Achievement of interventions on HIV infection prevention among migrants in China: A meta-analysis

Abstract In China, migrants with acquired immunodeficiency syndrome (AIDS) have become a serious problem in the field of AIDS prevention. This study aimed to evaluate the efficacy of interventions for human immunodeficiency virus (HIV) infection prevention for migrants in China and to identify factors associated with intervention efficacy. A computerized literature search of the Chinese National Knowledge Infrastructure, Wan Fang, and PubMed databases was conducted to collect related articles published in China. Only self-control intervention studies or studies containing sections regarding self-control interventions wherein the method of intervention was health education were included. Rev Manager 5.3 software was used to analyze the intervention effects in terms of knowledge, attitude, and behavior indexes. Relative to pre-intervention, the HIV interventions showed statistically significant efficacy in terms of sexual transmission of HIV, condom use for HIV prevention, change in attitude towards HIV/AIDS patients, incidence of commercial sex behavior, and recent condoms use during sex (P < .01). Moreover, the baseline rate of migrants, intervention time, peer education, region, and education background were factors influencing the efficacy of the intervention. Significant improvement in terms of knowledge of sexual transmission of HIV and attitudes and behaviors among migrants was observed; however, based on the findings of previous studies, the interventions should be customized for different people from different districts in China. Further research is needed to evaluate subgroups of migrants in China according to their baseline characteristics.


Introduction
As a malignant disease, acquired immunodeficiency syndrome (AIDS) causes severe harm to human health and hinders economic development. AIDS is characterized by rapid dissemination and high mortality rates. Since AIDS is now globally widespread, it has become a serious public issue and an urgent social problem (Liu & Tang, 2010). Moreover, movement of individuals and populations is an important factor in the spread of the human immunodeficiency virus (HIV) (Soskolne & Shtarkshall, 2002). In many countries, including China, migrants have been identified as a high-risk population of HIV transmission, who may become 'bridge populations', transferring HIV from high-risk populations to a low-risk population. In 2016, the number of migrants in China exceeded 245 million (National Health and Family Planning Commission of the People's Republic of China, 2017), with 48.3% females and 51.7% were males. The average age of migrants was 29.8. In the whole migrants population in China, the generation born in the eighties accounted for 56.5%, and the generation born in the nineties accounted for 18.7%, and this number kept increasing. With rapid urbanization in China and the transfer of surplus rural labor force, population flow has become active unprecedentedly and is currently a prominent social phenomenon. The migrants are defined as populations that have left their domiciles and live in other places (National Health and Family Planning Commission of the People's Republic of China, 2013). Most migrants are sexually active, and their cultural level is low: their knowledge of sexually transmitted diseases (STDs) and HIV/AIDS and self-awareness are weak. These factors predispose them to several complex sexual and reproductive health risks. In their hometowns, these people are subject to very strict social control. In a strange city, however, this tight external control is weak, making it easier for them to indulge in risky behaviors (Lin, Fang, Lin, Li, & Su, 2006). Combined with China's urban and rural isolation system, the social status of migrants is very unstable; in addition, immigrant mentality is prominent, which results in them having weak social interactions and poor community inclusion with the urban society (Liu, Ma, Xiong, Li, & Wang, 2002). Thus, migrants tend to display characteristics of marginalized groups and are more likely to practice risky sexual behavior. The prevalence of HIV infection is higher among migrants than among the general population (Fitzgerald et al., 2003). Research conducted by the U.S. Centers for Disease Control andPrevention in 1987 and1992 reported that the rate of HIV infection among migrants in urban communities was 2.6%-5%, which was 10 times higher than that in ordinary residents (Holmberg, 1996). In China, of the HIV-infected patients detected in urban areas including Shanghai, Shanxi, and Zhejiang from 1995 to 2000, more than two-thirds were migrants (Luo, Mo, & Wu, 2002; National Health and Family Planning Commission of the People's Republic of China, 2017;Zhu, Wu, & Yu, 2001). Thus, this population is not only at high risk of acquiring HIV infection but also transmitting it. Moreover, because mobility is difficult to manage (Lin, Fang, Li, Xu, & Liu, 2005), migrants with AIDS have become a serious problem in the field of AIDS prevention.
Because of the lack of specific and inexpensive drugs and effective vaccines, health education intervention is considered an effective measure to curb the spread of the AIDS epidemic (Li et al., 2004). Many studies have shown that health education intervention can effectively improve the level of knowledge of AIDS prevention, induce changes in risky behavior, and reduce the spread of HIV infection.
Currently, several studies investigating migrants in the context of HIV prevention have been carried out, but the effects of the interventions have varied greatly. In this study, we used a meta-analysis to evaluate the effect of interventions conducted in previous studies to explore the effects of health education on HIV/AIDS prevention among migrants in China and provide evidence in support of conducting AIDS-related health interventions for migrants.

Search strategy
We performed a systematic literature search using several strategies: (1) electronic database searches, such as PubMed, Wan Fang, and China National Knowledge Infrastructure (CNKI), and keywords such as 'AIDS', 'HIV', 'Migrants', 'Intervention', and 'Health Education' were used; (2) requested for articles sent to researchers; and (3) reviewed reference sections of articles obtained in the searches. Studies matching the selection criteria and available as of February 10, 2018, were included in the analysis.

Study selection
Studies were selected if they met the following criteria: (1) they were self-control intervention studies or studies containing sections regarding self-control interventions; (2) they targeted migrants in China; (3) they provided information needed to calculate effect sizes; and (4) the evaluation indicators included HIV/ AIDS knowledge, attitude, and behavior change. We excluded studies in which the target study population was sex workers. Duplicate studies, reviews, local reports, conference abstracts, and presentations were excluded. Two co-authors (Rui Z and Ling C) independently extracted relevant studies following the inclusion criteria. In cases of missing data, we contacted the corresponding authors. Disagreements were resolved by consensus in a panel meeting (Rui Z and Ling C). The characteristics of the records included in the meta-analysis are shown in Table 1.

Statistical analysis
Odds ratios (ORs) were used as the measure for summary statistics of the efficacy of the HIV intervention. Owing to the initial assumption of between-study heterogeneity, a random effects model, which incorporates both within-and betweenstudy variability, was applied to combine the log OR across studies. A series of subgroup analyses were conducted by stratifying the original studies by baseline rate, time of intervention, peer education, region, and education background. All statistical analyses were performed using Review Manager 5.3 (Cochrane). Moreover, P-values less than .05 were considered statistically significant. All statistical tests were two-sided.
Table 1 depicts the basic information of included articles, included author, published year, region, sample size before/after intervention, gender before/after intervention, average age before/after intervention, marital status before/after intervention, education level before/after intervention, intervention measures, and outcome indicators. Table 1 indicated that studies were conducted in 22 provinces of China, and included 28,732 participants before the intervention and 24,381 participants after the intervention. The participants aged 15-65 years, and the average age of the participants was shown in Table 1, most of the participants were educated until junior high school or less. They are engaged in construction and manufacturing industries, and they always work in construction sites and small factories. For young migrants, they were more likely to work in small factories and engaged in manufacturing industries; the work in manufacturing industries was Journal des Aspects Sociaux du VIH/SIDA easier than that in construction industries, but the wage was lower and they only can support themselves. However, elder migrants were more likely to work in construction sites to gain much higher wages and support their family. Intervention measures included promotional material, lectures, quizzes, peer education, and condom distribution, and the time of intervention ranged from immediately to 1 year. Furthermore, 10 studies (Huang & Qiu, 2015;Ma, 2015;Peng & Zhong, 2015;Wang et al., 2008;Xie & Yang, 2011;Xu, Wang, et al., 2006;Zhao & Xu, 2015;Zhou & He, 2011) did not report the duration of the intervention.

Sensitivity analysis
We used a fixed effects model to combine the ORs of the five indicators, and the effect size was similar to the results of the random effects model. Thus, the combined result had good stability, indicating that the results were reliable.

Stratified analyses
No single methodological quality feature was associated with intervention efficacy. However, the intervention effects did vary by intervention characteristics. We performed a stratified analysis according to baseline rate, time of intervention, peer education, region, and education background; the results showed that they are all factors that influence the efficacy of intervention, but the efficacy of the intervention differed according to these factors (Table 3).

Discussion
This study investigated the efficacy of interventions from three aspects, HIV/AIDS knowledge, attitude, and behavior, using five indicators. Sexual transmission has become the major route of HIV transmission in China (Ministry of Health of the People's Republic of China, The Joint United Nations Program on HIV/ AIDS, World Health Organization, 2012); therefore, we selected HIV/AIDS transmission route and prevention method as indicators 1 and 2. Social inclusion of patients with AIDS is key for HIV/AIDS prevention, and acceptance of such migrants had a positive effect on the prevention of AIDS; thus, changing the attitude towards these patients was selected as indicator 3. The ultimate goal of the intervention was to promote people to make behavioral changes; therefore, the incidence of commercial sexual behaviors and recent condom use were selected as indicators 4 and 5, to serve as indicators of behavioral change. In addition, because of differences in survey instruments in each study, we selected the most commonly used indicators in order to improve the utilization rate of data and reduce the bias.
We were very encouraged to find that the HIV interventions were efficacious, not only in terms of improving knowledge but also in terms of attitude and behavior among migrants in China. However, after stratification by influencing factors, different factors seemed to have different effects on intervention efficacy. Our study selected articles from the past 10 years, and systematically evaluated intervention methods; we found that the efficacy of the intervention differed by baseline rate, time of intervention, peer education, region, and education background.
We found that interventions with lower baseline rates were more efficacious at improving HIV/AIDS knowledge, attitude, and behavior than interventions with higher baseline rates (Table 3).
Therefore, interventions should be designed differently if the target population involves people with low awareness or those never accepted in the society; this intervention would not be very effective for people with high awareness, unless targeted measures would be used. For behavioral indicators, the longer the duration of the intervention, the better the efficacy (Table 3); we found that interventions lasting 3-6 months was more efficacious at improving HIV/AIDS knowledge and attitude than interventions lasting fewer than 3 months or more than 6 months. Thus, people would be tired of lengthy health education interventions. The optimum duration should be selected on the basis of the target population's knowledge and attitude. An intriguing finding was that interventions conducted in developing central and western regions were more efficacious than in developed eastern regions (Table 3) for knowledge and attitude indicators. The reason for this discrepancy is that eastern cities are more liberal, AIDS education has been increasingly publicized, and the people thus have relevant knowledge. Accordingly, the effect of the intervention is not obvious. Compared to the east, the central and western regions are closed and backward, and health education is not often publicized; thus, an educational intervention can effectively improve people's knowledge of AIDS. In future, studies should pay attention to the economic status of the area in which an intervention is planned. However, behavioral interventions in developing central and eastern regions were more efficacious than those conducted in western regions (Table 3), which may be due to the developed economy in eastern regions, where people have a higher education and more knowledge. Moreover, people in eastern regions are more likely to change their behaviors after health education, which is related to the high baseline rate in terms of HIV/AIDS knowledge and attitude. Education background also has an effect on intervention efficacy (Table 3): the interventions were more efficacious in people educated until junior high school and below in terms of the behavioral indicators, which suggested that they preferred to change their behaviors after the health education. Regarding the interventions for the knowledge and attitude indicators, the intervention effects were greater for people educated until junior high school and below, who accounted for 50%-80% of the population; thus, there was an obvious increase in HIV/AIDS knowledge and attitude among people with low education who readily accepted the health education. However, for people with extremely low education, though the intervention efficacy was greater than in people with high education, the poor cultural bias and weak background knowledge significantly negatively impacted the efforts at improving AIDS/HIV knowledge and attitude. Therefore, the intervention methods should be designed according to the target population. We need to use popular and easy ways to intervene for people with low education backgrounds, to make them easily accept AIDS/HIV-related knowledge and change their HIV/AIDS-related attitudes. In terms of peer education, there were some differences compared to that observed in previous studies. Our study findings were similar to those of Ngo, Ha, Rule, and Dang (2013), who reported that peer education is an effective means of preventing AIDS. Bond and Wolf (1998) also thought that people with a similar age, background, social economic status, and sex can have a common topic of discussion, and thus share information, which is the goal of education. Our study suggested that the efficacy of peer education as an intervention was uncertain and that the efficacy of peer education needs further study in light of the changes in HIV epidemiology.
Some achievements have been observed after intervention; however, with an increase in the number of people with AIDS, the results are different: the epidemiology of HIV had changed, along with the knowledge and attitude of migrants with HIV/ AIDS. Thus, the existing intervention methods are no longer relevant; we should explore more precise measures and adopt different methods according to different people and districts, so as to design precise and optimum interventions. For example, with the widespread of social media, including WeChat, QQ, and Weibo, the internet intervention was widely used in HIV/AIDS health education. We can make a personalized intervention on HIV/AIDS among young migrants through social media. Additionally, for old migrants, more paper-made health education materials were needed, and easier, interesting papermade propaganda was carried out for migrants with low education. For the developed eastern region, convenient network gives the chance to implement health education through social media, and intervention should include more strategies about how to reduce high-risk behavior of HIV/AIDS, while more knowledge about the harm of HIV/AIDS should be included in the health education in the developing western region.
Considering that most of the migrants had a junior high school degree, HIV/AIDS health education should be strengthened in the future, especially in rural areas and junior high school students, with an aim to enable them to recognize the threat of AIDS and learn how to protect themselves. Learning from the experiences of some countries such as Cambodia, AIDS prevention interventions should be strengthened in rural areas, including establishing farmers' life school, paying more attention to migrant families, as well as improving AIDS-related knowledge and urban life skills training within rural communities (United Nations Development Program, 2002). These methods can make farmers acquire the necessary city life skills and health knowledge, and reduce the high-risk behavior typically observed; furthermore, this approach could effectively prevent HIV transmission within the hometowns of these migrants upon their return (Haour-Knipe, 2009). This is the first time to evaluate the efficacy of HIV intervention in Chinese migrants. Several reviews (Crepaz et al., 2006;Crepaz et al., 2014;Globerman et al., 2017) studied the efficacy of HIV preventive interventions for people living with HIV in highincome settings, these studies focused on the specific methods of intervention among HIV patients, suggested that comprehensive risk counseling and services, provided skills building were effective in reducing sexual risk behaviors. While our study focused on the overall efficacy of HIV/AIDS interventions and the influencing factors of efficacy among migrants in China, our study suggested that the intervention conducted among migrants in China was effective in altering their knowledge, attitude, and behaviors on HIV/AIDS. Additionally, in China, the methods of intervention were numerous, including panel, video, consults, knowledge contest, lecture, etc., and no consistent criteria are there to classify these methods of intervention; therefore, it is difficult for us to evaluate the efficacy of specific methods of intervention. Maybe we can carry out a meta-analysis that aims to evaluate the efficacy of intervention methods on HIV in a further study. Futhermore, comprehensive risk counseling and services, provided skills building, group-level intervention could be carried out in the intervention of HIV/AIDS prevention in future intervention studies.
Some limitations of this review warrant comment. Our selection criteria were rather tight, which resulted in a limited number of studies. After the stratified analysis, the heterogeneity within most subgroups still existed. Prompting factors influencing the efficacy of HIV/AIDS interventions for migrants are more complicated. The composition of the migrant population is complex: different careers, flow areas, education background, workplace, age, and income are likely to impact the HIV/AIDS knowledge received, which could be the cause of the heterogeneity. Due to the mobility of this population, interventions are widespread and thus difficult to track, which leads to a large rate of missing data. This affects the evaluation of the intervention effects, especially in the long term. The majority of the studies targeted migrants in large cities; thus, it is not clear to what extent our findings are generalizable to migrants in small-and medium-sized cities.

Conclusions
The aforementioned results corroborated the strong evidence in support of efficacy of HIV interventions for migrants in China. Further research on how interventions can be adapted to different settings and subpopulations would be valuable, especially in terms of behavior indicators.

Ethics approval and consent to participate
The study's protocol and data collection procedure were approved by the Institute of Public Health and Management. Our study was a meta-analysis, we collected data from published articles, we did not contact the participants, and we could not know the personal information, so our research does no harm to participants and their information was secure.

Availability of data and materials
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.