Implementing an effective intervention to reduce HIV transmission and demonstrating that an intervention is efficacious, are complementary, but very different processes. Strategic selection of a population is a complex process whose details are rarely presented in research reports. The goal of this article is to describe the process of site selection for a randomized controlled trial (RCT) for a community popular opinion leader intervention [1–3] to promote safer-sex social norms, and to prevent transmission of HIV and sexually transmitted diseases (STD) in China.
The number of new STD reported in China increased from 13 in 1977 to 460 000 in 1997, and to over 800 000 in 2000 [4,5]. Simultaneously, HIV has been identified in each of the 31 provinces of China, with current estimates of 700 000 HIV-infected persons . Sexual contact has increased over time to become a major mode of HIV transmission . China is experiencing rapid social change: the social norms regarding sexuality have shifted, placing the population at increased risk for HIV infection [8–12]. This study describes the selection of a population engaging in a substantial number of high-risk sexual acts. During this period of rapid social change and rising HIV/STD rates, The Community Popular Opinion Leader (CPOL) Program was implemented; an intervention that recruits popular opinion leaders who shape social norms, and trains them to influence their community norms.
The study was reviewed and approved by the Ethical Review Committee (i.e. Institutional Review Board - IRB) at the Chinese Academy of Preventive Medicine (re-organized as the Chinese Center for Disease Control and Prevention in 2007) and by IRB at the University of California, Los Angeles.
Understanding prerequisites for intervention research
Populations being selected for conducting scientifically sound intervention studies must meet certain prerequisites. Some criteria were established because of the principles of research design. For example, to have a cost-efficient design and demonstrate risk reduction, it is scientifically efficient to select research participants who are demonstrating risky behaviors at the initial assessment. If participants were not at risk at the time of recruitment, there would be no opportunity for these participants to show improvements in risk reduction as a consequence of the intervention.
To observe change in a valid way, researchers must be able to track a high percentage of the target population over time. A highly mobile population (e.g., sex workers at karaoke bars in China) cannot usually be tracked longitudinally with high follow-up rates unless massive research funds are available for outreach. In addition to the risk and stability of the target population, we must also consider whether the size of the population is large enough to ensure sufficient statistical power for analysis, yet small enough for feasible implementation. Furthermore, factors that may lead to contamination across intervention categories, including non-recognizable boundaries between recruitment sites, and sociodemographic characteristics across and within venues, must be considered in site and population selection.
While some site and subject selection criteria are based on issues critical to study design, criteria related to feasibility of implementation are also key for the population study. For example, if the stakeholders and gatekeepers to a population are not enthusiastic about the intervention, it would be much more difficult, if not impossible, to implement it . Similarly, if the site is remote and can only be accessed with great time, expense, and difficulty, it may not be feasible to conduct a large trial, especially one involving collection and processing of biological specimens. Table 1 summarizes the intervention prerequisites related to population selection.
Refinement of selection criteria
Based on the intervention prerequisites described above, we developed a more refined set of selection criteria for sites and target populations. Our primary selection criteria were: (1) population size in each site (n ≥ 200); (2) stable population (at least 80% of subjects can be followed-up for 2 years); and (3) high baseline STD prevalence or high risk behaviors (in > 15% of community members). Our secondary selection criteria were: (1) an adequate number of study venues; (2) organizational capacity and enthusiasm of gatekeepers and communities; (3) degree and intensity of social networks; (4) likelihood of contamination between potential intervention and control sites; (5) socio-demographic similarities across various sites; and (6) clear boundaries between study sites.
Identification of populations at risk
The existing research and surveillance literature in China has identified five high-risk groups for HIV/STDs: (1) sex workers and their clients [14,15]; (2) injecting drug users and their sexual partners [16,17]; (3) men having sex with men (MSM) ; (4) STD patients and their spouses ; and (5) migrants . Each of these populations was investigated and eventually eliminated for a variety of reasons: sex workers because of their high mobility; injecting drug users because of frequent incarceration and social instability; MSM because they are highly stigmatized, making them sometimes invisible and often inaccessible; persons attending STD clinics because of stigmatization and social instability; and migrants because they are highly mobile.
In order to find potential risk groups for the study, the collaborative investigators conducted a series of interviews and focus groups with key informants that included public health officials, health education specialists, and HIV prevention researchers. Four subgroups were identified as potentially at high risk for HIV/STDs. The following groups were considered likely to meet the study criteria and were examined to confirm the assumption of risk: (1) long-distance truck drivers; (2) self-employed small business workers (e.g., market vendors); (3) construction workers; and (4) factory workers. The four subgroups identified from the key informant interviews were also linked to migrant populations (e.g., construction workers who work in a city for several years at a time or market owners who continue to be considered migrants despite living in a community for between five and fifteen years). Each of these populations has sufficient income to provide access to sex workers. Some key informants did not confirm frequent risk activities among factory workers or self-employed businessman; however rates of STD reported in various provinces suggested emerging risk. The rate of social change was so rapid that it was unclear whether factory workers would now demonstrate risk. Thus, it was necessary to investigate the appropriateness of these populations for the RCT.
Strategies for selecting appropriate target populations
Quantitative and qualitative methods were used during visits to the potential site venues to validate information being given to researchers. National and provincial STD and HIV sentinel statistics were collected and analyzed from the five provinces or autonomous regions under consideration (Anhui, Beijing, Inner Mongolia, Jiangsu, and Fujian).
Qualitative methods, mainly informal open-ended interviews, social mapping, and observations were used to obtain data from long distance truck drivers, villagers, construction workers, small business owners and employees, and factory workers. Teams of investigators visited public health officials in the selected provinces to review available STD and HIV behavioral surveillance data, and to estimate the degree of enthusiasm and support for implementing an STD/HIV–related project. Data from local STD clinics, hospitals, and anti-epidemic stations were collected to provide a general idea of the characteristics of the populations in the region who attended STD clinics (by gender, age, residence, occupation, and cause) in the sites being investigated. Informal discussions were also held with STD health care providers to collect more detailed information, such as type of employment, which was not included in the official data.
A brief, self-administered, anonymous risk assessment was conducted to collect more detailed information on the specific potential target populations to verify site selection criteria, socio-demographic data, and high-risk sexual behaviors. Efforts were also made to understand any factors in the lifestyle of the target population that could affect the conduct of an intervention trial.
Table 2 presents the results of the study for the study criteria in the seven discrete potential target populations for the RCT from five geographic areas, including: (1) village migrants from Fuyang; (2) construction workers in Beijing; (3) truck drivers in Inner Mongolia; (4) construction workers in Jiangsu; (5) factory workers in Jiangsu; (6) factory workers in Fuzhou; and (7) market workers in Fuzhou.
Village migrants from Fuyang, Anhui
China is currently experiencing a huge population migration from villages to cities. Male villagers usually work in cities as construction, factory, or mining workers, and visit female sex workers. Females work as factory workers, but may be sexually exploited and even lured into sex work .
Researchers visited 11 villages near Fuyang, in Anhui Province. Carrying out a study in these rural villages would be possible only if: (1) there is a large number of villages with migrants; (2) a large number of migrants within each village; and (3) a sufficient period during which migrants annually visited their villages to enable the implementation of the CPOL intervention program. There were sufficient villages (> 2500); at least 30–50% of each village were migrants. However, recruiting villages as a strategy to reach migrants was inefficient because migrants did not return home frequently enough to receive the intervention. Therefore, enrolling this population into an RCT was not deemed feasible.
Construction workers in Beijing
There were more than 50 large construction sites in Beijing employing between 100 and 300 workers for projects requiring about two to three years for completion. Because construction workers are away from home for long periods of time and have disposable income, we anticipated that visits to sex workers would be common.
Construction managers, migrant workers, gate guards, taxi drivers, and other people working in and around four construction sites in Beijing were informally interviewed. Construction workers represented a relatively stable population because they returned to Beijing every year, for a period of 6 months to 1 year, to work in the same construction teams, and for the same construction company. Some informants and their friends from the same province had worked in the same construction team for 2 years or more.
Although there were expensive hotels or beauty shops nearby where sex workers were known to frequent, migrant construction workers had fewer opportunities to engage in high-risk sexual behaviors compared to workers in other occupations in Beijing. Many restrictions were placed on construction workers, and they worked seven days per week from 0700 to 1900 h. In addition, construction companies usually withheld 70 to 90% of their salaries each month until work on the building site had been completed. As migrants, they were under the added restriction of having to live and eat on the construction sites because of a curfew imposed on them by the company; workers were strongly advised by their bosses not to leave the construction compounds at any time. Leaving the compound would be possible grounds for their arrest if they lacked a Beijing hukou (local work and residence permit). This would result in their expulsion to their own province. Construction workers therefore entertained themselves on the construction sites, and only occasionally ventured outside the construction compounds to buy essential everyday items. Given these observations suggesting unexpectedly low risk behaviors, the research team did not think construction workers in Beijing suitable for the RCT.
Truck drivers in Inner Mongolia
In Inner Mongolia, long-distance truck drivers transport coal for hundreds of miles. Along the road, there were many restaurants and inns that provided sexual services as well as food. The number of large trucking companies reported to be operating in Inner Mongolia was 10, each having between 150 and 700 truck or lorry drivers.
Truck drivers in three coal mining companies were evaluated for RCT suitability. The mining companies were 100 km apart, and employed between 500 and 700 truck drivers operating in between 300 and 500 vehicles. Coal drivers worked in 12-h shifts. Most drivers were male, 20 to 30 years old, 60% were married, and most were local residents. Drivers returned home to their families every three to four days, and every one to two days in the busy season.
There were indications that truck drivers were a high-risk group. We observed a high number of cheap hotels and inns priced between 20 and 30 RMB per night (about US$ 2.50–3.75) inclusive of one meal along the trucking routes. Informal discussions with hotel owners revealed that it was common for truck drivers to stay overnight at these inns. Data collected from seven local public and private STD clinics indicated that drivers were their main source of patients.
Most truck drivers were relatively stable, having worked for two to three years for the same company. It was difficult for drivers to find the time to be together and to talk to each other, as most were continuously on the road; existing social networks were not clearly identified, suggesting that this population may benefit less from the particular intervention that we were testing (CPOL). In addition, lack of easy accessibility to the National STD Laboratory in Nanjing prevented this site from being selected for the study.
Construction workers in Jiangsu
Although construction sites in Beijing were not deemed viable for the intervention, variations in employment policies across provinces led us to examine alternative construction sites. The team of investigators visited two construction sites in Jiangsu. The first site had between 200 and 500 construction workers; most had worked for the same company over the past three years. Workers on the construction site were temporary migrant workers, mostly male and from Zhejiang, Anhui, Sichuan, and Jiangsu provinces. They lived in eight to nine-person dormitories, ate their meals on the construction sites, worked seven days per week from 0700 until 1900 h, not including overtime, and had a 2100 h curfew imposed on them every night. Only 50% of their salary was released every month, the balance being placed in a voluntary salary savings program. The second construction site we visited had 1700 male migrant construction workers. About 80–90% of them had worked for the same company for two to seven years. Workers lived in small dormitories that held between six and eight persons. They worked seven days per week from 0730 to 1730 h and had no curfew.
We did not see entertainment facilities near either construction site. Occasionally, workers would go to the downtown area, which has a red-light district, to eat and relax. Downtown was about two and 10 km away from the two construction sites, respectively. Travelling this distance required considerable effort by the workers, since there are no public buses and few taxis. Our observations suggested that both lifestyle and fiscal inhibitions lead to high risk sexual behaviors. Therefore, due to the large number of workers that would require follow-up given likely low event rates, and due to the relative lack of research access, this site did not meet the criteria for high risk and was not suitable for our RCT.
Factory workers in Jiangsu
In many middle-sized cities in Jiangsu, there are joint-venture factories. Each factory has many single workers. Managers of these factories employ mostly female migrants and local workers because they are perceived to be better at detail-oriented work compared to their male counterparts. The company provides dormitory-style accommodations for its workers, which creates stable social networks and possibilities for changing social norms in the CPOL.
Three factories were examined as potential sites. Factory workers at a local computer company had typically worked for the company for two years or longer, since the time of recruitment in their home provinces. Migrant workers living in these dormitories, however, reported low-risk sexual behaviors, and access to the factory was severely restricted to outsiders due to company concerns about espionage of the production process. Eighty percent of the dormitory residents were female, unmarried, and aged 20 to 30 years (unmarried young women not engaged in sex work are unlikely to be sexually active in Chinese society).
Managers and workers in the second factory reported that the company had 1500 workers; about 70% were 19–35 year-old males. Seventy percent were unmarried, and most of them lived near or around the factory complex. Some workers who lived in the dormitory had a 2300 h curfew imposed on them. It was observed that there were few recreational or entertainment facilities (i.e., karaoke bars, discos, and saunas, which are known by local health workers to host sex workers) located near the factory. The nearest entertainment facilities were in the downtown area, which was difficult for workers to visit because few buses operated after 6pm. Some workers cohabitated with other workers or had several girlfriends.
The third factory had approximately 900 workers. Most workers (90%) were male migrants from Anhui, Jiangsu, and Sichuan provinces, and were stable, having worked for the factory for three to four years. Workers spent some of their leisure time in the company's recreation rooms, which had a library, gym, and computers. Workers revealed that they had visited hairdressers, karaoke bars, and computer game bars surrounding the factory and the area. Some workers reported having many casual girlfriends, and sex workers were available.
Examination of these three factories demonstrated great variation in size, demographic profile of workers, and level of risk. If these factories had been selected as study sites, the criteria on comparability across randomized venues would not have been met. Therefore, these factories were eliminated from consideration for the RCT.
Factory workers in Fuzhou
Fuzhou, a coastal city in Fujian Province, has experienced rapid economic development, and reports high STD rates. Workers in a shoe factory, workers in a steel factory, and food market vendors in Fuzhou were assessed and compared, using a self-administered questionnaire.
The shoe factory had more than 2000 workers; 80% were migrant women, whereas the steel factory employed 300 workers, most of whom were male (> 50% were married). A self-administrated anonymous risk assessment reported that sexual risk behaviors were relatively low among both the male and female factory workers at both factories, suggesting their unsuitability for our RCT.
Market workers in Fuzhou
Market workers are often stable migrants who have lived in the city for between five and fifteen years. Normally a market has 50 to 150 stalls, with a total of 100 to 300 stall owners and employees.
Social activities for market workers are usually centered within a few blocks of each market, as market workers usually live close by and socialize within that area. Maps have shown that karaoke bars and beauty parlors are often located near the markets. There are about 150 markets in Fuzhou city that, given their number and geographic distribution, makes them promising as study venues. The proximity of Fuzhou to Nanjing's national laboratory for STD testing made specimen transport quite feasible.
Interviews with market managers and a few market employees suggested that this group might have high-risk sexual behaviors. A self-administrated anonymous risk assessment questionnaire was implemented among 180 market employees. Overall and gender-specific sexual risk behaviors were significantly higher among market workers than factory work (Table 3). Sexual risk was measured as having multiple sexual partners in the past year. This target population met all the key design and feasibility criteria, and was selected as the community site with food markets as potential intervention venues.
Targeting HIV preventive interventions to populations at high risk is an international priority. The selection of an appropriate population must be systematically conducted in order to ensure successful implementation of any prevention RCT. This article describes the process of setting and verifying criteria for identifying high-risk groups for a behavioral intervention trial. A systematic examination of the appropriateness of seven potential target populations/settings was implemented in China among truck drivers, factory workers, migrant villagers, market workers, and construction workers in five regions.
Market workers in Fuzhou were identified as the optimal target population. There were several key factors: (1) self-reported risk among market workers; (2) concurrent evidence of risk in official statistics of STDs for Fuzhou City; (3) a relatively stable population with an average period of current residency from two to five years; (4) sufficient size and number of markets to support a randomized trial; (5) evidence of social networking and communal activities among the market workers; (6) clear boundaries and geographic distances between the markets to render contamination from intervention to control markets unlikely; (7) supportive governmental and nongovernmental gatekeepers and stakeholders with demonstrated ability to assist the project; and (8) feasible procedures for transporting and processing laboratory samples and data within a 24-h period. Other potential populations/settings were rejected because they failed to meet one or more of the criteria established for this process. Ethnographic research and a pre-baseline survey of STDs verified that food market vendors were appropriate for the intervention study .
It is important that research communities and journals begin to document the process of community selection in order to advance effective, cost-efficient interventions. Selecting a target population is a time-consuming and rigorous process. The population criteria used for this study are applicable to other intervention trials, because the criteria were based on common prerequisites for intervention studies. Although many successful intervention trials have typically implemented the procedures described herein, procedures were often not documented and shared with other investigators. Systematically and routinely documenting successful procedures will save time and increase the number of successfully implemented trials in the future. Thus, we hope the documentation of this process will assist future investigators in successfully implementing intervention trials.
Sponsorship: This study was funded by NIH/NIMH grant number 1U10MH61513, which is a 5-country Cooperative Agreement being conducted in China, India, Peru, Russia, and Zimbabwe. Each site has selected a different venue and population with which to implement the prevention program entitled Community Public Opinion Leader (C-POL) Intervention. The Steering Committee for this trial is Carlos Caceres, M.D. (Peru); David Celentano, Sc.D. (U.S./India); Thomas Coates, Ph.D. (U.S./Peru); Tyler Hartwell, Ph.D. (U.S./RTI); Danuta Kasprzyk, Ph.D. (U.S./Zimbabwe); Willo Pequegnat (NIMH); Mary Jane Rotheram-Borus, Ph.D. (U.S./China); Suniti Solomon, M.D. (India); Godfrey Woelk, Ph.D. (Zimbabwe); and Zunyou Wu, M.D. (China). This article is based on a China country specific pre-ethnographic study. This study also received support from the UCLA Center for AIDS Research (CFAR) grant #AI28697.
1. Kelly JA, St Lawrence JS, Diaz YE, Stevenson LY, Hauth AC, Brasfield TL, et al
risk behavior reduction following intervention with key opinion leaders of population: an experimental analysis. Am J Public Health 1991; 81:168–171.
2. Kelly JA, Murphy DA, Sikkema KJ, McAuliffe TL, Roffman RA, Solomon LJ, 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; 350:1500–1505.
3. Sikkema KJ, Kelly JA, Winett RA, Solomon LJ, Cargill VA, Roffman RA, et al
. Outcomes of a randomized community-level HIV
prevention intervention for women living in 18 low-income housing developments. Am J Public Health 2000; 90:57–63.
4. Zhu H. AIDS report from China. China Population Today 1998; 15:18.
5. Chen XS, Gong XD, Liang GJ, Zhang GC. Epidemiologic trends of sexually transmitted diseases in China. Sex Transm Dis 2000; 27:138–142.
6. State Council AIDS Working Committee Office and U.N. Theme Group on HIV
AIDS in China. A joint assessment of HIV
/AIDS prevention, treatment and care in China. Beijing: China Ministry of Health, 2007.
7. Wu Z, Sullivan SG, Wang Y, Rotheram-Borus MJ, Detels R. Evolution of China's response to HIV
/AIDS. Lancet 2007; 369:679–690.
8. Jones RH. Mediated action and sexual risk: searching for “culture” in discourses of homosexuality and AIDS prevention in China. Cult Health Sex 1999; 1:161–180.
9. Gil VE. China update: HIV
increasing. AIDS Soc 1993; 4:1–6.
10. Gil VE, Wang MS, Anderson AF, Lin GM, Wu ZO. Prostitutes, prostitution and STD
transmission in mainland China. Soc Sci Med 1996; 42:141–152.
11. Liu H, Xie J, Yu W, Song W, Gao Z, Ma Z, et al
. A study of sexual behavior among rural residents of China. J Acquir Immune Defic Syndr 1998; 19:80–88.
12. Portsmouth S, Stebbing J, Keyi X, Jianping Z, Guohua P. HIV
and AIDS in the People's Republic of China: a collaborative review. Int J STDs AIDS 2003; 14:757–761.
13. Rogers SJ, Doino-Ingersoll J, Hayes-Cozier R, Weisfuse I. Evaluating HIV
prevention community planning. Eval Health Prof 1996; 19:465–487.
14. Rogers SJ, Ying L, Xin YT, Fung K, Kaufman J. Reaching and identifying the STD
risk of sex workers in Beijing. AIDS Educ Prev 2002; 14:217–227.
15. van den Hoek A, Yuliang F, Dukers NH, Zhiheng C, Jiangting F, Lina Z, et al
. High prevalence of syphilis and other sexually transmitted diseases among sex workers in China: potential for fast spread of HIV
. AIDS 2001; 15:753–759.
16. Zheng X, Tian C, Choi KH, Zhang J, Cheng H, Yang X, et al
. Injecting drug use and HIV
infection in southwest China. AIDS 1994; 8:1141–1147.
17. Nguyen AT, Nguyen TH, Pham KC, Le TG, Bui DT, Hoang TL, et al
. Intravenous drug use among street-based sex workers: a high-risk behavior for HIV
transmission. Sex Transm Dis 2004; 31:15–19.
18. Choi KH, Gibson DR, Han L, Guo Y. High levels of unprotected sex with men and women among men who have sex with men: a potential bridge of HIV
transmission in Beijing, China. AIDS Educ Prev 2004; 16:19–30.
19. Liu H, Detels R, Yin Y, Li X, Visscher B. Do STD
clinics correctly diagnose STDs? An assessment of STD
management in Hefei, China. Int J STD
AIDS 2003; 14:665–671.
20. Anderson AF, Qingsi Z, Hua X, Jianfeng B. China's floating population and the potential for HIV
transmission: a social-behavioral perspective. AIDS Care 2003; 15:177–185.
21. NIMH Collaborative HIV
Prevention Trial Group. Selection of populations represented in the NIMH Collaborative HIV/STD Prevention Trial
. AIDS 2007; 21