Bangladesh Behavioral Surveillance Surveys, 1998-2000

This document presents valuable data about HIV/AIDS-related knowledge, attitudes, and behaviors, based on the behavioral surveillance surveys conducted in Bangladesh 1998- 2000.

Introduction to Behavioral Surveillance Surveys

FHI's Behavioral Surveillance Surveys (BSS) provide valuable data about HIV/AIDS-related knowledge, attitudes and behaviors. The BSS methodology is a monitoring and evaluation tool designed to track trends in HIV/AIDS-related knowledge, attitudes and behaviors in sub-populations at particular risk of HIV infection, such as female sex workers, injection drug users (IDUs), migrant men and youth. Based on classic HIV and sexually transmitted diseases (STD) serologic surveillance methods, BSS consist of repeated cross-sectional surveys conducted systematically to monitor changes in HIV/STI risk behaviors. A key benefit of the methodology is its standardized approach to questionnaire development, sampling frame construction and survey implementation and analysis. BSS findings serve many purposes: They yield evidence of project impact, provide indicators of project success, highlight persistent problem areas, identify appropriate intervention target populations, identify specific behaviors in need of change, function as a policy and advocacy tool and supply comparative data concerning behavioral risks.

BSS have been conducted in more than 20 countries — primarily in Africa and Asia — since 1992, and they are increasingly being used in Latin America and the Caribbean. Since 1999 they have been used in cross-border sites in Asia and Africa, where they are proving beneficial for understanding the pandemic from a regional instead of a purely country-specific perspective. In several countries, multiple rounds of BSS already have been implemented, with the trend data used to formulate new programs and to adapt existing ones.

Introduction to Bangladesh BSS

Bangladesh has been considered a low prevalence country for many years, since the first case of HIV infection was recorded in 1987. Debates about the future course of the epidemic required sound data for decision-making. In 1998 the Government of Bangladesh (GoB), with support from UNAIDS and technical advice from FHI, undertook the first wave of expanded surveillance for HIV. This included screening for HIV, syphilis and risk behaviors among selected high risk groups. These surveys were executed by the International Centre for Diarrhoeal Disease Research, Bangladesh (ICDDR,B) on behalf of the Bangladesh AIDS Prevention and Control Programme (BAPCP). The next wave took place in 2000, through CARE, Bangladesh, and a third wave is in progress at the time of writing.

The Bangladesh Behavioral Surveillance Surveys (BSS) serve as a tool to inform the National AIDS Program as well as donors of the risk factors among various groups, to advocate for increased investment in prevention and to measure the progress of various interventions. This report highlights the findings from the first and second waves of the Bangladesh BSS, conducted in 1998 and 2000.

Study Design and Methodology

The first wave of surveillance was an exercise in setting up sentinel sites and populations. The BSS was executed mostly as convenience samples of high risk groups, with the exception of brothel sex workers. The second wave began probability sampling for all groups, except hijras, while the third wave is designed to execute probability sampling in all groups. Over the course of the first three waves of BSS, the types of groups have been expanded, sites added and methods and indicators have been improved and standardized.

Study Populations

For both practical and ethical reasons, the National Surveillance Advisory Committee considered whether a prevention program existed or was planned for each group before including it. In the first wave, the populations included were IDUs, female street-based sex workers, men who have sex with men (MSM), long-distance truck drivers and their helpers, and brothel sex workers. A small test survey was conducted of hijras. The second wave included IDUs, female street-based sex workers, male sex workers, hijra (transgender) sex workers, rickshaw pullers and brothel sex workers.

Study Sites

Because Bangladesh has registered brothels throughout the country (18 in 1998 and 15 in 2000), these were sampled nationally in both waves. Previous studies had shown that the highest concentrations of IDUs were in Dhaka and Rajshahi, a city in N. Bengal bordering India. Street-based female sex workers were most accessible in Dhaka and Chittagong, the nation's largest port. Long-distance truckers (drivers and helpers) were found in and around Dhaka. MSM were interviewed at cruising sites, but during the second wave only male sex workers (MSW) were included. Rickshaw pullers, an important client group for sex workers, were sampled in Chittagong. Hijra sex workers were easier to find in Dhaka than elsewhere.

Table 1. Surveyed populations, sites and sample sizes, 1998-99 and 2000

Survey populations	Site	1998-99	2000
Brothel sex workers	National	1147	867
Street-based sex workers (female)	Dhaka	518	583
Street-based sex workers (female)	Chittagong	-	521
Street-based sex workers (male)	Dhaka	207	582
MSM (at cruising sites)	Dhaka	200	-
Hijras	Dhaka	150	336
Truckers	Dhaka	411	-
Rickshaw pullers	Chittagong	-	411
IDUs (street-recruited)	Dhaka	430	682
IDUs (street-recruited)	Rajshahi	450	512

Sample size

Sample sizes were determined for the first wave based on estimates from previous studies of key indicators (e.g. a measure of condom use or needle sharing) and the degree of confidence required to detect a change of about 10 percent in the next wave of surveillance. During the second round, sample sizes were adjusted to reflect findings of the first round.

Sample design

Maps were constructed of the locations at which target groups could be found by teams consisting of members of the target groups and others. During the first wave, interviewers then visited these locations at specific times of day when they could find the required participants. They continued until reaching the required sample size. During the second wave, mapping was more systematic and worked ward-by-ward through each city, recording how many target group members were seen. In addition, they sampled and inquired about differences by day of week and time of day, seeking to know if there were great differences in this regard. This was required because the second wave used a time-location sampling strategy. Where there were real differences -- for example, many more sex workers on a particular day than during the rest of the week -- the site was considered to comprise two primary sampling visits (PSUs). Each different time-location combination received a unique PSU number and these were randomly selected to compose the final sample. Teams then went to the site for four-hour periods and attempted to interview all target group members present. Lists were kept of the numbers of persons seen and interviewed, as well as those who were not interviewed, either because they had already been interviewed (duplicates), simply refused or left before the interviewer had time to reach him or her. Time-location sampling is a little more complex than simple convenience sampling, but is an excellent way to produce probability samples of hard-to-reach groups. This allows results to be generalized more reliably to the group as a whole.

The brothel sample was constructed differently and the same way in both waves. As the brothels were located all over the country, time and cost for travel had to be conserved. Estimates were made before visiting of the number of sex workers in each brothel, based on all available information from NGOs, previous and current research. The total number required for the sample size was then divided proportionately among the brothels and a target sample size per brothel estimated. Teams of interviewers then visited each brothel and spent the first day conducting a room census. Depending on the sample size needed, every second, third or fourth used room was selected. The number of sex workers staying in each room was recorded and, using a small device for randomization called a Kisch card (printed on the questionnaire), one of the women was randomly selected for interview. When all samples were completed, there were a few cases in which the original estimates were incorrect and a single person returned to those brothels to acquire a few more interviews.

Interviewers were selected to participate based on their experience with the target groups. Many were members of these groups themselves–such as ex-addicts, sex workers, hijras or MSM–while others had worked previously with these groups as researchers or NGO workers. Interviewers were trained for at least five days so they would understand how to talk about sex, needle use and other sensitive topics. A structured questionnaire, previously tested for clarity, was administered. All respondents were read a consent statement to which they had to agree before the interview and were given a short educational talk with condom demonstration following the interview.

Results

The main socio-demographic characteristics are shown below in Table 2. The MSM sample in 1998 is separated into MSW and non-sex working MSM due to marked differences between these subgroups.

Table 2. Main socio-demographic characteristics of samples

Group	Age (mean)		Years Ed (mean)		Currently Married (%)		Yrs in Sex Work (mean)		Years Injecting (mean)
	Wave 1	Wave 2	Wave 1	Wave 2	Wave 1	Wave 2	Wave 1	Wave 2	Wave 1	Wave 2
Brothel SW	22.1	24.0	0.9	0.8	3.0	1.3	4.9	5.8	-	-
Female Street SW —Dhaka	21.8	22.5	1.3	1.3	28.0	8.1	3.3	3.3	-	-
Female Street SW-Chittagong	-	22.8	-	1.6	-	25.5	-	3.8	-	-
Male sex workers-Dhaka	22.2	24.9	7.8	7.8	3.4	13.3	-	8.9	-	-
MSM -Dhaka	33.2	-	11.7	-	48.0	-	-	-	-	-
Hijra SW —Dhaka	28.0	26.2	2.8	3.8	9.0	6.0	7.7	9.8	-	-
IDU- Dhaka	32.3	35.6	2.3	2.4	42.0	65.7	-	-	3.3	5.2
IDU-Rajshahi	34.8	34.5	4.4	4.0	79.0	73.6	-	-	5.0	4.3
Truckers	27.0	-	3.7	-	48.0	-	-	-	-	-
Rickshaw pullers	-	29.2	-	1.6	-	67.9	-	-	-	-

With the exception of non-sex working MSM, literacy and educational levels are very low in all of these groups, a major social factor contributing to their vulnerability.

Knowledge indicators

Table 3. Proportions of each group who could mention any of the main modes of HIV transmission.

Group	Male-female sex		Male-male sex		Shared needles		Blood transfusion		Mother-to-child
	Wave 1	Wave 2	Wave 1	Wave 2	Wave 1	Wave 2	Wave 1	Wave 2	Wave 1	Wave 2
Brothel SW	59	72	1	0	13	36	12	37	3	14
Female Street SW —Dhaka	44	72	11	1	10	39	20	28	6	15
Female Street SW-Chittagong	-	55	-	6	-	3	-	2	-	4
Male sex workers-Dhaka	43	72	55	66	33	47	29	50	5	17
MSM -Dhaka	64	-	63	-	49	-	52	-	34	-
Hijra SW —Dhaka	39	30	18	24	3	5	5	13	2	0
IDU- Dhaka	45	64	3	10	60	92	5	33	2	13
IDU- Rajshahi	33	47	14	9	50	76	14	7	2	2
Truckers	52	-	6	-	4	-	3	-	0	-
Rickshaw pullers	-	25	-	5	-	2	-	3	-	0

Levels of knowledge remain very low, though there appears to be improvement in male and female sex workers and IDUs. Similarly, very few considered themselves at high risk for HIV infection--fewer than 2 percent in any group during wave two, except hijras at 11 percent and street workers in Chittagong at 9 percent. Generally, the data showed a great deal of confusion and denial concerning their own assessment of risk.

Behavioral indicators

Due to differences in sampling in most groups between waves, sound statistical comparisons on behavioral indicators can only be calculated for brothel sex workers. Nonetheless, the principal behavioral indicators for all groups are shown in Table 4 for sex workers and Table 5 for client groups. Table 6 shows specific risks related to drug use for IDUs. Brothel sex workers are compared across waves in the following figures.

Table 4. Main indicators: Sex Workers

Groups	Mean no. of clients last week		% used condom last time with client		% 100% condom use last week with clients		% able to show condom to interviewer		% 100% condom use last week with personal partners
	Wave 1	Wave 2	Wave 1	Wave 2	Wave 1	Wave 2	Wave 1	Wave 2	Wave 1	Wave 2
Brothel sex workers	16.3	18.5	22.0	21.0	3.3	<1	47.0	52.8	9.0†	2.4
Street sex workers-Dhaka	9.2	12.5	8.0*	24.5	19.0	1.0	15.0	25.8	10.0	2.1
Street sex workers-Chittagong	-	13.6	-	24.2	-	4.1	-	15.6	-	18.4
Male sex workers-Dhaka	3.0	6.2	25.0	41.6		2.7	26.0	55.9	14.0‡	3.6‡
Hijra sex workers	13.3	12.9	5.0	9.4	0	0	5.0	11.0	3.0‡	<1‡

* for all clients yesterday.
† for main personal partner only.
‡ for male and hijra sex workers only; the figure refers to last time.

Despite some minor differences in indicators, the data as a whole show little significant change in safe sex behaviors, except perhaps for male sex workers. Sampling differences, however, do not allow statistical comparisons, except for brothel sex workers, as shown in the following figures.

Additional indicators collected in Bangladesh aimed at monitoring the level of participation in HIV prevention programs as well as the prevalence of violence against sex workers, an issue recognized as one that perpetuates a dangerous and insecure environment in which to learn about and practice safer sex. In 2000, participation in HIV prevention programs was reported by 48 percent of Dhaka street sex workers, none in Chittagong, 22 percent of brothel women, 43 percent of male sex workers and only 2 percent of hijras. Also during wave two, the surveys revealed that as many as 60 percent of the Dhaka street sex workers, 52 percent of those in Chittagong, 18 percent of the women in brothels, 60 percent of hijra sex workers and 16 percent of male sex workers had been forced into sex by police or street thugs the previous year. These figures are important for developing advocacy to address the severe marginalization of these groups, a major component of their vulnerability to HIV infection.

Table 5. Main indicators: Client groups

Groups	% went to sex worker last month		% used condom with sex worker last time		% with non-regular,non-commercial partner last month		% used condoms with non-regular, non-commercial partner last time		% bisexually active last year
	Wave 1	Wave 2	Wave 1	Wave 2	Wave 1	Wave 2	Wave 1	Wave 2	Wave 1	Wave 2
Truckers	36.0	-	14.0	-	13.0	-	11.0	-	11.0	-
IDUs-Dhaka	18.0	32.3	10.0	15.1	7.0	10.0	8.0	19.2	7.0	8.5
IDUs-Rajshahi	15.0	27.8	20.0	25.0	5.0	27.5	24.0	18.6	2.0	3.3
Rickshaw pullers	-	93.2	-	22.2	-	29.8*	-	30.1	-	61.2

In addition to their role as a bridging group between commercial sex workers and the general population, bisexually active men may intensify transmission through the common practice of unprotected anal intercourse. Among the 21 percent of rickshaw pullers reporting anal intercourse in the past week, only 17 percent used condoms. A high proportion of the bisexual activity among truckers took place with hijras who themselves report extremely low condom use. Anal intercourse as a risk factor is not confined to male-to-male sex; 46 percent of Dhala's female street sex workers and 39 percent of the brothel sex workers also report anal intercourse the previous week.

Table 6. Main indicators for IDUs: needle sharing behavior and participation in needle exchange programs.

Groups	Mean no. injections last week		% who shared at all last week		Mean % of injections shared last week		% participating in needle exchanges
	Wave 1	Wave 2	Wave 1	Wave 2	Wave 1	Wave 2	Wave 1	Wave 2
IDUs-Dhaka	20.4	15.8	93.0	74.8	62.0	14.1	0	81.0
IDUs-Rajshahi	21.7	17.1	96.0	55.5	73.0	16.3	0	55.9

Despite sampling differences across waves, the strong effect of needle exchanges on overall sharing is evident. Consistent non-sharing is less reduced than the mean percent of injections. Subsequent waves of BSS are required to confirm this effect.

Brothels in Bangladesh have been gradually closing through the years, and three closed between the first and second waves of BSS. Sampling methods were the same across waves and allow comparison. Figure 1 shows some of the main indicators for both waves in the brothels, indicating small incremental changes in any use and access, but not in consistent use of condoms. The availability of fewer brothels has led to higher numbers of clients per sex worker.

Figure 1. Changes in main indicators among brothel sex workers

Conclusions

• While Bangladesh remains a low HIV prevalence country, ample risk behaviors have been documented requiring the rapid mobilization of resources for many at-risk groups.
• Trends are showing little change in consistent condom use between waves one and two among female sex workers, but more sex workers are using condoms.
• Male sex workers appear to be improving their levels of safer sex in response to prevention programs.
• All sex workers experience considerable levels of violence, with the least among those in brothels.
• The clients of sex workers so far sampled exhibit high levels of risky sexual behavior, suggesting their potential role as bridging groups.
• IDUs are learning about the danger of sharing needles and are utilizing need exchanges. Consistent non-sharing is still low, especially in Dhaka.

Acknowledgments

The Bangladesh BSS were executed by:

International Centre for Diarrhoeal Disease Research, Bangladesh (ICCDR-B)
CARE, Bangladesh

Administered by:

Bangladesh AIDS Prevention and Control Programme (BAPCP); Institute of Epidemiology, Disease Control and Research (IEDCR)

With technical assistance from:

Family Health International

Funded by:

UNAIDS
Department For International Development (DFID)
US Agency for International Development (USAID)
Family Health International

This executive summary is based on the following reports:

• Report on the Sero-Surveillance and Behavioural Surveillance on STD and AIDS in Bangladesh 1998-1999. AIDS and STD Control Programme, Directorate General of Health Services, Ministry of Health & Family Welfare, Govt. of the People's Republic of Bangladesh.
• Report on the Second National Expanded HIV Surveillance, 1999-2000. Bangladesh, Aids and STD Control Programme, Directorate General Of Health Services, Ministry of Health & Family Welfare, Govt. of The People's Republic Of Bangladesh. September, 2000.