The study was conducted from April 2005 to December 2005. Preliminary qualitative information was collected through census of the fishing spots and in-depth interviews of both key informants as well as boatmen themselves. This information helped guide the development of the questionnaire for the quantitative survey. Survey data was collected through structured interviews.
Preliminary data collection
As a part of rapport building, initial meetings were conducted with relevant people of the community. These meetings allowed the study team to identify local vulnerable groups and key informants who were linked with boatmen. Follow-up meetings were conducted with selected key informants during the process of data collection.
A mapping exercise was conducted to construct a census of boats and boatmen in Teknaf. Information was collected through observations made by trained interviewers. Preliminary information on number, location, and types of 'ghats' (landing areas/platforms for boats) was obtained from key informants. Field workers visited each 'ghat', counted boats, boatmen, and made observation for four to five hours at each ghat to identify the types of boatmen according to their activities. Each 'ghat' was observed twice. Mapping information was crosschecked with prior information obtained from key informants.
The mapping exercise identified three types of boatmen based on their activities: non-fishing boatmen, day-long fishing boatmen, and deep sea fishing boatmen. Thirty-three non-fishing boatmen from fifteen boats were identified at two ports in Teknaf. These boatmen originate in Teknaf and travel to Myanmar transporting either passengers or goods. Day long fishing boatmen were defined as non-migrant boatmen who fish in the local areas but do not go to the deep sea for fishing. Deep sea fishing boatmen are those who spend 7–10 days in the deep sea on fishing expeditions.
In-depth interviews were conducted with selected key informants; elderly boatmen, boat owners, NGO workers, male and female pimps of sex workers, journalists, hotel managers, and border traders. Information provided by different key informants was compared to identify contradictions and points of consistency regarding information provide.
Another series of in-depth interviews were conducted in order to further elaborate the risk behaviour linkages of Teknaf boatmen and obtain more information about their lifestyle and day-to-day experiences. These interviews were conducted with members of the following groups: boatmen (n = 17), spouses of boatmen (n = 11), female commercial sex worker (CSW) from different venues, including street, hotel, and home-based workers (n = 31), injecting drug users (n = 11), and transport workers (n = 5). In-depth interviews were conducted using flexible guidelines, which were developed after obtaining preliminary information from mapping. The recorded interviews were transcribed. Content analysis was done manually with the transcripts. Similar themes and sub themes were identified and categories. To describe linkages between patterns, themes, and experiences, we developed a data matrix with different cells to reflect the relationships identified within the data.
Cross-sectional survey
For the sample size estimation, we assumed 3% condom use rate, as last time condom use rate during sex with CSW by transport worker was reported to be 3% [19], with precision of +/-2%, and a level of confidence of 95%. Based on these parameters, the sample size required was 279. Assuming a non-response percentage of 5%, 294 boatmen would need to be contacted. Values of other key indicators, such as sexual intercourse with CSW, are not known. However, a sample of 279 interviews would provide adequate precision with an estimated prevalence of sex with a commercial sex worker of 50% (+/-5.8%) and greater precision on lower prevalence estimates. As mapping exercises identified 33 non-fishing boatmen and about 900 deep sea fishing boatmen, we attempted to enroll all non-fishing boatmen and one third of the deep sea fishing boatmen. There were two ports for deep-sea fishing and each deep sea fishing boats consisted of three types of boatmen such as pilot, enginemen, and other crewmembers. There were 10 to 12 boatmen in each boat. We targeted 25 boats from each port and recruited 6 boatmen from each boat covering three types of crewmembers. We also enrolled 104 non-migrant local boatmen who were involved with the local fishing trade to have a comparison as we assumed that they have lower risk of HIV infection due to cross border mobility.
Data was collected by 8 trained interviewers having minimum a bachelor degree, who were previously involved in the national behavioral surveillance for HIV. In addition, fifteen days intensive training was given to the data collectors on vulnerable groups in the study site, risk behaviour for HIV, mode of transmission of HIV infection, and data collection techniques including data collection tools. The cross-sectional survey collected demographic information, looked at boatmen's mobility patterns, and assessed boatmen's knowledge about HIV transmission and their own perceptions of risk. The survey measured vulnerability through data regarding boatmen's sexual risk behavior, including number, gender, and types of sexual partners and condom usage with different types of partners. The survey questionnaire consisted of seven sections. The first section was on socio demographic profile of the boatmen. The second section inquired about knowledge of boatmen related to HIV infection particularly on modes transmission, risks, and prevention. The third section included questions regarding experience of symptoms of sexually transmitted infections and its care. In section 4, use of condom was investigated. Section 5 and 6 addressed the issues of sexual practices and sexual partners. The last section was on sexual violence.
Boatmen were defined as males aged 18 and over who had worked in the past six months as crewmembers in a boat based in Teknaf. Boats were defined as those used primarily for fishing for transport of passengers, or for trade of goods. Participation in the study was on a voluntary basis. All participants in the study received a simple explanation about the objectives of the study and oral consent was obtained prior to the in-depth qualitative interviews or the quantitative survey. Confidentiality was strictly maintained throughout the interview process. Permission to tape record interviews were obtained from the participants prior to the interview took place. During the few in-depth interviews where tape recording was not possible detailed notes were taken instead. All surveys were conducted anonymously. Prior to the implementation of the study, official approval was obtained from the Research Review Committee (RRC) and the Ethical Review Committee (ERC) of ICDDR, B. The ERC is an independent body house at ICDDR, B to look after the ethical aspects of research protocols of the Centre and comprised of experts mostly outside the Centre.
Data was entered using Epi-Info for Windows Version 3. Range and consistency checks were done to ensure accuracy. Data was analyzed using SPSS version 10. Both bi-variate and multivariate analysis were done. Descriptive analysis included preparation of frequency cross-tabulations, calculating means, medians and 95% confidence intervals on estimated proportions for categorical variables. In bi-variate analysis demographic characteristics of boatmen are presented and Student-t test was done to see the overall significance of different categories. Also, chi-square/fisher's exact test performed and crude odds ratios with 95% CI obtained to see the association between sexual intercourse with non-marital partners and other variables such as marital status, age, education, income, type of boatmen, activities of boatmen, knowledge of mode of transmission of HIV, and individual risk perception. In multivariate analysis logistic regression was done to assess the influence of co-factors such as marital status, age, education, income, type of boatmen, activities of boatmen, knowledge of mode of transmission of HIV, and individual risk perception on dependent variable that is intercourse with female non-marital partners in last 12 months. To show statistical significant associations between impendent and dependent variables adjusted odds with 95% CI were calculated and p-values were obtained using logistic regression model.