Introduction

Thailand has developed more than neighboring countries of Myanmar, Cambodia and Laos, resulting in increased labor migration from these countries to Thailand. To sustain a growing economy and to compete in international markets for low cost export of goods, Thailand is dependent upon a cheap migrant labor force. Estimates of almost 2.5 million laborers have been made, including both documented and undocumented workers [1]. The majority of workers are estimated to be undocumented. Most workers are young adults, a majority male, and many come from areas of their countries that are close to the Thai border [2]. In provinces with a sea coast, male migrants predominate due to their employment on fishing boats. Many of the male migrants in coastal provinces were single, particularly Cambodian men. Women in the coastal provinces were likely to be married (80 % in 2008) and have followed their husbands to Thailand. In contrast, inland provinces where many migrants participate in factory work have only a slightly larger proportion of women compared to men [3]. Women in the non-coastal provinces involved mainly in factory work were more likely to be single (54 % in 2008) than women in the coastal provinces. Once workers come to Thailand, most remain for long periods of time [3]. There is some seasonality among the plantation workers, but little in other industries.

There is some evidence that social integration and participation of migrants in their receiving communities may benefit health. On arrival in a new area, migrants may work to develop ties with the community and to create new social networks [4]. These networks may include members of the migrants own community as well as members of the host society. There is some evidence that contacts within their own community may be beneficial to preserve their mental health [4, 5] and to avoid economic exploitation [6]. Furthermore, contacts within the host community may promote access to further economic opportunities as well as health information and services. It has been suggested that integrated individuals may be subject to social controls that may promote the adoption of healthful behaviors and prevent behaviors with defined health risks [7, 8]. A comparative study of European countries showed that in countries with strong policies favoring integration of migrants, negative pregnancy outcomes were significantly reduced [9]. A mechanism through which integration policies may be protective may be through the increased social participation of the migrants in the receiving society. Although social integration has been incorporated into health research in other areas, it has received less attention in AIDS prevention research.

Research on social networks and HIV prevention has shown that AIDS knowledge and preventive behaviors are influenced by persons in one’s social network. In rural Malawi, a unique data set on informal conversations about AIDS in communities has shown that men and women evaluate sources of risk and sexual pleasure as well as advice from official sources and devise local, sometimes innovative strategies to avoid HIV infection [10]. Indeed, interventions that rely on conversations between friends have been shown to be of value in prevention of sexually transmitted diseases [11–13]. We posit that increased social participation of migrants will result in improved knowledge of AIDS and lead to increased condom use. Contacts with other migrants as well as with Thai nationals may be valuable in this process.

Previous research on migrant laborers in the Asian region has identified several factors related to AIDS knowledge and HIV risk behaviors. Factors related to AIDS knowledge include greater knowledge among males and those with more education and less knowledge among seafarers and agricultural workers as well as those who have recently come to Thailand [14–16]. Factors related to risk behavior include single marital status, long stay in Thailand, casual, paid or regular partner, work in the seafarer or seafood industry, low self efficacy, alcohol use, injecting drug use, peer norms, lack of family restraint, and low perceived vulnerability to HIV/STI infection [17–23]. We evaluate the role of known risk factors, as well as the role of social integration, in influencing migrants’ knowledge of HIV and condom use behaviors.

Constructs for hypothesis testing inn this study were drawn form the Health Belief Model [24] and social cognitive theory [25]. HIV preventive behaviors are posited to be related to (a) perceived susceptibility to HIV infection, (b) cues to action including social integration, AIDS knowledge, and participation in AIDS prevention activities, (c) barriers toward condom use such as reduced pleasure and (d) modifying variables including relationship factors and demographics such as age, marital status, country of origin, education and occupation. Self efficacy, an essential part of social cognitive theory, is the person’s belief that he or she can undertake an activity successfully [25]. Self efficacy for condom use with partners is also hypothesized to increase condom use.

Methods

Data

Data for the study were drawn from the Phase 2 Baseline survey of the project on the Prevention of HIV/AIDS among Migrant Workers in Thailand (PHAMIT) conducted by the Institute for Population and Social Research of Mahidol University in collaboration with the Raks Thai Foundation in 2010. Financial support was provided by the Global Fund to Fight AIDS, Tuberculosis and Malaria. The target group for this survey was male and female laborers who had worked in Thailand for at least 3 months whose country of origin was Myanmar, Cambodia or Laos. The survey was conducted in the ten Thai provinces containing the largest numbers of migrant workers, stratified by country of origin to endure inclusion of sufficient numbers of workers from the three sending countries. Because of the unregistered status of many workers, respondents were selected using a snowball sampling technique. After sample areas were selected, the field data collection coordinator initiated the sampling by randomly choosing index (seed) respondents from a scattering of locations in the sample area. This index respondent then referred the team to another potential respondent who met the selection criteria until ten or eleven individuals were selected per sub area. Further details of the sampling procedure can be found in the survey report [2]. Sampling weights were used in all analyses to insure that the total sample represented the migrant population in size of group from each sending country.

Data were gathered through personal interviews with men and women age 15–59 years of age. The questionnaire was prepared in four languages: Thai, English, Burmese and Cambodian. The interviews were conducted by trained, experienced, interviewers who were fluent in Thai language as well as the language of the migrant laborers in their area. An IRB clearance was obtained through Mahidol University.

Measures

Dependent Variables

Measures of Sexual Behavior and Condom Use

Had a sexual partner. Data were coded to indicate if the respondent (1) had a regular partner (yes/no), (2) a nonregular partner (yes/no), or a paid partner (yes/no). A regular partner meant a married partner or a partner who was living with the respondent although there was no marriage registration or marriage ceremony. Nonregular partners were casual partners excluding paid sex workers. During the interview, the interviewer provided these definitions to the respondent, when he/she asked questions related to regular or nonregular partners.

Consistent condom use with partner. Consistent condom use was measured with four levels always (4), most of the time (3), sometimes (2), never in the last 12 months (1). Consistent condom use was measured for each type of partner. Dichotomous variables were created for the logistic analyses for each type of partner. Because consistent condom use was low for regular partners, the variable was coded ever used a condom (1) or never used a condom (0). In contrast, because condom use was high with nonregular and paid partners, the variable was coded always used a condom (1) or used a condom less than always (0).

Integration and Knowledge Factors

AIDS knowledge. The AIDS knowledge score is the number of correct answers to 19 questions on AIDS transmission and treatment. The questions included transmission through casual contact, blood, needle sharing, mosquito bites, sex with a healthy looking person, and mother to child transmission, breastfeeding, antiretroviral use by HIV positive mothers and prevention though condom use and reducing the number of partners,

Social Integration factor. Social integration was measured using duration of residence in the community (years in Thailand), ability to speak the Thai language to communicate with local Thais (good, fair, poor, not at all), acquisition of a Thai nickname, an indicator of interaction with the Thai community (yes/no), and the participation in social events with the Thai (number of events) and with the migrant community (number of events). A factor analysis with a varimax rotation was estimated for these variables. One significant factor emerged and an index was created using the regression method for factor scores. The factor was then coded into deciles to facilitate interpretation.

Participation in an HIV Prevention Meeting (Yes/No)

Health Belief Model factors

AIDS Susceptibility. Each respondent’s evaluation of his/her level of risk of HIV infection was coded as high, low and not at risk.

Barrier Reduce pleasure. Respondents were asked if condoms reduce pleasure (yes/no). This variable was measured for each type of partner.

Self Efficacy. Convince a partner to use a condom. Respondents were asked “Can you convince your partner to use a condom? “(yes/no). This variable was measured for each type of partner.

Modifying Demographic Factors

Modifying factors include age, gender, marital status, country of origin and education. Dummy variables representing occupation were included in early models but were dropped due to lack of significance.

Statistical Methods. The statistical methods used were multiple linear regression for the continuous variable AIDS knowledge and multiple logistic regression for the condom use variables. In building the models, an hierarchical model building procedure was used. Using our conceptual model, we divided the variables into three groups: social integration and information (social integration index, attendance at an HIV prevention meeting, AIDS knowledge), health behavior variables (susceptibility, reduce pleasure, convince partner) and modifying factors (demographic variables). We gave equal priority to the first two groups of variables and a lower priority to the modifying factors. In each analysis we show the full model, and a reduced model when a set of variables was eliminated. Wald statistics were used to determine significance of groups of variables for the logit models on condom use and an F statistic for the linear regression on AIDS knowledge. In the model for condom use with non regular partners, country of origin and marital status could not be included due to lack of variance.

Results

The demographics of the study population are shown in Table 1. The average age of a migrant was about 28, with a range from 15 to 59. About 40 % of males were married, compared to 48 % of females (p < 0.01). The largest group of migrants were from Myanmar (82 %) with smaller proportions from Cambodia (10 %) and Laos (9 %). About 38 % of migrants had less than a primary education, 43 % a primary education and 19 % more than a primary education. Men had more education than women (p < 0.01).

Table 1 Weighted sample characteristics
Full size table

Descriptive statistics on the AIDS, sexual behavior and integration are also shown in Table 1. Out of 19 questions on AIDS, the migrants answered 13 on average correctly. About 38 % of the respondents had participated in an AIDS program. More women (41 %) than men 34 % participated in an AIDS program. (p < 0.01). Most respondents had low levels of perceived susceptibility to AIDS. The integration factor is measured in deciles of integration. Women had lower scores on integration than men (p < 0.01).

The majority of respondents reported a regular partner (84 %). More females (94 %) than males (76 %) reported a regular partner (p < 0.01). The level of condom use was low for regular partners. About 40 % of respondents reported that they could persuade a regular partner to use a condom, with more males (47 %) than females (32 %) reporting that they could do this (p < 0.01). More men (37 %) than women (10.9 %) reported than condoms reduce pleasure with their regular partner (p < 0.01).

About 6.5 % of men and 0.1 % of women reported nonregular partners. For men, condom use with nonregular partners was at a moderate level with 56 % of men reporting that they always use condoms and 23 % reporting that they never use condoms with these partners. About 6 % of males reported a paid partner. Condom use was high with the paid partners with 75 % reporting that they always use a condom. Almost all of the men reported that they could convince a nonregular (84.5 %) or a paid (83.2 %) partner to use a condom. The majority of these men also reported that condoms reduce pleasure with nonregular (70 %) and paid (64 %) partners.

Table 2 shows the multiple regression analysis of factors related to AIDS knowledge among migrant workers. Both the integration factors and the modifying demographic factors were significant when tested as a group (p < 0.05). Migrants who had higher scores on social integration had higher HIV knowledge scores (p = 0.01). The AIDS knowledge score increased about 0.10 for each decile of integration. In addition, migrants who had participated in a local meeting related to HIV prevention answered more than three more questions correctly than migrants who had not attended an AIDS meeting (p < 0.01).

Table 2 Multiple linear regression analysis of factors related to AIDS knowledge among migrant workers
Full size table

Older workers had a somewhat higher level of AIDS knowledge. They scored 0.04 higher for each year of age (p < 0.01). Gender and marital status were not significantly related to AIDS knowledge. Migrants from Myanmar and Cambodia had lower levels of AIDS knowledge than those from Laos. Migrants from Myanmar answered almost one less question correctly on average while those from Cambodia answered about one and a half fewer questions correctly (p < 0.01). Migrants with lower levels of education had lower level of AIDS knowledge compared to those with more education (p < 0.01). Migrants with less than a primary education answered about 2.4 fewer questions correctly than those with more than a primary education. Those with a primary education answered one fewer questions correctly than those with more than a primary education.

The multiple logistic analysis of factors related to ever using a condom with a regular partner is shown in Table 3. The integration knowledge factor variables and the Health Belief Model variables were both significantly related to condom use with regular partners (p < 0.05). The modifying demographic factors were not significant as a group. In the final model, the degree of social integration was positively related to condom use (AOR:1.09, 95 % CI: 1.01, 1.12). AIDS knowledge was positively related to condom use (AOR: 1.16, 95 % CI: 1.06, 1.27). Participation in an AIDS prevention meeting was also related to condom use (AOR1.72, 95 % CI: 1.15, 2.58).

Table 3 Multivariate logistic regression analysis of factors related to ever using a condom with a regular partner among migrant workers
Full size table

Among the Health Belief Model variables, the individual variable AIDS susceptibility was not significantly related to condom use. The barrier that condoms reduce pleasure was significantly related to condom use (AOR 1.76, 95 % CI: 1.17, 2.65). In addition, self efficacy had a large positive association with condom use (AOR: 6.01, 95 % CI: 4.00, 9.04).

Table 4 shows the factors related to always using condoms with nonregular partners for males. Women reported too few partners in these categories for analysis. All three groups of variables were significantly related to condom use with nonregular partners. For males with nonregular partners, the social integration variable was positively related to condom use (AOR 1.26, 95 % CI: 1.02, 1.57). The individual variables AIDS knowledge and participate in an AIDS meeting were not significant. Among the Health Belief Model variables, only AIDS susceptibility was significant. (AOR 0.10, 95 % CI: 0.01, 0.76) Among the demographic modifying factors, age was not related to condom use. Men with less than a primary education (AOR 0.01, 95 % CI: 0.00, 0.09) and men with a primary education (AOR 0.13, 95 % CI: 0.03, 0.59) were less likely to use condoms with nonregular partners compared to men with more than a primary education.

Table 4 Logistic regression analysis of factors related to always using a condom with nonregular partners for male migrant workers
Full size table
Table 5 Logistic regression analysis of factors related to always using a condom with paid partners for male migrant workers
Full size table

Of the three groups of variables, only the Health Belief model variables were significantly related to condom use with paid partners (p < 0.05) (Table 5). Among the individual variables in the final model, AIDS susceptibility was negatively related to condom use (AOR 0.09, 95 % CI: 0.02, 0.37). Self efficacy was also related to condom use (AOR 5.79, 95 %CI 1.89, 17.80).

Conclusions

In summary, the results indicate that several factors were related to AIDS knowledge and the consistency of condom use. First, the social integration of migrants was positively related to both aspects of HIV prevention—AIDS knowledge and condom use with regular and nonregular partners. Demographic factors, including age, education, and country or origin had an important effects on AIDS knowledge. Migrants with lower levels of education had less knowledge of AIDS and less condom use with nonregular partners. Exposure to AIDS education had a positive effect on AIDS knowledge and condom use. Health Belief Model factors were associated with condom use with regular, paid and unpaid partners.

While reported condom use was high with paid partners, only the Health Belief Model variables were significantly related to condom use with paid partners. This may be because of the successful 100 % condom use program in Thailand [26] which created strong pressure for brothel owners to insist that clients use condoms with sex workers. It is possible that condom use in brothels may be due as much to structural factors than to the individual motivation of clients.

The demographic results point toward the need for a focus on AIDS education among the workers with the lowest levels of education. Cambodians and Burmese also had lower levels of AIDS knowledge than Lao migrants. The AIDS prevention program “Prevention of HIV/AIDS among Migrant Workers in Thailand” (PHAMIT), supported by the Global Fund to Fight AIDS, Tuberculosis and Malaria through the Raks Thai Foundation and seven other collaborating NGOs have provided AIDS prevention programming in migrant communities for several years. The AIDS programs that many of the migrants attended had a positive association with both AIDS knowledge and condom use, suggesting that the migrants are learning from the program.

Female migrants had lower levels of social integration than male migrants. A comparison by gender of the five variables included in the integration factors showed that women had similar durations of residence in Thailand and similar levels of participation in community events with Thai and migrant persons. However, the females were less likely to speak Thai or to have a Thai nickname. This indicates a lower level of interaction with Thai persons that may be due to more segregation by country of origin in the parts of the labor force that women are likely to work in. Earlier research has also shown that kin networks are more important for women than men in migration, perhaps due to their greater vulnerability in the migration process [3]. This factor may also decrease the number of contacts within the Thai population among women. More outreach to women in these communities may be needed to promote health behaviors.

The study has some limitations. The cross sectional design limits the causal inferences that can be made from the study. Furthermore, the study relies on self reports of sensitive behaviors that cannot be verified. Finally, the number of nonregular and casual partners was small.

In conclusion, social integration has been associated with positive health outcomes in this and other settings. However, although migrants are staying in Thailand for long periods of time, there are obstacles to their integration. There is no clear path to a permanent residence status so most migrants stay on in an irregular status [3]. Research has shown that public opinion among the Thai population is not negative with regards to the regular or documented migrant workers and with migrants who live in refugee camps along the borders. However, public opinion was very negative for undocumented workers with out a legal status [25]. Migrants are not allowed to form their own labor unions and few join Thai labor unions. Migrants receive health care from government and private sources, though they report that they do not receive equal treatment with Thai people in the community [27]. Qualitative research with migrants revealed that Thais still resented and insulted migrant workers and did not try to understand the migrant culture [28]. Limiting the spread of HIV in migrant populations as well as other positive health outcomes may be enhanced by promoting the integration of migrants into the community as well as into school and work settings. Interactions with others in the community can promote learning about health and health services and pathways to protect migrant health. Facilitating migrant integration into migrant social networks as well as Thai social networks may promote migrant health.