Abstract
College-age men were recruited using Facebook™ advertisements (ads), as well as traditional recruitment methods, for a randomized controlled trial to compare immunological responses to human papillomavirus vaccine administered in two dosing schedules. This study compares enrollees who were recruited through traditional recruitment methods versus social networking sites (SNSs), including Facebook. Potential participants were recruited using flyers posted on and off campus(es), and distributed at health fairs, classes, sporting, and other campus events; e-mails to students and student organizations; and print advertisements in student newspapers and on city buses. Facebook ads were displayed to users with specific age, geographic, and interest characteristics; ads were monitored daily to make adjustments to improve response. A total of 220 males, aged 18 to 25 years enrolled between October 2010 and May 2011. The majority of participants (51%) reported print advertisements as the method by which they first heard about the study, followed by personal contact (29%) and Facebook or other SNSs (20%). The likelihood of a SNS being the source by which the participant first heard about the study compared with traditional methods was increased if the participant reported (a) being homosexual or bisexual or (b) posting daily updates on SNSs. Facebook and other SNSs are a viable recruitment strategy for reaching potential clinical trial participants among groups who typically use social media to stay connected with their friends and hard-to-reach groups such as young men who self-identify as homosexual or bisexual.
Introduction
Human papillomavirus (HPV) infection is the most common sexually transmitted infection (STI) affecting more individuals each year than the seven other leading STIs combined. Men and women aged 15 to 24 years have the highest prevalence (approximately 79 million) and incidence of HPV infection (approximately 14 million new cases annually in the United States; Satterwhite et al., 2013). Although frequently asymptomatic, long-term HPV infection is the primary cause of cervical cancer and genital warts (Steben & Duarte-Franco, 2007), and is implicated in a sizeable proportion of cancers of the oropharynx, vulva, penis, and anus (D’Souza & Dempsey, 2011; Heffernan, O’Neill, & Timon, 2010; Moscicki, 2008; Plotkin, Orenstein, & Offit, 2008). Annually, the cost of treating cervical intraepithelial neoplasia, head and neck malignancies, recurrent respiratory papillomatosis, and genital warts exceeds $1.7 billion (Owusu-Edusei et al., 2013).
The HPV vaccine is a highly effective strategy for preventing HPV infections that result in genital warts and HPV-related cancers (Moscicki, 2008). The quadrivalent HPV vaccine was first licensed in June 2006 for use in females aged 9 to 26 years (Centers for Disease Control and Prevention [CDC], 2011a) and for males of the same ages in 2011 (Barr & Sings, 2008; CDC, 2011b; Schuchat, 2011). Three doses are required for full protection and are licensed to be given at Day 0, then 2 and 6 months later. HPV vaccine uptake remains low (CDC, 2011a) despite its proven clinical efficacy (Rambout, Hopkins, Hutton, & Fergusson, 2007), extensive research to increase acceptance of the vaccine, and significant promotional campaigns. Four years after the introduction of the vaccine, less than half of the eligible female adolescents (49%) had initiated HPV vaccination with only 32% completing the three dose series (CDC, 2011a). As of 2012, only 18% of adolescent boys had received ≥1 dose of HPV vaccine (Cullen, Stokley, & Markowitz, 2014).
Among young men, previous research has reported that HPV vaccine acceptability and intention to receive the vaccine were directly related to perceived susceptibility to HPV infection, perceived benefits, and effectiveness of the HPV vaccine, as well as belief that one’s male friends will receive the HPV vaccine, but were inversely related to perceived barriers to receiving HPV vaccine (Gerend & Barley, 2009).
The purpose of this study was to compare the characteristics of college-age men who participated in a HPV dosing trial designed to test the noninferiority of an alternate dosing schedule of 0, 2, and 12 months (Lin, Zimmerman, Nowalk, Huang, & Raviotta, 2014). The health belief model (Janz, Champion, & Strecher, 2002) was used as a framework for the design of the recruitment strategies that used both traditional and electronic recruitment methods, including Facebook™ advertisements (ads). Generally, social networking sites (SNSs) have been determined to be a valuable and cost-effective means of recruiting young populations (Park & Calamaro, 2013). Specifically, Facebook ads have been successfully used to recruit children with Klinefelter syndrome (Close, Smaldone, Fennoy, Reame, & Grey, 2013) and young women for chlamydia screening (Ahmed et al., 2013). Facebook is also popular among health educators; for example, a systematic examination of SNSs identified 178 sexual health promotion activities, of which 71% used Facebook as the delivery platform (Gold et al., 2011). Moreover, in a review of the use of SNS for public health research, Capurro et al. (2014) conclude that SNSs are well suited to research about public health topics that are typically considered socially undesirable (alcohol, tobacco, or illegal drug use and sexual behavior/health), and/or focus on traditionally hard-to-reach groups.
Among young adult (18-29 years) Internet users, a high percentage (72%) report using SNSs such as Facebook and MySpace™ (Lenhart, Purcell, Smith, & Zickuhr, 2011) with millions of active Facebook users of all ages reported (Facebook.com, 2014). This popularity, combined with Facebook’s unique ad targeting mechanisms, suggests that Facebook ads could be a successful addition to clinical trial recruitment campaigns (Scott, 2013).
Our research questions asked (a) how did Facebook ads and other SNSs compare with traditional recruitment methods to enroll 18- to 25-year-old men for a clinical HPV vaccine efficacy trial and (b) did the individuals who first heard about the study through Facebook or other SNSs (via friends) differ from those who first heard about it using traditional recruiting methods. This information may be useful to other researchers who are using online methods to recruit participants from specific population subgroups for clinical trials.
Method
Eligibility
The study protocol was approved by the University of Pittsburgh Institutional Review Board. Targeted enrollment was 220 males, ages 18 to 25 years with fewer than five lifetime sexual partners (to reduce the likelihood of HPV infection), no history of HPV infection (genital warts) or vaccination, no autoimmune disease nor immunosuppression, no hospitalization in the past year, no receipt of blood products or immunoglobulins within 90 days, no participation in other drug studies within 30 days and no receipt of other vaccines within 8 days. The protocol required four visits to the study site for an intake survey, blood draw, and vaccination dose 1, dose 2, dose 3, and final blood draw and survey. Participants also completed online postvaccination surveys to assess presence and severity of vaccine side effects. Participants who completed all four visits received $130 compensation.
Theoretical Framework
The theoretical model used in this research study was the health belief model (Janz et al., 2002). The authors designed the advertising campaign, enrollment procedures, and study logistics to leverage perceived susceptibility, perceived benefits, perceived barriers, self-efficacy, and cues to action. Prior to licensure of the HPV vaccine for males aged 9 to 26 years, most promotional campaigns had focused on young women and prevention of cervical cancer. Lack of knowledge of their own risk of orogenital cancers from HPV infection, as well as having not been directly targeted by marketing campaigns may have attenuated young men’s awareness of the benefit of HPV vaccination. The advertising campaign was designed to increase perceived susceptibility to HPV infection by using ads composed of a color photo of an attractive subject next to a variety of call-to-action messages. Other design considerations addressed the constructs of the health belief model. For example, by targeting college classes and other groups of young men, the authors hoped to increase the perceived benefits of vaccination; by providing a simple path to enrollment via online screening, the authors hoped to increase individuals’ self-efficacy toward vaccination and by offering the vaccine free of charge, the authors hoped to eliminate the barrier of the high cost of vaccination. Reminder e-mails served as cues to action.
General Recruitment and Enrollment
Recruitment was focused on college campuses within 10 miles of the clinical facilities. No attempt was made to limit exposure to either traditional or online methods by location. Recruitment occurred from October 13, 2010, through May 4, 2011. Recruitment followed the process illustrated in Figure 1. Any male who contacted the research team through telephone or e-mail, was given a scripted description of the study protocol and inclusion criteria. If the individual expressed an interest in participating, he was directed to the online screening page, as was any individual who clicked a Facebook ad. The study team was prepared to complete the online process for any individual without access to the online study screening page. The screening page provided an overview of the study and inclusion criteria and directed interested individuals to the online screening form. Those who completed the screening form were immediately informed of their eligibility status on the next web page. Contact information for each eligible participant was automatically e-mailed by the site to the research team who contacted him to schedule his first visit. Randomized study IDs and study groups (standard dosing schedule of 0, 2, and 6 months vs. alternate dosing schedule of 0, 2, and 12 months) were assigned at that time the web form was submitted.
Recruitment and enrollment for human papillomavirus vaccination trial among college-age men.
Traditional Recruitment
Traditional recruitment methods included all methods that did not involve content posted on the Internet such as the following:
Flyers posted on and off campus in locations such as residence halls, student health centers, student unions, public bulletin boards, coffee shops, academic buildings, and businesses frequented by students
Flyers distributed at health fairs, classes, sporting, and other campus events
E-mails to students, leadership of student interest groups, and to any accessible mailing lists of potentially eligible young men
Student newspaper ads
Display ads on city buses
Notices in student newsletters
Announcements in classes by teachers or study staff
All printed advertisements and flyers were thematically similar and included a large title, at least one image of a young adult or couple, a brief description of the study, and the study’s phone number, e-mail address, and website address. Male and female portraits were selected that represented the target population in age, race, and overall appearance as well as, portraits of opposite sex and male same sex couples.
Facebook Recruitment
Facebook ads ran nearly concurrently with traditional recruitment efforts. The first ad was placed on November 5, 2010. Ads were purchased and formatted using the Facebook ads online interface and were placed as right column banners beside the Facebook content. Copy was limited to a 25-character title, a 135-character body including spaces, and a small image. Four variations of body text were created and the images were the same as those used for the print ads. The graphic design of the images was adjusted to fit the thumbnail size and horizontal orientation while maintaining thematic consistency with the print ads. The domain of the destination URL was displayed as a subheading.
The audience was targeted using the following key words: live in the United States, in Pittsburgh, Pennsylvania, exactly ages 18 to 25 years inclusive, male, student attending one of five local universities, single or in a relationship, speak English, and not already connected to PittVax (the title of our Facebook page).
As a part of its advertisement system, Facebook provided to advertisers interactive tools that reported near real-time ad performance and allowed adjustment of most placement parameters. The daily budget was set at $10 per day at a cost per impression of no more than $0.32. The ad’s click through rate was monitored daily and when it reported a decline from its usual performance, it was refreshed using different images and/or different parameter specifications to yield the highest number of clicks possible within the study budget. During the entire campaign, 15 different ads were placed.
Data Collection
Demographic and other data were collected from the baseline and postvaccination surveys. Specific to the social media aspects of this study, participants were asked how they first heard about the study (with response choices provided), how many friends they encouraged to participate in the study and what method they used to do so (with response choices provided), whether they stay connected to people through Facebook, Twitter™, or other SNSs, how often they post updates and read others’ updates. Facebook metrics that were collected and used in this analysis were the following: impressions = the number of times ads were displayed, clicks = the number of times an ad was clicked on by a viewer, cost = the dollars spent, and cost per click = total cost divided by number of clicks, cost per enrollee = total cost divided by 220.
Statistical Analysis
Data collected from the online surveys were retrieved from the web reports and imported into SPSS Version 20.0.0. Variable categories (e.g., age, year in school, sexual orientation) were combined because of small sample sizes in some cells. One participant did not provide some survey data and was dropped from analyses. Comparisons of characteristics between groups were made using chi-square analyses with alpha set at p < .05. Three regression analyses were conducted with the outcome variable for each model being first hearing about the study via social media and the reference groups being first hearing about the study via (a) personal contact, (b) print advertisements, and (c) traditional methods, which consist of personal contact and print advertisements combined. Independent variables were age group (18-21 vs. 22-25 years), race (White vs. non-White), sexual orientation (heterosexual vs. homosexual/bisexual), and frequency of making posts on an SNS (≤1/month, 1-3 times/week, ≥1/day). Variables were entered into backward regression analysis models to determine which factors were most likely to predict method of recruitment.
Results
Recruitment took place from October 13, 2010, to December 18, 2010, and from January 3 to May 4, 2011. Facebook ads ran from November 5, 2010, to December 18, 2010, and from January 3, 2011, to June 1, 2011, with a total of 21 million impressions resulting in 3,884 clicks on the Facebook ads. All recruitment efforts produced interest among 428 young men who completed the screening form. Of those, 311 were eligible and 220 enrolled, 44 of whom reported first hearing about the study through Facebook or another SNS.
The direct costs of the Facebook ads were $4,820, $1.24/click, and $110/enrollee, and did not include time spent managing the ads and the campaign. The direct cost of recruiting 111 young men who first heard about the study through print ads and flyers was approximately $6,758 or $61/enrollee although this cost did not include the personnel cost of receiving permission to post flyers and distributing them. The remaining 65 participants reported first hearing about the study from a source without direct costs (word of mouth, announcements in class, etc.). Although the time spent on various recruitment efforts by the research team was not collected, anecdotally, the amount of time spent on the Facebook ads campaign was considerably less than the time spent on the print ad campaign.
One half of enrollees were randomized to the standard dosing group (109) and one half (111) to the alternate dosing group. Table 1 reports the demographic characteristics and survey responses of participants overall and by method by which they first heard about the study. Participants were distributed across the 18- to 25-year age range with 9% to 19% of the enrollees in each age group; 19% were non-White and 15% were homosexual or bisexual; 36% were freshmen or sophomores in college, 25% were juniors and seniors, 29% were professional students; and 10% were not in school. The reason most frequently cited for enrolling in the study was the incentive payment (34%) followed by the desire to protect oneself from HPV (27%), the free vaccine (14%) and to help with a scientific experiment (13%). The second most important reason cited was evenly divided among the incentive payment, desire to protect oneself from HPV and the free vaccine (24%), with desire to help with a scientific experiment reported by 16% of respondents.
Descriptive Characteristics of Study Enrollees.
Note. HPV = human papillomavirus. Entries in boldface indicate statistical significance (p < .05).
Social networking sites = Facebook, Facebook ads, Twitter, or other social networking site.
One participant did not provide complete survey information and was not included in the analysis.
The majority of participants (51%) reported print advertisements as the method by which they first heard about the study, followed by personal contact (29%) and Facebook or other SNSs (20%). While 90% of participants reported that they stay connected with friends via social media, they more often reported reading posts on SNSs than making posts themselves.
Enrollees who first heard about the study through social networking were more likely to report being homosexual/bisexual; p = .008 and to more frequently post on SNS (p = .002) than were enrollees who first heard about the study by a non-SNS method. Enrollees were also divided by whether or not they referred friends to the study. Men who referred friends to the study were slightly younger (p = .049) and more likely to be a freshman or sophomore in college (p = .014; data not reported).
Results of the three regression analyses are reported in Table 2. The likelihood of an SNS being the source by which the participant first heard about the study compared with personal contact increased if the participant reported being homosexual or bisexual, and reported more frequently making one or more posts on SNS per day (p ≤ .037). When compared with print advertisements, the only variable that significantly increased the odds that a participant first heard about the study on SNS was reporting making posts to SNSs at least once a day (p = .007). In the third model, print advertisements and personal contact were combined into “traditional” methods and compared with SNS; sexual orientation and frequency of posting to SNS were related to SNS being the first means of hearing about the study (p < .030). Thus, those who were more active participants in SNS were more likely to have first heard about the study by this medium.
Likelihood (Odds Ratios and 95% CIs) of Enrolled Participant First Hearing About the Study Through Social Networking Sites Compared With Other Methods of Recruitment, by Logistic Regression.
Note. CI = confidence interval; social networking sites = Facebook, Facebook ads, Twitter, or other social networking site. Entries in boldface indicate statistical significance (p < .05).
Traditional methods = print advertisemt combined with personal contact.
Discussion
Existing scholarly literature describing and/or evaluating online recruitment campaigns is growing (Battistella, Kalyan, & Prior, 2010; Berry & Bass, 2012; Williams, Proetto, Casiano, & Franklin, 2012). The increasing use of the Internet and popularity of Internet media platforms, including SNS have identified that online campaigns are a viable recruitment option for clinical research (Lohse, 2013; O’Connor, Jackson, Goldsmith, & Skirton, 2014; Park & Calamaro, 2013; Piotrowski, 2012; Scott, 2013). Moreover, studies exclusively using Facebook (Fenner et al., 2012; Richiardi, Pivetta, & Merletti, 2012) are beginning to appear in the literature.
The Facebook ads used in this study stimulated interest among 3,884 young men who clicked on the ad to read it. At the conclusion of enrollment, 44 participants (20%) reported first hearing about the study through Facebook or another SNS via friends; this method was the only introduction to the study for 31participants (14% of total participants). Thus, using social media to advertise the study reached a small, but not insignificant, group that may not have been included had the study relied solely on traditional methods.
The direct cost per participant ($110) was approximately 1.8 times higher for electronic advertising compared with paper forms (newspaper and bus ads, posted and distributed flyers). However, the total cost per participant for each method of recruitment was not directly comparable because the personnel time required for each method was not measured. Anecdotally, posting and distributing flyers around college campuses was time intensive compared with posting, monitoring, and changing Facebook ads. A recent study that used Facebook ads to recruit HIV-positive individuals for a single survey, estimated their cost per participant to be significantly less at $3.56 (Yuan, Bare, Johnson, & Saberi, 2014). This difference likely reflects the difference in time and effort required to complete a 7- to 13-month study with two blood draws and three vaccine injections versus a 30-minute survey.
Part of the appeal of social media advertising is the exponential increase in message exposures potentially gained through social sharing. Snowball recruitment or sampling the process of using a group of source individuals with a given set of characteristics to reach out to their social network for similar individuals and in turn, those individuals reaching out to their social networks (Sadler, Lee, Lim, & Fullerton, 2010). Some 84% of online young adults use Facebook (Duggan & Smith, 2013). With a reported 150 to 200 friends per college-age user (Fox, 2011), the potential for reaching individuals targeted for research in this age group is vast. Bhutta (2012) reported a successful snowball enrollment campaign using Facebook where 15 groups provided potential exposure to 37,463 individuals. O’Connor et al. (2014) recruited 529 older mothers using a Tweet/retweet recruitment campaign.
SNSs could not have been used successfully as the sole method of participant recruitment in this study. The majority of college-age males in this study reported that they stay connected with friends through SNS; 64% of the study participants reported reading posts one or more times per day and nearly 20% reported making posts at least once per day, commensurate with previous reports (Pempek, Yermolayeva, & Calvert, 2009). Yet few participants referred friends through social media channels and opted for more private methods of referral such as face-to-face conversations or sending personal messages to friends. Thus, the multiplicative impact of social media on friend referral patterns was not observed in this study. Given that 17% of participants first heard about the study through a referral from a friend, our online recruitment campaign might have been more productive if a concerted effort had been made to encourage participants to share study information through social networking.
Acceptability of HPV vaccine among heterosexual, homosexual, and bisexual college-age men is high if perceived risk of HPV infection and benefits of HPV vaccination are also high (Liddon, Hood, Wynn, & Markowitz, 2010; Reiter, Brewer, McRee, Gilbert, & Smith, 2010; Zou et al., 2014). The authors identified the second most important reason for enrollment in the study was “to protect myself or others from infection” and reflected an awareness of perceived susceptibility. Others have reported that protection of their female partners from HPV infections was not a predictor of HPV vaccine acceptability among heterosexual college-age men (Gerend & Barley, 2009). Conversely, 91% of participants believed that HPV is an important problem in their community and nearly 100% of participants believed that their participation would be beneficial to others. This apparent conflict seems to reflect findings of McCann, Campbell, and Entwistle (2010) who report that expressed interest in participating in health care trials stems from a willingness to help others and contribute to medical knowledge, whereas actual participation is related to perceived personal benefit.
Consistent with others’ findings of the value of SNS to recruit difficult-to-reach populations (O’Connor et al., 2014; Park & Calamaro, 2013), it was identified that SNSs were more likely to reach homosexual/bisexual men as a point of first contact than personal contact (word of mouth) or the combined traditional recruitment strategies. This finding that SNS may be a more effective advertising/recruitment strategy for this subpopulation of men is especially valuable for both HPV vaccination strategy and risk communication. The likelihood of anal cancer among homosexual/bisexual men is nearly 20 times that of heterosexual men (Daling et al., 2004) and HPV is implicated in a sizeable proportion of cancers of the oropharynx, penis, and anus (D’Souza & Dempsey, 2011; Heffernan et al., 2010; Moscicki, 2008; Plotkin et al., 2008). SNS may prove to be an effective means to increase perceived severity of and susceptibility to HPV infection among homosexual and bisexual men.
Strengths and Limitations
This study demonstrated the ability to use Facebook ads and other SNSs to recruit young men for research on a vaccine of low acceptance. The generalizability of this study is limited by its observational methodology, small sample size, limited geographic scope, and narrow age range. Hence, the participants may not represent the broader population. However, the sampled population was likely to have access to both social and conventional media and is among the age group likely to use SNS. The study was conducted when Facebook ads were still an immature medium; since that time, Facebook has introduced new advertising methods that may differ in their effectiveness at reaching college-age males. Additionally, other SNSs with advertising opportunities have multiplied as has the use of mobile devices for Internet access. While conventional mass media marketing and health promotion theory can inform efforts to influence individuals through social media, it is likely that new methodologies and techniques will be required to optimize interventions conducted through this emerging communication medium.
Conclusions
Social media recruitment should be considered by researchers hoping to reach college-age men, as the majority of the participants in this study reported using social media to stay connected with their friends. As the use of social media broadens, this is a viable resource for reaching potential participants. Further research to increase social sharing of potential research opportunities and the benefits of participation may increase the effectiveness of social media recruitment.
Footnotes
Authors’ Note
The views expressed herein are those of the authors and not those of the funder.
Declaration of Conflicting Interests
The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Dr. Zimmerman has research grants from Sanofi Pasteur, Inc. and Pfizer, Inc. Dr. Lin has research funding from Sanofi Pasteur, Inc. and Pfizer, Inc. and consults for MedImmune, LLC. Dr. Nowalk has research funding from Pfizer, Sanofi Pasteur Inc. and consults for MedImmune, LLC.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This investigation was supported by an investigator-initiated research grant (IISP ID No. 38206) from Merck & Co, Inc. The project described was also supported by the National Institutes of Health through Grant Numbers UL1 RR024153 and UL1TR000005.