Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
AIDS Educ Prev. Author manuscript; available in PMC 2014 February 7.
Published in final edited form as:
PMCID: PMC3917310

Attitudinal, Behavioral, and Biological Outcomes of a Community Popular Opinion Leader Intervention in China

Mary Jane Rotheram-Borus, PhD, Li Li, PhD, Li-Jung Liang, PhD, Yi Wen, MD, Zunyou Wu, PhD, and the NIMH Collaborative HIV/STD Prevention Trial Group


The effects of a community popular opinion leader (CPOL) intervention were examined among market vendors in a city on the eastern coast of China. Employees of 40 food markets were enrolled in a study that provided HIV-related education and tests, and treatment for sexually transmitted diseases (STDs). Twenty markets were randomly assigned to a CPOL intervention (N = 1695) and 20 markets to a control condition (N = 1616). Market employees in the intervention condition reported positive attitudes regarding STD/HIV prevention and more frequent discussions about safe sex than those in the control condition. Compared to baseline, the prevalence of unprotected sexual acts and new STDs were significantly lower within each study condition 24 months later. Although the CPOL intervention achieved its goal of shifting attitudes within food markets, the gains did not lead to the expected behavioral and biological outcomes.

Sexually transmitted diseases (STDs) are a major cause of acute illness, infertility, long-term disability, and death, with severe medical and psychological consequences (World Health Organization, 2001). More than 340 million new STD cases are reported throughout the world every year (World Health Organization, 2007). Consistent with global trends, STDs are also rising in China. Chlamydia and herpes, followed by syphilis, trichomoniasis, and gonorrhea are the primary bacterial STDs in China (Detels et al., 2003; Wu et al., 2007a). Incidence rates of selective STDs have increased 3.7 times, from 13.8 per 100,000 in 1990 to 50.7 per 100,000 in 1998 and 58.1 in 2002 (Chen, Gong, Liang, & Zhang, 2000; Zhang, 2002). The incidence of AIDS (0.75/100,000) and HIV (2.55/100,000), however, remained low in 2007 despite a nearly twofold increase over the 2004 incidence (Chinese Center for Disease Control and Prevention, 2008).

The Community Popular Opinion Leader (CPOL) intervention model was developed by Jeffrey Kelly and colleagues using diffusion of innovation theory (Rogers, 1983). By identifying popular opinion leaders (POL) and training the POL to diffuse social norms to friends and neighbors within their community, significant reductions in HIV-related risks have been demonstrated among gay men, adolescents, and women in low-income housing in the United States (Kelly et al., 1991; Kelly et al., 1992; Kelly et al., 1997). Gay men in the United Kingdom, however, did not reduce HIV-related sexual acts when the CPOL intervention did not reach at least 15% of POLs in a community and was not sustained for at least one year (Elford, Bolding, & Sherr, 2001; Elford, Sherr, Bolding, Serle, & Maguire, 2002; Flowers, Hart, Williamson, Frankis, & Der, 2002).

Given the failure of the CPOL intervention to impact risk outcomes when core elements of the intervention were not adhered to (Elford, Bolding, & Sherr, 2004; Hart, Williamson, & Flowers, 2004; Kelly, 2004), it is critical to document that the CPOL intervention nevertheless achieves its intended mediating mechanisms: changing social norms, building social support, and enforcing positive attitudes regarding HIV prevention. Based on substantial ethnographic research (NIMH Collaborative HIV/STD Prevention Trial Group, 2007; Lieber, Chin, Li, & Rotheram-Borus, in press), we identified norms that needed to be shifted in order to reduce STDs in China. These norms included increasing treatment-seeking at STD clinics; increasing conversations regarding sexuality among partners; and increasing the use of condoms in extramarital relationships.

Food markets in a city of about 3 million persons on the eastern coast of China were identified as communities with relatively high prevalence levels of bacterial STDs and herpes, yet a relatively low rate of HIV infection (Detels et al., 2003; NIMH, 2007). In the CPOL intervention markets, social networks were identified (e.g., mahjong players, gamblers) and POLs within each network were selected, recruited, trained, and encouraged to diffuse HIV prevention messages to their friends and neighbors. The support for POLs decreased over time, but continued intermittently for two years.

This article evaluates whether POLs effectively shifted community norms in intervention markets and the impact of these shifts on three outcome indicators: 1) attitudes toward safe sex and preventive care; 2) unprotected sex acts with non-spousal partners; and 3) STD incidence. Given the important role of POLs in the CPOL intervention model, we also evaluate the effects of the intervention on POLs and typical market employees in the intervention condition.


Study Background and Study Venues

This study is part of a National Institute of Mental Health Collaborative HIV/STD Prevention Trial that took place in five countries (China, India, Peru, Russia, Zimbabwe) with populations at risk of contracting HIV and STDs (NIMH, 2007). In China, the study was conducted from 2002 to 2006 among market vendors in a city on the eastern coast. In China, a market is defined as a place where citizens buy fresh meat, vegetables, fruits, and household goods. A typical market has between 80 and 200 stalls, with a total of 150 to 300 stall owners and employees. Social activities for market vendors usually center within a few blocks of each market, as most vendors live close to work and socialize within that area. Selection was based on the size and geographic location of the markets (i.e., market venues had to be sufficiently distant from each other to prevent contamination across the intervention and control markets). All markets were separated by at least two kilometers unless there was a physical barrier that prevented contact across markets (e.g., one set of markets was closer than 2 km, but a large, fenced, divided highway separated the markets). In total, 40 local food markets were selected as venues from among 95 possible markets (Wu et al., 2007b).

Ethnographic observations, pilot studies of HIV-related risk behaviors, and intervention piloting preceded the mounting of the randomized controlled trial. This article uses data collected from the baseline and 24-month follow-up of the main trial, with supplementary measures on knowledge, attitudes, and behaviors regarding sexuality and behaviors of the POLs in local markets.

Participants and Recruitment

Participants were 18 to 49 years old and were randomly selected from 40 food markets, with a refusal rate of less than 8%. Each participant provided written voluntary informed consent and completed a baseline assessment. The baseline assessment included an in-person interview, a physical exam, collection of blood and urine specimens and vaginal swabs for STD/HIV testing, and syndromic assessment of potential STD. The same assessments were given at baseline and 24 months later.

Originally, 3,912 participants were eligible and included in the intervention outcome analysis. Among them, 601 did not participate in the site-specific assessment at the 24-month follow-up and were excluded from analysis, which resulted in 3,311 subjects in this study (85% follow-up rate).


Food markets were paired based on STD rates at baseline and randomly assigned to either the CPOL intervention or the control condition. Participants from both the control and intervention conditions received HIV/STD educational brochures, HIV/STD testing and treatment, and access to free or affordable condoms at each assessment interview. In addition, the community-level intervention was undertaken in intervention condition venues. Local intervention teams identified, recruited, trained, and engaged POLs from each intervention venue to effectively communicate HIV/STD prevention messages to other market employees, friends, and acquaintances during the course of everyday conversations. POL participants were identified by means of study staff ethnographic observation, nominations made by venue gatekeepers and other key informants, nominations made by other market employees after the baseline interview, and self-nomination.

Approximately 25% of the market vendors who regularly presented in the venue were selected as POLs, and each POL was invited to attend a group training session every week for four weeks, led by four intervention facilitators. The skills training–based sessions taught POLs how to deliver HIV/STD prevention intervention messages to others during daily conversation. POLs agreed to have these conversations with co-workers and friends following the training period. There were support sessions for POLs every other month over the 24-month follow-up period. A total of four group training sessions and 11 reunion sessions were conducted during the 24-month study period.


Assessments were conducted for each participant at baseline and 24 months. A complete assessment included a computer-assisted personal interview (CAPI), which was administered in a private office, face-to-face interview, and a biological specimen lab test for STDs. CAPI was developed to automatically incorporate skip patterns and logistic checks to reduce human error, and was used because of its feasibility with the vendor population (Li et al., 2007). In addition to basic demographic information (age, gender, marital status, education, and self-reported discretionary income per month), three main outcomes were measured and analyzed in this study: behavioral outcome, biological outcome, and attitudinal outcome.

Behavioral Outcome was measured as any unprotected sexual acts with non-spousal partners in the three months prior to the assessment. Participants were asked about their number of sexual partners in the past three months. For those persons with multiple partners (i.e., more than one), we further queried how many times the participant had sex with each non-primary partner and how many times condoms were used during the past three months. If the number of times having sex was greater than the number of times using condoms with non-primary sexual partners, the participant was coded as having unprotected sex within the past three months. With regard to sensitive sexual activities, interviewers clarified answers such as the number of sexual partners and condom use, and the confirmed number was used in the analysis.

Biological Outcome was measured through laboratory tests. Venous blood samples were collected to test for syphilis, herpes simplex virus type 2 (HSV-2), and HIV. Urine specimens were also collected for chlamydia and gonorrhea testing. Trichomonas testing was performed on women only by vaginal swabs. A composite binary variable was constructed to indicate whether or not a new case of at least one of the six STDs was detected at a follow-up visit. All initial tests were performed at the local Hospital for Sexually Transmitted Diseases and the STD laboratory at the National Center for STD and Leprosy Control in Nanjing, China. Quality control retests were performed by the Study Reference Lab at Johns Hopkins University. Participants with non-viral positive STD results received treatment and counseling; HIV-positive participants received counseling and were referred to the local China CDC for CD4 assessment and free antiretroviral treatment; and HSV-2-positive participants received treatment referral and counseling. All treatment procedures followed the study protocol and China CDC guidelines.

Attitudinal Outcome was measured through the following key intervention messages that were assessed at the end of the study: 1) one should have regular annual physical check-ups; 2) one should have open conversations about sexual health with sexual partners; 3) one should use condoms during sexual intercourse to prevent STDs outside or before marriage (including before marriage and extramarital sex); 4) HIV is a serious health problem; 5) STDs are serious health problems in your community. Responses to each question ranged from 1 (strongly agree) to 5 (strongly disagree). Each of these five variables was constructed as 1 if the response was at least agree and 0 otherwise.

Statistical Analysis

Differences in demographic and baseline variables were tested using chi-square (χ2) and t tests for categorical and continuous variables, respectively. Chi-square tests were also used to determine whether differences in each of the five key intervention messages between the control and intervention were statistically significant. Generalized estimating equation (GEE) models (Diggle, Heagerty, Liang, & Zeger, 2001), were used to examine the intervention effect on the primary behavioral and biological outcomes. Covariates included intervention status as a binary variable (0 for control and 1 for intervention), time (baseline and 24-month), and an intervention status-by-time interaction. Models also included participant-level random effects to account for the correlation between repeated measures at baseline and 24 months. Odds ratios and associated 95% confidence intervals were calculated. Similar comparisons were repeated for CPOL versus non-CPOL within the intervention condition for the acceptance of key intervention messages as well as the primary behavioral and biological outcomes. All analyses were performed using SAS version 9.1 (SAS Institute Inc., Cary, NC).


Demographic characteristics of the 1,616 participants in the control condition and 1,695 participants in the intervention condition are presented (Table 1). Across conditions, around 57% were women and over 40% were 31 to 40 years old. The mean age for the control (M = 36, SD = 7.5) and intervention conditions (M = 37, SD = 7.7) were statistically different (P = 0.006), but unlikely to be clinically or epidemiologically relevant. Men were slightly older in the intervention than the control condition (P = 0.003), whereas women were similar in age between the two conditions. Half of study participants had received only a primary school education, and around 38% of both conditions had received a junior high school education.

Table 1
Demographic characteristics (N = 3,311)

Acceptance of protective attitudes was significantly higher in the intervention compared to the control condition at the 24-month follow-up assessment (Table 2). Significantly (P < 0.001) more intervention than control participants endorsed regular physical check-ups, agreed that open conversation about sexual health between sex partners is necessary, consistently used condoms during sexual intercourse outside of or prior to marriage, and believed that HIV and STDs were serious health problems in their community.

Table 2
Agreement with key intervention messages at 24 months (intervention vs. control group)

The CPOL intervention was associated with significantly greater changes among the POL diffusing HIV prevention messages in the intervention communities, compared to the non-POL in the intervention market venues. As shown in Table 3, almost all the POLs and non-POLs in the intervention endorsed regular physical check-ups, open conversation about sexual health with sex partners, and that condoms should be used to prevent STDs during sexual intercourse outside of or prior to marriage. Only around half of POLs agreed that HIV and STDs are serious health problems in their community, although this was significantly higher than among non-POLs.

Table 3
Agreement with Key Intervention Messages at 24-months (CPOL vs. non-CPOL Groups)

The comparison of any unprotected sexual acts in the past three months between baseline and 24-month follow-up for intervention versus control and CPOL versus non-CPOL groups is presented in Table 4. The estimated odds of having any unprotected sex acts outside of marriage decreased significantly from baseline for participants in both the intervention and control conditions (OR = 0.66; 95% CI: 0.50–0.85; P = 0.0015; OR = 0.82, 95% CI: 0.62–1.08, respectively). The decreases were similar across conditions. Within the intervention condition, the estimated odds of having any unprotected sex acts for the POL participants at 24 months was 45% smaller than at baseline (OR = 0.55; 95% CI: 0.31, 0.98; P = 0.0425), in contrast to a 31% decrease among non-POL at the 24-month follow-up assessment (OR = 0.69; 95% CI: 0.51–0.93; P = 0.0132).

Table 4
Comparison of any unprotected sex in the past three months between baseline and 24-month follow-up (intervention vs. control groups; CPOL vs. non-CPOL groups)

Table 5 compares the rates of any new STDs between baseline and the 24-month follow-up for the intervention versus control condition and for the POL versus non-POL participants within the intervention condition. The estimated odds of acquiring any new STDs was 50% less at 24 months versus baseline for participants in both the control (OR = 0.49; 95% CI: 0.41–0.58; P < 0.0001) and the intervention condition (OR = 0.50; 95% CI: 0.42–0.60; P < 0.0001). There was a higher percentage of the non-POL participants who had new STDs at baseline and 24-month follow-up assessment compared to the POL within the intervention condition. However, the estimated odds of getting any new STDs at 24 months was significantly reduced from baseline for both the POL (OR = 0.44; 95% CI: 0.30–0.64; P < 0.001) and the non-POL (OR = 0.52; 95% CI: 0.42–0.64; P < 0.001) participants. No primary behavioral or primary biological outcome was observed.

Table 5
Comparison of any new STDs between baseline and 24-month follow-up (intervention vs. control groups; CPOL vs. Non-CPOL groups)


An evaluation of the CPOL intervention was mounted in 40 market venues with fidelity to the core elements of the original CPOL intervention trials in the United States. The employees in the CPOL intervention venues were significantly more likely to endorse normative beliefs and social support for STD/HIV prevention than were reported among employees in the venues in the control condition. Employees in both conditions, however, decreased their STDs at a very high rate and the rate did not differ significantly across condition. Based on field experience, community members embraced the POL trainings, had very high attendance, acquired status in their local communities for being involved in the project and continued to have conversations over two years regarding safe sex and prevention. Examination of Tables 35 indicate that CPOLs who received the training, compared to non-CPOLs in intervention markets, significantly decreased their STD incidence and improved condom use with non-spousal partners (as well as their attitudes). There were significantly more positive norms towards conversations with sexual partners, annual sexual check-ups, and condom use in the non-POLS and CPOLS compared to the control markets. The differences were clinically relevant as well as significant. These changes document the added benefits of training.

These data present many challenges for public health administrators and researchers. In almost every HIV-related randomized controlled trial to date, there have been significant improvements among participants in the control conditions. Norms regarding the ethical conduct of research require that we provide free educational information and condoms, access to STD evaluations and treatment, individual counseling when an STD was identified, training for pharmacists to deliver efficacious treatments, and that researchers observed interactions in local food markets for an extended period of time. These are significant public health interventions, and it is reasonable to speculate that a good strategy for controlling an epidemic is to study the setting in detail. Access to prevention strategies and repeated self-examination of one’s risky behaviors are significant public health interventions that may go a great distance in reducing HIV and STD risk globally.

National changes in the visibility of HIV prevention, simultaneous with the rapid emergence of more settings for acquiring STD, occurred over the course of this study. For example, the number of newspaper articles on HIV prevention increased dramatically and exposure to mass media expanded substantially (Li et al., 2009). These natural evolutions in the contexts and settings for risk may also decrease our ability to study risk reduction interventions.

The discrepancies between self-report of risky behaviors, rates of STD, and endorsements of beliefs and attitudes regarding STD/HIV are substantial. The source(s) for these discrepancies were not resolved by this study. Pilot studies indicated that the more educated the participant, the greater the probability of distortion of one’s sexual history and the higher the probability of an STD in a person reporting no sexual contact (Detels et al., 2003). Strong cultural beliefs that have been maintained for centuries are difficult to break, even over the course of many conversations.

In order to conduct a randomized controlled trial for a community-level intervention with a 24-month follow-up period, it is necessary to identify a stable community with high risk (Wu et al., 2007a). Those at risk for or having relatively high rates of STDs are usually highly unstable populations (e.g., truck drivers shifting routes often in order to maximize loads and profit; sex workers rotating at three-month intervals around an urban center to provide new opportunities for local customers). There are two stable populations in China who experience generalized epidemics of HIV who may have been studied: injecting drug users and, only recently, homosexual men. There were significant ethical barriers to including such participants in this study: drug users faced the possibility of forced treatment if identified as participants in this trial and homosexual men were highly closeted, particularly in 2003 to 2004 when this trial was being conducted.

Local health authorities in China embraced the approach, as it empowered local POLs to take charge of their health and well-being. Observers from every health province perceived the approach to be quite different from traditional health education programs, in which participants typically took notes during a lecture by a local health expert. The strategies of active participant involvement, creation of games and role-play scenarios to practice new skills, and leaders sharing leadership within intervention sessions were novel experiences in China. Diffusing prevention messages with engaging activities and experiences that foster local community empowerment is a highly acceptable public health practice. However, it did not change STD rates dramatically.


This study was funded by National Institute of Mental Health (NIMH) grant number U10MH61513, a five-country Cooperative Agreement being conducted in China, India, Peru, Russia, and Zimbabwe.


  • Chen XS, Gong XD, Liang GJ, Zhang GC. Epidemiologic trends of sexually transmitted diseases in China. Sex Transm Dis. 2000;27(3):138–142. [PubMed]
  • Chinese Center for Disease Control and Prevention. Office for Disease Control and Emergency Response. Epidemic analysis of statutory STDs in China. Beijing: China CDC; 2008.
  • Detels R, Wu Z, Rotheram MJ, Li L, Guan J, Yin Y, Liang G, et al. Sexually transmitted disease prevalence and characteristics of market vendors in eastern China. Sex Transm Dis. 2003;30(11):803–808. [PubMed]
  • Diggle P, Heagerty P, Liang K-Y, Zeger S. Analysis of Longitudinal Data. Oxford University Press; USA: 2002.
  • Elford J, Bolding G, Sherr L. Peer education has no significant impact on HIV risk behaviours among gay men in London. AIDS. 2001;15(4):535–538. [PubMed]
  • Elford J, Bolding G, Sherr L. Popular opinion leaders in London: a response to Kelly. AIDS Care. 2004;16(2):151–158. [PubMed]
  • Elford J, Sherr L, Bolding G, Serle F, Maguire M. Peer-led HIV prevention among gay men in London: process evaluation. AIDS Care. 2002;14(3):351–360. [PubMed]
  • Flowers P, Hart GJ, Williamson LM, Frankis JS, Der GJ. Does bar-based, peer-led sexual health promotion have a community-level effect amongst gay men in Scotland? Int J STD AIDS. 2002;13(2):102–108. [PubMed]
  • Hart GJ, Williamson LM, Flowers P. Good in parts: the Gay Men’s Task Force in Glasgow--a response to Kelly. AIDS Care. 2004;16(2):159–165. [PubMed]
  • Kelly JA. Popular opinion leaders and HIV prevention peer education: resolving discrepant findings, and implications for the development of effective community programmes. AIDS Care. 2004;16(2):139–150. [PubMed]
  • Kelly JA, Murphy DA, Sikkema KJ, McAuliffe TL, Roffman RA, Solomon LJ, Winett RA, et al. Randomised, controlled, community-level HIV-prevention intervention for sexual-risk behaviour among homosexual men in US cities. Community HIV Prevention Research Collaborative. Lancet. 1997;350(9090):1500–1505. [PubMed]
  • Kelly JA, St Lawrence JS, Diaz YE, Stevenson LY, Hauth AC, Brasfield TL, Kalichman SC, et al. HIV risk behavior reduction following intervention with key opinion leaders of population: An experimental analysis. Am J Public Health. 1991;81(2):168–171. [PubMed]
  • Kelly JA, St Lawrence JS, Stevenson LY, Hauth AC, Kalichman SC, Diaz YE, Brasfield, et al. Community AIDS/HIV risk reduction: the effects of endorsements by popular people in three cities. Am J Public Health. 1992;82(11):1483–1489. [PubMed]
  • Li L, Rotheram-Borus MJ, Batterham P, Lin Y, Wu Z, Detels R, Guan J. Measuring sexual risk using audio computer-assisted self-interviewing (ACASI) versus computer-assisted personal interview (CAPI) in China. International Journal of Sexual Health. 2007;19(1):25–30.
  • Li L, Rotheram-Borus MJ, Lu Y, Wu Z, Lin C, Guan J. Mass media and HIV/AIDS in China. J Health Commun. 2009;14(5):424–438. [PMC free article] [PubMed]
  • Lieber E, Chin D, Li L, Rotheram-Borus MJ, Detels R, Wu Z, et al. Sociocultural contexts and communication about sex in China: Informing HIV/STD prevention programs. AIDS Educ Prev. 2009;21(5):415–429. [PMC free article] [PubMed]
  • NIMH Collaborative HIV/STD Prevention Trial Group. Methodological overview of a five-country community-level HIV/sexually transmitted disease prevention trial. AIDS. 2007;21:S3–S18. doi: 10.1097/1001.aids.0000266453.0000218644.0000266427. [PubMed] [Cross Ref]
  • Rogers E. Diffusion of Innovations. 2. New York, NY: Free Press; 1983.
  • World Health Organization. Global prevalence and incidence of selected curable sexually transmitted infections: Overview and estimates. Geneva: World Health Organization; 2001.
  • World Health Organization. Global strategy for the prevention and control of sexually transmitted infections : 2006 – 2015 : breaking the chain of transmission. Geneva: World Health Organization; 2007.
  • Wu Z, Rotheram-Borus MJ, Detels R, Li L, Guan J, Liang G, et al. Selecting at-risk populations for sexually transmitted disease/HIV intervention studies. AIDS. 2007a;21(Suppl 8):S81–87. [PubMed]
  • Wu Z, Rotheram-Borus MJ, Li L, Guan J, Detels R, Yin Y, et al. Sexually transmitted diseases and risk behaviors among market vendors in China. Sex Transm Dis. 2007b;34(12):1030–1034. [PubMed]
  • Zhang J. Epidemiological situation, risk factors, prevention and control strategies of sexually transmitted diseases in China. Lit Inform Prev Med. 2002;8:637–640.