BOTSWANA'S HIV EPIDEMIC AND ACCESS TO HIV SERVICES
Botswana is a politically stable middle-income nation of 1.7 million people1 in southern Africa. Botswana's HIV prevalence has been among the highest in the world since 1995,2 with the prevalence in pregnant women increasing from 6% in the first survey in 19903 to 37.4% in the 2003 national HIV surveillance.4 During the 15 years of the HIV epidemic, many factors thought to contribute to the explosive spread of HIV in this population have been described, including social norms allowing for concurrent sexual partners,5-7 relative infrequency of marriage,1,4 high population mobility and relative wealth,5,6 intergenerational sex,8,9 little female empowerment in sexual relationships,6,8 frequent separation of couples employed by the government and mining industries,6 and the rarity of male circumcision.10,11
When HIV was spreading rapidly in Botswana, few Batswana knew their HIV status. HIV enzyme-linked immunosorbent (ELISA) tests have been available through the country's public health system since the mid-1990s, but there have been barriers to HIV testing in this environment, including reluctance among patients and providers to discuss HIV, shortage of trained counselors, concerns about confidentiality in small communities, results taking several weeks to return, and population mobility, leading to many unclaimed results. A survey in Botswana's 2 largest cities in 2000 revealed that 50% of respondents wanted an HIV test but that none wanted to be tested in a government facility (unpublished data, BOTUSA Project, 2000).
Botswana's government is committed to HIV-related health programs; in recent years, the availability of health care for HIV and AIDS has improved substantially. Botswana began Africa's first free national program for prevention of mother-to-child transmission (PMTCT) in 2001 and the continent's first national public antiretroviral (ARV) program in 2002. As of October 2004, more than 25,000 people were receiving ARV therapy. Isoniazid prophylaxis (IPT) against tuberculosis (TB) for HIV-infected persons became widely available in 2003. Despite advances in care, uptake of HIV testing within the health care system remained relatively low in the first years of service expansion. In 2002, 49% of pregnant women were tested for HIV (unpublished data, Botswana National PMTCT Program, 2002), and only 6.4% of patients with TB had an HIV test result recorded in the national electronic TB register.12
VOLUNTARY COUNSELING AND TESTING NETWORK
In 2000, the Government of Botswana and the Centers for Disease Control and Prevention (CDC) determined that providing voluntary counseling and testing (VCT) outside the health care system was a high priority, and the first freestanding VCT facilities, termed Tebelopele, or “look into the future,” opened in April 2000. Since 2003, 16 Tebelopele centers have provided free anonymous HIV rapid testing with same-day results for the public. Since 2000, a social marketing campaign has encouraged testing at Tebelopele and has included promotion by Batswana leaders, educational and entertainment sessions, billboards, radio and television announcements, and brightly painted vehicles displaying messages encouraging HIV testing.
Funding and Staffing
Testing supplies, staff salaries, training, information systems, social marketing, and running costs were funded by the CDC. Counselors have backgrounds in social sciences or humanities, nursing, and teaching, and many are university educated. All undergo 8 weeks of training, including HIV basics, counseling, and rapid HIV testing. Yearly operating costs are approximately $US 3.5 million. In October 2004, Tebelopele became Botswana's largest independent nongovernmental organization (NGO), funded by the US government.
Counseling and Testing Process
Tebelopele uses a standardized counseling and testing protocol that takes approximately 1 hour. Pretest counseling includes risk assessment and preparation for receiving results. Rapid tests are performed by counselors, and posttest counseling includes risk reduction plans and condom distribution for all clients. Disclosure of HIV status, testing of sexual partners, and referrals to support and health services are discussed as appropriate.
The CDC and Botswana Ministry of Health developed an algorithm for parallel rapid HIV testing using 2 different test kits and validated its use in Botswana. The sensitivity and specificity of the algorithm compared with ELISA were 100% and 99.6%, respectively (unpublished data, BOTUSA Project, 2000). Use of rapid tests by counselors was studied within Tebelopele and found to produce accurate and valid results compared with tests performed by laboratory technicians.13 First-line testing includes 1 Determine kit (Abbott Diagnostics, Abbott Park, IL) and 1 Unigold kit (Trinity Biotech, Bray, Ireland). If the tests are discordant, they are repeated. If they remain discordant, Oraquick (Abbott Diagnostics) is used as a tiebreaker. Quality control is done using ELISA on blood drawn from clients at selected centers each month.
Information collected anonymously from clients includes demographics, sexual history, reasons for seeking an HIV test, and the test result. Data are entered into an Epi-Info 6.04b (January 1997) database. Data are merged, and reports are produced at the national level.
During 2000 through 2002, there were changes in the way condom use (“always,” “sometimes,” or “never”) and sexual partner data were collected. Both questions initially asked clients to recall the past 6 months and were changed to include only the past 3 months, but changes to the 2 questions were made in different years. During 2003 through 2004, both referred to the past 3 months, and our analysis of partner and condom data uses only data from 2003 through 2004 (67.1% of the total data set).
This analysis describes Tebelopele clients, identifies factors associated with HIV infection, and describes the impact of the expansion of HIV care and treatment on the use of VCT services. Univariate analysis and multivariate logistic regression were done using SPSS 12.0 for Windows (SPSS, Inc., Chicago, IL). All demographic variables were used in the multivariate model of demographic risk factors for HIV. Because all had statistically significant associations with HIV status, all were also used in the multivariate model examining behavioral variables of interest and their association with HIV status. Because of the close relation between condom use and the number of partners reported in other studies, each variable was adjusted for the others and for demographics. Multivariate analysis included only first-time adult clients of the Tebelopele network.
HIV Testing and Quality Control
Client specimens sent for ELISA testing received the same result as the rapid test algorithm in nearly all cases (these data are not available for every year): 1184 (99.5%) of 1190 specimens in 2001, 1235 (99.0%) of 1247 specimens in 2003, and 1343 (99.2%) of 1354 specimens in 2004.14 Initial dual-parallel rapid test results were discordant for 0.1% of clients in 2000 through 2001 and for 0.9% of clients in 2003. In 2003, 73.7% became concordant when repeated, and a tiebreaker was used to provide a result for the remaining tests.14
Visits to Tebelopele Centers
Between April 2000 and September 2004, 157,423 visits to Tebelopele VCT centers occurred. Of all visits, 2.9% had no HIV result recorded, 1.3% were by children (aged <15 years), and 66 were missing gender or age. These were excluded from analysis. Of the remaining visits, 22.3% were by clients who reported having been tested at a Tebelopele before, and thus appeared more than once in the data set.
Repeat Tebelopele Clients
There were 33,563 clients who were previously tested at a Tebelopele. Based on self-report, 31,484 (93.8%) had a negative test result at their previous visit. Among those, 1113 (3.5%) had a positive test result on the return visit. Among unmarried sexually active clients, those who had visited Tebelopele before were more likely to report having always used condoms in recent months than were first-time clients (67.1% vs. 50.2%, odds ratio [OR] = 2.02; P < 0.001).
First-Time Tebelopele Clients
Only first-time adult clients (n = 117,234) were included in the remaining analysis. Among this group, 43,485 (37.1%; 95% confidence interval [CI]: 36.8% to 37.4%) were HIV-positive. Figure 1 shows the number of visits each month, daily client/counselor ratios, HIV positivity, percentage of clients being tested because of illness, and expansion of Tebelopele and HIV care and treatment services.
Testing demand increased steadily during Tebelopele's first 2 years. The introduction of ARV therapy in January 2002 was followed by a more rapid increase in testing, especially of clients who reported illness as the reason for testing. By the beginning of 2003, more than 20% of clients reported that their reason for seeking a test was illness or wanting access to IPT or ARVs. The percentage of clients who were HIV-positive increased from 26.3% before the launch of the national ARV program to 38.8% after the program began. Among those seeking a test for health reasons, 77.7% were HIV-positive.
Other reported reasons for seeking a test, including family planning, interest in PMTCT services, and testing before marriage or before having sex with a new partner, represented ≤5% of visits each, with no change over time.
Characteristics of Clients
Demographics, Behavior, and Reasons for HIV Testing
Table 1 summarizes client characteristics and reported behaviors by gender. Most clients were unmarried, reported having only 1 sexual partner, and reported using condoms during the last 3 months. Male clients were more likely to have no education or university education, to have had more sexual partners, and less likely to report never using condoms. Female clients were more likely to be unemployed and to state the reason for seeking an HIV test as concern about partner behavior, partner illness, or an unborn child.
HIV Testing History
Of new Tebelopele clients, 16.2% had been previously tested for HIV in another setting, and 12.1% of those reported that their prior test result had been positive. Of those who tested positive in the past, 38.7% expected a negative result at the current visit, and 11.7% actually received a negative result.
Risk Factors for HIV
Figure 2 shows HIV prevalence by age and gender. Prevalence among women aged 15 to 19 years was more than 9 times that of men in the same age group, and women had a higher prevalence until the age of 30 to 34 years. In multivariate logistic regression (Tables 2, 3), among women, factors significantly associated with being HIV-positive included increasing age; being unmarried or widowed; low educational level; being unemployed, a laborer, or a businessperson; testing with a partner; having no sex partners in the last 3 months; using condoms never or sometimes; and seeking a test because of health concerns or partner behavior or symptoms. The same factors were associated with HIV-positive status for male clients, as were having only 1 sex partner in the last 3 months and being in the uniformed services.
Among 3274 clients reporting never having had sex, 136 (4.2%) were HIV-positive. The average age of these clients was 29 years. Among clients reporting having had sex, 21.1% reported no sex in the last 3 months. Among these recently abstinent clients, 36.5% reported illness as the reason for seeking a test (compared with 16.9% of clients who had had sex recently; P < 0.001) and 45.5% were HIV-positive.
Clients who came for testing as part of a couple made up 8.2% of all clients; this percentage was stable over time. Most (61.2%) clients testing as part of a couple were unmarried; 8.4% were being tested because they intended to marry, and 5.7% were being tested before having sex with the partner with whom they sought a test. Of people visiting because of intent to marry, 36.7% were visiting with their partner. There was no change over the 5-year period in the percentage of couples who were testing before marriage or before having sex.
People visiting as part of a couple were more likely to be professionals (19.0% vs. 14.7%; P < 0.001), to have a tertiary education (26.3% vs. 22.6%; P < 0.001), a higher mean age (34 vs. 31 years) and were more likely to be HIV-positive (43.3% vs. 36.5%; P < 0.001) than people visiting alone. Discordant results were found in 23.1% of couples. Married couples were slightly less likely to be discordant than unmarried couples (22.0% vs. 25.4%; P < 0.001). Among discordant couples, half had an HIV-positive man and half had an HIV-positive woman.
Overall, 44.2% of sexually active clients reported “always” using condoms in the past 3 months, 28.4% reported “sometimes” using condoms in recent months, and 23.2% reported “never” using condoms in recent months. Unmarried sexually active people with only 1 recent partner were more likely to report never using condoms than were people with more partners (19.3% vs. 8.9%, OR = 2.4; P < 0.001), and this group had high HIV prevalence (51.4%).
Comparison of Voluntary Counseling and Testing Clients With the Botswana Population
We compared demographic data on VCT clients with 2001 census data.1 VCT clients included a higher proportion of girls and women and lower proportions of people between 15 and 19 years old and older than 40 years of age than Botswana's population as a whole. VCT clients were more likely to have a tertiary education and less likely to have no education. The distribution of marital and employment status and urban versus rural residence among VCT clients was similar to that of the general population.
Acceptance of Tebelopele Initiative
Tebelopele has been well used and has performed at least 1 HIV test for 11% of Botswana's adult population in the first 5 years of its operation. A population-based survey of adults in 2003 revealed that of people who had ever been tested for HIV, 37% had been tested at a Tebelopele.7 The acceptability of Tebelopele was especially notable among young people: 73% of men and 42% of women aged 15 to 24 years who had been tested for HIV had done so at a Tebelopele.7
Voluntary Counseling and Testing During Expansion of HIV Care and Treatment
The striking increase in HIV prevalence in Tebelopele centers nationwide after the opening of the first ARV clinics (see Fig. 1) corresponded to an increase in visits by ill people who wanted HIV testing to access medical treatment. HIV prevalence nationwide was stable during 2000 through 2004, ranging from 35.4% to 37.2% among pregnant women in annual surveillance.4
As seen in Figure 1, during the 2.5 years of Tebelopele's rapid expansion (April 2000-September 2002), average client-to-counselor ratios increased from 2 to 4 to 5 per day. Because counselors are expected to see approximately 8 clients per day, this indicates that excess testing capacity still existed at the end of the rapid expansion. The ARV program began in January 2002 and rapidly expanded between April 2003 and August 2004. During this period, average client-to-counselor ratios increased to 7 to 9 clients per day per counselor, suggesting that the current saturation of the Tebelopele network was driven by increasing access to ARV therapy.15
Social Reasons for Seeking an HIV Test
It is concerning to note that in 4 years of active promotion of couple testing, seeking an HIV test with a partner, especially before having sex for the first time or before marriage, has not become a more common reason for testing. Counseling and testing of couples has been advocated as a way to ensure accurate sharing of essential health information within sexual partnerships and to provide a safe and supportive environment for such discussions. Our data indicate that 80% of clients have at least 1 partner and are having sex with that partner, but only 5% to 10% of clients tested with their partner, and of those, only 5% to 10% did so before having sex. Even among clients stating marriage as the reason for a test, only one third were tested with their potential spouse, and these percentages were stable over time. Tebelopele's social marketing campaigns encouraging couples to test together at key points in their relationship seem to have had limited effectiveness.
Risk Factors for HIV
Among VCT clients, increasing age was the single greatest risk factor for being HIV-positive among both genders, with female clients having a high HIV prevalence 5 to 10 years earlier than male clients. This is thought to reflect older men often having younger female sexual partners.8,9,16
Number of Sexual Partners
Some clients reported never having had sex, but 4.2% of these were HIV-positive. This anomaly, reported for 136 clients, may reflect data collection or entry errors. Alternatively, these clients may have had sexual exposures to HIV that they did not consider as “having sex,” they may have not have reported their sexual behavior truthfully, or they may have contracted HIV through a nonsexual route. No data on nonsexual potential exposures are available. Among VCT clients reporting having had sex but not during the past 3 months, HIV prevalence was high, and many of these clients reported illness as the reason for seeking a test. Many of these clients were likely abstinent because of illness rather than as a method of HIV prevention.
Most VCT clients reported only 1 recent sexual partner, which agrees with other behavioral data collected recently in Botswana.7,17 Among unmarried sexually active men, having only 1 partner was associated with never using condoms and with having HIV infection, however, suggesting that discordant couples may cease condom use after some period of fidelity without the benefit of HIV testing.
In multivariate analysis accounting for condom use and number of partners, having more sexual partners was associated with a significantly lower risk of infection for men. This finding remained true if people seeking a test because of health concerns were excluded from the data set and if people who had previously been tested for HIV outside of Tebelopele were excluded from the data set. No interaction effect was found between the number of partners and condom use on HIV status. At least 2 factors probably account for this finding. First, there may be residual confounding by condom use. People with more partners who reported always using condoms during the 3 months reported on here may have used condoms consistently for a longer period than had their single-partner counterparts, thus reducing their lifetime HIV risk. Second, the number of partners reported in this short time frame is probably a poor reflection of lifetime sexual network exposure, as it has been in other studies.9,18 People reporting 1 recent partner may have been exposed through their sexual networks to similar numbers of HIV-infected people as those reporting multiple recent partners. The average number of concurrent sexual partners within the past 12 months has been found in population-based studies of sexual behavior and HIV in major African cities to be unrelated to HIV prevalence,19 although the reported number of lifetime partners was an important predictor of regional variation in the prevalence of HIV infection in a 4-city analysis of population-level data from African cities.20 Our data cannot estimate lifetime number of partners or lifetime sexual network exposure to HIV and do not demonstrate a correlation between higher numbers of sexual partners and HIV risk.
Married VCT clients had a significantly lower prevalence of HIV infection, but only 16% of VCT clients were married. This low frequency of marriage is similar to HIV surveillance data, which indicated that 88% of pregnant women in 2003 were unmarried,4 and to census data.1 Other reports from Africa have shown higher HIV prevalence among married persons than among single persons, possibly because of decreased condom use and frequent sex between discordant married people, but these data were collected in countries in which more of the population is married.21-23
The relative infrequency of marriage among all age groups in Botswana has been said to result, in part, from the significant loss of legal rights that women in Botswana have historically experienced within marriage. These laws were recently reviewed by the Botswana legislature, and in late 2004, the Abolition of Marital Power Bill was passed “to ensure full gender equality under the law with respect to all civil marriages in Botswana.”24,25 Tebelopele data suggest that such efforts to make marriage a more attractive option for young Batswana may contribute to reduced HIV transmission.
Use of Condoms
Condom use was high in this data set and was similar to that of other recent population-based behavioral surveys in Botswana.7,15 These data demonstrate a substantial benefit of reported consistent condom use. This concurs with findings in Uganda indicating that consistent condom use reduces the risk of HIV infection.26 Furthermore, clients previously tested at a Tebelopele were much more likely to use condoms than were first-time clients, suggesting that HIV testing, risk reduction counseling, and providing free condoms may contribute to safer sexual behavior. These data highlight the importance of continuing attempts to market and educate the population about condoms as well as ensuring the availability of condoms as a vital part of HIV prevention efforts in this setting.
Other Risk Factors
Higher educational level in this sample was associated with a lower risk of HIV; in other African populations, higher education has been a risk factor for infection, perhaps because of additional mobility and resources conferred by education and employment.27 In Botswana, where the mobility and wealth of the population are higher overall than in other African countries,5,6 increased education may lead to less risky sexual behavior that outweighs the effect of greater resources on sexual choices.
Although marriage and condom use were associated with a lower risk of having HIV, age and educational level were the most significant predictors of HIV status in this sample. These data suggest that Botswana's HIV prevention efforts should focus on factors that are modifiable, including improving the overall educational level of the population, making marriage a more attractive option, encouraging 100% condom use, and increasing the number of people who know their HIV status and that of their sexual partner(s) and have an opportunity to choose behaviors accordingly.
It is encouraging to note that most VCT clients report only 1 recent sexual partner but discouraging to note less condom use and higher HIV prevalence among these clients. For much of Botswana's population, having 1 sexual partner is insufficient protection against HIV because of the high risk that any chosen partner may be infected with HIV. To provide effective protection, having 1 partner must be combined with consistent condom use until both partners have had negative HIV test results. These data provide impetus to emphasize strongly that short-term monogamy does not equal protection from HIV in the future education of Botswana's highly HIV-affected population.
HIV testing in Botswana is increasingly accepted, and demand for testing is high. The availability of HIV care and treatment services is transforming the knowledge of one's HIV status from a burden to an asset, especially for those with concerns about their health. Future social marketing of VCT should emphasize testing for those who are not ill, continue to encourage couple testing, and focus on normalizing HIV testing at pivotal points in interpersonal relationships, including before having sex for the first time, before ceasing condom use, before marriage, and before pregnancy.
The authors acknowledge the hard work of the Tebelopele staff, who did all the counseling, testing, and logistical and data management that allowed us to write this paper.
1. Government of Botswana Central Statistics Office. 1991 and 2001 Population and Housing Census
. Gaborone, Botswana: Central Statistics Office; 1992 and 2001.
3. UNAIDS. Botswana Epidemiological Fact Sheet on HIV/AIDS and Sexually Transmitted Infections, Update 2002
. Geneva, Switzerland: UNAIDS; 2002.
4. Botswana National AIDS Coordinating Agency (NACA). 2003 Second Generation HIV/AIDS Surveillance Technical Report
. Gaborone, Botswana: NACA; 2003.
5. Hope K. Population mobility and multi-partner sex in Botswana: implications for the spread of HIV/AIDS. Afr J Reprod Health
6. Macdonald D. Notes on the socio-economic and cultural factors influencing the transmission of HIV in Botswana. Soc Sci Med
7. Pappas-Deluca K, Koppenhaver T, and the Makgabaneng Listenership Survey Group. Makgabaneng Radio Serial Drama Listenership Survey Report. BOTUSA Project
. Gaborone, Botswana; 2005.
8. Mandevu R. Botswana: where young girls are “easy prey.” Special report: women and HIV. AIDS Anal Afr
9. Gregson S, Nyamukapa C, Garnett G, et al. Sexual mixing pattens and sex-differentials in teenage exposure to HIV infection in rural Zimbabwe. Lancet
10. Langeni T. Male circumcision and sexually transmitted infections in Botswana. J Biosoc Sci
11. Kebaabetswe P, Lockman S, Mogwe S, et al. Male circumcision: an acceptable strategy for HIV prevention in Botswana. Sex Transm Infect
12. Botswana National TB Program 2002 Annual Report
. Gaborone, Botswana; 2003.
13. Kenyon T, Alwano MG, Jikijela C, et al. The accuracy of HIV rapid testing as performed by counselors compared with lab technicians, Botswana Tebelopele VCT centers. Presented at: XII International Conference on AIDS and STDs in Africa; 2001; Ouagadougou.
14. BOTUSA Project. Tebelopele Voluntary Counseling and Testing Centre Annual Reports 2001-2004
. Gaborone, Botswana: BOTUSA.
15. Alwano MG, Roels T, Mokomane M, et al. The effect of introducing antiretroviral therapy in the public setting on utilization of voluntary counseling and testing services in Botswana, 2001-2003 [abstract MoOrD1035]. Presented at: XV International AIDS Conference; 2004; Bangkok.
16. Kenyon TA, Mwasalla VI, Molosiwa R, et al. The epidemiology of HIV infection among clients attending Tebelopele voluntary counseling and testing (VCT) centers in Botswana, 2000-2001 [abstract WePeC6157]. Presented at: XIV International AIDS Conference; 2002; Barcelona.
17. Government of Botswana Central Statistics Office and Family Health International. Botswana AIDS Impact Survey (BAIS)
. Gaborone, Botswana: Family Health International; 2001.
18. Johnson KM, Alarcon J, Watts D, et al. Sexual networks of pregnant women with and without HIV infection. AIDS
19. Lagarde E, Auvert B, Carael M, et al, for the Study Group on Heterogeneity of HIV Epidemics in African Cities. Concurrent sexual partnerships and HIV prevalence in five urban communities of sub-Saharan Africa. AIDS
20. Auvert B, Ferry B. Modeling the spread of HIV infection in four cities of sub-Saharan Africa [abstract WeOrC1310]. Presented at: 14th International AIDS Conference; 2002; Barcelona.
21. McFarland W, Mvere D, Katzenstein D. Risk factors for prevalent and incident HIV infection in a cohort of volunteer blood donors in Harare, Zimbabwe: implications for blood safety. AIDS
. 1997;11 (Suppl 1):S97-S102.
22. Crampin AC, Glynn JR, Ngwira BM, et al. Trends and measurement of HIV prevalence in northern Malawi. AIDS
23. Clark S. Early marriage and HIV risks in sub-Saharan Africa. Stud Fam Plann
24. Tautona Times
, no. 46. Weekly electronic press circular of the Office of the President. Gaborone, Botswana, December 11, 2004.
26. Ahmed S, Lutalo T, Wawer M, et al. HIV incidence and sexually transmitted disease prevalence associated with condom use: a population study in Rakai, Uganda. AIDS
27. Smith J, Nalagoda F, Wawer MJ, et al. Education attainment as a predictor of HIV risk in rural Uganda: results from a population-based study. Int J STD AIDS