HIV/Sexual Risk Reduction Interventions in China: A Meta-Analysis

Abstract

The current study was a meta-analysis of the efficacy of educational, psychosocial or behavioral interventions that were conducted in China to promote HIV/sexual risk reduction. A total of 26 intervention studies qualified for the meta-analysis. Outcome variables for which effect sizes were calculated included condom use with different types of sexual partners, HIV/AIDS knowledge, condom use knowledge, intentions of condom use, condom use self-efficacy, and others. Mean weighted effect sizes were calculated for each outcome measure across reviewed studies; effect size for each outcome measure was weighted by their inverse variance; fixed effects and random effects meta-analytic procedures were used. The Q statistic was used to examine whether the effect sizes were homogeneous in nature and moderating analysis (i.e., the Q_b statistic) was used to compare the effect sizes of intervention studies that were different in a number of categorical variables. The reviewed interventions were successful in improving HIV knowledge (d=0.706), condom use knowledge (d=0.620), attitudes toward people living with HIV/AIDS (PLWHA; d=0.625) and in increasing condom use with regular partners (d=0.477), condom use with casual partners (d=0.444), general condom use (d=0.408), and condom use self-efficacy (d=0.584) among target audiences. In addition, moderating analyses on three most examined variables, including HIV knowledge, condom use, and attitudes toward PLWHA, demonstrated that interventions that reported the conduction of formative research and process evaluation, that were peer-led, and that included only one follow-up were more likely to report a positive impact on condom use behavior among target audiences (p<0.001), HIV knowledge (p<0.001), or attitudes toward PLWHA (p<0.001).

Introduction

A n estimated 740,000 people were living with HIV in China as of 2009, and 48,000 new cases occurred in 2009.¹ Although injection drug use and commercial plasma collection were the major routes of infection in China, sexual transmission has recently become a primary mode of HIV transmission, and increased the spread of HIV infection from most-at-risk populations to the general population.² Among the estimated 740,000 people who are living with HIV in 2009, 44.3% were infected by heterosexual transmission and 14.7% by homosexual transmission; among the 48,000 new cases in 2009, 42.2% were attributable to heterosexual contact and 32.5% to homosexual transmission.¹ In addition, the high prevalence of unprotected bisexual behavior among Chinese men who have sex with men (MSM) have exerted a strong impact on the HIV epidemic.^3
–5 Due to traditional Chinese culture and social pressure, many Chinese men who have sex with men (MSM) hide their sexual orientation and get married/have sex with women. Therefore, bisexually active men play a significant bridging role in transmitting HIV/sexually transmitted diseases (STDs) to the general population through heterosexual activities.^3
–5

A number of prevention intervention programs have been conducted to reduce HIV-related sexual risk-taking behaviors among various Chinese populations. Several narrative reviews have compared and integrated the components of the intervention studies, including location, sampling, design, intervention strategies, and outcomes. But we know little about the efficacy of these interventions in reducing HIV-related sexual risk-taking behavior. No study has quantitatively synthesized the existing HIV intervention literature and examined whether these behavioral interventions are effective (efficacious) in reducing sexual risk behavior among Chinese populations. “Future preventive interventions, however, will only have the greatest chance of being effective if lessons learned from past intervention efforts are seriously considered.”^6(p335) Meta-analysis has been introduced to examine the efficacy of intervention studies in behavioral, health, and social sciences.⁷ Meta-analysis, which is a statistical technique similar to research synthesis, provides more objective appraisals of the evidence than conventional narrative reviews due to four primary advantages, including using structured discipline on literature searches, representing key findings in a manner that focuses on effect sizes rather than on qualitative summaries or statistical significance, coding study characteristics systematically, and handling large number of study studies.⁷

In the current study we sought to quantitatively appraise the overall efficacy of interventions on reducing HIV/sexual risks in China and to conduct a meta-analysis to examine efficacy of the HIV interventions in changing HIV-related knowledge, attitudes, and behaviors among various Chinese populations. We also sought to identify intervention and study characteristics that were associated with the efficacy of HIV behavioral prevention interventions in China.

Methods

Sources of literature

Comprehensive literature searches were conducted to obtain relevant studies. Studies were retrieved from (1) computerized databases, including PsycINFO, MEDLINE, and Academic Search Complete. Dissertation Abstracts, using combinations of the keywords HIV, AIDS, sexual behavior, intervention, campaign, HIV knowledge, safer sex, safer sexual behavior, condom use, HIV testing, China; (2) reference sections of relevant manuscripts; (3) the reference list of the review study conducted by Hong and Li.² To exhaust potential intervention studies, this process was iterated between August 2010 when we started the study and upon completion of the initial coding in October 2011 until no new articles were identified.

Inclusion/exclusion criteria

A study staff conducted an initial screening based on title and abstract review. Remaining citations were then screened by two independent coders based on a thorough reading of the full-text articles. Studies were included in this meta-analysis if they met the following criteria: (1) studies that were conducted in China to quantitatively examine any educational, psychosocial or behavioral interventions promoting HIV/sexual risk reduction; (2) studies that were published in peer-reviewed English journals prior to October 2011; (3) studies that provided information needed to calculate effect sizes; (4) studies that employed an experimental (e.g., randomized controlled trial) or quasi-experimental design with an appropriate control group; (5) studies that reported an empirical outcome evaluation that assessed the impact of the campaign/intervention on the following educational, psychological, and behavioral outcomes. The educational outcomes included HIV knowledge and condom knowledge; psychological outcomes were condom use intentions, condom use self-efficacy, attitudes toward HIV, and attitudes toward people living with HIV/AIDS (PLWHA); the behavioral outcomes included condom use, and HIV testing. Some studies also assessed rates of HIV infection. Therefore, HIV infection was included as one outcome variable.

Those interventions focused on only occupational prevention of blood-borne pathogens, antiretroviral treatment programs, testing for only other STDs, or contraceptive-only interventions were excluded. Also, since the focus of this review was on primary HIV prevention interventions, secondary interventions among HIV-infected individuals were excluded. In addition, interventions targeting solely injection drug use or solely on reducing HIV-related stigma, or solely improving attitudes toward people living with HIV/AIDS (PLWHA) were excluded. Studies were excluded if they used a within-subjects design with no control/comparison condition. Studies that were written and/or published in Chinese were excluded from the current review because they were difficult for readers outside of China to access.⁸

Article coding

Two independent coders coded all the articles in the final data set on nine dimensions of interest, including location, intervention year, sample, target population, formative research, use of theory, intervention description, process evaluation, and outcomes. Methodological quality (MQ) for each study was also coded on 12 items derived from Miller and Wilbourne's⁹ Methodological Quality Rating Scale (MQRS) and Kennedy and collegues'¹⁰ study. Sample items include “prospective cohort,” “random assignment of participants to the intervention,” “follow-up rate of 80% or more,” “control/comparison groups equivalent on sociodemographic measures,” “control/comparison groups equivalent at baseline on outcome measures,” “outcome data collected by personnel blind to treatment condition.” One point was awarded for each of these 12 items. Holsti's¹¹ formula was used for calculating intercoder reliability. The intercoder reliability ranged from 0.80 to 1.0. The two coders discussed disagreements with the first author and arrived at agreement on all discrepancies.

Publication bias

Publication bias was first examined by generating funnel plots of the weighted mean effect size by standard error.¹² Funnel plot asymmetry was tested by rank correlation test to examine the correlations between standardized effect size estimates and their variances¹³ and by linear regression to regress standardized effect size estimates against their precisions that was defined as the inverse of the standard error).¹⁴

Effect size calculation

The standardized mean difference (d), which is defined as the mean differences in groups divided by the pooled standard deviation, was employed as the effect size indicator. However, this effect size index suffers from a slight upward small sample size bias. Therefore, effect sizes were corrected for such a bias using recommended formulas.⁷ Available data (e.g., F or t values, summary statistics) were used to calculate effect sizes. When odds ratios were reported, they were transformed to d using the natural logarithm; and when results were reported in percentages, they were converted to odds ratios first and then to d following appropriate formulas.¹⁵

Multiple effect sizes were calculated for individual studies when more than one educational, psychosocial, or behavioral outcomes were reported. These effect sizes were coded and presented separately. If a study contained multiple comparison conditions, the average of the effect sizes was used in analysis. If a study targeted different populations and assessed a same outcome for these different target audiences, the effect sizes were also coded and presented separately. If a study contained more than one follow-up time points of data, the data from the longest follow-up were used to calculate the effect size. In addition, effect sizes with a positive sign indicated that the intervention condition outperformed the control or comparison condition and effect sizes with a negative sign indicated that the reverse.

In the studies that examined condom use, some included general condom use (i.e., without specifying the types of sexual partners), some others asked questions in reference to specific types of partners (e.g., regular, casual). Accordingly, the current study coded the studies that examined condom use into “condom use with regular partners,” “condom use with casual partners,” and “general condom use.” The final set of outcome variables for which effect sizes were calculated included condom use with regular partners, condom use with casual partners, general condom use, HIV/AIDS knowledge, condom use knowledge, attitudes toward PLWHA, intentions of condom use, condom use self-efficacy, attitudes toward HIV, and HIV infection.

Meta-analysis

Effect size for each outcome measure was weighted by its inverse variance.⁷ Mean weighted effect sizes were calculated for each outcome measure across reviewed studies and the Q statistic was used to examine whether the effect sizes were homogeneous in nature. Random effects procedure was then employed based on the results of heterogeneity analysis. The Q_b statistic was employed to compare the effect sizes of intervention studies that were different by a number of design features including inclusion of formative research (yes/no), use of a theory (yes/no), inclusion of process evaluation (yes/no), type of intervention facilitators (peer-led versus professional-led), and number of postintervention follow-up surveys (one versus two or more). A significant Q_b statistic indicates that the design feature can significantly explain the variance and difference in effect sizes.⁷ SPSS Macros for Meta-Analysis (SPSS Inc., Chicago, IL) were used to conduct all data analyses.⁷

Results

Characteristics of studies

Initial searches resulted in 242 abstracts that were examined more closely for their potential relevance to this review. A total of 70 articles that had the potential to be included in the review were retrieved in full text and examined. Based on our exclusion criteria, a number of articles were subsequently excluded, including 16 studies that used pretest–posttest one group design; 11 (15.7%) studies that had English abstracts but were published in Chinese journals; 5 studies (7.1%) that focused on occupational prevention of blood-borne pathogens, biologic HIV prevention approaches and antiretroviral treatment programs; 6 studies (8.6%) that focused on conceptual development and methodological overview rather than on outcome evaluations of HIV interventions; 1 study (1.4%) that focused only on drug use practices; 3 studies (4.3%) that focused on HIV-infected heterosexual men; 2 studies (2.9%) that focused solely on reducing HIV-related stigma and improving attitudes toward PLWHA. Consequently, a total of 26 articles were included in the final data set for review. On average, studies received 7.3 of 12 possible points for study design and methodological quality. Methodological quality scores for each study are shown in Table 1.

Table 1.

Characteristics of Studies and Effect Sizes of Outcome Measures

Authors (year of publication)	Location	Year of study	Sample size	Age of participants (mean unless specified otherwise)	Target audience	Formative research	Guiding theory	Intervention description	Process evaluation	Outcome & effect size	MQ
Cai et al. (2008)¹⁶	Shanghai	2004–2006	1729	17.22	Senior high school students	None	None	Peer-led and school-based intervention; 1-month and 12-month follow-up	None	HIV Knowledge d=0.45 Intention to use condoms d=0.13	10
Cheng et al. (2008)³⁵	Shanghai County, Henan Province	2003	1174	16.22	Senior high school students	None	None	Teachers-led school-based HIV/AIDS education through life-planning skills training program; Participatory training methods (group discussion, role playing, brainstorm, case study, games, etc.) which combined information education with effective skills building were used; 3-month follow-up	None	HIV/AIDS knowledge d=0.55 Attitude toward PLWHA d=1.34 Condom use self-efficacy d=0.48 Communications on HIV/AIDS d=0.51	6
Gao and Wang 2007²¹	Chendu, Sichuan	2005	160	25.88	Gays, money boy-commercial sex workers	In-depth case studies, focus groups, special workshops	HBM, SCT, TRA, Diffusion model, subcultural model, leadership-focused model, communication model, empowerment model, socio-political model, social structural model, social action model, cultural approach	Peer-led and gay bar-based participatory entertainment-education (PEE) strategy (e.g., gay bar-based melodrama series on condom use) were developed; Three small media materials such as logo, coaster, making-friend card were provided; Outdoor entertainment games (e.g., sports, play cards and mahjong) were organized. 5-month follow-up	None	Safe sex knowledge d=0.53 Attitudes toward HIV d=0.53 Condom use with casual sexual partners d=0.62 Condom use with regular sexual partners: d=0.63	5
Huang et al. (2008)²⁵	San-ming, Fujian	2006	3068	17.4	Senior high school students	None	None	Peer-led and school-based, the intervention was integrated into the regular school health education schedule, one session per week for 4 weeks, 3-month follow-up	80% reported to be actively involved in the intervention sessions	HIV/AIDS-related knowledge d=0.86 Attitude toward PLWHA d=0.26	6
Jiang et al. (2010)³⁹	Fujian, Xinjiang, Guangxi, Shanxi	1999–2007	735	Unknown	General population	None	None	Field workers-led and community-wide implementation of HIV/AIDS/STI interventions; promotion of political commitment and social mobilization; policy formulation; STI/HIV/AIDS institution capacity building and staff training; integrated efforts across different sectors. 8-year follow-up	Rapid cross-internal evaluation, midterm external evaluation	HIV knowledge d=0.44 Condom use during the last intercourse d=0.19	3
Lau et al. (2007)²²	Dazhou, Sichuan Province	2003–2005	400	Unknown	Establishment-based and street-based FSW	Pilot tests	None	Peer-led community-based Intervention activities include mobilization of community resources, seminars and training for keepers of entertainment establishments and FSWs, publicity events and advertisements via different mass media, outreach education by peer educators, voluntary counseling and testing services, free STD checkup and support services, social marketing of condoms. 12- and 24-month follow-up	None	HIV knowledge d=1.37 Consistent condom use with clients d=0.79 Consistent condom use with regular partners d=0.69	7
Lau et al. (2008)³²	Hong Kong		280	53.6% were 21–30 years old.	MSM	None	None	Internet-based behavioral intervention; Participants received periodic emails that contained the messagesMonitoring of behaviors with interactive feedback and online peer counseling6-month follow up	None	HIV knowledge condom useEfficacy of the intervention was not supported as no statistically significant between-group differences in HIV risk behaviors and perceptions were observed d=0	7
Lau et al. (2010)³³	Hong Kong	January 2005–March 2006	301	≥25	Male Cross-border truck drivers	None	None	Experienced field workers-led; Voluntary counseling and testing (VCT) plus information dissemination approach (VCT-ID) vs. the information dissemination only approach (ID);Educational pamphlets were disseminated.1- and 2-month follow-up	None	Consistent condom use with nonregular partner: d=0.33Consistent use with regular partner d=0.11HIV/STD-related knowledge d=0.66	9
Li et al. (2010)¹⁷	Shanghai	March–June 2006	2,237	14.2	Children of migrant workers	None	None	Peer-led intervention integrated into regular school health education schedule in the intervention group; group discussion, games, case study and brainstorming3-month follow-up	Five training sessions and The median duration of interventions was 4 hours	HIV knowledge: d=1.24Attitude toward PLWHA d=0.15Protection self-efficacy d=0.17	9
Li et al. (2008)³⁰	Nanjing, Jiangsu Province	2001	358	20	College students		Protection motivation theory	Trained graduate students and faculty members delivered the intervention, a cultural adaption of the PMT-based “Focus on Kids” (FOK) Curriculum,Intervention games and group activities lasted for five afternoons with about 2.5 hours in each afternoon.6-month follow-up	None	HIV/STD knowledge d=0.50Condom use knowledge d=0.29Condom use self-efficacy: d=0.80Intentions of condom used d=0.08	7
Li et al. (2006)³⁴	Nanning, Guangxi Province	2004	278	23.8	FSW		Social cognitive theory and the theory of reasoned action	Clinicians or health workers delivered the individual-oriented voluntary counseling and testing (VCT) intervention program that was developed and evaluated by the CDC's Project RESPECT.Interactive approach, discussion and questions, didactic messages, modeling and condom use skills practice, STD testing, distributing free condoms6-month follow-up	None	HIV/STD knowledge d=0.40Condom use self-efficacy d=0Consistent condom use the last three episodes d=0.44STD infection: d=0.29	7
Li et al. (2011)³¹	Nanjing		304	16.3	High school students	Qualitative and quantitative data gathering	Protection motivation theory	Graduate students and faculty members from a university in Nanjing delivered a culturally adapted school-based FOK HIV prevention program to high school students.10 students in each group participated in games and role-play activities.6-month follow-up	None	HIV knowledge d=0.38	8
Liao, et al. Zeng and Liao (2010)³⁷	Hainan	2006–2008	9336	unknown	Rural students of Primary schools	None	None	An experimental curriculum was delivered by trained teachers to 5th graders. The curriculum included four lessons: adolescent development, HIV transmission and prevention, drug prevention, risk behavior rejection.Participatory activities such as brain storming, role playing, group discussion, case study and games were used.16-month follow-up	None	HIV knowledge d=0.26Attitudes toward HIV d=0.07	7
Lin et al. (2010)³⁸	Beijing	Unknown	300	22.37	Female migrants	None	Protection motivation theory	Psychology faculty members and graduate students in a local university and outreach workers in a local women's association delivered a six-session (each approximately 1.5 hours) risk reduction intervention. Activities include games, group discussions, videos, rehearsals, role-plays, brainstorming, homework assignments.4-month follow-up	None	HIV Knowledge d=1.57Condom use self-efficacy d=0.97Condom use response efficacy d=1.11Intention to use condom d=0.69Consistent condom use d=0.07	8
Lou et al. (2006)¹⁸	Shanghai	March 2003–December 2003	1337	Unknown, only age ranges were reported	High school and college students	Focus group	None	Messages on sexual and reproductive health knowledge, service information, counseling, and discussion were delivered through a special website that included web pages, online videos, Bulletin Board System, email counseling and expert mailbox.10-month follow-up	Students' participation and feedback related to the intervention was evaluated.	Condom knowledge d=0.41HIV/AIDS Knowledge d=0.36	8
NIMH Collaborative HIV/STD Prevention Trial Group (2010)²⁶	Fuzhou	2002–2007	2717	Unknown	Vendor markets	Preliminary ethnographic studies	Diffusion of innovation theory	Intervention and comparison conditions received HIV/STD prevention brochures and pamphlets, educational materials, information about HIV/STD counseling, testing and treatment.But the Intervention condition received C-POL intervention and the community-level intervention.C-POLs also wore logos on T-shirts, hats or other apparel24-month follow-up	Quality control procedures to ensure consistency and fidelity over time	Unprotected sex in the past 3 months d=− 0.035Incidence of STDs (including HIV) d=0.02	11
Rotheram-Borus et al. (2011)⁴¹	A large eastern coastal city	2002–2006	3912	36	Migrant food market vendors	None	Diffusion of innovation theory	Community popular opinion leader (C-PDL)-led intervention; Pilot C-PDL training and reunion sessions taught skills in delivering and diffusing theory based HIV/STI prevention messages to friends and acquaintances during everyday conversations12- and 24-month follow-up	None	Unprotected extramarital sex among those who experienced one or more STIs at baseline d=0.13STIs Incidence d=0.22	11
Tian et al., 2007²⁸	Mouding and Dayao Counties in Yunan Province	January 2003–January 2004	2736	30 for villagers14.23 for students	Villagers and students in rural counties	None	None	Health workers and teachers delivered a Web-based intervention that featured the subjects of sexually transmitted infections and AIDS.A total of 52 different programs on the website changed weekly.Randomized selection of counties for intervention and control groups.12-month follow-up	None	HIV/AIDS knowledgeVillagers: d=0.22Students: d=0.26	5
Wang and Keats (2005)²³	Sichuan		450	26.37	Male participants in Yi, Tibetan, and Han cultural groups	In-depth case studies, focus groups, and special workshops, pilot testing	Behavior change theories in HIV (e.g., TRA, SCT, etc.)	Peer-led diffusion of role model storiesWorkshops, group sessions, the oral role model stories strategy, peer outreach,Workshops, group sessions, the oral role model stories strategy, peer outreach,3-month follow-up	None	Correct knowledge about safe/unsafe sex: d=1.26Attitudes toward HIV d=0.60Condom use with casual partners: d=1.02Condom use with regular partners: d=0.69	5
Wang et al (2005)¹⁹	Suburban Shanghai	May 2000–January 2002	2227	18.5	Unmarried youth aged 15–24	None	None	Family planning workers, counselor, and project staff delivered community-based comprehensive sex education distribution of educational reading materials (brochures, pamphlets, books), screening of educational videos, lectures, peer group discussions, and provision of reproductive health services and counseling.20-month follow-up	On average, participants in the intervention group reported receiving 8.7 brochures or books; 79% participants viewed educational videos (average, 1.7 videos per Participant); 25% attended group discussions (average, 1.4); and 21% attended the only lecture. Only 11% of participants utilized the counseling services; the average number of counseling sessions per participant was 2.6.	Used a condom every time d=0.63	9
Wang et al. (2006)²⁰	Suburban Shanghai	May 2000–January 2002	2,227	18.5	Unmarried youth aged 15–24	None	None	Family planning workers, counselor, and project staff delivered community-based comprehensive sex educationEducational reading materials, educational videos, lectures, peer group discussions, provision of health services and counseling20-month follow-up	Intervention group reported receiving 8.7 brochures or books.79% participants viewed educational videos (average, 1.7 videos per participant)25% attended group discussions (average, 1.4); and 21% attended the only lecture.Only 11% of participants utilized the counseling services; the average number of counseling sessions per participant was 2.6.	HIV knowledge Males: d=0.20 Females: d=0.22	9
Wu et al. (2002)³⁶	Fuyang, Anhui	1997–1999	3354	Unknown	Both health workers and villagers	A pilot survey to provide information to which base the first workshop.Pilot testing of educational materials	Diffusion of innovation	A training of trainers strategy: health workers received initial workshops and then diffused intervention programs through secondary and tertiary workshops at county and township levels to health workers and villagers and the community.Printed training manuals, flyers, newsletters, billboards, blackboards, radio, TV and loudspeakers were used to distribute intervention program. Role-playing was conducted in counseling and videos were showed after discussions. Cases studies were used.18-month follow-up	94% of the participants participated in workshops and 92% received training manuals.	Health workers knowledge: d=1.41Attitudes toward people with HIV/AIDS: d=1.51Condom use during most recent sexual intercourse d=0.67Villagers HIV Knowledge d=1.08Attitudes toward people with HIV/AIDS: d=1.60Condom Use during most recent sexual intercourse d=1.12	5
Wu et al. (2007)⁴⁰	Guangdong and Guangxi	2002–2003	852	≥11	IDUs	In-depth interviews, group discussions and observations were used to collect information about the target population including drug use information, purchasing and sharing of needles, sexual behavior, and awareness of AIDS-related knowledge	None	Health workers-led community-based interventions that involve handing out delivered educational pamphlets, displaying educational posters, delivering lessons about drug abuse and HIV/AIDS, viewing a photo exhibition and educational videos.12-month follow-up	69.5% in Guangdong and 46.8% in Guangxi received needles.Educational pamphlets were received by 60.8% and 59.4% of participants respectively, and 70.9 and 69.6% had seen educational posters, respectively. More than half of the participants in Guangxi had participated in a face-to-face counseling session with a health worker (52.7%) or a peer educator (55.2%), and in Guangdong these figures were 61.7 and 66.0%, respectively	Consistent condom use in Guangdong d=0.79Incidence of HIV d=0.31	8
Wu et al. (2010)²⁹	Yunnan	January 2006–January 2007	2624	Unknown	Dai farmers	None	No behavior change theory, but based on Participatory action research (PAR)	Community-based intervention provided training sessions to health educators (voluntary monks and women). Training modes included lecturing, video shows, game-playing, panel discussion.Monks then put up posters and distributed propaganda materials. Also they preached among worshipping villagers.Women team included posters, propaganda materials, public speeches, song and dance performances, and true stories being incorporated into a popular local drama “Zanha.”12-month follow-up	Questionnaire survey to pinpoint problems after each training session	Condom use with regular partners d=0.18Condom use with nonregular partners d=0.40Intention to use condom d=0.14Attitudes toward PLWHA d=0.12	8
Ye et al. (2009)²⁷	Fujian	2006–2008	893	17.5	Senior high school students	None	None	Peer-led HIV/AIDS education for intervention group and teacher-led health education curriculum for control groupOne-month and 24-month follow-up	None	HIV knowledge d=0.34Attitudes toward PLWHA d=0.16Condom useES=0.15	7
Zhou et al. (2009)²⁴	Leshan and Deyang in Sichuan	2002–2005	2066	29.49	IDUs	None	None	Peer-led community-based information, education and communication (IEC) activities, volunteer counseling and testing (VCT), condom distribution, needle-syringe program (NSP), methadone maintenance therapy (MMT), etc.12- and 24-month follow-up	≤30% in 2003 but ≥80% in 2005 in intervention city received intervention education and services	HIV awareness increase d=0.25Condom use with casual partners d=0.26Condom use with CSWs/clients d=0.08	5

MQ, methodological quality; HBM, health belief model; SCT, social cognitive theory; TRA, theory of reasoned action; PLWHA, people who live with HIV/AIDS; FSW, female sex workers; MSM, men who who sex with men; STI, sexually transmitted infection; STD, sexually transmitted disease; PMT, protection motivation theory; IDUs, injection drug users; CSWs, commercial sex workers.

The 26 studies had a cumulative sample size of N=46,055 (median N per study=1337). Most studies (κ=17, 65.4%) reported mean age of the participants. On average, the study samples of the 17 studies that reported mean age were 22.21 years old (standard deviation [SD]=6.26). Most of the intervention studies were published after the year of 2001 (approximately 50% of publications occurred after 2008). Five studies were conducted in Shanghai,^16

–20 4 studies in Sichuan Province,^21

–24 3 in Fujian Province,^25
–27 2 in Yunnan Province,^28–29 Nanjing,^30,31 and Hong Kong,^32,33 respectively. Other study locations include Guangxi,³⁴ Henan,³⁵ Anhui,³⁶ Hainan,³⁷ and Beijing,³⁸ with only one study conducted in each of these locations. Two studies were conducted in multiple provinces.^39,40 And one study used “a large eastern city” to label the location.⁴¹ The target population for these studies included high school students (κ=5),^{16,25,27,31,35} female sex workers (κ=2),^22,34 market workers (κ=2),^26,41 farmers/villagers (κ=1),²⁹ both villagers and students (κ=1),²⁸ unmarried youth (κ=2),^19,20 injection drug users (κ=2),^24,40 men who have sex with men (κ=2),^21,32 college students (κ=1),³⁰ truck drivers (κ=1),³³ children of migrant workers (κ=1),¹⁷ ethnic cultural groups (κ=1),²³ general population (κ=1),³⁹ female migrants (κ=1),³⁸ rural primary school students (κ=1),³⁷ both high school and college students (κ=1),¹⁸ both health workers and villagers (κ=10).³⁶ In addition, the 26 reviewed studies conducted an average of 1.27 (SD=0.45) postintervention follow-up surveys (76.9% one follow-up and 23.1% two follow-ups).

In the current review, only 35% studies (κ=9) reported using a theory or a conceptual model as a guiding framework for the intervention. The theories and models reported in these studies included the health belief model,⁴² the theory of reasoned action,⁴³ social cognitive theory,⁴⁴ diffusion model,⁴⁵ protection motivation theory,⁴⁶ subcultural model,⁴⁷ leadership-focused model,⁴⁸ communication model,⁴⁹ empowerment model,⁵⁰ and social action model.⁵¹ Also, only 31% intervention studies (κ=8) reported conducting some formative research during the development of intervention. The methodologies of formative research in these studies included pilot testing, focus groups, in-depth case studies or interviews, special workshops, and group discussions. In addition, 42% (κ=11) of the studies reported conducting some type of process evaluation, including intensive interviews with prevention workers and focus groups with participants. Some studies reported the percentage of participants who read educational materials, were involved in intervention sessions, were familiar with educational materials, or received intervention manuals. Some studies reported duration or length of the intervention; for example, the days, hours, or sessions of workshops or seminars.

In the current review, two studies (8%) reported using printed logos along with other printed materials to deliver relevant intervention messages.^21,26 Most studies tended to use peers (κ=9, 35%), or professionals/paraprofessionals such as faculty members, health workers, physicians, nurses, fieldworkers, and project staff (κ=12, 46%) to facilitate the intervention delivery. Two studies (8%) used popular opinion leaders to disseminate the intervention messages and 3 (12%) used the Internet as an intervention delivery channel. In addition, most of the studies employed a combination of printed materials (e.g., brochures, pamphlets, books, flyers, newsletters, billboards, blackboards, posters), face-to-face (e.g., workshops, seminars, voluntary counseling and testing programs, didactic lectures, group discussions, role plays, peer education, or outreach), and visual/audio materials (e.g., videos, audio tapes, VCDs and cartoon folders) to deliver the intervention messages. Only one study³⁶ mentioned that radio and television were used to deliver intervention programming. In addition, a number of studies used a variety of creative art items such as coaster and making-friend cards as well as other preventive activities such as distribution of free condoms, free needles, and other sterile injection equipment.

Furthermore, the current review found that condom use (κ=16, 62%), HIV knowledge (κ=21, 81%) and attitudes toward PLWHA (κ=6, 23%) were the most common outcome measures, followed by condom knowledge (κ=4, 15%), condom use self-efficacy (κ=4, 15%), intention to use condoms (κ=3, 12%), attitudes toward HIV (κ=3, 12%), and HIV/STD infection (κ=3, 12%), communication on HIV/AIDS (κ=1, 4%). Since studies that included general condom use, condom use with casual partners, and condom use with regular partners were coded separately, κ=24 rather than κ=16 effect sizes were included in the data analysis related to condom use. In addition, one study assessed attitudes toward PLWHA for two different target populations, therefore, κ=7 rather than κ=6 effect sizes were included in the data analysis related to this variable. Table 1 presents the characteristics of the reviewed studies and the effect sizes for each outcome variable.

Efficacy of interventions

Mean weighted effect sizes, 95% confidence intervals, and significance tests on the outcome variables were shown in Table 2. The significant Q statistics demonstrate that all of these mean effect sizes exhibited heterogeneity, which indicates that the variance of the sample of effect sizes is greater than what is expected from sampling error alone. Compared to control conditions, HIV prevention/sexual risk reduction interventions improved HIV knowledge (d=0.706, 95% confidence interval [CI]: 0.477–0.935, κ=21; Fig. 1), condom knowledge (d=0.620, 95% CI: 0.204–1.036, κ=4), attitudes toward PLWHA (d=0.625, 95% CI: 0.172–1.077, κ=7), and attitudes toward HIV (d=0.385, 95% CI: −0.023–0.794, κ=3). Of those studies that assessed condom use (Fig. 2), interventions significantly increased condom use with regular partners (d=0.477, 95% CI: 0.090–0.863, κ=6), condom use with casual partners (d=0.444, 95% CI: 0.155–0.660, κ=9), general/unspecified condom use (d=0.408, 95% CI: 0.155–0.660, κ=10), and condom use self-efficacy (d=0.584, 95% CI: 0.262–0.905, κ=4). Also, HIV prevention/sexual risk reduction interventions lowered HIV/STD infections (d=0.159, 95% CI: −0.007–0.325, κ=3). Random effect sizes ranged from d=0.159 (attitudes toward HIV) to d=0.706 (HIV knowledge). Forest plots of effect sizes and 95% confidence intervals of the interventions' impact on the two most common outcome measures (i.e., HIV knowledge and condom use) were displayed in Figs. 1 and 2.

FIG. 1.

Forest plot of effect size and 95% confidence interval (CI) of the impact of interventions on HIV knowledge.

FIG. 2.

Forest plot of effect size and 95% confidence interval (CI) of the impact of interventions on condom use.

Table 2.

Mean Weighted Effect Sizes (d) and Homogeneity Test Statistics

		Weighted mean d (95% CI)
Outcome measures	κ	Random effects	Q statistics
HIV knowledge	21	0.706 (0.477, 0.935)^a	Q=1480.07^a (p=0.0000)
Condom knowledge	4	0.620 (0.204, 1.036)^b	Q=54.88 (p=0.0000)^a
Intention of condom use	3	0.316 (−0.045, 0.676)	Q=13.845 (p=0.001)^a
Condom use with regular partners	6	0.477 (0.090, 0.863)^c	Q=92.27^a (p=0.0000)
Condom use with casual partners	9	0.444 (0.178, 0.710)^b	Q₅=61.87 (p=0.0000)^a
General condom use	10	0.408 (0.155, 0.660)^b	Q=331.77 (p=0.0000)^a
Attitudes toward PLWHA	7	0.625 (0.172, 0.1.077)^b	Q=615.55^a (p=0.0000)
Condom use self-efficacy	4	0.584 (0.262, 0.905)^a	Q₃=56.60 (p=0.0000)^a
Attitudes towards HIV	3	0.385 (−0.023, 0.794)	Q₂=31.30 (p=0.0000)^a
HIV/STD infection	3	0.159 (−0.007, 0.325)	Q₂=16.10 (p=0.0003)^a

p<0.001.

p<0.01.

p<0.05.

CI, confidence interval; PLWHA, people living with HIV/AIDS; STD, sexually transmitted disease.

Publication bias

Visual inspections of the funnel plots did not show obvious evidence of asymmetry in funnel plots of the effect sizes of the interventions for HIV knowledge (Fig. 3) and condom use (Fig. 4). However, the Begg's¹³ rank correlation test indicated a significant association between effect sizes and variances for condom use (r=0.654, p<0.001), but not for HIV knowledge (r=0.465, p=0.08). Egger and colleagues'¹⁴ regression asymmetry test indicated that the intercept from the regression analyses was significantly different from zero for condom use (intercept=0.323, p=0.045), but not for HIV knowledge (intercept=0.303, p=0.245).

FIG. 3.

Funnel plot of Cohen's d effect size by standard error for HIV knowledge.

FIG. 4.

Funnel ploto Cohen's d effect size by standard error for condom use.

Moderator tests

Given the facts that HIV knowledge (κ=21), condom use (κ=25) and attitudes toward PLWHA (κ=7) were examined most frequently in the reviewed studies and the effect sizes for these three variables were heterogeneous, moderator tests were conducted on these three variables. Different categories of condom use, including condom use with regular partners, condom use with casual partners, and general condom use, were combined for the purpose of these moderating analyses. These moderating analyses help understand what characteristics of intervention design were associated with greater effects on these three outcome measures. Sample size and age of the target audience were examined for their association with effect sizes of HIV knowledge, condom use, and attitudes toward PLWHA. However, intervention effect sizes were not significantly correlated with age and sample size.

Effect sizes of three outcomes measures were examined across intervention characteristics, including inclusion of formative research (yes/no), use of a theory (yes/no), inclusion of process evaluation (yes/no), type of intervention facilitators (peer-led versus professional-led), and number of postintervention follow-up surveys (one versus two or more). As shown in Table 3, interventions that reported some type of formative research were significantly more likely than those did not report any formative research to have impacted HIV knowledge (p<0.001), attitudes toward PLWHA (p<0.001), and condom use (p<0.001). In addition, interventions that were theory-based were significantly more likely than those that used no theory to have impacted HIV knowledge (p<0.001) and attitudes toward PLWHA (p<0.001). However, theory-based interventions (d=0.353) and those used no theory (d=0.380) had similar impact on condom use (p=0.437). Interventions that reported process evaluation were significantly more likely than those that reported no process evaluation to have impacted condom use (p<0.05) and attitudes toward PLWHA (p<0.001). However, interventions that reported no process evaluations were significantly more likely to have impacted HIV knowledge (p<0.001). In addition, peer-led interventions were significantly more likely than professional-led interventions to have impacted HIV knowledge (p<0.001) and condom use (p<0.001). However, professional-led interventions were significantly more likely to have impact on attitudes toward PLWHIA (p<0.001). Finally, interventions that included one follow-up survey were significantly more likely than those that included more than one follow-up surveys to report a significant impact on HIV knowledge (p<0.001), condom use (p<0.001) and attitudes toward PLWHIA (p<0.001).

Table 3.

Mean Weighted Effect Sizes by Categorical Moderating Variables

	HIV knowledge					Condom use					Attitudes toward PLWHA
Variable	N	κ	d	95% CI	p	N	κ	d	95% CI	p	N	κ	d	95% CI	p
Formative:
Yes	4636	5	1.145	(1.078, 1.211)	^a	12,297	10	0.550	(0.501, 0.600)	^a	6708	2	1.568	(1.473, 1.663)	^a
No	26,210	16	0.513	(0.485, 0.542)	^a	18,908	15	0.193	(0.149, 0.237)	^a	9996	5	0.256	(0.112, 0.412)	^a
Q_b=291.35					^a	Q_b=112.05				^a	Q_b=583.68				^a
Use of theory:
Yes	3835	6	1.376	(1.299, 1.452)	^a	15,135	10	0.353	(0.311, 0.393)	^a	6708	2	1.569	(1.474, 1.664)	^a
No	28,379	15	0.510	(0.582, 0.537)	^a	16,070	15	0.339	(0.284, 0.394)	^a	9996	5	0.256	(0.208, 0.304)	^a
Q_b=434.01					^a	Q_b=0.151				0.698	Q_b=583.68				^a
Process evaluation:
Yes	14,265	8	0.360	(0.252, 0.468)	^a	22,619	10	0.385	(0.339, 0.432)	^a	14637	5	1.569	(1.474, 1.664)	^a
No	17,944	13	0.426	(0.391, 0.461)	^a	8586	15	0.311	(0.264, 0.357)	^a	2067	2	0.256	(0.208, 0.304)	^a
Q_b=232.13					^a	Q_b=4.99					Q_b=11.83				^a
Messenger:
Peer-led	10,393	6	0.775	(0.730, 0.820)	^a	12,293	11	0.504	(0.438, 0.570)	^a	8822	4	0.200	(0.148, 0.252
Professional-led	21,821	15	0.525	(0.494, 0.557)	^a	18,912	14	0.297	(0.259, 0.334)	^a	7882	3	1.173	(1.099, 1.248)	^a
Q_b=79.30					^a	Q_b=28.49				^a	Q_b=440.19				^a
Follow-up:
1 follow-up	26,685	15	0.64	(0.613, 0.670)	^a	20007	13	0.430	(0.386, 0.474)	^a	15811	6	0.561	(0.516, 0.606)	^a
2 follow-ups	5529	6	0.45	(0.388, 0.5128)	^a	11198	12	0.242	(0.193, 0.292)	^a	893	1	0.160	(0.022, 0.298)	^a
Q_b=30.23					^a	Q_b=30.99				^a	Q_b=29.39				^a

p<0.001

PLWHA, people living with HIV/AIDS; CI, confidence interval

Discussion

Because of the magnitude of the HIV epidemic in China, numbers of HIV/sexual risk reduction interventions to promote behavior change among at-risk Chinese populations have been increasing during the past decade. However, very little quantitative research has examined the efficacy of these interventions. The purpose of this study was therefore to conduct a meta-analysis of the efficacy of educational, psychosocial, or behavioral interventions to promote HIV/sexual risk reduction in China.

Overall, the results of this meta-analysis revealed that the interventions reviewed were successful in improving HIV knowledge, condom knowledge, attitudes toward PLWHA, and in increasing condom use and condom use self-efficacy. The results were consistent for all outcome measures. While the exiting primary interventions have targeted a variety of populations, including both high- and low-risk groups, more efforts are needed in the future to focus on the most-at-risk populations such as MSM, female sex workers, and migrants to effectively curtail the HIV epidemic in China.

Approximately one third of the studies reported conducting formative evaluation research and those interventions that reported the conduction of formative research were more likely to have influenced HIV knowledge, attitudes toward PLWHA, and condom use behaviors among target audience. These findings are consistent with the recommendations in the field of HIV prevention.⁵² Future intervention studies in China should recognize the importance of conducting and reporting formative evaluation research. Formative research allows researchers to investigate potential barriers and facilitators among target audience for intervention adoption, obtain input from participants regarding the content, design, and delivery modality of the intervention, and optimize the feasibility, acceptability, and efficacy of interventions.^53
–55 Formative research usually presents the first opportunity for the researchers to interact with study communities, so it can help investigators to establish collaborative partnerships with the communities and participants.⁵⁶

Similarly, interventions that reported process evaluations were more likely to have a positive impact on condom use and attitudes toward PLWHA. This finding was consistent with previous studies in other cultural settings.^57,58 However, the current study found that process evaluation was not significantly associated with improved HIV knowledge. One possible problem with the data on process evaluation in the current review was the lack of specific information on the nature and scope of the process evaluation. Future research is needed to examine this issue more closely by examine the type/scope of process evaluation activities that could be more influential on outcome variables, and more useful in improving the efficacy of HIV intervention prevention programs.

Approximately one third of the interventions in the current review were theory-based. According to Cappella,⁵⁹ behavioral theories, when properly applied, allow researchers to identify the determinants that influence a behavior change among a given target audience. Likewise, information processing theories,⁶⁰ and message effects theories⁶¹ can help make intervention messages more persuasive. Through moderating analyses, this review did find evidence that theory-guided interventions had stronger effects on outcome measures such as HIV knowledge and attitudes toward PLWHA compared with non-theory–guided studies. However, the current study found that theory-guided interventions do not have stronger effects on condom use behavior than non-theory–based studies, which is not consistent with previous literature.⁶² One possible explanation is that some of the theories and models reported might not be the appropriate theories to study the change of condom use behavior or make intervention messages more persuasive in this regard. Another possible explanation is that the short duration of follow-up (e.g., 6-month) for most of these interventions was not sufficient to observe a significant change in behavior, although the intervention may indeed be efficacious in changing the condom use behavior. The third possible explanation is that most of these theories and models were originally developed in Western countries and they might not be optimal in changing condom use behavior among some Chinese populations, who may be quite different in attitudes and beliefs in terms of sexuality and condom use behavior. Given the fact that theory-based interventions have been demonstrated to be more effective in achieving intervention goals in multiple cultural settings across the global,⁶³ future HIV prevention efforts in China should integrate theories of behavior change, information processing and message effects into development of intervention programs, whenever feasible and appropriate. However, these theories and models should be specifically developed or culturally adapted for Chinese populations.

The data in the current study suggested that peer-led interventions might be more effective than professional-led interventions in improving HIV knowledge and condom use, while professional-led interventions were more effective in changing attitudes toward PLWHIA. Peer-based HIV risk reduction interventions have shown effects among young people and students in other countries^64
–66 and MSM⁶⁷ because of the normative influence of peers on decision-making.⁶⁶ For the same reason, peer-led interventions might be more effective in China for youth, students, MSM, or other populations to whom a clear peer social circle or social network can be well defined. Likewise, because of the authority-oriented culture in China in which individuals are inclined to defer to authority figures including persons who hold a higher social status and/or who possess greater knowledge, professional-led interventions could therefore better impact the decision-making process of individuals regarding their behavior and attitudes, especially those who have a low socioeconomic status or were low-educated. Future studies need to test the impact of various intervention facilitators in Chinese setting so we can maximize the intervention effect.

Finally, this study suggests that interventions that included only one follow-up survey had significant larger effects on improving HIV knowledge, attitudes toward PLWHA, and condom use than those included two follow-ups. This finding suggests a short-term effect for most of the interventions and also a possible decay of intervention effect over a longer period of follow-up. Future efforts are needed to develop interventions that can demonstrate a long-term effect and be sustainable over time.

Several limitations of the current meta-analysis should be noted. First, there were only 26 studies qualified for this meta-analysis, and even smaller number for most of the effect size calculations, which limits the accuracy of weighted mean effect sizes for some outcome measures such as condom knowledge, different measures of condom use, condom self-efficacy, intentions of condom use, and others. Second, due to the small sample size, moderator tests were limited to three outcome measures. In addition, the small sample size also made it unfeasible to conduct analysis on some important moderators such as target populations. Third, the current review only included published studies and only English publications. While such inclusion criteria were justified based on the access concern, we might have missed some studies/reports that also reported the efficacy of relevant interventions in China. In addition, some studies might remain unpublished due to nonsignificant findings, therefore, the so-called file drawer problem or publication bias⁷ might exist in this study. Fourth, some of the studies in the review did not provide sufficient information on some of the outcome measures (e.g., attitude toward HIV) or key design characteristics (e.g., dose of intervention exposure). In conclusion, this meta-analysis provides encouraging preliminary evidence that HIV prevention/sexual risk reduction interventions conducted in China improved HIV knowledge, condom knowledge, attitudes toward PLWHA among target audience. Also, these intervention significantly increased condom use and condom use self-efficacy among target audience. These studies support the use or adaptation of psychosocial, educational, and behavioral interventions targeting various at-risk or vulnerable populations in China. Future research is needed to improve the feasibility and efficacy of the interventions by employing a culturally appropriate theoretical or conceptual framework, employing a rigorous research design and methodology including formative research and process evaluation, and targeting factors at both individual and community levels. Future research also needs to develop interventions that will demonstrate a long-term and sustainable effect among various populations across different settings in China.

Footnotes

Author Disclosure Statement

No competing financial interests exist.

References

UNAIDS. 2009. Estimates for the HIV/AIDS epidemic in China. http:// www.unaids.org.cn/download/2009%20China%20Estimation%20Report-En.pdf">http:// www.unaids.org.cn/download/2009%20China%20Estimation%20Report-En.pdf. 2012 June 25.

Hong

, Li

. HIV/AIDS behavioral interventions in China: A literature review and recommendation for future research. AIDS Behav, 2009; 13:603–613.

Fan

, Lu

, Ma

et al. Behavioral and serologic survey of men who have sex with men in Beijing, China: Implication for HIV intervention. AIDS Patient Care STDs, 2012; 26:148–155.

Liao

, Kang

, Jiang

et al. Bisexual behavior and infection with HIV and syphilis among men who have sex with men along the east coast of China. AIDS Patient Care STDs, 2011; 25:683–691.

Liu

, Qu

, Guo

, Sun

. Factors that influence risky sexual behaviors among men who have sex with men in Liaoning Province, China: A structural equation model. AIDS Patient Care STDs, 2011; 25:423–429.

Noar

. Behavioral interventions to reduce HIV-related sexual risk behavior: Review and synthesis of meta-analytic evidence. AIDS Behav, 2008; 12:335–353.

Lipsey

, Wilson

. Practical Meta-Analysis. Thousand Oaks, CA: Sage Publications, 2001.

Guo

, Li

, Stanton

. HIV-related behavioral studies of men who have sex with men in China: A systematic review and recommendations for future research. AIDS Behav, 2011; 15:521–534.

Miller

, Wilbourne

. Mesa Grande. A methodological analysis of clinical trials of treatments for alcohol use disorders. Addiction, 2002; 97:265–277.

10.

Kennedy

, Medley

, Sweat

, O'Reilly

. Behavioral interventions for HIV positive prevention in developing countries: A systematic review and meta-analysis. Bull World Health Organ, 2010; 88:615–623.

11.

Budd

, Thorp

, Donohew

. Content Analysis of Communication. New York: MacMillan, 1967.

12.

Sterne

, Egger

. Funnel plots for detecting bias in meta-analysis: Guidelines on choice of axis. J Clin Epidemiol, 2001; 54:1046–1055.

13.

Begg

, Mazumdar

. Operating characteristics of a rank correlation test for publication bias. Biometrics, 1994; 50:1088–1101.

14.

Egger

, Davey Smith

, Schneider

, Minder

. Bias in meta-analysis detected by a simple, graphical test. BMJ, 1997; 315:629–634.

15.

Sanchez-Meca

, Marin-Martinez

, Chacon-Moscoso

. Effect-size indices for dichotomized outcomes in meta-analysis. Psychol Methods, 2003; 8:448–467.

16.

Cai

, Hong

, Shi

et al. Long-term follow-up study on peer-led school-based HIV/AIDS prevention among youths in Shanghai. Int J STD AIDS, 2008; 19:848–850.

17.

, Huang

, Cai

et al. Evaluation of a school-based HIV/AIDS peer-led prevention programme: The first intervention trial for children of migrant workers in China. Int J STD AIDS, 2010; 21:82–86.

18.

Lou

, Zhao

, Gao

, Shah

. Can the Internet be used effectively to provide sex education to young people in China? J Adolesc Health, 2006; 39:720–728.

19.

Wang

, Hertog

, Meier

, Lou

, Gao

. The potential of comprehensive sex education in China: Findings from suburban Shanghai. Int Fam Plan Perspect, 2005; 31:63–72.

20.

Wang

, Ann

, Shah

, Li

. The impact of a community-based comprehensive sex education program on Chinese adolescents' sex-related knowledge and attitudes. J HIV AIDS Prev Child Youth, 2006; 7:43–64.

21.

Gao

, Wang

. Participatory communication and HIV/AIDS prevention in a Chinese marginalized (MSM) population. AIDS Care, 2007; 19:799–810.

22.

Lau

JTF

, Wang

, Chen

et al. Evaluation of the overall program effectiveness of HIV-related intervention programs in a community in Sichuan, China. Sex Transm Dis, 2007; 34:653–662.

23.

Wang

, Keats

. Developing an innovative cross-cultural strategy to promote HIV/AIDS prevention in different ethnic cultural groups of China. AIDS Care, 2005; 17:874–891.

24.

Zhou

, Zhang

et al. A quasi-experimental study on a community-based behaviour change programme among injecting drug users in Sichuan, China. Int J STD AIDS, 2009; 20:125–129.

25.

Huang

, Ye

, Cai

et al. Study on peer-led school-based HIV/AIDS prevention among youths in a medium-sized city in China. Int J STD AIDS, 2010; 19:342–346.

26.

NIMH Collaborative HIV/STD Prevention Trial Group. Results of the NIMH collaborative HIV/sexually transmitted disease prevention trial of a community popular opinion leader intervention. J Acquir Immune Defic Syndr, 2010; 54:204–214.

27.

, Huang

, Li

et al. HIV/AIDS education effects on behavior among senior high school students in a medium-sized city in China. Int J STD AIDS, 2009; 20:549–552.

28.

Tian

, Tang

, Cao

, Zhang

, Li

, Detels

. Evaluation of a web-based intervention for improving HIV/AIDS knowledge in rural Yunnan, China. AIDS, 2007; 21,Suppl 8:S137–142.

29.

, Zhang

, Shan

. An HIV/AIDS intervention programme with Buddhist aid in Yunnan Province. Chin Med J, 2010; 12:1011–1016.

30.

, Stanton

, Wang

et al. Cultural adaptation of the Focus on Kids program for college students in China. AIDS Educ Prev, 2008; 20:1–14.

31.

, Zhang

, Mao

, Zhao

, Stanton

. Effect of social cognitive theory-based HIV education prevention program among high school students in Nanjing, China. Health Educ Res, 2011; 26:419–431.

32.

Lau

JTF

, Lau

, Cheung

, Tsui

. A randomized controlled study to evaluate the efficacy of an Internet-based intervention in reducing HIV risk behaviors among men who have sex with men in Hong Kong. AIDS Care, 2008; 20:820–828.

33.

Lau

, Tsui

, Cheng

, Pang

. A randomized controlled trial to evaluate the relative efficacy of adding voluntary counseling and testing (VCT) to information dissemination in reducing HIV-related risk behaviors among Hong Kong male cross-border truck drivers. AIDS Care, 2010; 22:17–28.

34.

, Wang

, Fang

et al. Short-term effect of a cultural adaptation of voluntary counseling and testing among female sex workers in China: A quasi-experimental trial. AIDS Educ Prev, 2006; 18:406–419.

35.

Cheng

, Lou

, Mueller

et al. Effectiveness of a school-based AIDS education program among rural students in HIV high epidemic area of China. J Adolesc Health, 2008; 42:184–191.

36.

, Detels

, Ji

et al. Diffusion of HIV/AIDS knowledge, positive attitudes, and behaviors through training of health professionals in China. AIDS Educ Prev, 2002; 14:379–390.

37.

Liao

, Jiang

, Yang

, Zeng

, Liao

. A life-skills-based HIV/AIDS prevention education for rural students of primary schools in China: What changed? What have we learned? Biomed Environ Sci, 2010; 23:409–419.

38.

Lin

, Li

, Stanton

et al. Theory-based HIV-related sexual risk reduction prevention for Chinese female rural-to-urban migrants. AIDS Educ Prev, 2010; 22:344–355.

39.

Jiang

, Wang

, Yang

et al. Integrated response toward HIV: A health promotion case study from China. Health Promot Int, 2010; 26:196–211.

40.

, Luo

, Sullivan

et al. Evaluation of a needle social marketing strategy to control HIV among injecting drug users in China. AIDS, 2007; 21,Suppl 8:s115–122.

41.

Rotheram-Borus

, Wu

, Liang

et al. Reductions in sexually transmitted infections associated with popular opinion leaders in China in a randomized controlled trial. Sex Transm Infect, 2011; 87:337–343.

42.

Janz

, Becker

. The health belief model: A decade later. Health Educ Q, 1984; 11:1–47.

43.

Ajzen

, Fishbein

. Understanding Attitudes and Predicting Social Behavior. Englewood Cliffs, NJ: Prentice-Hall, 1994.

44.

Bandura

. Social Foundations of Thought and Action: A Social Cognitive. Englewood Cliffs, NJ: Prentice Hall, 1986.

45.

Rogers

. Diffusion of Innovations, 3rd. New York: The Free Press, 1982.

46.

Rogers

. A protection motivation theory of fear appeals and attitude change. J Psychol, 1975; 91:93–114.

47.

Fischer

. Toward a subcultural theory of urbanism. Am J Sociol, 1975; 80:1319–341.

48.

Friedman

, Jarlais

, Ward

. Social models for changing health-relevant behaviour. DiClemente

, Peterson

. Preventing AIDS: Theories and Methods of Behavioral Interventions. New York and London: Plenum Press, 1994; 95–116.

49.

Freire

. Pedagogy of the Oppressed. New York: Seabury Press, 1983.

50.

Werner

, Sanders

, Weston

, Babb

, Rogriquez

. Questioning the Solution: The Politics of Primary Health Care and Child Survival with an In-Depth Critique of Oral Rehydration Therapy. Palo Alto, CA: HealthWrights, 1997.

51.

Minkler

, Wallerstein

, Wilson

. Improving health through community organization and community building. Glanz

. Health Behavior and Health Education: Theory, Research, and Practice. San Francisco: John Wiley & Sons, 2008; 291–298.

52.

Atkin

, Freimuth

. Formative evaluation research in campaign design. Rice

, Atkin

. Public Communication Campaigns, 3rd. Thousand Oaks, CA: Sage Publications, 2001; 125–145.

53.

Gittelsohn

, Harris

, Whitehead

et al. Developing diabetes interventions in an Ojibwa-Cree community in northern Ontario: Linking qualitative and quantitative data. Chronic Dis Can, 1995; 16:157–164.

54.

Young

, Johnson

, Steckler

et al. Data to action: Using formative research to develop intervention programs to increase physical activity in adolescent girls. Health Educ Behav, 2006; 33:97–111.

55.

Rogers

, Tross

, Doino-Ingersol

, Weisfuse

. Partner notification with HIV-infected drug users: Results of formative research. AIDS Care, 1998; 10:415–429.

56.

Johnston

, Whitehead

, Simic-Lawson

, Kendall

. Formative research to optimize respondent-driven sampling surveys among hard-to-reach populations in HIV behavioral and biological surveillance: Lessons learned from four case studies. AIDS Care, 2010; 22:784–792.

57.

Valente

. Evaluating communication campaigns. Rice

, Atkin

. Public Communication Campaigns, 3rd ed. Thousand Oaks, CA: Sage Publications, 2001; 105–124.

58.

Noar

, Carlyle

, Cole

. Why communication is crucial: Meta-analysis of the relationship between safer sexual communication and condom use. J Health Commun, 2006; 11:365–390.

59.

Cappella

. Integrating message effects and behavior change theories: Organizing comments and unanswered questions. J Commun, 2006; 56,suppl:S265–S279.

60.

Stephenson

, Southwell

. Sensation seeking, the activation model, and mass media health campaigns: Current findings and future directions for cancer communication. J Commun, 2006; 56:S38–S56.

61.

Rothman

, Bartels

, Wlaschin

, Salovey

. The strategic use of gain and loss framed messages to promote healthy behavior: How theory can inform practice. J Commun, 2006; 56:S202–S220.

62.

Noar

. Challenges in evaluating health communication campaigns: Defining the issues. Commun Methods Meas, 2009; 3:1–11.

63.

Fisher

, Fisher

. Theoretical approaches to individual-level change in HIV risk behavior. Peterson

, DiClemente

. HIV Prevention Handbook. New York: Plenum, 2000; 3–55.

64.

Brieger

, Delano

, Lane

, Oladepo

, Oyediran

. West African youth initiative: Outcome of a reproductive health education program. J Adolesc Health, 2001; 29:436–446.

65.

Cupples

, Zukoski

, Dierwechter

. Reaching young men: Lessons learned in the recruitment, training, and utilization of male peer sexual health educators. Health Promot Pract, 2010; 11,Suppl 3:19S–25S.

66.

Mosena

, Ely

, Ho

, Ruch-Ross

. Peer advocates for health: A community-based program to improve reproductive health knowledge and lifestyle choices among adolescent males. Int J Mens Health, 2004; 3:221–240.

67.

Morin

. Effects of a behavioral intervention to reduce risk of transmission among people living with HIV—The Healthy Living Project randomized controlled study. J Acquir Immune Defic Syndr, 2007; 44:213–221.