Health communication is the practice of using social media to promote healthy behaviors.1 The explosion of new communication technologies has created new opportunities for promoting health.2 This explosion is reflected in major public health initiatives focused on enhancing health communication in the United States3 and China.4 Health communication interventions are especially helpful in improving health knowledge, supporting health services, and spurring behavior change. Health communication has been shown to increase demand for health services, reduce risk behaviors, and improve health service delivery.5,6
Many health communication interventions draw on the concept of social marketing, which is the systematic application of commercial marketing concepts and techniques to the planning, execution, analysis, and evaluation of programs.7,8 Companies often use evidence-based social marketing principles to develop health communication intervention tools for health improvement.7,8 Social marketing has emerged as a standard method for developing and implementing health communication interventions.9 However, the social marketing approach has generally been a “top-down” approach, relying mostly on experts.10 This approach may neglect the knowledge, creativity, and power of nonexperts. Insufficient community engagement during the development of social marketing campaigns may result in ultimately less appealing and less effective campaigns.
Crowdsourcing, the process of shifting individual tasks to a group, has the potential to overturn the top-down paradigm of corporate health communication projects. Crowdsourcing differs from conventional health communication methods in 2 ways.10 First, rather than using experts for idea generation, crowdsourcing methods assign idea-generating tasks to the community, promoting direct community engagement. Second, a crowdsourcing approach can be used to design more effective images, policy, and videos.10,11 By using some principles of community-based participatory research (CBPR) (involving members of a study population as active and equal participants in the phases of intervention development), crowdsourcing can tap community wisdom to generate new messages to promote condom use among populations that have been challenging to reach through conventional health communication methods. China provides a favorable setting for evaluating crowdsourcing because of 2 trends. First, government policies encourage innovation, and specifically crowdsourcing, to improve health.12 Second, China has large populations among second and third-tier cities, providing a range of in-person and online crowds. In addition, most of the current condom promotion strategies in China were only limited to community-based intervention (ie, distribute condoms in venues),13 and promote condom use through social media would be essential.
The purpose of this study was to compare the effectiveness of a crowdsourced video to a social marketing video in promoting condom use among high-risk men who have sex with men (MSM) in China. And a noninferiority randomized controlled trial (RCT) was chosen for the following reasons: First, there is a well-established literature on social marketing approaches, and social marketing approach has emerged as a standard method for developing interventions.9 Second, a previous study proved that crowdsourced video was noninferior to social marketing intervention in promoting human immunodeficiency virus (HIV) testing among MSM,14 but whether it is also noninferior to social marketing intervention in promoting condom use is still unclear. Third, the rate of condomless sex is high among MSM in China.15 Specifically, given crowdsourcing is a bottom-up approach,10 it can substantially increase community engagement and reduce stigma.16,17 We anticipated that a crowdsourced intervention has a high likelihood of motivating behavior change among MSM, including condom use.
MATERIALS AND METHODS
This noninferiority RCT evaluated whether a crowdsourced video is noninferior to a social marketing video in promoting condom use among MSM in China. A control group without a video intervention was not included. The study was conducted between November of 2015 and February of 2016. Before participant recruitment, a pilot study with 150 MSM was conducted to evaluate the survey instrument and inform sample size calculation.
The study protocol was approved by Chinese (Guangdong Provincial Centre for Skin Diseases and STI Control) and American (University of North Carolina at Chapel Hill and University of California, San Francisco, No. 15–1522) institutional review boards (Supplement A, http://links.lww.com/OLQ/A313, Protocol), and has been published.18
This study used a 1-minute crowdsourced video intervention and a social marketing video intervention. The crowdsourced video was developed through a crowdsourcing contest in 2015. The reasons for using crowdsourcing contest include: (1) crowdsourcing method is a bottom-up approach, and it uses some principles of CBPR, and it has strong potential to engage the people to participate in; (2) crowdsourcing approach was considered to a time-saving, cost-efficient, and useful approach for getting a variety of potentially compelling and effective health communication messages.19–21 We followed the following steps to develop the crowdsourced intervention. First, the eligibility of the crowdsourced video (1-minute short video, relevant, and has the potential to promote condom use among Chinese MSM) was determined by the organizers, and a call for entry was drafted and discussed. Second, the call for entries was publicized on the group websites (gay websites), and through lectures and interactive feedback sessions at college campuses (introduced the crowdsourcing contest through lectures and answered relevant questions to audiences after the lectures). The call for entries was also shared publicly on WeChat (an instant messaging system similar to Facebook and Twitter, with about 1 billion users). Anyone was eligible (not restricted to MSM) to submit a video, whereas a prize was only provided to the finalist video. After entries had been collected, a group of expert panel was recruited from public health, business, and research sectors selected the finalist video from all entries, by providing each of the video entries with a score of 1 to 10 (1, worst; 10, best). The judges identified the contest winner based on the capacity to reach untested individuals, generate excitement, and community responsiveness. Overall, 11 eligible videos were received and scored. The finalist video emphasized the protective function of condoms by showing a wall protecting against cartoon virus (http://v.qq.com/x/page/j0171qo8h75.html). The 1-minute social marketing video was shot by a marketing company in Jinan (Shandong, China) following a script was written by social marketing experts in San Francisco, CA and approved by young MSM in the gay community (http://v.qq.com/x/page/c016616uiyl.html). This social-marketing video was specifically designed for this study. It emphasized sexual health as love, with 2 men negotiating on condom use before having sex.18
Setting and Participants
This is an online RCT, and participants were recruited across China. We partnered with Danlan (Beijing, China), an organization that runs Blued (a gay partner-seeking mobile app with over 27 million users) to recruit the participants online. Banner advertisements were placed on the mobile app, inviting participants to join the survey. This organization also sent announcements through its social media platforms, WeChat and Weibo. All participants who clicked the link for the survey were screened for eligibility. Inclusion criteria included: born biologically as a male, anal sex with a man at least once during their lifetime, condomless anal or vaginal sex in the last 3 months, and at least 16 years of age. After meeting the screening criteria, participants were asked to sign the informed consent form electronically and to provide a mobile telephone number (only for follow-up purpose) and finish an online baseline survey.
After the baseline survey, eligible participants were randomized in a 1:1 ratio using computer-based randomization in Qualtrics (Provo, UT). This is a blind study, participant recruitment, randomization, and intervention delivery were all computer-based, and the participants did not know whether the video they watched is a crowdsourced video. After finishing the baseline survey, all eligible participants viewed either the crowdsourced video or the social marketing video (without noticing which video they watched). The videos were embedded in the baseline survey, and they have to watch the video before they can submit the survey. They completed 3-week and 3-month follow-up surveys after video watching to assess their condom use after the intervention.
At 3 weeks and 3 months postintervention, the research group sent each participant a text message with a survey link. The surveys included 30 questions about the primary outcome (condom use) and secondary outcomes. Upon completion of the 3-week and 3-month postintervention surveys, participants received a mobile credit valued at US $16 (including US $8 for the baseline survey) and US $8, respectively.
The primary outcome of this study was self-reported condomless sex with a man or woman in the 3 weeks and 3 months after intervention. Secondary outcomes for both 3 weeks and 3 months postintervention included condomless sex with a man, condomless sex with a woman, improvement in condom use social norms, improvement in condom self-efficacy, condom negotiation, HIV testing, and syphilis testing.
The baseline survey collected data on sociodemographic characteristics including age, education, annual income, student status, marital status, self-identified sexual orientation, and disclosure of sexual orientation to healthcare provider(s).
Questions on condom use social norms investigated a participant's perception of their friends' attitudes toward condom use and safe sex. Each participant was asked to answer 6 survey items (5-point Likert scale: 1, strongly disagree; 5, strongly agree).22 Mean social norm scores were compared with evaluate whether participants experienced a change in social norms after viewing the intervention. Questions on condom use self-efficacy investigated a participant's intent to use condoms, also measured at both baseline and at the 3 weeks and 3 months follow-up, using 7 survey items each graded on a 5-point Likert scale (graded as above).23 Condom negotiation was defined as an attempt to convince an unwilling partner to use a condom.
The sample size for this noninferiority RCT was determined to assume an equal probability of reporting condomless sex in the crowdsourced video and social marketing video arms. Assuming a 50% probability of condomless sex in each arm, a 1-sided α of 2.5%, a noninferiority limit of 10%, and loss to follow-up of 10%, a total of 1170 individuals was required (585 in each arm) to have 90% power (1-β).
Participant demographic characteristics and sexual risk behaviors were described in each of the intervention arms. The primary endpoint was evaluated using the difference in proportions between the 2 arms of participants still engaging in condomless sex within 3 weeks and 3 months after either intervention (crowdsourced minus social marketing), with a noninferiority margin of +10%. The upper limit of a Wald 2-sided 95% confidence interval (CI) was used to evaluate noninferiority.
For the primary outcome, a complete-case analysis was conducted only for participants who completed the 3-week and 3-month follow-up surveys. A multiple imputation method was conducted as a sensitivity analysis. Covariates in the imputation model were intervention arm, age group, education, home province, sexual orientation, the number of partners in the last 3 months (before baseline) and condom use during a first sexual encounter with another man. Statistical analyses were conducted using SAS version 9.4 (SAS Institute, Cary, NC).
Effect modification was assessed using a linear probability model that included an interaction term between intervention arm and the specified covariate. The purpose of the effect modification is to analyze whether the intervention would have a different effect in different subgroups of MSM. The 4 prespecified covariates (whether watched a condom promotion video in the last 3 weeks before intervention, whether watched a testing promotion video in the last 3 weeks before intervention, number of male partners in the last 3 months before intervention, and whether were students) were evaluated.
Results were reported following standardized guidelines (Supplement B, http://links.lww.com/OLQ/A314, CONSORT Checklists). The study was registered with ClinicalTrials.gov (NCT02516930).
Overall, the study link was clicked 7892 times. Of these, 341 withdrew from the survey before eligibility screening, 5747 did not meet the inclusion criteria, and 194 withdrew before reaching the informed consent, and 413 people did not complete the baseline survey. From the 1197 people who finished our online survey, 24 people were further excluded (5 duplicates, and 19 invalids). Of the remaining 1173 individuals, 578 and 595 were randomly assigned to either the crowdsourced or the social marketing intervention group (Fig. 1). Of the 1173 participants, 907 (77%) completed our 3-week follow-up online survey, and 791 (67%) completed our 3-month follow-up online survey. The response rate was comparable in both arms. The participants who responded to follow-up were similar to those who did not at 3 weeks, except marital status and sexual orientation (Supplement C, http://links.lww.com/OLQ/A315).
Overall, participants were recruited from 269 cities in 32 provinces of China. Of the 1173 participants, the majority were older than 20 years (71%), never married (83%), and had an annual income less than US $15,000 (94%). Most of the participants self-identified as homosexual (70%) and over a third were students (36%). All participants identified as men, and no participant identified as transgender. Demographics and behaviors were similar between the 2 intervention arms (Table 1).
Primary Condom Use Outcome
Of the 907 respondents at 3 weeks, 299 (33%) reported engaging in condomless sex in the 3 weeks after watching their assigned condom promotion video. In the crowdsourced video arm, 146 (33.6%) of 434 reported condomless sex compared with 153 (32.3%) of 473 in the social marketing video arm. For the complete-case analysis (907 respondents), the estimated difference in proportions between arms was +1.3% (95% CI, −4.8% to 7.4%). The estimated difference in imputation analysis was +1.3% (95% CI, −4.1% to 6.6%) (Fig. 2).
Of the 791 respondents at 3 months, 537 (68.0%) reported engaging in sex with male only, 28 (3.5%) reported engaged in sex with female only, and 65 (8.2%) reported engaging in sex with both male and female. Four hundred two (50.8%) reported engaging in condomless sex after watching their assigned condom use promotion video. Among the 365 people who reported engaged in condomless sex with male partner in the past 3 months, 245 (67.1%) engaged in condomless sex with regular partner only, 83 (22.7%) engaged in condomless sex with casual partner only, and 37 (10.1%) engaged in condomless sex with both regular and casual partners. In addition, 58 men engaged in condomless sex with female partners in the last 3 months. In the crowdsourced video arm, 196 (52.1%) of 376 reported condomless sex compared to 206 (49.6%) of 415 in the social marketing video arm. For the complete-case analysis, the estimated difference in proportions between arms was +2.5% (95% CI, −4.5% to 9.5%, the noninferiority criteria were met. No significant modification of the intervention effect was observed (Supplement D, http://links.lww.com/OLQ/A316).
Secondary Outcomes at 3 Weeks
For the complete-case analysis, the estimated differences in proportion of condomless sex with a male or female partner between arms were +0.7% (95% CI, −5.2% to 6.6%) and +2.3% (95% CI, −0.8% to 5.3%) for the crowdsourced arm and social marketing arms, respectively (Fig. 3).
The complete-case analysis showed that the estimated differences between the 2 intervention arms for condom social norms and condom use self-efficacy, respectively, were +1.0% (95% CI, −5.6% to 7.5%) and +5.6% (95% CI, −1.1% to 12.0%). The postintervention condom negotiation rates were also similar between the 2 intervention arms, with a difference of −3.3% (95% CI, −9.8% to 3.2%). The estimated differences in proportions for HIV and syphilis testing between arms from the complete-case analysis were −0.7% (95% CI, −5.6% to 4.3%) and +2.4% (95% CI, −1.9% to 6.6%), respectively.
The secondary outcomes investigated at 3 months postintervention are listed in Supplement E, http://links.lww.com/OLQ/A317. They were similar to those at 3 weeks postintervention. At 3 weeks and 3 months postintervention, consistent increased mean total scores for condom use social norm and condom use self-efficacy were observed (Supplement F, http://links.lww.com/OLQ/A318). Supplement F also indicated that there was no interaction between the other videos watching and the intervention on promoting condom use.
This RCT demonstrated that a crowdsourced video was not inferior to a social marketing video in promoting condom use among high-risk MSM in China. Promoting condom use among MSM is challenging.24 However, by engaging the community in developing novel and creative solutions,25 crowdsourcing has the potential to create effective interventions that are more acceptable to the community.26 Our study extends previous research in condom use promotion among MSM by using crowdsourcing, recruiting only high-risk MSM at greatest risk for HIV, and evaluating both short- and medium-term effects of the intervention.
Our results showed that the video developed through crowdsourcing contest was not inferior to the social marketing video in promoting condom use. This finding is consistent with the sparse literature on using crowdsourcing as a health communications tool for intervention development.25 However, the previous study indicated that health communication tools such as video interventions usually have short-term effects, observable immediately after viewing.27 One potential method to increase the effect duration is to deliver the intervention through social media platforms more frequently. In addition, because all participants engaged in condomless sex within 3 months before the intervention, and about half of them consistently used condom within 3 months after the intervention, indicated that both interventions successfully increased condom use among Chinese MSM.
At 3 months postintervention, we also found evidence of persistent effects from the crowdsourced intervention, both in the primary outcome (with roughly half of the participants engaging in condomless sex in the 3-month postintervention) and some secondary outcomes, including condom use self-efficacy and social norms total scores. This was especially important, considering the inclusion criteria for our study required participants to have had condomless sex in the 3 months before the study. Our results were promising in terms of the long-term effectiveness of a crowdsourced intervention. However, further research would be useful to determine the optimal frequency of campaigns.
This study has several policy and implementation implications. First, crowdsourcing contests are adaptable to many settings to develop local health campaigns. The multisectoral networks and infrastructure necessary for the implementation of such contests are commonly found in a wide range of low- and middle-income countries. By using such networks, crowdsourcing can be used to collect wisdom from large numbers of people to develop health communication tools that are responsive to local challenges. Furthermore, the crowdsourcing contest model used in this study and models using networks could be useful in settings where civil society organizations are constrained or less able to inform public health programs directly.7
Three potential limitations of our study merit discussion. First, the self-selection processes for trial participation itself is an intervention. Especially, the recruited participants were primarily MSM who were young and well educated,28 cannot represent all MSM in China, and even cannot represent the registered gay dating app users. However, we anticipated that the bias of the self-selection process would be balanced between the 2 intervention groups. Second, one third of participants were lost to follow-up at 3 months, which could have introduced a selection bias. However, both those who did and did not respond to the follow-up survey were similar in sociodemographic and sexual risk behaviors, and the imputation results accounting for nonresponse closely matched the complete case data. Third, all the behaviors measured in our study were self-reported, and social desirability bias may be a concern. However, because all of the surveys conducted in our study were computer based,29 we anticipated that the strength of this bias was minimal. Fourth, some of our assumptions for sample size calculation (10% loss to follow-up) was not met, as around one third of the participants lost to follow up in 3 months. This may reduce the power of the currently reported study. However, based on the remaining samples, our study did achieve a noninferiority. Fifth, the sample size calculation was based on the primary outcome and may do not have enough power to detect the effect of modification.
Although our study demonstrates that a crowdsourced video is noninferior to social marketing tool in promoting condom use among Chinese MSM, research on crowdsourcing is still very limited. Future studies on crowdsourcing implementation should aim to refine crowdsourcing methods, use versatile strategies to promote crowdsourcing contests, and induce and sustain community engagement during the entire crowdsourcing process. In addition, studies to evaluate the long-term effects of multiple crowdsourcing communication tools will be critical, as the effect of a single message intervention is very likely to fade over time.30
1. Babrow AS, Mattson M, Thompson T, et al. Theorizing about health communication. Handbook of health communication 2003; 35–61.
2. Kreps GL, Neuhauser L. New directions in eHealth communication: Opportunities and challenges. Patient Educ Couns 2010; 78:329–336.
3. Parrott R. Emphasizing “communication” in health communication. J Commun 2004; 54:751–787.
4. Zhao Y. Communication in China: Political economy, power, and conflict. Rowman & Littlefield Publishers, 2008.
5. Ralph LJ, Berglas NF, Schwartz SL, et al. Finding teens in TheirSpace: Using social networking sites to connect youth to sexual health services. Sexuality research and social policy 2011; 8:38–49.
6. White RO, Eden S, Wallston KA, et al. Health communication, self-care, and treatment satisfaction among low-income diabetes patients in a public health setting. Patient Educ Couns 2015; 98:144–149.
7. Wei C, Herrick A, Raymond HF, et al. Social marketing interventions to increase HIV/STI testing uptake among men who have sex with men and male-to-female transgender women. Cochrane Database Syst Rev 2011; CD009337.
8. Bull SS, Levine DK, Black SR, et al. Social media-delivered sexual health intervention: A cluster randomized controlled trial. Am J Prev Med 2012; 43:467–474.
9. Gordon R, McDermott L, Stead M, et al. The effectiveness of social marketing interventions for health improvement: What's the evidence? Public Health 2006; 120:1133–1139.
10. Zhang Y, Kim JA, Liu F, et al. Creative contributory contests to spur innovation in sexual health: 2 cases and a guide for implementation. Sex Transm Dis 2015; 42:625–628.
11. Qin Y, Liu F, Tang W, et al. HIV self-testing among high-risk men who have sex with men in China: A cross-sectional study. Lancet 2016; 388:S76.
12. China SCo. State Council Guidance on the accelerating of the building of innovation supporting platform of China. China: State Council of China, 2015.
13. Duan Y, Zhang H, Wang J, et al. Community-based peer intervention to reduce HIV risk among men who have sex with men in Sichuan province, China. AIDS Educ Prev 2013; 25:38–48.
14. UNAIDS, WHO. Ethical considerations in biomedical HIV prevention trials. Geneva: UNAIDS, 2012:80.
15. Tang W, Pan J, Jiang N, et al. Correlates of Chlamydia and gonorrhea infection among female sex workers: The untold story of Jiangsu, China. PloS one 2014; 9:e85985.
16. Zhang TP, Liu C, Han L, et al. Community engagement in sexual health and uptake of HIV testing and syphilis testing among MSM in China: A cross-sectional online survey. J Int AIDS Soc 2017; 20:21372.
17. Zhang W, Schaffer D, Tso LS, et al. Innovation contests to promote sexual health in China: A qualitative evaluation. BMC Public Health 2017; 17:78.
18. Liu C, Mao J, Wong T, et al. Comparing the effectiveness of a crowdsourced video and a social marketing video in promoting condom use among Chinese men who have sex with men: A study protocol. BMJ Open 2016; 6:e010755.
19. Ess Hv. Success criteria and strategies for Crowdsourcing www.slideshare.net2010
[cited 2017 may 18, 2017]. Available from: https://www.slideshare.net/searchbistro/harvesting-knowledge-how-to-crowdsource-in-2010
accessed Sep 17, 2010.
20. Brabham DC. Crowdsourcing as a model for problem solving an introduction and cases. Convergence 2008; 14:75–90.
21. Parvanta C, Roth Y, Keller H. Crowdsourcing 101: A few basics to make you the leader of the pack. Health Promot Pract 2013; 14:163–167.
22. DeHart DD, Birkimer JC. Trying to practice safer sex: Development of the sexual risks scale. J Sex Res 1997; 34:11–25.
23. Brafford LJ, Beck KH. Development and validation of a condom self-efficacy scale for college students. J Am Coll Health 1991; 39:219–225.
24. Musinguzi G, Bastiaens H, Matovu JK, et al. Barriers to condom use among high risk men who have sex with men in Uganda: A qualitative study. PloS one 2015; 10:e0132297.
25. Tang W, Han L, Best J, et al. Crowdsourcing HIV test promotion videos: A noninferiority randomized controlled trial in China. Clin Infect Dis 2016; 62:1436–1442. [published Online First: 2016/05/01].
26. Ranard BL, Ha YP, Meisel ZF, et al. Crowdsourcing—Harnessing the masses to advance health and medicine, a systematic review. J Gen Intern Med 2014; 29:187–203.
27. Vidanapathirana J, Abramson M, Forbes A, et al. Mass media interventions for promoting HIV testing. Cochrane Database Syst Rev 2005; 3.
28. Han L, Bien CH, Wei C, et al. HIV self-testing among online MSM in China: Implications for expanding HIV testing among key populations. J Acquir Immune Defic Syndr 2014; 67:216–221.
29. Mao J, Liu C, Wong T, et al. Crowdsourcing Versus Social Marketing Video Campaigns to Promote Condom Use: A Non-inferiority Randomized Controlled Trial Among MSM and Transgender Individuals in China. Hongkong: 2016 International Behavioral Health conference, 2015.
30. Tucker JD; SESH Study Group. Crowdsourcing to promote HIV testing among MSM in China: Study protocol for a stepped wedge randomized controlled trial. Trials 2017; 18:447.