HIV PREVENTION FOR MIGRANTS IN TRANSIT: DEVELOPING AND TESTING TRAIN

Abstract

This study was a pilot investigation of the feasibility, acceptability, and effects of TRAIN (Transit to Russia AIDS Intervention with Newcomers) a three-session HIV preventive intervention for Tajik male labor migrants performed in transit. Sixty adult Tajik male labor migrants on the 5-day train ride from Dushanbe to Moscow were randomly assigned to either the intervention or a control condition. Each initially completed an in-person survey then another 3 days later (immediately postintervention), and participated in a cell phone survey three months later. All participants came to all intervention sessions, were satisfied with the program, and completed all postassessments. In comparison with the controls, the TRAIN group reported significant increases in condom use with sex workers and non-sex workers, condom knowledge, worry about HIV/AIDS, talking with persons about HIV/AIDS, talking with wife about HIV/AIDS, community activities, and religious activities. HIV/AIDS prevention performed in transit is feasible, accceptable, and potentially efficacious in diminishing HIV risk behaviors in labor migrants.

Labor migration has recently attracted attention as a contributor to the global HIV/AIDS epidemic. Evidence from several different geographic regions indicates that migrant workers are at elevated risk of acquiring HIV/AIDS (Anderson, Qingsi, Hua, & Jianfeng, 2003; Fitzgerald, Chakraborty, Shah, Khudr, & Duggan, 2003; Li, Morrow, & Kermode, 2007; Organista, Carrillo, & Ayala, 2004; UNAIDS, 2008). Public health officials are concerned that migrant workers may serve as a “bridge” spreading HIV/AIDS to the general populations in both sending and receiving countries (Kramer et al., 2008).

Central Asia and eastern Europe report the fastest rates of HIV/AIDS growth in the world (UNAIDS, 2004). Massive political, social, cultural, and behavioral changes, along with economic upheavals and collapse of the public health infrastructures, have created circumstances conducive to the rapid spread of HIV (Kelly & Amirkhanian, 2003). Tajikistan, a former Soviet republic, has a highly mobile population. Following the civil war (1992-1997), which caused massive devastation, Tajikistan became one of the largest regional labor exporters globally (Erlich, 2006). Over 1 million Tajik citizens work outside of their country. Most migrant workers are male (80%), married (70%), and between 30 and 40 years of age (Olimova & Bosc, 2003).

The primary destination for Tajik labor migrants is Russia, which has historical links with Tajikistan. With its rapid economic growth in the past decade, Russia has needed laborers for the most difficult, dirty, and dangerous work, the so-called 3D jobs. Because of the arrangement between the two countries, Tajik citizens can travel back and forth to Russia as seasonal migrant workers without having to obtain a visa. In Russia, Tajik workers also have established diaspora networks (Sodiqov, 2010).

The majority of Tajik migrants live and work in Moscow, carrying goods in the bazaars, working in construction sites, sweeping the streets, and working in food services. Far from their families, most Tajik male migrants have sex with multiple partners, including sex workers (who are known to have rates of HIV that are 30-120 times higher than the general population in Moscow (Shakarishvili et al., 2005). In January 2009 there were 1,422 registered HIV-positive cases in Tajikistan, of which 145 (10.2%) were migrant workers (Tajikistan HIV/AIDS Prevention Center, 2008).

HIV prevention efforts with labor migrants have largely focused on disseminating information through education sessions, brochures, and counseling for those getting HIV tested (Organista et al., 2004). “Off-the-shelf slogans dreamt up by donors”(Horton & Das, 2008) have been critiqued, but country-specific empirically driven solutions are still wanting. Indeed, very few HIV prevention efforts for labor migrants have been empirically tested (Olshefsky, Zive, Scolari, & Zuniga, 2007; Organista et al., 2004). A 1-session intervention with truck drivers in India used an information-motivation-behavioral skills model and found mixed support for effectiveness (Cornman, Schmiege, Bryan, Benziger, & Fisher, 2007). Changes in condom use with marital partners were explained by intervention effects on attitudes, norms, behavioral skills, and intention to use condoms, with no such changes with nonmarital partners.

In Tajikistan through the Global Fund in 2005, “migrant cabinets” were established in 30 districts, providing free health services to migrants, including sexually transmitted infections (STI) treatment and HIV/AIDS prevention. However, no available evidence indicates their utilization or effectiveness. To build an empirically driven HIV prevention intervention for male labor migrants, we conducted developmental ethnographic and survey studies (Golobof, Luo, Bahromov, & Weine, 2008; Weine, Bahromov, & Mirzoev, 2008; Luo, Golobof, Bahromov, & Weine, 2008). These studies provided descriptive and contextual knowledge on male migrant workers and indicated that: (a) male migrants are taking some, but not enough, precautions to prevent HIV; (b) male migrants’ ability to protect themselves from acquiring HIV is compromised by harsh living and working conditions, poor access to health care (including health promotion, particularly HIV prevention and voluntary counseling and testing), alcohol use, and little spousal communication regarding sexuality; and (3) male migrants have important sources of strength, including men’s prosocial identities as part of their family clan, Tajik village network, and Muslim religion; however, these affiliations often tend to not address HIV prevention and tend to encourage HIV risk behavior such as concurrent sex with sex workers and alcohol use with sex. On the basis of these studies we concluded that to be potentially effective and feasible, HIV prevention interventions should not only provide migrants with HIV knowledge and prevention skills but should address those difficulties and strengths specific to the social and familial experiences of married male labor migrants.

We learned from migrants of the challenges of conducting HIV preventive interventions in Tajikistan and in Moscow. The majority of Tajik migrants are in Tajikistan only during the winter season (December to March).

Before departure migrants are busy with their families, preparing to travel, and distributed all over the country. After arrival in Moscow it is difficult to gather a large group of migrants, as they are busy working and are spread throughout the Moscow metropolis. The feedback we received from migrants was that an intervention conducted in transit would likely be more successful in terms of engaging a sizable group. They suggested that we consider the train.

Train 227 travels from Dushanbe to Moscow, a route it has followed since the Soviet era (Zotova, 2006). The rail cars are from the Soviet era, with six-passenger compartments containing lower and upper berths a long narrow corridor. The ride usually takes 4 days and crosses the borders of four different countries: Uzbekistan, Turkmenistan, Kazakhstan, and Russia. Crossing so many borders makes the journey very difficult for the migrants because of bribe expectations and harassment from border guards, customs officials, and transport police. Migrants take as little money with them as possible—just enough for frugal meals and paying bribes to whoever may turn up and demand to be shown papers (Zotova, 2006).

Despite these and other difficulties, the train is a major mode of travel to Moscow for Tajik male labor migrants and presents advantages for implementing HIV prevention for several reasons. First, the train contains more than 600 passengers and the vast majority of them are Tajik male labor migrants. Second, migrants in transit have 4 “free” days during which they can be engaged in preventive interventions. Third, each train contains a diverse group of labor migrants from different regions, religions, ages, and workplaces, a much broader spectrum than could be found at any point in Moscow or Dushanbe. In comparison with rail transport, airports and airplanes posed even more obstacles in terms or where and how to intervene. This suggested the possible appropriateness of intervening on the train from Dushanbe to Moscow.

METHOD

Overview of Research Design

The central research questions addressed in this pilot study were (a) Was the Transit to Russia AIDS Intervention with Newcomers (TRAIN) feasible and acceptible to migrants in transit by train? (b) Was there a change in condom use with sex workers or non-sex workers as a function of TRAIN over time? (c) Were there changes in other variables? (d) Did any of these other variables appear to mediate TRAIN effects? This study also considered recommendations for HIV prevention programs with labor migrants and the implications for researching HIV preventive interventions for labor migrants in transit.

SAMPLE

The sample for this research consisted of 60 Tajik migrant workers who were recruited on the Dushanbe-Moscow train according to the following criteria. Participants had to be male migrants from Tajikistan between 18 and 45 years of age who were married or engaged to a woman in Tajikistan. They had to be seasonal or temporary migrant workers who planned to return to Tajikistan at least once per year; 4) embarking on first, second, or third work visit to Moscow. In addition, participants had to self-report at least two of the following risks: (a) recent unprotected sexual contacts in Tajikistan or Russia, (b) known present or past STI diagnosis (c) present or past tuberculosis (TB), (d) present or past intravenous drug use, (e) present or past alcohol abuse. We excluded persons who did not understand the nature and extent of participation in the study and the risks and the benefits of the participation in the study and who were not able to provide a cell phone number for follow-up assessment.

The recruited and enrolled sample consisted of 60 Tajik married male migrants who were proportionally selected from four regions of Tajikistan so as to roughly reflect the regional diversity of Tajik migrants in Moscow (Erlich, 2009): 25 (42%) from Dushanbe, 13 (22%) from Khatlon, 10 (17%) from Sughd, and 12 (20%) from Gorno-Badakhshan. The sample consisted of 48 (80%) Sunni Muslims and 12 (20%) Ismaili Muslims. The average age of participant was 29.5 years (SD = 5.9). Education level was 53 (88%) high school, 1 (2%) college, and 6 (10%) university. All participants reported that they financially supported their families back home. Average number of persons they supported financially was 6.6 (SD = 2.5). The average income was 17,216 rubles earned per month (SD = 4,158) of which they sent home an average of 7,992 rubles per month (SD = 8,715).

TRAIN

TRAIN is a three-session HIV prevention program designed for use on the train from Dushanbe to Moscow. The TRAIN manual was drafted collaboratively by Dr. Bahromov and Dr. Weine with the support of other members of the Tajik-U.S. collaborative team and with input from Tajik migrants and diaspora community leaders in Moscow. The manual is based on the aforementioned empirical findings from the developmental research as well as other research on HIV preventive interventions and on HIV and migration. Conceptually, it integrates the WHO, HIV prevention model (World Health Organization, 2009), with migration theory (UNAIDS, 2001), social ecological theory (Latkin & Knowlton, 2005), peer education theory (Kelly et al., 2006), and a social cognitive model of behavioral change (Bandura, 1994). TRAIN was designed to reduce HIV risk behaviors involving sexual contacts in Tajik male labor migrants by enhancing (a) the migrants’ risk awareness and HIV prevention skills through increasing HIV knowledge, condom knowledge, knowledge about risk behaviors in Moscow, and risk communication and (b)) the migrants’ preparedness for life in Moscow through positive changes in social and cultural orientation, peer communication, family communication, and help seeking.

Written permission to conduct the intervention on the TRAIN was obtained by Tajikistan government’s Department of Railway Transport in response to a written request from the Ministry of Health of Tajikistan. The letter of permission was kept on the train and shown to the border guards at all the crossings.

TRAIN was held in the dining car during the times when meals were not being served (10-12 a.m.; 2-4 p.m., and 8-10 p.m.). The dining car was divided into two rooms, each with eight tables and a total of 32 chairs. For group sessions, we reserved the room closest to the kitchen so as no to be disturbed by others who were coming into the dining car for food. Dr. Bahromov conducted the intervention with the assistance of a male health educator from the Tajikistan HIV/AIDS Prevention Center. Both were fluent in Russian and Tajik and had prior experience conducting training and research with Tajik migrant population. TRAIN was conducted with three groups of 10 persons each.

TRAIN was a time-limited intervention of three sessions over 3 days of a journey that lasted 4 days. Each meeting began with casual conversation over tea for 15 minutes, then 90 minutes of didactic presentation, role-playing, and group discussions and concluded with a 15-minute wrap-up. TRAIN included standard HIV education and awareness intervention components with a focus on Tajik labor migrants, including migration orientation (and specific HIV risk behaviors for migrants in Moscow such as concurrent sex), information about help seeking with community supports in Moscow, spousal communication about HIV prevention, and communication with community members about HIV prevention.

Session topics were as follows: Session 1: Tea and Welcome, Introducing TRAIN, Getting to Know One Another, Understanding Risks Facing Tajik Migrants in Moscow; Session 2: Tea and Welcome, Social Life of Migrants, Understanding HIV/AIDS, Using Condoms Correctly and Consistently; Session 3: Tea and Welcome, Talking About HIV Prevention with Family, Talking About HIV Prevention with Peers, Review of TRAIN.

PROCEDURES

Each train to Moscow consists of approximately 20 passenger rail cars (11 Platskartny [third class] cars of approximately 54 migrants each and 9 Kupe [second class] cars of approximately 36 migrants each) with a total of 630 passengers. We randomly selected five cars to participate in the intervention and five cars to serve as controls. On each car we created a list of passengers to randomly identify six migrants who met research criteria.

Oral informed consent to participate was obtained as approved by the Institutional review boards of the University of Illinois at Chicago and the Ministry of Health of Tajikistan. We did our best to find a space to speak individually with potential subjects (either in their berth or in the corridor), then verbally described the study to them, answered their questions, and invited them to participate in the study. Subjects were also told that the researchers would keep their participation confidential. Because of the crowded conditions on the train, subjects were also told that other people on the train would know that they were participating in a research project.

Those persons who gave oral consent were asked to complete the pretest survey, and upon completion men in each of the cars were notified whether their car had been randomly assigned to the intervention or to the control group. Subjects in the group were told that we would ask that group participants not talk with others about what participants said in the group.

Table 1 shows that there were no significant sampling differences between the TRAIN and control groups. Those selected to participate in the intervention were issued an invitation with the exact time and place of the sessions on the train. Participants were each paid a total of $20 if they participated in all three surveys.

TABLE 1.

Demographic Variables for TRAIN and Control Conditions^a

	TRAIN (n = 30)	Control (n = 30)
Age	30.4 (5.9)	28.7 (5.9)
Education
High school	26 (87%)	27 (90%)
College	1 (3%)	0 (0%)
University	3 (10%)	3 (10%)
Region
Dushanbe	12 (40%)	13 (43%)
Khatlon	6 (20%)	7 (23%)
Sughd	6 (20%)	4 (13%)
GBAO	6 (20%)	6 (20%)
Religion
Sunni	24 (80%)	24 (80%)
Ismaili	6 (20%)	6 (20%)
Marital status
Married	27 (90%)	24 (80%)
Single	3 (10%)	6 (20%)
Financial support	6.9 (2.7)	6.3 (2.4)
(# of people)
Income	17,133 (3945)	17,300 (4426)
(Rubles / month)

Note. GBAO – Gorno – Badakhshan Autonomous Province.

^aNo significant differences.

All participants from both control and intervention groups were asked to complete the second survey in 3 days and to participate in a follow-up cell phone interview at 3 months. At the first follow-up assessment, the persons in the control group were asked if they had spoken with persons on the train from the TRAIN group: two (7%) spoke with a participant in the TRAIN group; two (7%) saw handouts from TRAIN group, and eight (27%) spoke about HIV/AIDS with other migrants on the train.

MEASURES

To assess TRAIN we compiled a survey instrument that combined items and scales from existing instruments that have been used with migrants with new items and scales focused on key issues concerning masculinity and migration. The survey consisted of items with either forced-choice (yes/no or Likert scale) responses or numerical response (e.g., estimate the percentage). It was designed to be self-administered or administered by cell phone interview in 45 minutes or less in order to be easily completed by the migrants in transit or at their place of destination. Question-naires were translated into two languages, Russian and Tajik, as almost every Tajik migrant is literate in one or both languages.

The survey consisted of questions in the following realms.

Demographic characteristics

ltems addressing age, educational level, family, work, finances, religion, and languages spoken were adapted from the CAFES and CHAMP surveys (McKay et al., 2004; Weine, Feetham, et al., 2008).

Marriage and Sexual Partners

Items addressing sexual history, HIV testing, condom use, sexual risk behaviors, wife and marriage (both in Tajikistan and Moscow), and sex workers were adapted from the AIDS survey and CHAMP survey (Agadjanian, 2004; McKay et al., 2004; Organista & Kubo, 2005).

HIV/AIDS

HIV/AIDS knowledge was assessed via 18 items, condom knowledge via seven items, worry about AIDS with 4 items, and talking with persons about AIDS and talkint with wife about AIDS with 1 item each. These items were adapted from the HIV knowledge survey and CHAMP survey with alphas .74-.87 (Agadjanian, 2004; Fitzgerald et al., 2003; McKay et al., 2004).

Community and religiosity

These constructs were assessed via scales of seven items for community participation and 7 items for religious activities that were adapted from the community and religiosity surveys on the basis of findings from the developmental research (Agadjanian, 2005; Weine, Feetham, et al., 2008). Alphas were .84 (religious activity) and .92 (community participation).

Group Process

Participants’ satisfaction with the intervention group was assessed via a self-report process measure adapted from the Metropolitan Area Child Study Process Measures (Tolan, Hanish, McKay, & Dickey, 2002).

ANALYTIC APPROACH

To address research question one, we provide descriptive information regarding feasibility and acceptability. To address research questions two and three, we conducted two-way mixed analysis of variance (ANOVA) tests comparing the initial and two postassessment values of the outcomes and potential moderators. To address research question four we tested for moderation using random intercept models (Hedeker & Gibbons, 2006). In this analysis, percentage of condom use (with sex workers and with non-sex workers) was the outcome variable. The pre-test value of condom use was included in the model as described above. The dependent variable we modeled was condom use at 3 months. All covariates introduced into the model were time-varying covariates, including AIDS knowledge, condom knowledge, worry about AIDS, talking with persons about HIV, and community participation. The likelihood ratio test was used to test for statistical difference between nested models that include all of the parameters of an initial model (A), plus additional parameters of an expanded model (B). The significance of the likelihood ratio test was calculated by comparing 2(logL_B−logL_A) to the chi-square distribution with degrees of freedom equal to the number of additional parameters in the initial model. Note that this analysis did not compare the various models with different covariates to one another but with the linear model.

RESULTS

FEASIBILITY

Recruitment, screening, obtaining informed consent, initial assessment, randomization and assignment to intervention and control groups proceeded without any problems. The overall implementation of the TRAIN groups went well; however, conditions on the train posed some logistical problems. When the train crossed a border, for example, all passengers had to be in their seats preparing documents for the passport check, making it necessary to interrupt intervention activities several times. Another difficulty concerned conducting sessions in the dining car, as sometimes the dining staff disturbed the group activities. These problems could be managed by scheduling the sessions at appropriate times.

ACCEPTABILITY

All participants came to all intervention sessions and stayed for the entire session. Table 2 indicates that in surveys conducted at both follow-up points the participants were highly satisfied with the intervention. All mentioned that they wanted to participate in another TRAIN group. Additionally all participants stated they would share the knowledge gained in TRAIN with their friends and peers. Another likely indicator of their satisfaction is that each of the participants completed all the postassessments.

TABLE 2.

Assessment of Group Process for TRAIN Participants^a (n = 30)

Items	Post 1b	Post 2c
I am better prepared for life in Moscow because of TRAIN.	3.8	4.0
I am better prepared to protect myself from AIDS because of TRAIN.	3.7	4.0
Other group members have caused me a lot of problems (reverse scored).	3.1	3.1
I got good ideas from other migrants.	3.5	3.5
TRAIN talks about things I can use in Moscow.	3.7	3.7
I am going to try some ideas I got from TRAIN.	4.0	4.0
The group meetings were helpful.	3.7	3.7
I would attend more TRAIN meetings	3.7	4.0
Train leaders understand what it is like to be a Tajik migrant.	3.9	3.9
I talk with my wife about what I learned in TRAIN.	3.6	3.9
I talk with other migrants about what I learned in TRAIN.	3.7	4.0
Migrants are better at making decisions because of TRAIN.	3.7	3.9
Total score	43.2	44.8
	(90%)	(93%)

Note.

^aNo significant differences.

^bPost 1: Immediately after intervention (assessed on train).

^cPost 2: Three-months after intervention (assessed in Moscow).

ANALYSIS OF POSSIBLE EFFECTIVENESS

Table 3 summarizes the analyses for possible changes in condom use and other variables as a consequence of TRAIN. In comparison with the controls, the TRAIN group demonstrated significant increases in condom use with sex workers (p < .001) and non sex-workers (p < .001), with no differences in the number of vaginal sex partners in the past month or use of sex workers in the past month.

TABLE 3.

Assessment of Outcome Variables for TRAIN and Control Conditions

	TRAIN (n = 30)		Control (n = 30)		Two-way Mixed ANOVAc
Variable	Pre	Post 1a & 2b	Pre	Post 1a & 2b
HIV/AIDS knowledge	6.4(1.6)	7.4(0.7) 7.2(0.8)	6.4(2.0) 6.3 (1.9)	6.3(1.9)	p <.1799
Condom use knowledge	9.0 (1.6)	11.3(0.7) 11.3(0.7)	10.4 (1.2)	10.4 (1.2) 10.7 (1.3)	p <.0001
Worry about HIV/AIDS	2.7(3.0)	3.9(0.5)	2.9(1.2)	3.1(1.1)	p <.0100
Talk with person about HIV/  AIDS	13(43%)	23(77%)	20(67%)	6(20%)	p <.0001
Talk with wife about HIV/  AIDS	6(20%)	23(77%)	5(17%)	6(20%)	P <.0005
Community activities	8.8(1.4)	13.0(0)	9.9(1.0)	10.5(1.6)	p <.0001
Religious activities	17.8(5.0)	19.0 (3.8)	15.4(4.7)	15.2(4.5)	p <.0396
Vaginal sex past month	7.0 (2.5)	3.7 (0.9)	7.6 (4.2)	3.7 (0.6)	p <.491
Sex with sex worker past  month	29(97%)	29(97%)	26(87%)	26(87%)	p <1.0
Condom with non-sex  worker past month (%)	0%(0)	33% (44)	0% (1.8)	8% (23.0)	p <.0001
Condom with sex worker  past month (%)	2% (15.2)	55% (44.2)	5% (15.2)	13% (34.6)	p <.0001

Note.

^aPost 1: Immediately after intervention (assessed on train).

^bPost 2: Three months after intervention (assessed in Moscow).

^cCondition × Time (Post 2).

The TRAIN group also had increases in condom knowledge (p < .001), worry about AIDS (p < .010), talking with persons about HIV (p <.001), talking with wives about HIV/AIDS (p < .0005), community activities (P2: p < .001), and religious activities (p < .0396). There were no differences in HIV/AIDS knowledge (although t tests demonstrated significant increases (p < .05) at both postassessments for the intervention but not the control condition).

ANALYSIS OF POSSIBLE MODERATION

To assess for the possible contributions of key variables, four linear random intercept models were considered. Each involved one time-varying covariate: AIDS knowledge, condom knowledge, worry about AIDS, talking with persons about HIV, and community activities. Each of these models was nested within the linear model that included no covariates. Separate analyses were conducted for condom use with sex workers and non-sex workers. The analyses compared each of the linear models with the covariate to the linear models (which had −2 log L of 1,148.3 and 1,108.5 respectively). The results of these covariate models are displayed in Table 4.

TABLE 4.

Random-Effects Models for Outcomes With Covariates (n = 60)

	AIDS Knowledge			Condom Knowledge			Worry About AIDS			Talking With Persons about AIDS			Community Activities
Parameters	Est.	SE	P<	Est.	SE	P<	Est.	SE	P<	Est.	SE	P<	Est.	SE	P<
Condom use with sex workers
β0 Intercept	−193.9	135.3	0.16	−54.6	104.4	0.6	−53.5	133.3	0.69	−139.9	39.7	.001***	−81.5	53.3	0.14
β1Group	112.5	76.4	0.14	26.6	67	0.69	22.2	72.3	0.76	68.4	25.8	.001***	54.4	20.8	.01**
β2Time	198	105.8	.06*	52.4	82.7	0.53	50.8	124.6	0.68	135.2	27.7	.001***	68.1	65.4	0.3
β3 Covariate	13.2	19.7	0.5	−11.1	24.5	0.65	−11.9	36.3	0.74	61.6	55	0.27	−2	8.3	0.81
β4 Covariate × Group	−9.4	11.2	0.4	5.4	15.2	0.72	6.9	20.3	0.73	−32.4	35.4	0.36	−4.4	6.4	0.49
β5 Covariate × Time	−14.1	14.91	0.34	11.4	19	0.55	12.4	32.7	0.71	−54.4	36.6	0.36	4.7	6.1	0.44
β6 Group × Time	−107.6	57.6	.06**	−24	48.7	0.62	−18.9	65	0.77	−63.7	16.7	.001***	−36.5	26.1	0.17
β7Covariate X Group  × Time	9.1	8.2	0.27	−5.3	11	0.63	−6.9	17.4	0.69	29	23.5	0.22	4.7	6.1	0.44
−2 log L	1128.8**			1117.3**			1148.1			1145			1146.6
Condom use with non-sex workers
β0 Intercept	−61.2	115.8	0.6	69.6	87.1	0.43	12.5	113.1	0.91	−20.2	33.5	0.55	−65.2	45.4	0.16
β1Group	27.8	65.4	0.67	−33.1	55.9	0.56	−8.7	61.4	0.89	5.9	21.7	0.78	28	17.8	0.12
β2Time	60	90.5	0.51	−70.1	69	0.31	−12.8	105.5	0.9	20.2	23.3	0.34	64	55.8	0.25
β3 Covariate	0.38	16.8	0.98	−30.3	20.5	0.14	−18.3	30.8	0.55	−50.7	46.2	0.28	0.93	7.1	0.9
β4 Covariate × Group	−0.38	9.6	0.96	13.9	12.7	0.28	8.7	17.2	0.61	25.8	29.7	0.39	−0.29	5.4	0.96
β5 Covariate × Time	−0.26	12.8	0.98	30.4	15.9	.06*	18.3	27.7	0.51	50.2	30.7	0.11	−0.64	5.2	0.9
β6 Group × Time	−26.7	49.3	0.6	34.1	40.7	0.4	8.9	55.1	0.87	−5.6	14	0.67	−26.8	22.3	0.23
β7Covariate × Group  × Time	0.26	7	0.97	−14.1	9.1	0.13	−8.7	14.7	0.56	−25.3	19/7	0.2	0	-	-
−2 log L	1,091.9**			1,074.8**			1,107.90			1,102.80			1,108.40

Note

^*l<p.

^**p<.05.

^***p<. 005.

AIDS Knowledge and Condom Knowledge

Regarding condom use with sex workers, the likelihood ratio tests comparing the significance of the AIDS knowledge subscale linear model with the linear model with no covariates yielded LRχ² = 1148.3–1128.8 = 19.5, which on 4 degrees of freedom is significant at p <. 05. For condom knowledge, LRχ² = 1148.3–1117.3 = 31.0, which on 4 degrees of freedom is significant at p <. 05. Regarding condom use with non-sex workers in Russia, the likelihood ratio tests comparing the significance of the AIDS knowledge subscale linear model with the linear model with no covariates yielded LRχ² = 1108.5–1091.9 = 16.6, which on 4 degrees of freedom is significant at p <. 05. For condom knowledge, LRχ² =1108.5–1075.8 = 32.7, which on 4 degrees of freedom is significant at p <. 05. This finding indicates that both AIDS knowledge and condom knowledge accounts for differences in condom use with non-sex workers for the intervention group.

Worry About AIDS, Talking with Persons About HIV, and Community Activities

Regarding both condom use with sex workers and non-sex workers, the likelihood ratio tests comparing the significance of each of these subscale linear models with the linear model with no covariates yielded LRχ² differences which were not significant at p <. 05. This finding indicates that worry about AIDS, talking with persons about HIV, and community activities did not account for differences in condom use with sex workers for the intervention group.

DISCUSSION

Findings from this study indicated that TRAIN could be practically implemented on the Dushanbe to Moscow train and that Tajik male labor migrants enthusiastically participated. There were some logistical difficulties involved with performing the intervention in the train dining car and with passport inspections during border crossings, but these appeared manageable.

Conducting the intervention on the train made it clear that Tajik labor migrants in transit face many difficulties posed by structural obstacles. The intervention and its leaders were not able to do anything to mitigate the difficult travel conditions or the vulnerability of the migrants to persons who may be looking to exploit them, both on and off the train. For example, during the train ride the migrants were harassed and bribed by border police from each country. These are examples of serious concerns regarding the human rights of labor migrants that need to be addressed by the responsible authorities (Amnesty International, 2009; Eurasianet, 2006). Despite the inability of TRAIN to address these issues, the intervention made an impact on HIV risk and protection amongst the intervention sample.

This is the first-known example of an HIV prevention trial performed in transit. It indicates that interventions performed in transit should be considered for labor migrants to Russia and other countries. For Tajikistan to avoid a large-scale public heath disaster, migrant workers must have access to effective HIV preventive services (International Organization for Migration, 2008). Delivering HIV prevention education to migrants in transit is a reasonable option to consider, as it presents several advantages compared with interventions delivered in either sending or receiving countries.

Findings from this study indicate that TRAIN was associated with changes in condom use with both sex workers and non-sex workers (e.g., regular partners) in Russia. Given that most men have unprotected sexual contact with both sex workers and regular partners, it is important to assess for both. The intervention did not appear to have an impact on HIV-risk exposure, as there was no evidence of change in the number of vaginal sex partners in the past month, or use of sex workers in the past month. This suggests that if the aim of the intervention is to go beyond diminishing high HIV risk behaviors (e.g., not using condoms with sex workers) and to include diminishing the sexual activities of migrants, then further development and modifications would be required.

TRAIN was associated with changes in condom knowledge, worry about AIDS, talking with persons about AIDS, talking with wives about AIDS, and community and religious activities. Despite the statistical change in communication with wives, in all likelihood, sustained changes in family communication regarding sexuality, HIV/AIDS, or condom use would also likely require intervention with the migrants’ wives, either separately, or with couples. The need for an HIV preventive intervention with migrants’ wives was the conclusion from a prior study of Tajik migrants’ wives (Golobof et al., 2008) and is also supported in the literature on gender and HIV (Gupta, Whelan, & Allendorf, 2003).

The fact that several members of the control group either heard about or read materials from the TRAIN intervention also suggests the possible advantages of in-corporating a social network model, where group participants are actively encouraged and instructed to share HIV prevention messages with others (Kelly et al., 2006). This suggests the possible appropriateness of multilevel interventions for labor migrants, which could include focuses upon labor migrants and migrants’ wives in the sending country, as well as by sex workers in the receiving countries, and diaspora community and religious leaders (Schensul & Trickett, 2009).

Regarding possible moderaters, HIV/AIDS knowledge and condom knowledge were shown to be moderators of the intervention effect, but not worry about AIDS, talking with persons about HIV, or community activities. This finding could reflect the relative importance of providing migrants with HIV prevention knowledge and condom prevention skills, which most have not yet received owing to lack of access to prevention in both Tajikistan and Russia. The failure to achieve significance of other possible moderators and of individual items in the models could be a function of the relatively small sample size of this pilot study or of the ways that the constructs were measured. Further rigorous assessment with a larger sample size is needed to fully evaluate the overall intervention model.

This study had several limitations. First, the small study sample is not necessarily representative of the large and diverse population of Tajik labor migrants in Russia. Second, the measures used to assess outcomes and moderators may be limited by linguistic or cultural misunderstandings. Third, follow-up was short term. Fourth, having a Tajik doctor travel with the migrants might have given them a sense of being protected, leading them to want to please the doctor by giving affirmative responses regarding the intervention.

Well-designed effectiveness studies of TRAIN and other HIV preventive interventions with migrants are needed. This study suggests certain elements to consider in transit. These include (a) a sound conceptual foundation that addresses aspects of migration as well as HIV/AIDS, (b) a rigorous sampling plan to assure randomization across key variables, (c) standardized measures focused on HIV and migration adapted for the sociocultural context, (d) attempts to limit contamination in transit, (e) maintaining intervention fidelity, and (f) longer follow-up assessments to examine possible intervention impact.

Several emerging issues were raised that could also be addressed in further research: (a) When an HIV preventive intervention is undertaken in an environment where the rights of migrants are being violated, how should intervention groups and leaders respond? (b) Regarding spousal communication, is it more appropriate to consider couples interventions or interventions with wives separately? (c) Are there subgroups of Tajik male labor migrants for which more tailored groups would be indicated, such as younger single men? (d) If a social network model were applied on the train, could the intervention effectively reach more migrants?