Thirty percent of Kenyan teenagers in schools in the city of Mombasa report that they have had sex with one or more partners, becoming sexually active at an average age of 15 for boys and 16 for girls. In Cambodia's five provincial capitals, three out of four members of the military and police forces admit using the services of a sex worker in the past 12 months. And 36 percent of truck drivers interviewed along Côte d'Ivoire's main highway say they have had sex with someone other than their wives or steady girlfriends during the previous year.
Getting people to talk directly and accurately about their own sexual behavior can be difficult. But in countries throughout the world, well-trained survey interviewers are doing just that in an effort to better understand the behaviors driving HIV epidemics and to assess whether interventions to reduce risky behavior are working. And people from all walks of life are opening up to these interviewers, revealing details about the most private aspects of their lives, in hopes of helping to curb the spread of the deadly virus that causes AIDS.
The data from these repeated behavioral surveys among cross sections of selected populations, which have been conducted in more than 10 countries, have proved so useful that the Joint United Nations Programme on HIV/AIDS (UNAIDS) included such surveys in its recommendations for "second-generation" surveillance of HIV/AIDS and sexually transmitted infections (STIs).
The first generation of recommendations had focused on measuring levels of HIV infection in "sentinel" groups, such as clients of STI or antenatal clinics. But experience with these epidemiological surveillance systems showed that HIV prevalence data alone cannot explain trends in the spread of the virus.
"Behavioral data are as essential as biological data for program design, monitoring and evaluation," said Dr. Michel Caraël, UNAIDS prevention team leader.
Standardized guidelines for behavioral data collection, developed jointly by UNAIDS and Family Health International (FHI),1 will help national AIDS programs gain a better understanding of the HIV/AIDS epidemics in their countries, says Dr. Thomas Rehle, FHI's associate director for evaluation, surveillance and epidemiological research. "Behavioral data complement the findings from HIV sentinel surveillance and allow a contextual analysis of the observed trends in HIV prevalence," he explained.
These guidelines recommend that systematic, cross-sectional behavioral surveys be carried out annually or biannually among groups that engage in high-risk behavior in countries at all stages of the HIV/AIDS epidemic.
The most extensive data from this type of survey has come from FHI's application of its behavioral surveillance surveys (BSS) methodology. Since conducting the first BSS in the city of Bangkok in 1993, FHI has assisted governments and local research organizations to adapt and apply the methodology in more than 20 countries.
Stephen Mills, associate director for technical support in FHI's Asia Regional Office in Bangkok, believes that interest in BSS is growing because the surveys fill two important gaps in HIV/AIDS evaluation. "The BSS help us understand whether the combined interventions of different groups are working together to change risk behaviors," he explained. "In the absence of HIV or in low-prevalence settings, they can also give us an early warning of where HIV may appear in the future."
Evaluating Programs
The BSS grew out of FHI's experience in evaluating hundreds of HIV/AIDS prevention projects in more than 60 countries during the past 12 years. Under the USAID-funded AIDS Control and Prevention (AIDSCAP) Project alone, FHI supported more than 165 knowledge, attitudes, beliefs and practices (KABP) surveys among the groups targeted by specific projects.
Although these surveys generated a great deal of interesting information, FHI evaluation specialists concluded that they were not the most efficient or effective way to evaluate the impact of prevention efforts on HIV risk behavior
Instead of measuring indicators of behavior change for many individual projects with overlapping target groups, FHI now consolidates its resources to assess the collective effort against HIV/AIDS in a city or region using a more methodologically rigorous approach than the typical KABP study. As a result, says FHI evaluation officer Christine Kolars Sow, "the data are more useful to more people."
While many welcome this approach, others lament the lack of data directly related to their own projects or activities. BSS data cannot show that any one intervention is responsible for trends in HIV risk behavior -- a limitation, Mills points out, that the methodology shares with KABP surveys. "The only way to show causality is through large and expensive quasi-experimental or experimental designs," he said. "These would rob scarce funds from prevention interventions."
This inability to attribute findings to specific projects is not necessarily a disadvantage, adds Mills, because no HIV/AIDS project works in isolation. "An appropriate strategy is multiple overlapping interventions that hit people from different sides," he said.
To assess a project's contribution, Mills suggests, evaluators can look at its role in a country's overall response to HIV/AIDS and at temporal relationships between specific interventions and behavioral trends. In Tamil Nadu, for example, the size of the USAID-funded AIDS Prevention and Control (APAC) Project and the fact that men in the target groups began to change their sexual behavior during its implementation suggest that the project has played a significant role in reducing HIV risk behavior.
Organizations can also assess whether interventions of any size are making a contribution to the collective effort against HIV/AIDS by monitoring whether their projects are being carried out as planned. FHI recommends collecting and tracking process indicators, such as number of people educated or trained and number of condoms sold or distributed. Actual coverage of target populations could be evaluated through rapid surveys to determine, for example, recall of a program's prevention messages.
The key to ensuring that interventions have an impact, says Dr. Tobi Saidel, technical officer for evaluation and surveillance in FHI's Asia Regional Office, is sound project design and careful project monitoring. "We already know from research that if state-of-the-art interventions are implemented according to recommended standards, they do result in behavior change," she said. "There is no need to prove that every time."
Ownership and Sustainability
In each country, many local partners are involved in designing the BSS and disseminating their results. But usually one organization -- often a private research firm, university or research institute -- is chosen to collect and analyze the data with technical assistance from FHI.
In Cambodia, where contracting with a private company or nongovernmental organization (NGO) was not an option because of the lack of a private sector, the national AIDS program's Planning, Monitoring, Evaluation and Research (PMER) Unit conducts the BSS. Some have urged the PMER to contract out various parts of the surveys, but many observers say the system works well.
Dr. Saidel believes that the PMER's assuming responsibility for BSS in Cambodia makes it likely that the surveys will be sustained. When she began to assist the PMER to coordinate BSS with its HIV surveillance system in 1998, she said, "What I saw was a group of people with a tremendous ownership over the data."
Government HIV/AIDS programs in two other Asian countries -- China, where Mills has served as a World Bank consultant to help design BSS for four provinces, and Vietnam -- are interested in collecting their own data. But others warn that Cambodia's experience is unusual. "It's really unique there because this is a new government, and they don't have a history of mistrust between government and the target groups," said Dr. Pamina Gorbach, an FHI consultant who has worked with the PMER since 1996.
"Often governments don't have very good rapport with disenfranchised groups," agreed Mills, noting that having government employees survey people about sensitive and sometimes illegal behaviors such as commercial sex and drug use could pose problems in many countries. In others, government agencies simply may not have the time or capacity to collect the data.
When the surveys are conducted by NGOs, whether they be private companies or nonprofit institutions, the biggest challenge is ensuring that BSS becomes an integral part of the country's surveillance system. Different models are emerging for various countries, but the common thread is the need to involve national AIDS program officials in designing, reviewing and disseminating results from the beginning.
Ensuring Quality
BSS is designed to gather information from groups rarely captured by traditional household surveys. This makes it much more difficult -- but not impossible -- to achieve a representative sample of the target group.
"With a little bit of extra effort in terms of mapping and finding out about the population in those sites, we can move from non-probability to probability sampling," Dr. Saidel said.
Mapping enables researchers to identify where target groups congregate and to estimate the size of the group in each site. Local organizations working with the target groups may already have lists of sites or maps showing the places where target groups can be found, but even then BSS researchers often need to do further research.
For the Kenyan BSS in the city of Mombasa, for example, lists were available of two kinds of sites where commercial sex workers meet their clients. Using one of the lists, BSS supervisors went to each of the bars and discos frequented by the more expensive sex workers and talked to bouncers and waiters to estimate the number of women working at each site.
"Initially, we did try getting the supervisors to take a physical count, but this was too subjective," said Carol N'Katha Kageenu of Steadman Research Services, the firm FHI hired to conduct the first round of the BSS in Kenya.
Visiting the almost 300 establishments where less expensive sex workers work would have been prohibitively expensive and time-consuming, so most of these estimates were obtained by phone. To learn about the size of this group, Steadman had to rely on the district health officer and his contacts.
"We tried to do this exercise with our own supervisors, but the bar owners were extremely hostile," Kageenu explained. "The bottom line is that you need to work closely with local leaders to obtain useful information."
Once this information on the target groups has been collected, FHI and its partners use creative approaches to probability sampling. A decision tree developed by FHI helps researchers design an appropriate sampling strategy.
Mapping, developing a sampling strategy, testing and adapting questionnaires, and other preparations for the first round of a BSS in a country take about three months. "By taking that extra time, you can get better data," Dr. Saidel said.
No matter how well designed a study may be, self-reported data on sex is often considered suspect. Yet experience with research on sexual behavior and comparisons of behavioral survey data with findings from other types of studies suggest that survey participants tend to tell the truth most of the time.
Ultimately, says Dr. Saidel, "getting people to tell the truth relates to the skill of the interviewers and management of the whole survey process."
BSS supervisors and interviewers must build trust among the target groups, create an environment where people feel confident that their privacy and anonymity will be maintained, and convince target group members that accurate research results will benefit them and their communities.
"You need to take the time to carefully train interviewers and prepare target groups," Dr. Saidel said. "All of this contributes to the validity of the data."
Learning From Results
During the past six years, FHI and its partners have disseminated BSS results and recommendations to various audiences in eight countries through publications, media interviews, and presentations at meetings, workshops and conferences. By tailoring the way the data were presented to meet the needs of specific audiences, they have encouraged policymakers and program planners to act on the recommendations drawn from BSS data.
The results have been encouraging. In most cases, the data are used extensively for program design and evaluation. And in some countries, BSS findings have also been used to develop policies that support HIV prevention and to advocate for policy reform.
Promising trends in risk behavior from Cambodia, Nepal and Tamil Nadu, India, suggest that prevention efforts are on the right track. In Tamil Nadu, for example, annual surveys conducted from 1996 to 1998 found that significantly fewer men reported having sex with anyone besides their wives, girlfriends or other regular partners. Among male factory workers who did have sex with other partners, the proportion using condoms during the last such encounter rose from 17 percent to 50 percent, and condom use with sex workers soared from 28 to 67 percent.
In Indonesia, on the other hand, the results of three rounds of BSS showed little or no behavior change. According to Mills, these trends are not unexpected. "Low prevalence and socioeconomic disruption make behavior change very hard," he said. "In such settings, people do not have HIV on their agenda, and thus it may take longer to produce behavioral change."
Nevertheless, the survey results and findings from qualitative research are being used to revise strategies for containing the HIV/AIDS epidemic in Indonesia. To Dr. Saidel, this experience is a perfect example of the BSS's value as a diagnostic tool.
BSS results are also being used to reorient interventions in Senegal, where the first-round results in 1997 showed that knowledge of HIV was much higher than knowledge of other STIs -- even though the latter are much more prevalent. "The BSS enabled us to verify that sexually transmitted diseases were not well-recognized, and a specific intervention has been carried out among students," said Idrissa Diop, director of Hygea, the research firm that carried out BSS in Senegal.
Data from the second round of the Senegal BSS revealed that only 9 percent of male workers and 5 percent of truck drivers had visited a sex worker during the previous year. No trend data are available for these two groups, which were not surveyed in the first round. But if subsequent rounds of BSS show that this unusually low level of commercial sex patronage is a trend, it may offer a partial explanation for Senegal's relatively low HIV prevalence of less than 2 percent.
The explanatory power of combining HIV behavioral and epidemiological information was perhaps best illustrated in Thailand. An analysis of five rounds of BSS data from Bangkok helped confirm that declines in HIV incidence and prevalence were the result of behavior change.
Combining quantitative and qualitative information can also make behavioral data even more useful for program design and evaluation. In Kenya, focus group discussions and in-depth interviews carried out to explore some of the questions raised by the first-round BSS results yielded a wealth of information about how and why risk behaviors persist. Youth spoke candidly about how a desire to preserve virginity can lead to experiments with unprotected anal sex and made it clear that their fear of pregnancy far overrides fear of HIV infection. Sex workers talked about the pressures of competition from girls as young as 14 selling sex on the street at cut-rate prices. This qualitative information will be used with BSS data to help strengthen behavior change interventions to address the real obstacles to HIV risk perception and reduction.
Meeting Evolving Needs
As interest in BSS has increased worldwide, FHI's evaluation specialists have been challenged to adapt the methodology to meet the needs of many different HIV/AIDS programs, donors and target groups. The flexibility of the methodology has enabled them to design BSS for a variety of situations and audiences, from assessing the contribution of a voluntary counseling and testing intervention in Zimbabwe to monitoring trends in risk behavior among mobile populations served by a cross-border prevention project in Indochina.
Adapting the BSS to gather information from women in marriages or other stable relationships is one challenge FHI faces in a number of countries. Since the BSS are designed to track risk behavior, the recommended indicators and sample questionnaires FHI has developed are not relevant for the many women at risk of HIV because of the behavior of their partners.
For such groups, "it's necessary to adapt the approach," Dr. Sow said. "We're doing this now in Zimbabwe and Nigeria, looking at issues of sexual negotiation and empowerment."
Other target groups that are relatively new to BSS -- drug users and men who have sex with men (MSM) -- pose different challenges. According to Dr. Saidel, little is known about the HIV risk behavior of these two groups in Asia. "The people who work with them will tell you it's a huge problem, and people aren't paying attention," she said.
In many parts of the world, research is needed to document levels of risk behavior among MSM and drug users and to assess the impact of this behavior on local HIV/AIDS epidemics. FHI has begun this work in collaboration with partners in Cambodia and India.
Working with MSM and drug-using communities and the organizations that serve them is the only way to carry out effective behavioral research among these groups, according to Dr. Saidel. Her research on indicators for drug users' risk behavior led her to the slums of Delhi, where she observed a large needle-exchange program run by an NGO called Sharan. From Sharan's members -- some former drug users themselves -- she learned about the wide range of potential risk behaviors, from actually passing a needle around to dipping needles into a shared vial or pot to mixing a drug with blood that may be contaminated.
Indicators of sexual behavior among drug users are also important to provide data for program implementation and advocacy.
"Condom use is a big area that gets ignored with drug users," Dr. Saidel said. "And the next thing you want to know is who are their partners. That might be a wake-up call to people who might be at risk."
Designing appropriate indicators is only one of the challenges of surveillance among disenfranchised groups, as FHI and its partners discovered during a first round of BSS in West Bengal, India, that includes MSM as a target group.
The West Bengal survey, funded by the United Kingdom's Department for International Development (DFID) as part of a comprehensive assessment of sexual health interventions in five Indian states, is one of the first to track sexual behavior among MSM. "This will move them beyond anecdotal information to knowledge of risk levels," Mills said.
But such information is particularly sensitive in cultures where homosexuality and bisexuality are not recognized, and Mills acknowledges that it may surprise many people. "We don't want a backlash against the MSM community in West Bengal, so we have to present the data very carefully," he said.
DFID's West Bengal Sexual Health Project worked with local NGOs and their clients in the MSM community to build consensus on the interpretation of this data and on the accompanying recommendations to policymakers.
Such collaboration is essential in disseminating the results of any BSS, Mills notes. "You have to make sure that good information leads to more informed choices."
Reference
- Family Health International and the Joint United Nations Programme on HIV/AIDS. 1999. Meeting the Behavioural Data Collection Needs of National HIV/AIDS and STD Programmes. A Joint IMPACT/FHI/UNAIDS Workshop: Report and Conclusions, May 1998.
|
BSS Experience Builds Capacity in Cambodia |
| In 1996, when Cambodia was just beginning to confront an explosive HIV/AIDS epidemic, the country's national AIDS program fit in one room.
"The national AIDS program was tiny," recalls Dr. Pamina Gorbach, who came to Cambodia that year with a team from the University of Washington to work with the program's staff on a study of the prevalence of sexually transmitted infections (STIs), including HIV, and the associated risk behaviors.
Three years later, the National Center for HIV/AIDS, Dermatology and Sexually Transmitted Disease (NCHADS) has its own building and a staff of 61 men and women. But most of those who worked on the 1996 study remain, including a core group of researchers in the Planning, Monitoring, Evaluation and Research (PMER) Unit. From that first behavioral study, their research has grown into a full-fledged series of annual behavioral surveillance surveys (BSS) that provide invaluable information about the behaviors that drive one of Asia's most severe HIV/AIDS epidemics.
First Steps
The 1996 research began as an STI prevalence study, conducted with assistance from Family Health International (FHI) and one of its partners in the AIDS Control and Prevention (AIDSCAP) Project, the University of Washington. Recognizing that little was known about STIs or STI risk behavior in Cambodia, the University of Washington added a behavioral component to the research.
The country was ruled by two prime ministers in an uneasy power-sharing arrangement that would lead to a coup a year later. Yet in spite of this political instability, Dr. Gorbach remembers, the timing was good. "It was a country in the process of being reborn, so we were there at the right time to get this started," she said.
The national AIDS program team was relatively young -- as was the leadership of most sectors of Cambodian society after the loss of a generation to genocidal civil war. But its members more than made up for their inexperience with intelligence, energy and commitment. Team members worked with Dr. Gorbach on every aspect of the study, learning skills that they would teach to others in subsequent years.
Three of the Cambodian researchers spent two months in the field with Dr. Gorbach, conducting 781 interviews in three sites. The next year, they drew on that experience to train and supervise 30 interviewers for Cambodia's first BSS.
A Head Start
Through the 1996 behavioral study, the researchers had already learned a great deal about which groups in Cambodia were at risk of HIV and other STIs and why. But additional research was required to gather the information needed to obtain a representative sample of each group. Much of his research was done on the streets and in public parks.
One of the interviewers -- a Ministry of Health (MOH) physician -- was particularly gifted. "She could talk to anyone," Dr. Gorbach said. Nevertheless, a pilot survey of female vocational students found no one who would admit to having sex, so that group was dropped from the BSS.
Like many other BSS, the Cambodian surveys used "cluster sampling" to obtain representative samples of the target groups. Team members made lists of the areas where members of each target group assembled, such as brothels, police departments, beer companies, workplaces and vocational schools, and numbers of individuals in each of these "clusters." Then they randomly selected clusters from each list and all members of those clusters were interviewed until the target sample size for each group was reached.
The researchers had a fairly good idea of what questions they would like to ask. They started with the questionnaire they had used for the 1996 survey and pared it down by about half for the BSS. The supervisors then did sample interviews with members of the target groups to pretest the questionnaire.
The questionnaire was also pretested during the field-based interviewer training. "The final pretest was really done by the people who were going to use the instrument," Dr. Gorbach said. "This gave them a lot of experience with the instrument, and it engaged them in the process of developing it." |
Each of the interviewers belonged to one of five provincial teams. All were employees of the provincial health departments of the MOH, but not the NCHADS. To avoid having provincial AIDS office staff members interview their own clients, the AIDS program borrows staff from other provincial health programs, such as malaria or tuberculosis control, for a few weeks.
Supervisors from the NCHADS's central office monitored the data collection in each province during the first two rounds of the BSS. This job includes ensuring the safety of the interviewers, checking all the forms, observing the interviewers to ensure that they are interacting appropriately with survey participants, and intervening when necessary to maintain privacy and confidentiality.
"For the first two years, four supervisors were there every day, watching the interviews, so we have really good quality data," Dr. Gorbach said.
Continuity and Change
Interviews for the 1999 BSS, funded through a World Bank loan for HIV/AIDS programs, began in mid-June. Most of the same interviewers and supervisors have returned to work on the second and third rounds of the survey, and all three rounds were supervised by field coordinator Dr. Heng Sopheab, giving it a remarkable continuity. As the NCHADS researchers have gained more experience with the methodology, they have assumed increasingly greater responsibility for different aspects of the survey.
The 1999 BSS marks the third round of the surveys in Cambodia and the first year of decentralization to the provincial level. Provincial AIDS office staff will serve as field supervisors, and the former supervisors who trained them will observe the interviews.
Decentralization will help free central level staff to manage an expanding portfolio of research and interventions. But according to Dr. Hor Bun Leng, NCHADS deputy director and head of PMER, their first-hand knowledge of the behavioral surveillance process has proved invaluable.
"It has helped the NCHADS a great deal in terms of capacity building to have experience in developing the BSS, conducting the interviews and doing the supervision," he said.
Using the Results
An analysis of two years of BSS results and a comparison with the 1996 behavioral study findings revealed encouraging progress in HIV/AIDS prevention. Consistent condom use during commercial sex had increased significantly among high-risk groups since 1996. For example, the proportion of sex workers reporting that they always used condoms with clients rose from 16 percent in 1996 to 53 percent in 1998.
This means, however, that only about half of the sex workers -- a group with an HIV prevalence of more than 40 percent -- use condoms consistently. Condom use with regular, usually non-commercial sex partners called sangsa (sweethearts) was low among all the groups surveyed, ranging from 5 percent among working men to 24 percent among sex workers. And a regression analysis showed that sex workers who have a regular partner or fewer clients are least likely to use condoms.
Findings such as these have been used to make adjustments in the NCHADS outreach program to sex workers and to develop guidelines for a national 100 percent condom policy based on a similar and very successful intervention in Thailand. Dr. Leng reports that the BSS results have also been used for strategic planning and advocacy.
"The most important and useful aspect of the BSS is the information on sexual behavior, which is surprising to the general public and to some conservative policymakers," he said. "With these results, we can easily gain support from policymakers in different ministries and institutions to create their own programs in response to the HIV/AIDS epidemic."
|
