A modified respondent-driven sampling (RDS) approach was used to recruit participants. Data collection occurred between October 2010 and May 2011. RDS is a systematic approach to identification and recruitment of hard-to-reach populations. The approach relies on referrals, where the initial “seed” or index person recruited is invited to identify and recruit others to participate.20
Recruitment occurred in community settings in the Denver CO metropolitan area and in a college community in Louisiana. Methods used were online personal channels and postings on popular blogs and websites, and advertisements in college and local newspapers in U.S. cities with higher than average combined incidence rates for STI and HIV.21,22
Recruitment was focused on African-American and Latino youth given the disparity in HIV and STI infection between these youth compared to other groups, although no racial or gender criteria were used in selection of participants.
In community settings, research assistants either approached people directly if they thought they might be eligible, or set up a table and waited for people to approach them. When recruiting online, three websites were accessed to better identify and reach youth of color, including Mi Gente, Black Planet, and Urban Chat. Recruiters posted information about the study to these sites, and responded to requests for more detail about the study. There were no inquiries from youth using the remaining two sites.
Finally, 16 local and school (community college, college and university) newspapers in geographic areas with the highest prevalence of chlamydia, gonorrhea and HIV among those aged 15–19 years were identified, and recruitment ads were placed in these papers. People responding to the ads sent an e-mail or voice-mail to study staff, which then enrolled them and encouraged them to recruit friends as described below. All participants, regardless of recruitment method, were screened using identical eligibility criteria (i.e., aged 16–25 years, a U.S. resident, owner of a Facebook page, willing to complete study behavioral risk assessments, and able to read and write in English).
Additionally, based on formative work for this project, only those people who agreed to sign up to receive news from (i.e., “like”) our Facebook study pages (intervention or control) would be able to see program content and conversations within their own newsfeed without going outside their profile page to engage with the study. Therefore “liking” the intervention or control Facebook page was an eligibility criterion. Once a person “likes” a group on Facebook, they become automatically linked to that groups’ page, and everything posted on the group page is broadcast to every network member's page in the form of an RSS (rich site summary, one process used to regularly update material online and share it with networks) feed. Those eligible were invited to participate.
Participants recruited by study staff were incentivized to recruit up to three friends to participate (Wave 1); this wave of recruits were incentivized to recruit up to three friends (Wave 2); this wave was incentivized to recruit up to three friends to participate (Wave 3). Each individual recruited by study staff and all people they recruited completed online consent prior to enrollment and then were considered part of the same network. Participants received a gift card valued at $5 per person for up to three people recruited into the study for a possible total of $15 for this effort.
All eligible participants, including seeds and all those referred through their networks, completed informed consent. They also completed a baseline behavioral assessment of sexual risk via an online tool generated and delivered through Zoomerang, a commercial online survey software program that allows users to create and publish surveys online. Zoomerang served as a third-party host for the data, and their third-party hosting agreements comply with all current IRB requirements related to privacy and data security.23
All participants were sent a link via e-mail on their Facebook page that would take them to the informed consent and online survey, which they could self-administer on their own computer. The survey took approximately 15 minutes to complete. Participants were given a gift card valued at $15 for completion of the baseline assessment. Study procedures were approved by IRBs at the University of Colorado and the Columbia Mailman School of Public Health.
Participants and those they recruited were randomly assigned as a network unit to either intervention or control status. The control page was called “18-24 News” intended as a play on the concept of sharing what was happening between 6 pm until midnight on the 24- hour clock (18:00–24:00) and what was interesting in the news to those aged 18–24 years. The intent of using this page as a control was to specifically avoid sexual health content.
The content for the intervention page, “Just/Us” on Facebook, was developed in concert with all members of the study team. Implementation was led by ISIS, Internet Sexuality Information Services in Oakland CA. Content was based on two fundamental ideas generated during this formative phase: that sexual health is a human right and function of social justice; and that youth need a space to share ideas and concepts with their peers, as well as professional experts. Content for the intervention page included eight broad topics related to sexual health (e.g., communication regarding sexual history; expectations for a healthy relationship; skills building for condom negotiation and condom use; and how to access STI testing). One week was devoted to each topic. The topics provided a framework for interactions between youth facilitators employed by ISIS and participants.
Youth facilitators would make multiple updates each day to the page in the form of video links, quizzes, and games as well as threaded discussions relevant to that weeks’ topic. At the end of 8 weeks, topics were recycled to ensure those enrolling at different times were exposed to all eight topics. Appendix A
(available online at www.ajpmonline.org
) shows sample elements from the Just/Us intervention page and a screen shot of the 18-24 News control page.
At the end of 8 weeks, participants were invited to complete a follow-up behavioral risk assessment. After completing this they could remain “friends” with Just/Us or 18-24 News on Facebook, but would be exposed to only the topics now recycling that they had already viewed. At 6 months, they were invited to complete their second follow-up assessment. Participants were offered an online gift card from Amazon, Jamba Juice, Walmart or Target valued at $15 for each assessment.
Measures included demographic characteristics of participants: age, gender, race, ethnicity, education, and ZIP code. The primary study outcomes were condom use at last sex (measured as the response to the question: Last time you had sex was a condom used? Yes or No) and proportion of sex acts protected by condoms in the past 60 days. (Participants were asked to estimate the number of times they had sex in the past 60 days, and then to estimate the number of times in 60 days they used a condom. Proportion of sex acts protected by condoms is the number of times an individual had sex protected by condoms in the past 60 days divided by the total number of times they had sex in 60 days).
Additional behavioral outcomes assessed were number of sex partners in the past 2 months (dichotomized as two or more partners compared to zero or one partner); intention to use condoms at the next sexual encounter (also dichotomized, yes or no); and whether the most recent sex partner was considered a “main” or primary partner or a casual partner. There were numerous factors measured on a 5-point scale from “never” to “all of the time” including whether participants were drunk or high during their last sexual experience; whether their friends on Facebook were likely to use condoms (peer norms for condom use); and whether they were confident they could use condoms (self-efficacy for condom use). Finally, at the 6-month follow-up, participants were asked to indicate whether they “liked” numerous Facebook pages, embedding Just/Us among three other choices to assess contamination—if controls “liked” Just/Us in large numbers, there would be a concern that they had been exposed to intervention content.
Analyses occurred between May 2010 and January 2012. Basic data on engagement with the Facebook pages and study elements were obtained through Google analytics, an open source web statistics site that generates information on how individuals interact with specific sites on the Internet. Statistical analyses were performed using SAS 9.2. The completed survey data included repeated measures from three time points: (1) a baseline at enrollment; and (2) a 2-month; and (3) a 6-month follow-up. The two study groups were evaluated for equivalency on study outcomes at baseline and demographic measures, including gender, race, ethnicity, U.S. region, age and education at enrollment using student's t-test and chi-square comparisons. Equivalency across groups was also assessed by recruitment method (face-to-face or Internet or newspaper) and by type of gift-card incentive chosen.
The unit of analysis was the individual, but observations were potentially non-independent because individuals were nested within networks, and networks were then assigned to treatment groups. Initial power estimates were established based on outcomes from previous work by some of the study team with youth online.24
Sample size estimates of 1156 with 578 per study arm were based on assumptions of baseline condom use of 55% with 90% power to detect differences of 10% between intervention at control groups with a CI of 99% (alpha=0.01), and intra-class correlations (ICC's) for network members of 0.15.
The ICC's at baseline were 0.15 for condom use at last sex and 0.13 for proportion of sex acts protected by condoms demonstrating that behaviors among people who already know each other are related, underscoring the need to account for the non-zero ICCs to avoid overestimating effects from exposure to intervention content. All outcomes among those sexually active were modeled utilizing a nested design with repeated measures techniques. The modeling, adjusted estimates and significance tests were performed using Proc Glimmix for binary outcomes and Proc Mixed for continuous outcomes to account for the nested structure of the data.
All outcomes were modeled in terms of the main effect of changes over time, the main effect of treatment, and the interaction between treatment and time. An interaction between time and treatment was considered evidence of an impact of the intervention and was interpreted in post hoc analyses comparing the adjusted least-squares mean estimates by treatment group at each of the three time points. Potential covariates included age, gender, race, ethnicity, region of the U.S., whether or not the participant was with a primary partner, size of the participant's captured network, method of recruitment, and incentives used to recruit the participant.