In the early years of the human immunodeficiency virus (HIV) epidemic in Africa, female sex workers (FSWs) were identified as a high-prevalence, high-incidence, core group among whom the extraordinary prevalence of other sexually transmitted infections (STIs) facilitated transmission of the virus to their clients, who then infected low-risk women not involved in the sex trade.1–5 The contribution of transactional sex to the overall dynamics of HIV transmission in large African cities was quantified when it became possible to measure anti-HIV antibodies in specimens obtained from clients having just paid for sex.6 In Accra, Ghana, and Cotonou, Benin, such studies revealed that 76% to 84% of prevalent cases of HIV infection among men had been acquired from FSWs.7,8 Interventions that reduce vulnerability and increase condom use would be expected to have a profound impact on HIV transmission, but can be achieved only imperfectly.
STI control among FSWs may further reduce HIV transmission. However, in resource-poor settings, syndromic management of symptomatic cases is neither sensitive nor specific in identifying women with cervical infections, especially FSWs for whom novel approaches are needed.9,10 Regular screening visits with point-of-care rapid and sensitive laboratory tests could improve performance, but this remains generally unavailable in low-resource settings.11 Periodical antibiotic treatment (PAT) of FSWs, regardless of the presence of symptoms or signs of cervicitis, has been proposed as a complementary approach. In South Africa, an uncontrolled study documented a reduction in STI prevalence after monthly azithromycin PAT.12 In a placebo-controlled trial in Kenya, monthly azithromycin among HIV-seronegative FSWs reduced the incidence of Neisseria gonorrhoeae (NG), Chlamydia trachomatis (CT), and, unexpectedly, Trichomonas vaginalis infections, but had no impact on HIV incidence.13 To determine whether PAT might be effective in controlling cervical infections in West African settings, we conducted a randomized controlled trial in Benin and Ghana. Secondary objectives were to determine the impact of PAT on HIV incidence among FSWs and on NG/CT prevalence among their clients.
MATERIALS AND METHODS
Study Population and Design
The study was designed as a cluster-randomized, double-blind, placebo-controlled trial conducted in Cotonou and Porto Novo (Benin) and in Accra (Ghana). The study population consisted of “seaters” FSWs who work from their homes in well-defined areas. “Seaters” are older, more stable, better organized, and have a much higher prevalence of HIV infection than FSWs who roam the streets or part-time sex workers.14,15 These communities (from 3 to 114 FSWs per site) were aggregated into 18 clusters (one cluster was composed of 1 to 10 FSW communities) based on geographical proximity. During a nine-month period (from March 2001 in Benin and from September 2001 in Ghana), FSWs from these clusters presenting to dedicated clinics were invited to participate. In addition, field workers visited FSWs at home or at prostitution sites to provide information about the trial and seek participation. Inclusion criteria were being currently active in sex work in one of the selected clusters and aged ≥18 years. Exclusion criteria were allergy to study drugs, pregnancy, or intention to leave the city within the next 3 months.
Within each city, the clusters were paired according to the FSW population size and demographic characteristics of FSWs. Within each pair, one cluster was randomly allocated to the intervention and the other to the placebo group using precoded envelopes. A baseline prevalence of gonorrhea of 25% was expected from previous studies in Cotonou and Accra, which had revealed prevalences of 21% and 34%, respectively.14,16 With an average of 35 FSWs per cluster, the required number of cluster pairs was estimated at 8.7 to detect a 50% reduction in NG prevalence, assuming 80% power and 5% α error.
Data Collection at Enrollment
At enrollment, participants were interviewed, a pelvic examination performed, and cervical swabs obtained for NG and CT polymerase chain reaction (PCR). PCR testing was performed at Hôpital Maisonneuve-Rosemont in Montréal using Amplicor (Roche Diagnostics, Canada). For NG PCR, all positive samples by Amplicor were confirmed with a 16SrRNA assay.17 Capillary blood specimens were blotted on a filter paper, stored at 4°C, and transported to the site laboratories. Dried blood spots were punched, eluted in phosphate buffer saline, and tested for HIV antibodies with Vironostika HIV Uni-Form II Ag/Ab (Organon Teknika, Netherlands). Nonreactive samples were considered HIV-seronegative. Reactive samples were tested with Genie II HIV-1/HIV-2 (Bio-Rad, France). Genie II dually reactive samples (to HIV-1 and HIV-2) and discordant samples (Vironostika reactive/Genie II nonreactive) were further tested by INNO-LIA HIV I/II Score (Innogenetics, Belgium). All participants were given free condoms, counseled on risk reduction, and invited to return 2 weeks later for HIV results and posttest counseling.
Monthly follow-up was conducted within FSW communities, alternating with clinic visits (Fig. 1). A questionnaire was administered each month. Cervical samples for NG/CT PCR were collected at clinic visits. Depending on date of enrollment, between 1 and 3 samples could be collected, categorized into the following intervals: 45 to 134 days, 135 to 224 days, and 225 to 315 days postenrollment. HIV serology was repeated during the last month of the study. At planned or intercurrent visits to the clinics, participants presumed to have cervicitis through the syndromic approach were given open-label single-dose ciprofloxacin and a 7-day course of doxycycline, regardless of their randomization group, as per the national STI management guidelines.
Study drugs were azithromycin 1 g and ciprofloxacin 500 mg and their identical placebos. Large-enough batches of either the active drugs or placebos were prepared for each cluster, numbered specifically for a given cluster, and distributed every month to all participating members of that cluster. The intervention consisted of a directly observed intake of azithromycin at months 1, 4, and 7, and ciprofloxacin at months 2, 3, 5, 6, 8, and 9 (Fig. 1). Participants assigned to the placebo group received placebos of azithromycin and ciprofloxacin following the same schedule. Neither study staff nor participants were aware of group assignment until the code was broken at the end of the study. The first month of the study was a run-in period during which the participants' motivation could be assessed and strengthened; administration of study drugs (or placebos) started 1 month later and was conducted on prostitution sites on months 1, 2, 4, 5, 7, and 8, and at the clinics on months 3, 6, and 9. Enrollment continued throughout, and participants enrolled later received the same drug or placebo as other members of their clusters at any given point in time. This regimen had been designed because the prevalence of NG was expected to be higher than that of CT, ciprofloxacin was expected to be more effective against NG than the 1 g dose of azithromycin needed for CT, and generic ciprofloxacin was far less expensive than azithromycin.
Cross-Sectional Survey Among Clients of FSWs
Two cross-sectional surveys were conducted among clients of FSWs, immediately before and after the intervention, at night-time, directly on prostitution sites, using the same approach as in previous studies.6,7 This was by necessity a convenience sample of men who accepted being questioned and tested at the time of a paid sexual intercourse. Urine samples were collected before the intercourse (for NG/CT PCR), and a questionnaire administered after the intercourse. A leukocyte esterase dipstick assay was performed on urine at the time of collection, and a combination of ciprofloxacin and doxycycline given to clients with positive results.6 Twenty condoms were given to all clients, but no other incentive was provided.
Data were analyzed with SAS version 9.1. Proportions were compared with the chi-squared test or Fisher exact test, whereas continuous variables were compared with the Wilcoxon rank sum test. We assessed the time trends in NG and CT prevalence separately in both study groups using chi-squared for trend. For assessment of the impact of the intervention, we used generalized estimation equations accounting for repeated measures over time and for the cluster effect. We compared the prevalence in both groups at enrollment and at different time intervals (45–134 days, 135–224 days, and 225–315 days postenrollment) as well as the difference in prevalence between the same time points and enrollment across the 2 groups. The latter comparison was our main impact measure and was estimated as the relative risk of the interaction term between time and group (antibiotics vs. placebo). These analyses were adjusted a priori for variables that were significantly associated with the group at baseline or with one of the study outcomes (age, HIV status, duration of sex work, price per intercourse, and condom use in last week). The main effect measure, group-time interaction ratio, compares the evolution over time of the prevalence of NG and CT between the intervention and control groups (computed as the ratio of the ratio of prevalence at time X to prevalence at baseline in the intervention group, divided by the corresponding measure in the control group). We also assessed potential interactions in treatment effect according to HIV status at baseline.
The study was approved by the ethics committees of the CHA universitaire de Québec, the Hôpital Maisonneuve-Rosemont in Montreal, and the Ministries of Health of Benin and Ghana. Informed and signed consent was obtained from all participants. Specimens were run and held apart from all personal identifiers other than a laboratory number. Apart from reimbursement of their travel expenses (US $4) and a complimentary soft drink at each planned visit at the clinics, participants were not given any incentive. CT and NG PCR tests had to be batched at the end of the study, and the results could not be used for individual management; by the time these results became available, most participants had received an antibiotic effective against the offending pathogen through syndromic management.
Of 659 FSWs approached for the study, 636 were enrolled (Fig. 2): 162 in Cotonou (8 clusters), 90 in Porto Novo (2 clusters), and 384 in Accra (8 clusters). Baseline demographic and behavioral characteristics of participants, according to study group allocation (n = 296 in intervention group and n = 340 in placebo group), are shown in Table 1. Distribution of participants in study groups was not balanced in each city owing to variations in size of clusters, and participants in the intervention group had been selling sex for longer than those in the placebo group. Distributions of all other variables were similar in both groups. Condom use with clients was similar in intervention and placebo groups but was more consistent among FSWs enrolled in Ghana where 100% (intervention group) and 98% (placebo group), respectively, alleged having always used condoms with recent clients compared with 47% and 51% in Benin. Baseline NG, CT, and HIV prevalence did not vary significantly between intervention and placebo groups. Overall prevalence was 14.7% for NG, 3.5% for CT, and 68.6% for HIV. Prevalence of NG and CT was higher in Cotonou and Porto Novo, whereas HIV prevalence was higher in Accra (Table 2). As anticipated, HIV prevalence increased with age, whereas the opposite was seen for CT. Baseline NG prevalence decreased as the price per intercourse increased. Low condom use was associated with a higher NG prevalence but not with CT and HIV.
At least one follow-up cervical sample was obtained from 80%  and 78%  of participants in the intervention and placebo groups, respectively (Fig. 2). Overall, 27%  of participants provided 1, 28%  provided 2, and 24%  provided 3 follow-up samples. Enrollment being continuous during the study period, the number of potential and actual follow-up visits depended on the date of enrollment. In addition, a participant could miss a visit and attend the subsequent one. Overall, 426/621 (69%) FSWs showed up for the first clinical follow-up visit, 324/554 (58%) for their second visit, and 238/453 (53%) for their third visit. Fifty-four participants (7%) stopped the trial prematurely (intervention group, n = 25; placebo group, n = 29) for the following reasons: moved out of the study area (n = 24), pregnancy (n = 9), death (n = 11), or other reasons (n = 10).
Adherence to Study Drugs and Tolerability
Overall, 71% (3111/4357) of the monthly doses were effectively taken. Adherence was higher among FSWs in the intervention group than in the placebo group (75% vs. 69%, P < 0.001). Azithromycin was taken by 91% of participants before their first clinical visit (intervention 91%; placebo 91%), by 75% within 3 months of their second visit (intervention 76%; placebo 74%), and by 60% within 3 months of their third visit (intervention 62%; placebo 58%). At least 1 dose of ciprofloxacin was taken by 89% of participants before their first visit (intervention 89%; placebo 89%), by 88% within 3 months of their second visit (intervention 92%; placebo 84%), and by 79% within 3 months of their third visit (intervention 79%; placebo 79%). None of the differences between trial arms were statistically significant.
At least 1 adverse effect was reported on 29% of occasions by women in the intervention group (vs. 23% in the placebo group; P = 0.05). The most common were weakness (intervention 8%; placebo 7%; P = NS), diarrhea (intervention 9%; placebo 4%; P = 0.005), epigastric pain (intervention 6%; placebo 2%; P = 0.01), and headache (intervention 3%; placebo 4%; P = NS).
FSWs were invited to seek medical attention at their clinic any time between scheduled visits, whenever needed. During the study, 314 ciprofloxacin tablets combined with a 7-day course of doxycycline (in addition to study drugs) were administered through syndromic management to 217 participants (34%). In the intervention group, 23% of FSWs received 1 course and 12% received at least 2 courses. In the placebo group, corresponding figures were 23% and 11%. Among participants who had at least 1 NG-positive result during the course of the study, 66% (42/64) of FSWs in the intervention group and 52% (46/89) in the placebo group received at least 1 dose of ciprofloxacin through syndromic management (in addition to study drugs). Among those with at least 1 CT-positive result, corresponding figures for doxycycline were 50% (12/24) and 67% (14/21).
Effect of Periodical Antibiotics on N. gonorrhoeae and C. trachomatis Prevalence in FSWs
NG prevalence decreased in both study groups (Table 3), but significantly so in the intervention group only from 13.5% at enrollment to 8.2% (first follow-up visit), 9.1% (second visit), and 5.6% (third visit) (P = 0.01, chi-squared for trend). Corresponding figures were from 15.8% to 12.3%, 11.9%, and 12.5% in the placebo group (P = NS). After adjusting for age, HIV status, duration of prostitution, price per intercourse and condom use, and accounting for prevalence at enrollment and cluster pairing effect, adjusted NG prevalence ratios (intervention vs. placebo) were 0.77 (P = NS), 1.07 (P = NS), and 0.49 (P = 0.05) at the first, second, and third follow-up visits, respectively.
As shown in Table 4, CT prevalence increased in the intervention group at the first follow-up visit (from 3.5% to 6.8%), whereas it decreased in the placebo group (from 3.6% to 1.8%; adjusted prevalence ratio 3.96; P = 0.0005). However, CT prevalence then decreased in the intervention group (3.0% and 1.6%) and remained stable in the placebo group (2.5% and 2.7%). Using the same approach as above, there was no significant impact of the intervention on adjusted CT prevalence ratios at 6 and 9 months. There was no heterogeneity in treatment effect according to HIV status for either of NG or CT.
Effect of Periodical Antibiotics on HIV Incidence in FSWs
End-of-study HIV serology was available for 118 (59.6%) of the 198 initially seronegative FSWs. There were 7 incident cases of HIV (4/65 in the intervention and 3/53 in the placebo groups; incidence rates respectively 19.3 and 23.0 per 100 person-years; P = NS).
N. gonorrhoeae and C. trachomatis Prevalence in Clients
NG prevalence decreased nonsignificantly from 2.2% (8/365) to 1.8% (9/514) among clients visiting FSWs randomized to the intervention group (P = 0.64) and from 4.0% (14/346) to 2.1% (12/575) among clients visiting FSWs in the placebo group (P = 0.08). CT prevalence also decreased nonsignificantly in both groups of clients: from 4.7% (17/365) to 3.3% (17/514) in intervention areas (P = 0.31) and from 4.0% (14/346) to 2.4% (14/575) in placebo areas (P = 0.17).
The West Africa Project to Combat AIDS and STIs set up dedicated clinics for the care of FSWs in Cotonou (1992), Accra (1997), Porto Novo (1999), and other West African cities. In 1998–9, NG and CT prevalence among FSWs were respectively 20.5% and 5.1% in Cotonou and 33.5% and 11.0% in Accra.14–16 The lower NG (14.7%) and CT (3.4%) prevalence that we documented at enrollment reflected the efficacy of our earlier interventions in reducing the burden of STIs among FSWs, through enhanced availability of condoms and better access to syndromic management. This, however, reduced the power of our study, a problem further compounded by decreasing prevalence of NG and CT over time in the control group, itself presumably reflecting study-related uptake of preventive measures. We were able to document a significant impact of the intervention on NG prevalence but only after 9 months, possibly because the exposure to potential reinfection was initially overwhelming. The latter can be estimated as follows. Participating FSWs had, on average, 90 clients per month, 3.1% of which were infected with NG initially. Assuming that 20% of intercourses were unprotected and a 50% per unprotected intercourse probability of male-to-female transmission of NG,18,19 approximately 28% of FSW would get reinfected within 1 month. In such a population with a high prevalence of HIV, the probability of reinfection may be even higher, as HIV enhances the susceptibility to gonococcal infection.20 Furthermore, some FSW may have been reinfected through nonpaying partners with whom condoms were generally not used.6 For this probability of reinfection to decrease, the short-term reduction in NG prevalence among FSWs immediately after each dose of PAT must first have an impact on the prevalence among clients and nonpaying partners. As the latter is influenced not only by incident cases but also by the duration of infectiousness of untreated prevalent gonococcal infections (up to 6 months),16 this reduction may require several months. It is only after the exposure to reinfection has been reduced that a sustained reduction in NG prevalence can be achieved among FSWs.
We could not document a beneficial effect of PAT on CT infection. Unless azithromycin treatment had a paradoxical effect on reinfection by altering the short-term immunity against CT,21,22 the increased risk at 3 months may have been a chance event. As only 3.4% of FSWs were infected with CT at enrollment, we had very low power to measure any protective effect. Furthermore, participants received at most 3 doses of azithromycin during the intervention, and the exposure to reinfection must also have been substantial, given the higher (4.3%) prevalence among clients and the longer duration of infectiousness, which probably compensates for the lower efficacy of transmission during an unprotected intercourse.18
As in Kenya,13 we could not document an effect of PAT on HIV acquisition. In both studies, HIV incidence among FSWs was only 2% to 3% per year either because the established interventions reduced the risk or through a selection process (many FSWs acquire HIV during their first year in the trade).14 It is also possible that in our study population, the contribution of gonococcal infections in enhancing HIV transmission was modest compared with that of bacterial vaginosis and genital herpes, which were not addressed by our intervention.
Accelerating the control of NG infection among FSWs through PAT, further than what can be achieved via enhanced condom use and syndromic management of symptomatic cases, would be a worthwhile endeavor, even if it does not reduce HIV incidence. FSWs represent an important reservoir of NG, the curtailment of which would be expected to lower prevalence among non-SW women, after a similar reduction has been obtained within the bridging population of male clients. It is plausible that the impact of PAT might be more important at some early stage of a comprehensive intervention for FSWs,11 when NG prevalence among FSWs and their clients is higher than those that we documented at the start of this study. Conversely, when, as occurred after this study was completed in Benin23 and Ghana and as documented also in Zimbabwe, NG and CT prevalence among FSWs further decreases to levels not much higher than those of the general adult population, PAT may become futile.24 Ciprofloxacin, a cheap and effective drug, remains an attractive agent for the treatment of gonococcal infections in West Africa, but its in vitro efficacy needs to be monitored closely, as resistance is rapidly emerging in east and southern Africa.25 Azithromycin should be avoided in settings where CT prevalence is relatively low, since its long half-life may facilitate the selection of resistant gonococcal strains.
Our study had limitations. Apart from the lower than expected baseline prevalence of NG, a further limitation was that one-third of participants received at least one course of ciprofloxacin and doxycycline as syndromic management of an episode of vaginal discharge, cervicitis, or pelvic inflammatory disease, which may have somewhat diluted the effect of the intervention. In addition, the dosing schedule used in this study might be too complex to be implemented in practice. In retrospect, we should have also attempted to organize the PAT of nonpaying regular partners of FSW, so as to reduce the probability of reinfection from these men.
In conclusion, PAT among FSWs can reduce the prevalence of gonococcal infection but only after such an intervention has been sustained over a relatively long period of time. Unfortunately, it probably will not further reduce the risk of HIV acquisition among this core group. Although PAT could be more effective in other circumstances, it can not be recommended at present as a routine strategy to control cervical infections among FSWs.
1. Kreiss JK, Koech D, Plummer FA, et al.. AIDS virus infection in Nairobi prostitutes. Spread of the epidemic to East Africa. N Engl J Med 1986; 314:414–418.
2. Piot P, Plummer FA, Rey MA, et al.. Retrospective seroepidemiology of AIDS virus infection in Nairobi populations. J Infect Dis 1987; 155:1108–1112.
3. Plummer FA, Nagelkerke NJ, Moses S, et al.. The importance of core groups in the epidemiology and control of HIV-1 infection. AIDS 1991; 5(suppl 1):S169–S176.
4. D'Costa LJ, Plummer FA, Bowmer I, et al.. Prostitutes are a major reservoir of sexually transmitted diseases in Nairobi, Kenya. Sex Transm Dis 1985; 12:64–67.
5. Nzila N, Laga M, Thiam MA, et al.. HIV and other sexually transmitted diseases among female prostitutes in Kinshasa. AIDS 1991; 5:715–721.
6. Lowndes CM, Alary M, Gnintoungbe CA, et al.. Management of sexually transmitted diseases and HIV prevention in men at high risk: Targeting clients and non-paying sexual partners of female sex workers in Benin. AIDS 2000; 14:2523–2534.
7. Cote AM, Sobela F, Dzokoto A, et al.. Transactional sex is the driving force in the dynamics of HIV in Accra, Ghana. AIDS 2004; 18:917–925.
8. Alary M, Lowndes CM. The central role of clients of female sex workers in the dynamics of heterosexual HIV transmission in sub-Saharan Africa. AIDS 2004; 18:945–947.
9. Alary M, Baganizi E, Guedeme A, et al.. Evaluation of clinical algorithms for the diagnosis of gonococcal and chlamydial infections among men with urethral discharge or dysuria and women with vaginal discharge in Benin. Sex Transm Infect 1998; 74(suppl 1):S44–S49.
10. Pepin J, Labbe AC, Khonde N, et al.. Mycoplasma genitalium
: An organism commonly associated with cervicitis among west African sex workers. Sex
Transm Infect 2005; 81:67–72.
11. World Health Organization. Periodic Presumptive Treatment for Sexually Transmitted Infections: Experience From the Field and Recommendations for Research. Geneva: World Health Organization, 2008.
12. Steen R, Vuylsteke B, DeCoito T, et al.. Evidence of declining STD prevalence in a South African mining community following a core-group intervention. Sex Transm Dis 2000; 27:1–8.
13. Kaul R, Kimani J, Nagelkerke NJ, et al.. Monthly antibiotic chemoprophylaxis and incidence of sexually transmitted infections and HIV-1 infection in Kenyan sex workers: A randomized controlled trial. JAMA 2004; 291:2555–2562.
14. Asamoah-Adu C, Khonde N, Avorkliah M, et al.. HIV infection among sex workers in Accra: Need to target new recruits entering the trade. J Acquir Immun Defic Syndr 2001; 28:358–366.
15. Nagot N, Ouangré A, Ouedraogo A, et al.. Spectrum of commercial sex activity in Burkina Faso: Classification model and risk of exposure to HIV. J Acquir Immun Defic Syndr 2002; 29:517–521.
16. Alary M, Mukenge-Tshibaka L, Bernier F, et al.. Decline in the prevalence of HIV and sexually transmitted diseases among female sex workers in Cotonou, Benin, 1993–1999. AIDS 2002; 16:463–470.
17. Diemert DJ, Libman MD, Lebel P. Confirmation by 16S rRNA PCR of the COBAS AMPLICOR CT/NG test for diagnosis of Neisseria gonorrhoeae
infection in a low-prevalence population. J Clin Microbiol 2002; 40:4056–4059.
18. Brunham RC, Plummer FA. A general model of sexually transmitted disease epidemiology and its implications for control. Med Clin North Am 1990; 74:1339–1352.
19. Hook EW, Hansfield HH. Gonococcal infections in the adult. In: Holmes KK, Sparling PF, Stamm WE, et al., eds. Sexually Transmitted Diseases. New York, NY: McGraw-Hill, 2008:627–645.
20. McClelland RS, Lavreys L, Katingima C, et al.. Contribution of HIV-1 infection to acquisition of sexually transmitted disease: A 10-year prospective study. J Infect Dis 2005; 191:333–338.
21. Brunham RC, Rekart ML. The arrested immunity hypothesis and the epidemiology of chlamydia control. Sex Transm Dis 2008; 35:53–54.
22. Brunham RC, Poubohloul B, Mak S, et al.. The unexpected impact of a Chlamydia trachomatis
infection control program on susceptibility to reinfection. J Infect Dis 2004; 192:1836–1844.
23. Ahoyo AB, Alary M, Ndour M, et al.. [HIV and sexually transmitted disease among female sex workers in Benin] [Article in French]. Med Trop 2009; 69:457–462.
24. Cowan FM, Hargrove JW, Langhaug LF, et al.. The appropriateness of core group interventions using presumptive periodic treatment among rural Zimbawean women who exchange sex for gifts or money. J Acquir Immun Defic Syndr 2005; 38:202–207.
25. Lewis DA, Scott L, Slabbert M, et al.. Escalation in the relative prevalence of ciprofloxacin-resistant gonorrhoea among men with urethral discharge in two South African cities: Association with HIV seropositivity. Sex Transm Dis 2008; 84:352–355.