Relatively little is known about the epidemiology of HIV in Togo, a small West African country with a population of 5.3 million, long isolated because of political constraints. UNAIDS estimates that national prevalence remained stable in recent years, with 3.3% of adults aged 15-49 years being HIV infected in 2007.1 Among pregnant women, prevalence was 4.2% in 2006 compared with 4.8% in 2003.2 No nationally representative population survey has been carried out to estimate prevalence among men. However, in neighbouring Ghana, 2.7% of adult women and 1.5% of adult men were HIV infected in 2003.3 More recently, a demographic and health survey in Benin estimated that only 1.5% of adult women and 0.8% of adult men in this other adjacent country were HIV infected.4 In Ghana and Benin, much higher HIV prevalences have been reported among female sex workers (SWs) (26%-74%) and their male clients (5%-16%).5-8 As part of an intervention aiming to reduce HIV transmission among core groups in Togo, we performed surveys of SWs and their clients, initially in Lomé (in 2003) and then in the whole country (in 2005), enabling us to describe the epidemiology of HIV among these vulnerable populations.
The West African Project to Combat AIDS and sexually transmitted infections (STIs) provided preventive and curative care to SWs and their clients in 9 West African countries, with funding from the Canadian International Development Agency. In Togo, this intervention began in 2002 and provided support to governmental health centers and nongovernmental organizations, one of which (Forces en Action pour le Mieux-Être de la Mère et de l' Enfant) had been supplying for over a decade condoms to SWs through a network of peer educators.
In 2003, mapping of prostitution environments was conducted in the Lomé area (prefecture of Lomé Commune and adjacent suburbs) and cities located within the administrative Maritime region: Agoenyivé, Ahépé, Akoumakpé, Gbatopé, Hahatoe, Kévé, Kouvé, Tabligbo, Toglékopé, Tsévié, and Vogan (Fig. 1). Mapping of prostitution sites (brothels, bars, hotels, and areas of street prostitution) and enumeration of SWs were performed by peer educators using the snowballing technique.9
The 2003 HIV survey, conducted in May to June, was limited to Lomé and Agoenyivé. Sites included in this survey were randomly selected, after weighing according to the number of SWs in each site. Within the 17 selected sites, all SWs were invited to participate by peer educators who made up to 5 visits. SWs consenting to participate came to a clinic in central Lomé, during daytime, where a questionnaire was administered and blood and cervical samples obtained. Transportation fees (≈ $3) were provided, as well as free condoms and treatment of any concomitant STI. Clients of SWs were recruited on the same 17 prostitution sites, between 19:00 and midnight. Number of clients to be recruited was proportional to the number of SWs enumerated on each site. Male field workers visited selected sites until the targeted number of clients was recruited. After the client left the SW' s room, a questionnaire was administered and a blood sample collected. Free condoms and treatment of concomitant STI were provided to clients.
For both SWs and clients, a few drops of capillary blood were deposited on a filter paper, which was air dried, individually stored in a zip closure plastic bag at 4°C and brought the next day to the Institut National d' Hygiène in Lomé. Samples were tested with Vironostika HIV Uni-Form II plus O (BioMérieux, Boxtel, The Netherlands). Those nonreactive were considered negative. Specimens reactive on the first enzyme-linked immunosorbent assay were tested with Immunocomb II HIV 1 &2 Bispot (PBS Orgenics, Courbevoie, France). Samples reactive on both enzyme-linked immunosorbent assays were considered positive. Discordant samples were tested with Immunocomb II HIV 1 & 2 CombFirm (PBS Orgenics), whose results were definitive. Cervical samples were tested for the presence of Neisseria gonorrhoeae and Chlamydia trachomatis using the polymerase chain reaction (Amplicor; Roche Diagnostics, Branchburg, NJ).5 All positive results for N. gonorrhoeae were confirmed with a 16S rRNA assay.10
In 2005, mapping and enumeration of SWs were extended to the whole country. For the HIV survey, procedures were the same as in 2003 except that, in addition to Lomé, SWs and clients were recruited, in November to December 2005, in the following urban centers (Fig. 1), in Plateaux region: Atakpamé, Anié, Gléï, and Kpalimé; in Central region: Aléhéridé, Blitta, Sokodé, and Sotouboua; in Kara region, the city of Kara; in Savanes region: Cinkassé and Dapaong. Most of these cities are located on the main north-south road that transects Togo and were selected because of accessibility and number of SWs. In the following additional cities, SWs were enumerated but not included in the HIV survey: Adidogomé, Agbodrafo, Agoenyivé, Agou, Agbodrafo, Ahépé, Akoumakpé, Alédjo/Falaise, Aného, Badou, Bafilo, Bassar, Dankpen, Gbatopé, Hahatoe, Kanté, Kévé, Kouvé, Mango, Niamtoungou, Notsé, Tandjouaré, Tabligbo, Toglékopé, Tsévié, and Vogan (Fig. 1).
Population denominators were obtained from census data.11 The study was approved by the institutional review board of the Togolese ministry of health, based on the anonymous linked approach. Study forms and samples were identified only by a number. An envelope was given to each participant with her/his study number. Those wishing to know the result of their HIV serology were invited to come to the local clinic a few weeks later where they presented the envelope with the study number. They received pretest and posttest counseling, and those with a positive HIV serology were referred to appropriate facilities.
Data were analyzed with Stata 8.0. Proportions were compared with the χ2 test or Fisher exact test when numbers were small. Logistic regression was used for multivariate analysis of risk factors for HIV infection, and results are presented as adjusted odds ratios with their 95% confidence intervals. Models were built up sequentially starting with the variable most strongly associated with the outcome and continuing until no other variable reached significance or altered the odds ratios of variables already in the model. When the final model was reached, each variable was dropped in turn to assess its effect. Different models were compared with the likelihood ratio test. We kept in the final model variables that significantly enhanced the fit at the P ≤ 0.05 level.
In 2003, HIV prevalence among Lomé SWs was 54.7% (211 of 386). Their age ranged from 12 to 50 years (median: 27 years). About one third (149, 37%) were from Togo, one third from Ghana (142, 35%), and the rest from other West African countries. As shown in Table 1, HIV prevalence was higher among older women, the Ghanaians, those without formal education, the divorced or widowed, those working in brothels, reporting a higher number of customers in the previous year (calculated from the average number of clients per week and number of weeks worked in the previous year), charging ≤500 FCFA (≈ $1) per intercourse, and those involved in sex trade for a longer period or who started selling sex later in their lives. Prevalence of gonococcal and chlamydial infections was 15.7% (53 of 337) and 7.7% (26 of 337), respectively.
In 2005, HIV prevalence was 45.4% (163 of 359) in Lomé (including the adjacent Maritime region), and risk factors for HIV were the same as in 2003 (Table 1). Outside Lomé, sampled only in 2005, HIV prevalence was 17.7% (100 of 566), and HIV was also associated with generally the same factors as in Lomé, but prostitution seemed more of a part-time activity (few women had more than 365 paid intercourses annually). In Lomé, prevalence of gonococcal and chlamydial infections decreased to, respectively, 2.9% (10 of 346; P < 0.001 comparing with 2003) and 4.3% (15 of 346; P = 0.08). Among SWs outside Lomé, 4.7% (17 of 360) were infected with N. gonorrhoeae and 3.1% (11 of 360) with C. trachomatis.
Table 2 shows the results of the multivariate analyses of risk factors for HIV infection among SWs. Level of education, price per intercourse, estimated number of paid intercourses in the previous year, and age at first transactional intercourse were no longer associated with HIV after adjustment. In 2003 and 2005, age, country of origin, being widowed, working from brothels, and duration of prostitution were associated with HIV. In 2005, a previous genital ulcer and working in Lomé were also independent risk factors.
Among clients recruited in Lomé (including the adjacent Maritime region), HIV prevalence was 13.3% (42 of 315) in 2003 and 8.3% (20 of 242) in 2005, whereas outside Lomé (2005 only), it was 3.9% (17 of 440). As shown in Table 3, within these 3 groups, HIV was more common among older men and those previously or currently married. HIV was more prevalent in men with a previous genital ulcer, but this was significant only among the Lomé 2003 clients. In multivariate analysis (Table 4), the independent risk factors in 2003 were older age and a history of genital ulcer, whereas paying a higher price became a significant protective factor. Among the 2005 clients, age and marital status were associated with HIV, as was residence in Lomé. When added to this final model, a history of past genital ulcer barely missed the limit of statistical significance (adjusted odds ratio 2.04, 95% confidence interval: 0.95 to 4.38, P = 0.07).
Table 5 displays characteristics of SWs and their clients in the various regions of Togo and HIV prevalence among the general adult population. For women, the latter was estimated from the means of prevalence among antenatal clinic attendees during the 2004 and 2006 national surveillance surveys.12 For men, we multiplied prevalence in women by 0.55, based on the male to female prevalence ratio of Ghana and Benin.3,4
The Lomé area, corresponding to 15% of the population of Togo, comprised 52% of all SWs in the country. The ratio of SWs per 1000 males aged 15-59 years was 13.0 in Lomé compared with 2.7 elsewhere. Compared with other regions, SWs in Lomé and Maritime region were older and worked more frequently out of brothels. There was little difference in duration of prostitution (data not shown). SWs in Lomé and Maritime region were busier, with a mean number of paid intercourses per year which is double that of other regions. There was a marked north-south gradient in HIV prevalence both among SWs and clients, mirroring variations in HIV prevalence within the general adult population.
In 2003, in Lomé, 87% (329 of 377) of clients claimed to have used a condom during the intercourse preceding their interview. In 2005, this proportion increased to 93% in Lomé, whereas it was 88% outside Lomé. Clients in Lomé bought sex less often than in other regions. From the SWs' and clients' estimates of the number of transactional intercourses, the ratio of client to SWs was calculated and was much higher in Lomé and Maritime region than elsewhere. The number of clients in each region can then be back calculated (number of SW × ratio of client to SWs). In the Lomé area, which comprised 68% of all men paying for sexual services in the country, almost one third of adult men had bought sex in the previous year compared with 6% in the surrounding Maritime region and <2% in the rest of the country.
Based on these estimates of the number of clients and HIV prevalence among clients, HIV prevalence among nonclient men can be calculated. From this prevalence ratio, the population attributable fraction (PAF) of prevalent cases of HIV among men acquired during transactional sex was estimated to be 32% in Lomé. Estimates of PAF for each of the other regions lacked precision given the small number of clients sampled; however, when all these regions were combined (Togo minus Lomé), PAF was only 2%.
By and large, risk factors for HIV infection among SWs were the same in Lomé (2003 and 2005) and in cities outside Lomé in 2005, even though the prevalence was much lower among the latter. These factors reflect cumulative exposure to HIV, which is a product of the duration of prostitution, the number of paid intercourses per year, the proportion during which a condom is not used, HIV prevalence among clients, and the cofactor effect of other STIs. Being widowed was also an independent risk factor, presumably reflecting the presence of AIDS in the deceased spouse. Among clients, age was a proxy for cumulative exposure, and the cofactor effect of genital ulcer disease was noted.
To our knowledge, this is the first study attempting to quantify sex work at a national level in an African country. Previous estimates were done only at a subnational level, in the capital city and/or selected regional centers.13 Our main finding is the profound variation between the capital city, Lomé, and the rest of the country in the dynamics of sex work and its contribution to HIV transmission. In Lomé, where the sex trade is concentrated, approximately one third of prevalent cases of HIV among men were acquired during transactional sex, whereas outside Lomé, the contribution of transactional sex seems marginal. This may account for the north-south gradient in HIV prevalence within core groups and pregnant women (as proxies for the general adult population), as HIV is amplified among the former before spreading to the latter via the bridging population of clients.
Lomé has a population of ≈816,000, followed by Sokodé (123,000), Kara (109,000), and Kpalimé (101,000). All other Togolese cities hold fewer than 100,000 inhabitants. We tried to enumerate SWs as accurately as possible throughout the country, but such estimates can only be imperfect, and we do not know whether the completeness of the enumeration varied between Lomé and smaller urban centers. The density of SWs in Lomé in 2005 (13.0 per 1000 males aged 15-59) was similar to that measured in 1997-1998 in Cotonou, Bénin (10.1 per 1000) and Yaoundé, Cameroun (14.4 per 1000).14 Enumeration may have been less comprehensive outside Lomé, for instance, because distant sites with few SWs were not visited, but this would not alter substantially the dichotomy with regard to populations involved in the sex trade.
We analyzed clients' data as per the location where the last transactional intercourse occurred rather than by area of residence, but this impacted little on our findings: in 2005, only 12% of clients enrolled in Lomé lived outside the capital, whereas 16% of those recruited in the Maritime region lived in Lomé. Most interregional sex trade corresponded to movements between Lomé and its suburban area within Maritime region. Overall, two thirds of clients of SWs in Togo live in Lomé and another fifth in the surrounding Maritime region, probably for several reasons. Young men migrate out of rural areas to the capital city creating a gender imbalance in the latter. Short-term migrants (traders, truck drivers, etc), also concentrated in Lomé, are prone to buy sex. A large city is by essence rather anonymous, which makes it less socially risky for a man to seek transactional sex. Traditional values and constraints also tend to disappear in urban centers. The result of these processes is that even though clients in Lomé bought sex less often each year than their counterparts in other regions, they were so numerous that the SWs, themselves concentrated in Lomé, ended up having roughly twice as many customers as elsewhere. At some earlier stage in the epidemic, this must have led to a higher exposure to HIV, higher HIV incidence and prevalence among SWs, and forward transmission to their clients creating a vicious circle which then impacted on transmission to the general adult population when clients had intercourse with non-SW female partners.
In 2005, clients in Lomé were somewhat more likely than others to claim having used a condom during the intercourse preceding their interview. This might reflect a better uptake of SW preventive interventions in the capital city, which probably occurred only after HIV prevalence had reached high levels among SWs. Fortunately, this will reduce substantially the frequency of HIV transmission from SWs to clients. There is in Lomé a professionalization of sex work with many full-time SWs working out of brothels compared with their part-time upcountry counterparts working from bars and hotels. The former are easier to reach during preventive interventions. Alternatively, the social desirability bias might have been more important among clients in Lomé, potentially more exposed to preventive messages resulting in a better grasp of the behaviour expected from them. However, the reduction in prevalence of gonococcal and chlamydial infections among SWs in Lomé, among whom in 2005 these infections were not more common than elsewhere despite their higher number of clients, does suggest that condom uptake was excellent at this specific stage of the HIV epidemic. Sex work in Lomé is changing rapidly: from 2003 to 2005, there was a substantial reduction in the estimated number of clients in the previous year among SWs working out of brothels (from 973 to 414 clients), bars/hotels (from 507 to 345), or from the street (from 797 to 280), probably reflecting a lower demand.
Togo has had an intervention program for SWs, both in Lomé and outside Lomé, since the early 1990s, implemented by a nongovernmental organization (Forces en Action pour le Mieux-Être de la Mère et de l'Enfant) and sponsored by various donors. Thus condoms have been generally available in sex work environments for a long period. This may be one amongst other factors explaining why the fraction of HIV prevalent cases attributed to contact with SWs in Lomé was only 32% compared with 84% in Accra and 76% in Cotonou.6,15 This relatively low PAF should not be a disincentive against SW interventions in Lomé, it may merely reflect the effectiveness of sustained interventions. Indeed, it can be calculated (number of intercourses × proportion without a condom × HIV prevalence among clients) that, in 2003, SWs in Lomé had an average of 13 unprotected intercourses with HIV-infected clients during the year, whereas this number decreased to 2 in 2005. For the rest of the country in 2005, this number was 1.
The programmatic implications of these findings are important. In Togo, interventions targeting SWs and their clients should concentrate efforts and resources on the capital city, trying to reach as many prostitution sites as possible and ensuring that condoms are always available within this broad diversity of brothels, bars, hotels, and areas of street prostitution. Number of SWs and clients to be reached are such that some kind of vertical program (dedicated clinics and full-time nursing officers and peer educators) is imperative. In the rest of the country, these interventions will have much less of an impact and should be integrated within the routine activities of the local health system through polyvalent health care workers providing care to SWs on top of other duties and part-time peer educators often working alone in small towns.
In conclusion, this first study quantifying sex work at a national level in Africa documented variations in HIV prevalence within Togo with a strong north-south gradient among SWs, their clients and antenatal clinic attendees, and a substantial concentration of the sex trade within the greater Lomé area, where transactional sex played a much greater role in HIV dynamics than in the countryside.
We are indebted to the nurses, peer educators, and field workers who contributed to the data collection and to Christian Audet for the map.