practice (documentation of assessment for alcohol use and SV, documentation of harm reduction counseling and referrals), b) sensitivity of provider assessment to detect SV (triangulated with client self-report of SV and disclosure to a provider), and c) triangulate client self-report of service use. All participants in the research study were asked for permission to review the clinical records to examine their clinical history and quality of care. Participants were assured that information would be kept confidential and their names would not be associated with the chart extraction data, only a code number. Participants could decline the chart review and still participate in the survey portion of this study.

Outcomes

Primary

Recognition of alcohol-related sexual violence and sexual risk

Twelve items⁴⁷ are used to calculate an overall mean knowledge score measuring recognition of what constitutes SV and attitudes related to the role of alcohol in risk for SV (e.g., “a person who is sexually assaulted after drinking alcohol should only blame him- or herself”, “when alcohol is involved in a sexual encounter, communication signals are easily misinterpreted”). Response options are on a 5-point Likert scale from “strongly agree” to “strongly disagree,” with a higher mean score indicating greater knowledge and recognition (baseline Cronbach α=0.46). These items are assessed at all time points.

Secondary

Self-efficacy to use harm reduction strategies

Using a 5-point Likert scale from “strongly agree” to “strongly disagree,” respondents select how confident they are that they could use seven specified harm reduction strategies to reduce risk for SV (“I am confident that I could use a sober buddy system to stay safer at parties and to make sure I get home,” “I am confident that I could talk to a student health center provider about how to safely tell a partner about having an STI if I were worried about their reaction”). Female participants are asked about an additional two harm reduction strategies specific to pregnancy risk reduction: “I am confident that I could talk to a student health center provider about using an intrauterine device (IUD) or implant if I felt I needed a form of birth control that my partner could not tamper with or detect” and “I am confident that I could talk to a student health center provider about keeping emergency contraception (EC, or the “morning after pill”, or Plan B) in a safe place if I were worried about my partner finding it“. Measured on follow-up surveys only, a higher mean score indicates feeling more efficacious at using these harm reduction strategies. Items are investigator-developed.

Knowledge of sexual violence-related services

Participants are asked at baseline, T2, and T3 surveys if they knew about 5 sexual violence resources, such as on- and off-campus sexual assault experts they could talk to and the National Sexual Assault Hotline. Women are additionally asked if they knew they could take emergency contraception (EC) after an unwanted sexual experience. Three additional, non-sexual violence-related resources are used as filler questions to obscure the resources of interest in the scale (e.g., “Did you know about the stress management resources available on your college campus?”). A summary score is calculated by adding one point for each additional resource endorsed as known by a participant (baseline Cronbach α=0.77). Items are based on earlier work by the investigators^48,49 and have been updated for use with this college sample.

Self-efficacy to use sexual violence-related services

Participants report their likelihood to use the resources related to sexual violence presented under knowledge of sexual violence-related services using a 5-point Likert scale from “very unlikely” to “very likely”; a higher mean score indicates greater likelihood to use those resources (baseline Cronbach α=0.89). Five items are asked of all respondents (e.g., “How likely would you be to contact an on-campus sexual assault expert if you needed information or help related to sexual assault?); women are asked one additional question on pregnancy prevention (“How likely would you be to seek out emergency contraception for pregnancy prevention if you had an unwanted sexual experience?”). Two additional, non-sexual violence-related items were also included (e.g., “How likely would you be to use the alcohol and drug abuse resources available on your college campus?”). This scale is based on prior work done by this research team for knowledge of sexual-violence related resources.^50–52

Intentions to intervene

Ten items (modified from earlier research)^48,53,54 are used to assess respondents’ proclivity to intervene when witnessing disrespectful and harmful behaviors among peers by asking them, “if you see a friend or peer doing any of the following things, how likely are you to try to stop it?” Scenarios presented include “fighting with a partner and starting to cuss at or threaten that person” and “constantly talking or calling their dating or sexual partner to find out who they were with, where they were, or what they were doing.” For each item, respondents rate their willingness to intervene on a 5-point Likert scale from “very unlikely” to “very likely”; responses are averaged across the 10 items, with a higher mean score indicating greater likelihood to intervene (baseline Cronbach α=0.92). Intentions to intervene is assessed on baseline, T2, and T3 surveys.

Positive bystander behaviors (among those witnessing SV)

Participants report if they have witnessed the same ten scenarios presented in the intentions to intervene items (e.g., “seen a friend or peer taking naked/sexual pictures of a girl/guy when they’re too drunk, high on drugs, or passed out to consent,” “seen a friend or peer shoving, grabbing, or otherwise physically hurting someone”). For those who have witnessed a scenario in the past 4 months, they are asked how they responded; response options included: doing nothing because it’s none of their business; doing nothing because it wasn’t safe; telling the person it was not OK; laughing or going along with it; interrupting the situation; or talking to a student resident assistant or a non-student adult (like a coach or professor) to ask for assistance or advice. Endorsement of at least one positive action (telling the person it was not OK, interrupting the situation, or talking to someone about it) is added across the 10 items to create a summary score, with a possible range from 0 to 10 (baseline Cronbach α=0.48). These positive bystander behaviors are assessed at baseline, T2, and T3.

Recognition of sexual and reproductive coercion

Eight items,⁴⁹ based on prior work by the investigators, assess respondents’ recognition of sexual and reproductive coercion at baseline, T2, and T3. Respondents rate each situation presented on a 5-point Likert scale from “not abusive” to “extremely abusive”; responses are averaged across the 8 items, with a higher mean score indicating greater recognition. Examples include “pressuring someone to have sex when they’ve said no” and “refusing to use a condom when someone asks to use one.”

Self-efficacy to obtain sexual consent

Respondents report how efficacious they feel at obtaining consent for sexual activity on a 5-point Likert scale from “strongly agree” to “strongly disagree,” using the perceived behavioral control subscale of the Sexual Consent Scale – Revised.⁵⁵ Of the 10 items, scenarios include “I feel confident that I could ask for sexual consent from a new partner” and “I would have difficulty asking for consent because it would spoil the mood.” Negative items are reverse coded and the 10 items averaged, such that a higher mean score indicates greater self-efficacy (baseline Cronbach α=0.83). These items are asked at baseline, T2, and T3.

Additional outcomes

Sexual violence victimization

Six items (modified)⁵⁶ assess the frequency of unwanted sexual experiences before college (baseline), since college (baseline), in the past 4 months (baseline, T2, and T3), and in the past 12 months (T3). Experiences include “has anyone fondled, kissed, or touched you sexually when you indicated that you didn’t want to?” and “has anyone made you have vaginal sex when you indicated you didn’t want to?” Participants report whether each situation occurred 0, 1, 2, 3, or 4 or more times in the specified time periods (baseline before college Cronbach α=0.78, since college α=0.80). An additional item assesses incapacitated multi-contact sexual experiences (“was there an occasion when a group of individuals had sex with you against your will or when you were incapacitated (e.g., by drugs or alcohol) and unable to object or consent?”). The frequency of the first 6 items and any occurrence of incapacitated multi-contact unwanted sex are added to create a summary score for sexual violence victimization.

Disclosure of sexual violence to health care provider

Participants who endorse any lifetime exposure to abuse, unwanted sexual experiences, partner physical violence, or stalking are asked immediately post-visit (exit survey) and on the T3 survey whether they had told their health provider about any of these experiences that they may have had.

Use of sexual violence-related services (restricted to those with history of SV at baseline)

Respondents are asked whether they have used 5 pre-specified sexual violence-related resources (e.g., calling the National Sexual Assault Hotline or visiting their website, were treated for potential STD exposure after an unwanted sexual experience) in the past 4 months (baseline, T2) or 12 months (T3). Women are additionally asked if they have taken emergency contraception after an unwanted sexual experience (will be analyzed both including and excluding emergency contraception). A summary score is calculated for total number of sexual violence-related services used.

Provider documentation of alcohol and sexual violence screening, counseling, and referrals

Following completion of T3 data collection, medical records review will be completed for sites and participants who consented to participate in this study phase. Participant charts will be reviewed for the duration of the study period (one year and 6 weeks following baseline data collection). Data to be abstracted include number and reasons for clinical visits, as well as provider documentation of screening for/discussion of, counseling regarding, and referrals related to sexual violence and alcohol misuse (as well as related referrals for substance use and intimate partner violence) during each visit during the study period.

Study Participation Timeline

The figure below illustrates the study flow for participants from randomization to timing of follow up surveys (Fig 1).

Figure 1. Two-arm Randomized-Controlled Trial Design and Data Collection Schedule.

CHC, college health center

Sample Size

Sample sizes needed to achieve sufficient statistical power to compare conditions for each of the outcomes were based on clinically meaningful differences between treatment groups with respect to changes in the outcomes across time (i.e. intervention effect). For the primary outcome (recognition of alcohol-related sexual violence and risk), the necessary sample size per arm was calculated based on Heo and Leon,⁵⁷ which primarily depends on the within-subject correlation (ICC). Based on information from preliminary studies, within-subject ICC ranged from 0.34 to 0.56. Assuming an ICC of 0.30, our enrollment of 2293 students across 28 schools, and assuming 80% retention, yields 80% power to detect a standardized effect size of 0.185 (similar to what we have observed in our prior studies). If our within-subject ICC is closer to the upper range of what we have observed in prior studies (i.e. 0.60), our detectable standardized difference would be 0.14.

Our sample size analysis reflects appropriately conservative adjustments for cluster-level randomization and incorporates the best available evidence using data from our previous randomized controlled trials in high schools and family planning clinics. We expect approximately 75% of the sample (n=1376, at Time 3 with 80% retention) to report witnessing any peer abusive behaviors in the prior 12 months based on our previous studies,52 thus we expect to have sufficient power to address our second hypothesis. While our third hypothesis is exploratory, we anticipate a 25% prevalence of recent SV victimization at baseline based on our clinic-based studies (primarily female, anticipated 573 subjects), thus we expect to have power to detect moderate effect sizes.

Assignment of Interventions

Randomization

Randomization of the clinics was stratified based on size (large, i.e., ≥3000 students served per year, versus smaller, <3000 students), and clinics from each stratum were allocated to either the intervention or control conditions, using computer-generated randomly permuted block allocation prepared using STATA statistical software under the direction of the lead study statistician. Clinic-level randomization was selected to minimize between-arm contamination of intervention and control conditions. Only the lead statistician was aware of the full randomization sequence; the PI and study team were unblinded to a school’s allocation only after that school was enrolled.

Data Collection, Management, and Analysis

Data Collection and Management

Male and female college students ages 18–24, literate in English, seeking services at any one of the 28 college health centers for any reason were assessed for eligibility. Interested participants provided consent in-person or online, and had opportunities to provide contact information, and an authorization form to review their clinic records; the clinic record authorization was not necessary to participate in the surveys. Participants then completed the 20-minute baseline survey (using REDCap,⁵⁸ an online survey program) prior to their clinic visit on an iPad or computer in a private location with a research assistant, or on their own phone, tablet or computer (if recruited online prior to a scheduled appointment). Participants were assigned a unique identifier based on their school and the time and date of data collection. Upon completion of the clinic visit and receipt of either GIFTSS intervention or alcohol-focused intervention, participants were directed to return to the research assistant to complete the Exit Survey, and received a $15 prepaid gift card. If the initial survey was completed online and not with a research assistant, exit surveys were scheduled to be sent an hour after their appointment with additional information on follow up with a research assistant to receive their gift card. The online survey contained an option to view or download a resource sheet containing relevant social services and college-specific health resources (including services available for problem drinking as well as SV). Based on ethical considerations, all participants (intervention and control) received this resource information. For students completing the surveys in the health centers, written information for relevant social services and college health was also available through research assistants, who offered them to students to take with them as needed.

Research assistants contact participants for follow up surveys. To minimize attrition, multiple forms of contact information were collected at enrollment. Contact methods included email, phone, and text follow-up as well as email reminders from student health services directors. The research assistants have access to a password protected email account to contact study participants to return for follow-up surveys. All email communication refers to a “College Health Survey” to maintain safety and confidentiality of the research participant.

Multiple precautions have been implemented to protect data collected during this study. All laptops and tablets are password protected. Survey data are collected via secure online tools and no data are stored on the individual data collection devices. Participant contact information is entered directly into a separate, encrypted file on a secure university-housed server that is accessible only to research team members who require access to participant contact information for facilitating follow up survey completion. Follow up surveys that are emailed to participants are distributed via REDCap and a secure service account maintained through UPMC.

Statistical Methods

The primary assessment of intervention effects will be based on intent-to-treat estimates. As-treated effect parameters will be estimated in secondary analyses and reported as exploratory. Point estimates will be calculated and hypothesis testing will be conducted in a generalized linear mixed model framework that will account for within-clinic and within-patient cluster effects as well as the handling of binary outcomes. Regression models will include a clinic-level covariate (clinic size) to account for the stratified randomization of clinics (larger vs smaller). The primary study hypotheses will be assessed with a two-tailed alpha set at 5%. No adjustments for multiple testing will be made as we assert that each of these hypotheses is of independent scientific interest. To assess differences at baseline between the participants in the intervention and control groups, demographics such as age, sex, race/ethnicity, year in school, and housing location will be compared while accounting for within-site clustering. Since randomization is done by site, the above demographics will also be compared across sites. Demographic variables significantly related to study arm, either at the patient or clinic-level, will be considered as covariates in the primary analyses. SAS software will be used for all statistical analyses.

Participation bias will be assessed by comparison of age and race/ethnicity of respondents participating in the study compared to the overall demographics of the college health centers as well as to the demographics of the colleges. Significant differences will be noted as potential validity threats. An attrition analysis will be conducted by comparing respondents who completed follow-up surveys with those who did not with regard to demographics as well as outcomes measured at baseline. Mechanisms for missing data will be investigated by comparing important covariates between students with and without missing data at each time point. We will characterize mechanisms of missingness, and if appropriate, we will utilize imputation methods such as multiple imputation via chained equations (MICE) to handle dropout. Ultimately, sensitivity analyses will be conducted comparing the results of our imputation methods to complete-case and available-data analyses.

For each of the study hypotheses, generalized linear mixed models will be used to determine whether the intervention results in increased levels of the outcomes across time, when compared to controls. For the primary outcome (recognition of alcohol-related sexual violence and sexual risk), the models will include variables for time, study arm, their interaction, clinic size, and random effects for clinic and participant-within-clinic. This will allow us to assess whether the interaction between time and study arm is significant and positive. In addition, we will include as covariates any subject- or clinic-level baseline characteristics related to study arm. Similar models will be repeated for each secondary outcome as well. For the secondary outcome of positive bystander behavior, the analysis will be restricted to participants who report witnessing any peer abusive behaviors in the 3 months prior to study enrollment (in our high school study among male athletes, this exceeded 75% of the sample, thus we do not anticipate a significant reduction in effective sample size); and for the exploratory outcome of use of sexual violence-related services, the sample will be restricted to participants who report SV victimization at baseline. The analyses for these secondary and exploratory hypotheses will be similar to that of the primary outcome, with the exception that the outcomes are binary. Therefore, a generalized linear mixed model with a logit link will be used. Because these participants are a post-randomization subset of the overall cohort, we expect to see imbalances at baseline at either clinic or subject levels and these covariates will be included in these secondary and exploratory analyses. Finally, an additional analysis for the “Use of sexual violence-related services” outcome will be conducted restricting to women, given that one of the components of that survey asks about “emergency contraception for pregnancy prevention”.

Following the definitive assessments of key study hypotheses, a series of exploratory analyses will be conducted. An intensity-adjusted analysis to reflect the actual receipt of the intervention will be conducted by replacing the original binary intervention indicator (0 or 1) with a continuous score variable in the primary models. This score will be based on answers to questions on the exit survey, given immediately after participants’ clinical visits, which includes questions about whether they received the palm-sized educational card, whether the provider reviewed the card with them, and whether they discussed specific topics such as sexual assault and healthy relationships during the college health center visit. For intervention arm participants, receiving the card and having at least one discussion point with their provider is scored as 1 (“full intervention intensity”). Receiving the card or having at least one discussion point with their provider is scored as 0.5, to indicate having received approximately half of the intervention. Participants who did not receive a card nor had any discussion with their providers are scored as 0. Control arm participants are assigned a score of 0, regardless of their exit survey responses.

We will also conduct a per-protocol analysis on the primary outcome in which only intervention-arm participants who received the full intervention as planned (corresponding to the intensity score of “1” above) are compared to all control arm participants.

The potential for differential impact of the intervention based on patient characteristics will also be evaluated, with a view toward identifying factors that may facilitate or impede the efficacy of the intervention, and to identify patient subgroups most receptive to the intervention or who may require more support. These factors include age, class (first year vs. upper years), sex, race/ethnicity, heavy and hazardous drinking). The outcome variable at T2 and T3 will be modeled as a function of the following variables: outcome at baseline, treatment group, potential moderator, interaction between the treatment group and moderator, a random effect for clinic, and a random effect for the subject. A significant interaction suggests presence of intervention effect heterogeneity; we will follow the approach of Kraemer⁵⁹ by focusing on effect size derivation rather than formal hypothesis testing. Additional exploratory analyses will examine differences in outcomes of interest by campus characteristics, including campus size, type of services participating in the study (counseling, medical, or both), and depth and breadth of alcohol misuse and SV prevention programming (occurring outside of the college health centers).

Monitoring

Data Monitoring

Despite the low risk associated with this behavioral intervention, we recognize the sensitivity of the topics (sexual violence and alcohol consumption) and have organized an internal data safety and monitoring process as well as an external advisory group. Summary reports of all data and safety monitoring activities are provided to the IRB at the time of annual renewals. Internal Data Safety and Monitoring Plan. The research team (PI and study team) meet weekly to review safety data. The tasks include:

1. Systematically review assessment materials to ensure that assessment is conducted appropriately and that participants disclosing abuse or violence receive appropriate connection to violence related services and that mandated reports are made by clinicians when appropriate.

2. Systematically review notes from research assistants to ensure that participants experiencing distress are being connected directly with the on-site health care provider, receiving educational materials, and being referred appropriately; this includes ensuring that all research assistants document asking each participant about emotional distress after completion of the survey.

3. Assess the quality of implementation and client feedback on the intervention. The PI will provide feedback and booster trainings for staff who express challenges with implementing the intervention and for instances where the intervention implementation does not appear to be going smoothly (based on feedback from clients on the exit survey, i.e., educational cards not being distributed by clinicians or clients reporting that clinicians did not discuss SV with them).

4. Monitor staff performance with regard to protection of privacy, confidentiality, maintenance of secure data bases, and study procedures designed to reduce the risk of distress and potential breaches of confidentiality.

5. Ensure that the PI (Miller), or a designated qualified individual, is available by phone in case research staff needs to confer regarding participant symptoms.

6. Review and report any adverse events associated with the study. Specifically, any adverse events will be reported to the University of Pittsburgh IRB following established protocol as well as to the National Institutes of Health (NIAAA).

External Advisory Group

Given the sensitivity of the topics that are the focus of this study (i.e., alcohol consumption and SV), an External Advisory Group which meets quarterly was also established. This external monitoring group consists of the Director of Health from Futures Without Violence (the national non-profit violence prevention organization that collaborated on development of the GIFTSS intervention) and advocates from two local victim service agencies in Pittsburgh who help lead the trainings. Group members were introduced to the details of the research protocol, but are not involved in the day-to-day implementation. They also have the authority to suspend the project should any challenges or adverse events arise, and to direct changes in research protocol should site specific problems emerge. For example, should a client experience emotional distress and the research assistant notifies the designated clinic staff but the research assistant reports that the staff was not responsive, this would be reported immediately to the PI as well as the External Advisory Group, and changes would be implemented before proceeding with any further recruitment or data collection.

RESULTS

Thirty-six colleges were approached for participation, with all but one responding to email and phone invitations. Twenty-nine of the 35 eligible colleges agreed to participate (83%, Fig 2. CONSORT), with 13 colleges assigned to intervention and 16 assigned to control. One school from the intervention arm withdrew following randomization but prior to enrolling any participants. In the intervention arm, 1,364 met eligibility criteria. Of those, 1,156 consented (85% of eligible) and 1,042 completed baseline surveys (90% of consented). In the control arm, 1,426 students were deemed eligible. In total, 1,331 provided informed consent (93% of eligible), and 1,252 completed baseline surveys (94% of consented). In all, 2293 participants from 28 colleges enrolled in the study at baseline.

Figure 2. CONSORT.

^aCollege did not have an eligible student health center

^bDue to recruitment methods used, number of potential participants approached cannot be calculated

Slightly more than half of the 28 participating colleges have undergraduate enrollment of ≥3000 students (57%); two-thirds (61%) are public colleges and almost half (43%) are located in an urban area (Table 2). The racial composition of the student body was predominantly non-Hispanic White, and did not vary by treatment arm. While almost all colleges (96%) offered both health and counseling services to their students, half (50%) of the colleges had only health services participate; the remainder had counseling centers only (11%) or both health and counseling centers (39%) participate in the study, which did not vary by treatment arm.

Table 2.

Characteristics of participating colleges

Characteristic	Total (n=28) % (n)	Treatment Arm			p value
		Intervention (n=12) % (n)	Control (n=16) % (n)
Undergraduate enrollment					>0.99a
<3000	42.9 (12)	41.7 (5)	43.8 (7)
3000+	57.14 (16)	58.3 (7)	56.3 (9)
Public status					.7047a
Public	60.7 (17)	66.7 (8)	56.3 (9)
Private	39.3 (11)	33.3 (4)	43.8 (7)
Campus location					.7022a
Non-urban	57.1 (16)	50.0 (6)	62.5 (10)
Urban	42.9 (12)	50.0 (6)	37.5 (6)
Racial composition of student bodyb	Median % (range %)
White	79 (34 – 88)	75.5 (34–88)	80 (57–87)		.2348c
African-American/Black	6 (2–21)	7 (4–14)	5 (2–21)		.0985c
Hispanic or Latino	3 (0–10)	3 (1–10)	2.5 (0–6)		.3181c
Asian	2 (0–34)	1.5 (0–34)	2 (0–4)		.8454c
Participating health and wellness centers					.4714a
Health services center only	50.0 (14)	58.3 (7)	43.8 (7)
Counseling center only	10.7 (3)	16.7 (2)	6.3 (1)
Both counseling and health services center	39.3 (11)	25.0 (3)	50.0 (8)

^aCalculated using a Fisher exact test

^b“Other race” is excluded here due to only 5 of the 28 schools reporting this prevalence

^cCalculated using the exact nonparametric Wilcoxon rank-sum test

Among participants in this study, race/ethnicity varied significantly by treatment arm; there were relatively more Asian (10%) and Black/African American (13%) and fewer White (58%) participants in the intervention arm compared to control (1%, 7%, and 74%, respectively). Most participants identified as female (72%), with no difference by treatment arm. The majority of participants were 18–21 years of age and in the first three years of undergraduate education. Almost all (96%) were enrolled full time. Campus residence halls, followed by off-campus housing (not including parental homes) were the most common housing locations for participants (Table 3).

Table 3.

Study Participant Demographics

Characteristic	Total (n=2293) % (n)a	Treatment Arm		χ2 p-valuec
		Intervention (n=1041) % (n)b	Control (n=1252) % (n)b
Race/ethnicity				0.00314
Asian	5.3 (122)	10.1 (105)	1.4 (17)
Black or African American	9.4 (216)	12.8 (133)	6.6 (83)
Hispanic or Latino	13.6 (312)	14.4 (150)	13.0 (162)
White	67.1 (1538)	58.4 (608)	74.3 (930)
Multiracial	3.1 (70)	3.3 (34)	2.9 (36)
Other	0.5 (12)	0.4 (4)	0.6 (8)
Gender				0.8407d
Male	26.5 (608)	28.0 (291)	25.3 (317)
Female	71.8 (1647)	70.8 (737)	72.7 (910)
Transgender female	0.1 (2)	0 (0)	0.2 (2)
Transgender male	0.1 (3)	0.1 (1)	0.2 (2)
I’m not on the binary	0.5 (12)	0.7 (7)	0.4 (5)
None of the above	0.1 (3)	0.2 (2)	0.1 (1)
Age				0.0259
18–19 years	39.6 (907)	40.6 (423)	38.7 (484)
20–21 years	39.3 (901)	35.7 (372)	42.3 (529)
22–24 years	16.5 (378)	20.0 (208)	13.6 (170)
Year in school				0.0309
1st year undergraduate	26.6 (609)	27.7 (288)	25.6 (321)
2nd year undergraduate	24.9 (570)	22.8 (237)	26.6 (333)
3rd year undergraduate	20.5 (469)	19.5 (203)	21.3 (266)
4th year undergraduate	16.2 (372)	15.5 (161)	16.9 (211)
5th year undergraduate	2.9 (66)	2.6 (27)	3.1 (39)
Graduate or professional school	6.5 (148)	10.6 (110)	3.0 (38)
Other	0.4 (10)	0.4 (4)	0.5 (6)
Enrollment status				0.3836
Full-time	95.7 (2195)	96.5 (1004)	95.1 (1191)
Part-time	1.5 (34)	1.5 (16)	1.4 (18)
Other	0.5 (12)	0.8 (8)	0.3 (4)
Housing location				0.6638
Campus residence hall	50.2 (1151)	48.7 (507)	51.4 (644)
Fraternity or sorority house	1.7 (40)	2.1 (22)	1.4 (18)
Other college housing	7.7 (176)	10.0 (104)	5.8 (72)
Parent/guardian’s home	4.0 (92)	3.8 (40)	4.2 (52)
Other off-campus housing	33.2 (761)	33.0 (343)	33.4 (418)
Other	0.8 (19)	0.9 (9)	0.8 (10)

^aPercentages may not sum to 100% due to small amounts of missing data

^bColumn percent

^cWald log-linear chi square tests, accounting for college-level clustering

^dThe statistical test was collapsed to Male, Female, and Other due to small cell size

DISCUSSION

To our knowledge, this is the first large-scale study on alcohol misuse and sexual violence exposure among college students seeking care in college health centers. The study will test the effectiveness of a brief harm reduction intervention in college student health centers to reduce risk for alcohol-related SV. Strengths of this study include the large number and heterogeneous nature of the college health centers involved in the study, the stakeholder engagement (practitioners, advocates, college students involved in development of the intervention and piloting), rigorous research design, and safety precautions to mitigate risk for breach of confidentiality. Notably, the original proposal for this study intended to recruit 20 campuses and 2200 students; our final number of campuses participating is 28, with 2293 students participating.

This study emerges from extensive community partnerships in Western Pennsylvania and West Virginia with college health practitioners, survivors of SV, and sexual assault victim advocates. Through this stakeholder input, we have learned that the most salient aspect of communicating risk related to SV with patients is ensuring that all patients receive information about alcohol-related SV, regardless of recent SV or alcohol exposure and regardless of whether they disclose any SV to a provider. That is, the stakeholders involved in developing this study have underscored the critical importance of integrating such risk communication into standard care without stigmatizing or singling out students who have experienced SV or have problem drinking. Given the high prevalence of both SV and alcohol consumption in this population, this study focuses on testing a brief intervention that is acceptable and feasible for integration into usual care that can be sustained and implemented with ease in college health centers more broadly. All partners have demonstrated a strong commitment to the study from the start, and will work collaboratively to ensure dissemination of the findings of this health center-based demonstration, should this trial produce positive results.

Some potential limitations of the study are 1) the reliance on self-report of alcohol use and sexual violence exposure; 2) the highly heterogeneous sample of campuses (from campuses with one provider to comprehensive health centers with over 30 staff); 3) the limited generalizability to college students across the nation given the study’s geographic focus; 4) the variable alcohol and sexual violence prevention programs available on each campus; and 5) the limited focus on sexual violence perpetration. First, this study relies on self-report of highly sensitive topics (alcohol and sexual violence). Due to the known under-reporting of sexual assault to authorities, studies on SV victimization tend to rely on participant self-report. Surveys are administered online or in a private area in each clinic on a tablet computer. Online data collection offers distinct advantages for SV and substance abuse related data collection, and has been recommended for obtaining high-quality data on this topic.^60–62 Second, the heterogeneity of the sample of participating campuses is both a strength and a possible limitation. Our randomization scheme takes campus size into consideration. Additionally, our analysis plan includes exploring differential intervention effects by campus size as well as type of service (counseling, medical, or both). We will also quantitatively and qualitatively assess differences in implementation of the intervention based on these health center characteristics. Demonstrating the feasibility of implementation of this intervention across different types of health centers will be helpful with considering dissemination to diverse college health centers, if the intervention proves to be effective. Third, the colleges in our sample from Western Pennsylvania and West Virginia may not be representative of college students across the nation. While over 30% of our sample represents underrepresented racial and ethnic groups, these distributions are still well below national estimates at degree-granting postsecondary institutions.⁶² Fourth, we are tracking the alcohol and SV prevention programs available on each campus (using environmental scans). Campuses that have extensive SV and alcohol prevention programming (which we assess with campus-wide environmental scans as well as quality assessment of each health center) may present a potential threat to validity, which we will consider in assessing intervention effects. Finally, we recognize that reducing SV on college campuses requires an unrelenting focus on reducing SV perpetration, which is not the focus of this intervention. The GIFTSS intervention was designed with a dual focus on prevention through social norms change (universal education for all students seeking care about reducing SV on campuses with an emphasis on positive bystander intervention behaviors and helping friends who may have experienced SV) combined with increasing safety and support for survivors of SV. If successful, the GIFTSS intervention may be one innovative component of comprehensive SV prevention and intervention on college campuses.

Table 1.

Individual Participant-Level Outcome Measures

Construct	Response Options	Surveys			Items
		Base line	T2	T3
Primary outcome measure
Recognition of alcohol-related sexual violence and sexual riska,47	5 point Likert scale ‘strongly disagree’ to ‘strongly agree,’ modeled as a mean score	X	X	X	A person who is drunk can still give consent to sexual activity. A person who is sexually assaulted after drinking alcohol should only blame him- or herself. Consensual drunk sex is a normal and harmless part of college life. Alcohol is the most common date rape drug (or substance). When a person is drinking alcohol, he or she is implying interest in engaging in sexual activity. If both partners are drunk and have sex, there is no way either partner can be accused of sexual assault or rape. The more alcohol a person has consumed, the less able he/she is to consent to sexual activity. Intoxication is not a defense against the charge of rape or sexual assault. If a person who has been drinking becomes sleepy or unconscious, he/she cannot give consent to any sexual activity. When alcohol is involved in a sexual encounter, communication signals are easily misinterpreted. As a general rule, alcohol makes sexual encounters easier and more enjoyable for both men and women. Alcohol use makes a person more vulnerable to sexual assault.
Secondary outcome measures
Self-efficacy to use harm reduction strategiesb	5 point Likert scale ‘strongly agree’ to ‘strongly disagree,’ modeled as a mean score		X	X	Please indicate how much you agree or disagree with the following statements: I am confident that I could look at online resources (like videos, websites) to find out what I can do to prevent sexual violence. I am confident that I could download the Circle of 6 safety app to keep me or my friends safe. I am confident that I could use a sober buddy system to stay safer at parties and to make sure I get home. I am confident that I could use the phone or use a computer/tablet in the student health center to contact a sexual assault advocate (hotline, online chat, etc.). I am confident that I could share information about sexual consent with friends. I am confident that I could talk to a student health center provider about how to safely tell a partner about having an STI if I were worried about their reaction. I am confident that I could talk to a student health center provider about using an intrauterine device (IUD) or the implant if I felt I needed a form of birth control that my partner could not tamper with or detect. I am confident that I could talk to a student health center provider about keeping emergency contraception (EC, or the “morning after pill” or Plan B) in a safe place if I were worried about my partner finding it. I am confident that I could talk about any challenges I am having in an intimate relationship (e.g. with my dating partner, girlfriend/boyfriend, hook-ups) with a student health center provider.
Knowledge of sexual violence related servicesc,48,49	Yes or No, modeled as summary score	X	X	X	Did you know about the National Sexual Assault Hotline (1-800-656-HOPE and www.rainn.org)? Did you know you can talk to the doctors, nurses, counselors, and staff at your college health clinic about sexual assault? Did you know that there are on-campus sexual assault experts that you can talk to? Did you know that there are off-campus sexual assault experts that you can talk to? Did you know that after an unwanted sexual experience you can get treated for potential STD exposure? Did you know that after an unwanted sexual experience you can take emergency contraception (EC)? Did you know about the stress management resources available on your college campus?* Did you know about the alcohol and other drug abuse resources available on your college campus?* Did you know about the stress management resources available on your college campus?* *not scored, used as filler items only
Self-efficacy to use sexual violence related servicesb	5 point Likert scale from ‘very unlikely’ to ‘very likely,’ modeled as a mean score	X	X	X	How likely would you be to: Use the stress management resources available on your college campus?* Use the National Sexual Assault Hotline (1-800-656-HOPE and www.rainn.org) if you needed information or help related to sexual assault? Talk to the doctors, nurses, counselors, and staff at your college health clinic if you needed information or help related to sexual assault? Contact an on-campus sexual assault expert if you needed information or help related to sexual assault? Contact an off-campus sexual assault expert if you needed information or help related to sexual assault? Use the alcohol and other drug abuse resources available on your college campus?* Seek medical treatment for potential STD exposure if you had an unwanted sexual experience? Seek out emergency contraception for pregnancy prevention if you had an unwanted sexual experience? *not scored, used as filler items only
Intentions to intervenec,48,53,54	5 point Likert scale from ‘very unlikely’ to ‘very likely,’ modeled as a mean score	X	X	X	If you see a friend or peer doing any of the following things, how likely are you to try to stop it? Spreading rumors about someone’s sexual reputation, like saying they are ‘easy’ Fighting with a partner and starting to cuss at or threaten that person Making rude or disrespectful comments about someone’s body, clothing or make-up Saying that it is OK to have sex with someone who is passed out or very drunk Making sexist, racist, and/or homophobic jokes Taking naked/sexual pictures of a girl/guy when they’re too drunk, high on drugs, or passed out to consent Constantly talking or calling their dating or sexual partner to find out who they were with, where they were, or what they were doing Showing other people sexual messages or pictures of someone on a cell phone or the internet Taking sexual advantage of someone who is drunk or high (like touching, kissing, having sex with) Shoving, grabbing, or otherwise physically hurting someone
Positive bystander behaviorsc,53,54	Have you seen or heard this: No, never I have seen or heard this, but not in the past 4 months Yes, I have seen or heard this in the past 4 months. [If yes to past 4 months]: How did you respond: I didn’t do or say anything, it’s none of my business I could not do anything as it wasn’t safe I told the person that acting like that was not okay (+1) I laughed or went along with it I interrupted the situation (on my own or with friends) (+1) I talked to a student resident assistant to ask for assistance or advice (+1) I talked to a non-student adult (like a coach or professor) to ask for assistance or advice (+1) Modeled as yes to any positive behavior (noted as +1) and then summed across the 9 items	X	X	X	The following questions ask about specific behaviors that you may have seen or heard among your fellow college students. In the past 4 months, have you: Heard a friend or peer spreading rumors about someone’s sexual reputation, like saying they are ‘easy’? Heard a friend or peer fighting with a partner and starting to cuss at or threaten that person? Heard a friend or peer making rude or disrespectful comments about someone’s body, clothing or make-up, such as mocking, taunting, or hooting? Heard a friend or peer saying that it is ok to have sex with someone who is passed out or very intoxicated? Heard someone making sexist, racist, and/or homophobic jokes? Seen a friend or peer taking naked/sexual pictures of a girl/guy when they’re too drunk, high on drugs, or passed out to consent? Seen a friend or peer constantly talking or calling their dating or sexual partner to find out who they were with, where they were, or what they were doing? Seen a friend or peer showing other people sexual messages or pictures of someone on a cell phone or the internet? Seen a friend or peer taking sexual advantage of someone who is drunk or high (like touching, kissing, having sex with)? Seen a friend or peer shoving, grabbing, or otherwise physically hurting someone? How did you respond?
Recognition of sexual and reproductive coercion49	5 point Likert scale from ‘not abusive’ to ‘extremely abusive,’ modeled as a mean score	X	X	X	How abusive do you think this is: Pressuring someone to have sex when they’ve said no. Refusing to use a condom when someone asks to use one. Preventing someone from using birth control. Trying to get someone pregnant when they don’t want to be Trying to make someone have an abortion when they don’t want one. Trying to make someone keep a pregnancy that they don’t want Accusing your partner of cheating when they want to use a condom. Threatening to leave someone if they don’t have sex (including oral, vaginal, or anal) Threatening to have sex with others if they don’t have sex (including oral, vaginal, or anal sex)
Self-efficacy to obtain sexual consentd,55	5 point Likert scale from ‘strongly agree’ to ‘strongly disagree,’ modeled as a mean score	X	X	X	I would have difficulty asking for consent because it would spoil the mood. I am worried that my partner might think it’s weird or strange if I asked for sexual consent before starting any sexual activity. I would have difficulty asking for consent because it’s not how I like to have sex. I would worry that if other people knew I asked for sexual consent before starting sexual activity that they would think I was weird or strange. I think that verbally asking for consent is awkward. I have not asked for sexual consent at times because I felt that it might backfire and I wouldn’t end up having sex. I believe that verbally asking for sexual consent spoils the mood. I would have a hard time verbalizing my sexual consent because it would be awkward. I feel confident that I could ask for sexual consent from a new partner. I feel confident that I could ask for sexual consent from a current partner.
Additional outcomes
Sexual violence victimization	Modeled as summary score of unwanted sexual experiences and incapacitated multi-person sexual experience
Unwanted sexual experiencese,56	0 1 2 3 4+ times Yes (≥1) or No	X	X	X	Baseline: Before college/Since you started college: T2: In the past 4 months: T3: In the past 12 months/In the past 4 months: How many times has anyone fondled, kissed or touched you sexually when you indicated that you didn’t want to? How many times has anyone tried to have sex with you (but it did not happen) when you indicated that you didn’t want to? How many times has anyone made you have vaginal sex when you indicated that you didn’t want to? How many times has anyone made you do oral sex or have it done to you when you indicated that you didn’t want to? How many times has anyone made you have anal sex when you indicated that you didn’t want to? How many times has anyone penetrated you with a finger or objects (vaginally, orally, or anally) when you indicated that you didn’t want to? Baseline [if >=1 times for any item]: Has any of this happened in the past 4 months?
Incapacitated multi-person sexual experienceb	Yes or No	X	X	X	Baseline: Was there ever: T2: In the past 4 month, was there: T3: In the past 12 months, was there/In the past 4 months, was there: …An occasion when a group of individuals had sex with you against your will or when you were incapacitated (e.g., by drugs or alcohol) and unable to object or consent?
Use sexual violence related servicesc,48,49	Yes or No, modeled as a summary score	X	X	X	Baseline: In the past 4 months, have you: T2: In the past 4 months, have you: T3: In the past 12 months, have you/In the past 4 months, have you: Called the National Sexual Assault Hotline or visited their website? Talked to the doctors, nurses, counselors, and staff at your college health clinic about sexual assault? Talked to an on-campus sexual assault expert? Talked to an off-campus sexual assault expert? Been treated for potential STD exposure after an unwanted sexual experience? [females only] Taken emergency contraception (EC) after an unwanted sexual experience?
Exit survey
Sexual violence disclosure to health center providerb	Yes or No				[If positive for lifetime exposure to emotional (cyber)abuse, unwanted sexual experience, partner physical violence, or stalking] Earlier when you took the first survey, you answered questions about hurtful experiences that may have happened to you. During your visit, did you share with your counselor or student health provider any experiences like these you may have had?

^aScale was modified to be gender-neutral

^bInvestigator-developed

^cItems were a combination of earlier work and updated for this college sample

^dPerceived behavioral control subscale only

^eScale was modified