METHODS

A cross sectional study was conducted at Tejgaon truck stand, one of the largest truck stands in Dhaka, the capital city, from October 1998 to April 1999. The study was approved by the institutional review board of the University of Alabama at Birmingham and the ethics committee of the Bangladesh Medical Research Council. The study group comprised 388 men in the trucking industry (drivers and helpers) working out of transport agencies based at Tejgaon truck stand. For an STD with a relatively low observed frequency of 3%, the estimate of true prevalence would have 95% CI of 1.3% to 4.7% in a sample of 388 subjects.

Owing to the impossibility of obtaining a list of the approximately 10 000 drivers/helpers operating at the truck stand, and the difficulty in locating those individuals, a two tiered sampling strategy was used. Initially, a census was conducted at the truck stand of employment agencies for truck drivers and helpers (a total of 185 agencies). A random sample of 38 of these 185 agencies was selected (using a random numbers table on epi-info). After obtaining approval from the selected agencies to recruit their workers for the study, subjects were recruited from each of the 38 trucking agencies. The recruitment objective was 10 subjects per agency; recognising that some individuals who consented to participate would not later show up at the clinic, more than 10 potential subjects were initially recruited at most agencies. The mean number from each agency who later showed up at the clinic was 10.2, the mode was 11, and the range was from 6 (one agency) to 14 (two agencies). The number of drivers and helpers from each agency varied according to whether drivers or helpers were available on the days of recruitment.

Over a period of 4 months, a team of three field workers visited the selected trucking agencies' office at Tejgaon truck stand to recruit subjects; each agency was visited several times. These offices were appropriate recruitment sites as all drivers and helpers need to go to them to obtain their trip assignments. Drivers and helpers in the office on recruitment days were invited to participate. Subjects who consented to participate were accompanied by a field worker to Paricharja clinic (a non-profit clinic located in the centre of the truck stand) or given an identification card (with the name of the subject's trucking agency and a subject ID number) and requested to come to the clinic on their own within a week.

The rate of refusal was estimated to be approximately 15–20%. This estimate is based on: (i) records of the number of men (22) who either agreed to participate but did not come to the clinic or who came to the clinic but left before it was their turn to be seen by the interviewer, and (ii) an informal estimate by the project field manager of the proportion of men (10–15%) who were approached but who either did not agree to participate or who were unwilling to even listen to a description of the study. As the latter group was not documented by the field workers, the rate of refusal is an estimate. We were unable to gather background data on those who did not participate as they frequently were in a hurry, having neither the time nor inclination to respond to questions.

At the clinic, subjects read (or had read to them) a consent form which they signed or marked with a thumb print. Research assistants then interviewed them about their lifestyle for approximately 40 minutes using a structured questionnaire, and a male physician conducted a comprehensive physical examination. Subjects provided both urine and blood samples at the clinic; the blood specimens were collected using standard venepuncture techniques.

After collection, the blood and urine samples were stored in a refrigerator until the end of the day when they were transported to the laboratory of the department of immunology at the Bangladesh Institute of Research and Rehabilitation in Diabetes, Endocrine and Metabolic Disorders (BIRDEM). There, blood and urine samples were centrifuged to obtain serum from the blood and pellets from the urine. These were stored in a −70°C refrigerator until the time of analysis. For gonorrhoea and chlamydia, DNA was extracted from the urine pellets using a commercial reagent Instagene Matrix (Bio-Rad, CA, USA); PCR and DNA enzyme immunoassays (DEIA) were performed using primer/probe kits (Diasorin srl, Italy). Taq DNA polymerase and deoxyribonucleotides (dNTPs) were used during DNA amplification (Gibco Brl, Life Tech, USA). Serological tests for syphilis were performed using rapid plasma reagin (RPR) with a dilution of 1:16 and Treponema pallidum using haemagglutination tests (TPHA) (Shield Diagnostic, UK). Further classification of syphilis into early latent, late latent, or previously treated categories was not possible because medical records were not available; we, therefore, used a minimum dilution of 1:16 to maximise the likelihood of active infection. Serum was also screened for antibodies to herpes simplex virus 2 (Diasorin srl Italy).

Statistical methods

Descriptive statistics (frequencies, means) were reported and contingency tables constructed. Multiple logistic regression was used to assess associations between infections and potential risk factors. In the regression analyses, conducted separately for infection with any bacterial STD (syphilis, gonorrhoea, or chlamydia) and with herpes simplex 2, the hypothesis driven model was determined by previous conceptualised risk factors for STD infection and pertinent background characteristics. Independent variables in the regression analyses included: (i) background characteristics (age, marital status, income, occupation, years in occupation, working interdistrict or local routes), (ii) recreational activities (alcohol and drug use) that can be conducive to high risk sexual relations, (iii) sexual behaviours (sex with a commercial sex worker, sex with a male partner, condom use, number of years of sexual intercourse), and (iv) markers of past high risk sexual behaviour (number of previous diagnoses with STD, and in the analysis of bacterial STD, infection with HSV-2). The latter were included to provide evidence beyond self reports of high risk sexual behaviour. Standard parametric methods, with the individual as the unit of analysis, were used as no significant differences were detected in sociodemographic characteristics of subjects according to the trucking agency employing them.

RESULTS

General characteristics of the study population

All 388 subjects participated in an interview about their background and lifestyle (table 1). Slightly over half (54%) were single. Almost two thirds were truck drivers and one third were helpers on trucks; 51% of the subjects currently worked primarily on interdistrict routes and 49% primarily on local routes. The majority had worked in the trucking industry less than 5 years. Most had attended school, though only 26% had gone beyond primary school and 54% were able to read and easily understand a newspaper. In formative research at the truck stand all drivers or helpers encountered were Muslim (the general population of Bangladesh is ∼88% Muslim, 10.5% Hindu, 0.6% Buddhist, and 0.3% Christian²²).

High risk sexual behaviours for STD were common. Condom use was very low, with approximately 73% never using, 5% using once, almost 20% using only occasionally, and 3% always or most of the time. The majority (54%) acknowledged having sexual relations with a female commercial sex worker (CSW) in the past year, and almost 40% had three or more sex partners in the past year. While 21% reported having a type of sexual relation (a “physical release”) with a male partner in their lifetime, only 7% had done so in the past year. For the 80 subjects who had ever had a male partner, sexual activities included: anal sex (95%), oral sex (4%), mutual masturbation (26%). Male partners of those 80 subjects included a friend (47.5%), a family member (5%), a male sex worker (15%), a neighbour (28.8%), and others (2.5%). The use of illicit drugs that are ingested or inhaled (particularly ganja and phensidyl) was common; alcohol consumption was also frequent, though to a lesser extent. Slightly over one third of the subjects reported having ever been diagnosed with an STD. Of 123 subjects who had been diagnosed with an STD in the past 2 years, 51% were diagnosed by an allopathic doctor the last time they were infected, 20% by a kobiraj (an indigenous doctor), 14% by self diagnosis, 9% by a pharmacist, 9% a homeopath, and 3% a friend (the percentages total more than 100% because 7% were diagnosed by more than one type of practitioner).

Prevalence of STD

Serological evidence of herpes simplex virus 2 (HSV-2) and syphilis were prevalent in this population; current gonorrhoea and chlamydia rates were at low levels (table 2).

There were only two co-infections with bacterial STD; one person had syphilis, gonorrhoea, and chlamydia and one person had both syphilis and gonorrhoea. Eleven of the subjects with HSV-2 also had a bacterial STD.

Logistic regression for current infection with any bacterial STD

Simple and multiple logistic regression analyses were conducted, examining associations with bacterial STD of (i) demographic characteristics, (ii) sexual risk factors, (iii) lifestyle factors that might contribute to sexual risk by reducing inhibitions (alcohol and drug consumption), and (iv) disease markers of high risk sexual behaviour (table 3).

In both simple and multivariate analyses the only significant risk factor for having a bacterial STD was having sex with a CSW in the past year.

Logistic regression analysis for infection with herpes simplex virus 2

Simple and multiple logistic regression analyses of associations with HSV-2 infections were also conducted (table 4).

In both the simple and multiple logistic regression analyses the only factor significantly associated with HSV-2 infection was being an interdistrict helper. In the multiple regression analysis, men who had a male sexual partner in the past year had much lower odds of infection than men who did not and the association approached significance (p = 0.052).

DISCUSSION

Truck drivers and truck helpers at Tejgaon truck stand had a higher prevalence of HSV-2 (25.8%) and serological syphilis (5.7%) than of gonorrhoea (2.1%) and chlamydia (0.8%). This compares with a population based study of 540 male slum dwellers in Dhaka that found higher rates of syphilis (11.5%), lower rates of gonorrhoea (1.5%) and similar rates of chlamydia (<1%).¹⁸ We believe the higher syphilis rate in the Dhaka slum dwellers is because of different testing procedures. In our study, to define seropositives the RPR had a dilution of 1:16, making it a more specific testing procedure with fewer false positives.

In the multivariate analysis the only significant association with bacterial STD was having sexual intercourse with a female CSW in the past year (OR=3.54, CI=1.29 to 9.72, p=0.014). Most of these bacterial infections were syphilis and high rates of syphilis infection have been found in studies of CSWs in Bangladesh.^23,24 Reported condom use was not protective, probably because of limited variation; only 105 subjects (27%) had ever used a condom, and out of these 94 had used them either occasionally or only once.

For HSV-2, the odds of infection were almost three times as high for interdistrict helpers as for interdistrict drivers (OR=2.51, CI=1.13 to 5.55, p=0.023). Drivers have a much higher income than helpers and thus may frequent a different type of CSW and different sexual networks. Of the 29 interdistrict helpers who had relations with CSWs, 69% went to floating CSWs compared to 20% of the 88 interdistrict drivers who had CSW exposure. Evidence from several studies in the country show higher rates of syphilis infection in floating sex workers than in brothel based sex workers^10,23,25; specific HSV-2 data do not exist. This suggests that men who have relations with floating sex workers may be at higher risk for at least certain types of sexually transmitted diseases. That interdistrict helpers were not also at higher risk for bacterial STD, may be attributable to our study population having much lower levels of bacterial infections than of HSV-2, making it harder to detect significant associations with risk factors.

The sexual behaviours of the truck drivers and helpers in this study are consistent with those of truck drivers in other contexts.^26,27 Sexual relations with CSWs were common: 54% of the subjects had sex with a CSW in the past year (57% of unmarried subjects, 43% of married subjects). Further, subjects frequenting CSWs were also having sex with other women: 50% of the married men in the sample had sex with both a CSW and their wife in the past year, and 25% of the 211 men who had sex with a CSW in the past year had 10 or more sexual partners in the past year. In addition, 82% of the 28 subjects who had a male sexual partner in the past year also had sex with a female CSW in the past year.

Few men used condoms. Only 24% of married subjects who had sex with a CSW in the past year had ever used a condom with a CSW (illustrating the risk of disease transmission to their wives). In male to male sexual relations, condom use was also rare, even though anal sex was the normative practice in those relations; only one of the 28 men who had a male partner in the past year had ever used a condom with a male partner. Condom use was low despite condoms being widely available in Bangladesh through government field workers, shops, and pharmacies. The most frequent responses given for not using condoms were that sex felt better without a condom (38%) and they did not see the necessity of using a condom (24%). Only 15% of our study subjects had ever used a condom for the particular purpose of preventing sexually transmitted diseases and only 13.7% were aware that condoms could prevent AIDS.

This study had several methodological limitations. The sample size of 388 may not have been sufficient to detect risk factors for STD infection, particularly for current infections. A second limitation was that a fully randomised sample could not be recruited as a comprehensive list of truck drivers employed at Tejgaon truck stop was unavailable. The most rigorous and viable sampling strategy was, therefore, to randomly select trucking agencies and then recruit truck drivers and helpers in those trucking agencies' offices. Given that all drivers and helpers need to go the trucking agencies' offices to obtain their next work assignment, we do not believe that any particular bias was introduced by recruiting subjects present in those offices on the days of recruitment.

A further limitation was that the sexual behaviour data were only available for the past year, especially important in analyses of lifetime exposure to HSV-2. Moreover, these data were collected by self reports in interviews, which present validity and reliability issues because of recall problems and a tendency for social desirability bias.^28–30 In a study in Bangladesh, Caldwell et al pointed to potential validity problems with self reports on sexual practices.¹² With heavily stigmatised practices such as having sex with other men (widely considered a sin) and drinking alcohol or taking recreational drugs (both proscribed by religious dictate and law), underreporting is very likely, though it may have been minimised by the clinical and private context of the interviews for this study.

The finding of the study indicating a high prevalence of HSV-2 and serological syphilis is a concern, especially because the association of these diseases with HIV transmission.^31,32 Should HIV enter into this low prevalence population it could spread rapidly. These results illustrate the importance of promoting condom use to men in the trucking industry; they clearly need to be made more aware of the role of condoms in preventing STD and HIV and of the importance of disease prevention. Public health campaigns targeting men in the trucking industry should strive to increase use in contexts of casual and, in particular, commercial sexual encounters.