Prevalence of HIV and other sexually transmitted infections, and risk behaviours in unregistered sex workers in Dakar, Senegal

Abstract

Objectives: 

To estimate the prevalence rates of HIV and other sexually transmitted infections (STI) among unregistered sex workers, and to describe their sociodemographic characteristics and sexual behaviours, and the reasons why they were not officially registered as sex workers, in order to design specific public health interventions.

Methods: 

A one-stage cluster-sample survey was conducted in Dakar in 2000. Unregistered sex workers were interviewed in randomly selected establishments (official and clandestine bars, brothels and nightclubs), and blood, endocervical and vaginal samples were collected for laboratory diagnosis.

Results: 

A total of 390 women with a median age of 29 years were recruited. One-seventh of them were under the legal age for prostitution in Senegal (21 years). The median length of prostitution was 24 months and 73.5% of the women stated regular prostitution. Three-quarters of the women were found to have markers for at least one infection. The prevalence rates were as follows: HIV-1, 6.0%; HIV-2, 3.6%; HIV-1+2, 0.4%; syphilis, 23.8%; gonorrhea, 22.0%; chlamydial infection, 20.0%; trichomoniasis, 22.4%; candidiasis, 19.0%; and bacterial vaginosis, 28.8%. The main reported reason for non-registration was ignorance of the legal system and its procedures (19.4%); 18.9% of the women refused to register. One-third of the women reported that their clients used condoms inconsistently or never.

Conclusion: 

This survey suggests that a multidimensional public health response is needed in Senegal, comprising legal information, downwards revision of the legal age for prostitution, and specific medical follow-up based on education, condom promotion and management of STI for non-registered sex workers.

Introduction

Commercial sex plays a central role in epidemics of sexually transmitted infections (STI), including HIV, by disseminating these infections among the general population via the clients of female sex workers [¹]. HIV/STI control programs therefore often target these women. Public health interventions combining condom promotion and STI treatment are considered cost-effective in this setting [^2,3] especially when the global HIV seroprevalence is low [⁴]. Higher HIV seroprevalence rates and lower condom use can be found in non-professional than in professional sex workers, emphasising the need to consider both populations in any HIV/STI control programs based on a comprehensive situation analysis [⁵].

In Senegal, a West African country, the HIV seroprevalence remains low (about 1.8% in 2000) in the general adult population [⁶]. The control of STI among female sex workers has been a major government concern for many years [⁷]. Since 1969, prostitution has been officially tolerated among women over 21 years of age, who must be registered and attend specific dispensaries for monthly medical visits that include a genital examination, laboratory tests, counselling and condom delivery. The efficacy of this system is believed to have contributed to the slow increase in HIV seroprevalence and to a reduction in other STI [^8,9]. However, many female sex workers remain outside this system, deliberately or through ignorance. In particular, women under the legal age for prostitution, many of whom start as early as 15 or 16 years of age, cannot register and are therefore deprived of specific medical follow-up. The other, probably numerous, reasons for non-registration remain to be studied. These unregistered sex workers in Senegal therefore might have a higher risk of acquiring HIV/STI than registered sex workers and their management represents the next step for the control of HIV epidemic [¹⁰].

In order to characterize this population before designing specific public health interventions, we conducted an epidemiological survey of HIV and other STI prevalence rates, and also studied sociodemographic characteristics, sexual behaviour, and the reasons for non-registration.

Methods

Study design

This cross-sectional survey was conducted in Dakar, the capital of Senegal, and its suburbs between January and September 2000. Women were recruited in official and clandestine bars (the latter having no licence are illegal in contrast with the former), brothels (illegal) and nightclubs. Given the clandestine nature of the activity studied, we chose a one-stage cluster sampling method which does not require sampling frames that list all the individuals [^11,12]. Clusters corresponded to the establishments in which subjects were enrolled. A census of these establishments was first carried out on the basis of the field knowledge of two welfare assistants who had each worked with Dakar female sex workers for more than 25 years, and information provided by local physicians, female sex workers and police. The required number of clusters was calculated as: C = (p × [1−p] × D)/(s² × b) where D = 1 + (b−1) × ρ, and p is the estimated proportion of the main variable of interest, D is its estimated design effect, b is its estimated average number of responses achieved per cluster, ρ is its estimated rate of homogeneity, and s is the desired standard error [¹²]. This survey focused on HIV/STI prevalence rates. It was estimated that p would be 0.3, b 5 and ρ 0.1; the required precision was 5%, and s was therefore given a value of 0.025. The estimate of p was based on known HIV/STI prevalence rates among registered female sex workers in Dakar (20.0% maximum) and increased because higher prevalences were expected in unregistered sex workers [^8,13]. The estimate of b was based on our knowledge of the field, and ρ was based on published data [¹²]. Thus 94 clusters were necessary. The establishments were then selected by simple random sampling out of the 183 establishments registered. After giving their written informed consent, all unregistered sex workers present in the selected establishments at the time of team visits were enrolled. The Senegalese national ethics committee approved the study.

Data collection

The field team included interviewers, physicians, laboratory technicians, welfare assistants (two of each) and female sex workers’ peer leaders. The welfare assistants, laboratory technicians, one physician and one interviewer were already involved either in the management of the registered female sex workers in Dakar and/or in social activities for female sex workers. Sociodemographic data were collected on a standard questionnaire, together with information on paid sexual activity, knowledge of STI/HIV infection, prophylactic behaviour, medical history, and the acceptability of public health intervention. The questionnaire was developed in conjunction with the field team and other medical and social personnel familiar with female sex workers including a sociologist, and then tested in a random sample of female sex workers. The interviewers were trained in the use of the questionnaire, and the interviewer and physician unfamiliar with female sex workers were trained about behaviours towards these women. Female sex workers’ peer leaders were charged to get in touch with unregistered sex workers. Empiric treatment for suspected STI was given immediately after the medical examination, during which blood, endocervical and vaginal samples were collected anonymously for laboratory diagnosis.

In general, the visits took place in the evening or at night, in the relevant establishments. However, when possible, the examinations and interviews took place at a neighbouring general dispensary or at the specific dispensary where all registered female sex workers in the Dakar region are monitored. Laboratory results were available 1 week after sampling; additional treatment for STI was given when necessary, and women found to be HIV-seropositive were referred to the national antiretroviral program after notification of their status by the well-trained welfare assistants [¹⁴]. Counselling was systematically provided, and condoms were offered to every participant.

Laboratory procedures

Direct microscopic examination of vaginal smears was performed to detect Trichomonas vaginalis and Candida albicans. In addition, C. albicans was also systematically detected by culture on Sabouraud medium. The diagnosis of bacterial vaginosis was based on the vaginal pH, the KOH test and Gram staining. Neisseria gonorrheae and Chlamydia trachomatis were detected in endocervical specimens by a PCR-based method (Roche, Basel, Switzerland). Genital ulcers were sampled and cultured for Haemophilus ducreyi on Mueller–Hinton medium enriched with horse blood. Serologic tests for syphilis included the rapid plasma reagin (RPR, Becton Dickinson, Mountain View, California, USA) and Treponema pallidum hemagglutination (TPHA, Sanofi Pasteur, Chaska, Minnesota, USA). Serologic screening for HIV was based on an enzyme-linked immunosorbent assay (EIA, Innotest, Innogenetics, Ghent, Belgium). All positive samples were confirmed and typed (HIV-1 or 2) using a line immunoassay (Innolia HIV-1+2, Innogenetics, Ghent, Belgium).

Statistical analysis

Data were entered and checked using EPI-INFO 6.04 (Centers for Disease Control and Prevention, Atlanta, Georgia, USA), and analysed with STATA Release 7.0 software (STATA Corporation, College Station, Texas, USA). The analysis took into account the cluster design by use of specific formulae for survey (SVY commands of the STATA 7.0 software), and the design effect was computed. Due to design effect, the confidence intervals (CI) in cluster-sample surveys are expected to be larger than in simple random surveys as individuals belonging to the same clusters often have similar characteristics. Also, the observations were weighted as the actual proportion of included/recorded establishments differed substantially between bars, brothels and nightclubs due to closing of some establishments between the census in 1999 and the survey in 2000, and absence of unregistered sex workers at the time of team visits in some other establishments. Qualitative variables were expressed as percentages and 95% CI. Continuous variables were expressed as mean when the distribution was normal, and otherwise as median and interquartile range (IQR).

Results

Study population

Three-hundred and ninety women were recruited in 80 establishments. Table 1 summarizes their characteristics. They were predominantly young. Sixty-four women (14.6%) were under the legal age for prostitution in Senegal (21 years). Almost all the women were Senegalese; the remainder originated from various other West African countries (Guinea, Cap Verde, Gambia, Guinea-Bissau, Mauritania, Mali, Liberia, Nigeria and Chad). Unmarried, poorly educated and otherwise unemployed women predominated. Most women had started prostitution recently. Overall, three-quarters of the women stated regular prostitution. Most reported few clients in the previous 7-day period.

Table 1: Characteristics of the 390 unregistered sex workers: Dakar, Senegal, 2000.

Prevalence rates of STI, including HIV, and other reproductive tract infections

The STI prevalence rates were high (Table 2). Overall, 39 women were HIV-seropositive (10.0%; CI, 6.0–14.0%). Serologic evidence of active syphilis (RPR+ and TPHA+) was found in 23.8% of the women. Likewise, a candidiasis was detected in 69 women (19.0%; CI, 14.8–23.1%) and a bacterial vaginosis in 109 women (28.8%; CI, 22.2–35.3%). Up to 73.5% of the women (CI, 65.4–81.6%) had markers of at least one infection. The median number of markers per woman was one (IQR, 0–2). One-hundred and eleven women (31.7%) had either gonococcal and/or chlamydial cervicitis, of whom 32.5% carried both organisms; 195 women (53.9%) had vaginitis, due to two organisms in 26.5% of them (trichomoniasis and/or candidiasis and/or bacterial vaginosis) and to three organisms in 1.8%. H. ducreyi was not detected in the only woman who had genital ulceration on physical examination. Only two of the 39 HIV-seropositive women knew they were infected.

Table 2: Prevalence of sexually transmitted infections among unregistered sex workers: Dakar, Senegal, 2000.

Reasons for non-registration

Fifty-one women (13.6%) had been officially registered in the past; of these, 37.7% said they had temporarily stopped prostitution and had not yet re-registered. The most commonly cited reasons for non-registration by the whole study population were ignorance of the legal system or its procedures, postponement of registration, lack of identity papers, rejection of the official system (by the woman or her partner), and a desire for discretion (Table 3).

Table 3: Reasons for non-registration among 375 unregistered sex workers: Dakar, Senegal, 2000.

Knowledge of STI/HIV and prophylactic behaviour

Nearly all of the women (95.7%) had heard about AIDS, but few felt they were at a high risk of acquiring HIV (n = 97; 23.8%) or other STI (n = 146; 36.5%). Among the women who considered themselves vulnerable to HIV or other STI, 60.5% and 61.2%, respectively, attributed their vulnerability to prostitution, while 23.9% and 31.1%, respectively, attributed it to either inconsistent condom use or condom fragility. On the other hand, among the women who did not feel vulnerable to HIV (n = 170; 43.3%) or other STI (n = 178; 44.6%), 74.0 and 74.6%, respectively, said they felt protected by condom use. Twenty-six per cent of the women did not mention heterosexual intercourse as a risk factor for HIV transmission, and 23.0% did not mention condom use as a preventive measure.

Thirty per cent of the women reported that they used condoms with their clients either inconsistently (16.4%) or never (13.9%) and this was supported by the ratio between the stated number of condoms used and the stated number of clients over the previous 7 days (less than one in 34.8% of women). Only 17.3% of the women stated they used condoms systematically with their regular boyfriend.

Acceptability of public health intervention

All but two of the women agreed to regular medical follow-up, and 23.4% said they would like to register with the authorities. When asked where they would like this follow-up to take place, 43.8% of women spontaneously cited the same dispensary as registered female sex workers, and 15.7% another medical facility; 38.4% reported no preference. When specifically asked, 88.4% said they would be willing to be monitored at the same dispensary as registered female sex workers, and 86.6% said they would accept another dispensary. The main reason for refusing to come to the same dispensary as registered female sex workers was the distance (56.8%) while the risk of stigmatization accounted for only 7.0%. Among the women who said they would refuse to come to another dispensary, 71.2% preferred the dispensary used by registered female sex workers.

To evaluate the validity of such a surprising preference for the dispensary used by registered female sex workers, we examined the facilities currently used by the unregistered sex workers for STI episodes: 30.6% reported going to general dispensaries, 25.4% to hospitals, and only 6.8% to the same dispensary as registered female sex workers. The choice was based on acquaintance/confidence in 33.1% of cases, proximity in 18.0%, and competence in 17.6%.

Almost all the women (95.8%) agreed to monthly medical follow-up. However, it is noteworthy that only 17.7% of the women came to get their laboratory results and necessary treatment in this survey.

Discussion

The definition of female sex workers is often difficult in contexts where payment for sex is common. Most relevant studies have relied on women's description of themselves as sex workers. In this survey, we included all unregistered women who stated they had relationships with multiple casual partners for a negotiated payment, irrespective of whether they considered themselves as sex workers. This resulted in the selection of women with a different pattern of exposure to that observed in other studies, notably with a shorter total length of prostitution and fewer clients per unit time [^15–18]. Also, we recruited women in establishments known to harbour sex workers, namely bars, brothels and nightclubs. The women thus recruited may not be representative of the overall population of unregistered sex workers, as they have more opportunities to meet potential clients than do home-based female sex workers for example.

Prevalence estimates and other information were obtained in a cluster-sample survey approach, given the impossibility of identifying all candidate unregistered sex workers. This design is particularly appropriate for this setting, which is frequent in developing countries and clandestine populations [^11,12]. The lower precision of this approach was counterbalanced by the use of appropriate formulae for sample-size determination and statistical analyses. However, precision also depends on the estimates used for sample-size determination. In retrospect, our estimates were in keeping with the situation actually observed. As regards the HIV seroprevalence estimate (our main objective), an average of 4.8 women with laboratory result per cluster were obtained (5 expected) and the design effect was 1.75 (1.40 expected). As a result, the precision of 5% required for the HIV seroprevalence estimate was reached, as for numerous other variables, and almost all the other variables were within 10% of precision.

The validity of answers obtained with questionnaires is debatable, as it is well known that female sex workers tend to reply in accordance with interviewers’ expectations. However, a Gambian study, in which most of the female sex workers were Senegalese, found that the accuracy of their statements depended on the topic [¹⁶]. For example, condom use was stated more accurately than the price per client. The accuracy of the answers to our survey was probably adequate for an epidemiological study, but the answers must be interpreted with care and the data should be considered indicative only.

Compared with recent data on registered female sex workers in Dakar, the HIV seroprevalence was significantly lower in these unregistered sex workers (10.0%; CI, 6.0–14.0% versus 18.9%; CI, 16.5–21.5%) [¹⁹]. Likewise, although non-significant, the prevalence of chlamydial infection was lower (20.0%; CI, 13.7–26.2% versus 28.5%; CI, 25.3–32.0%) [²⁰] in unregistered sex workers. In contrast, laboratory tests showed a significant higher prevalence of syphilis (23.8%; CI, 16.2–31.5% versus 11.1%; CI, 9.2–13.3%) [¹⁹] and a non-significant higher prevalence of trichomoniasis (22.4%; CI, 17.5–27.4% versus 15.4%; CI, 13.2–17.8%) [⁸] in our unregistered sex workers. Also, the prevalence of gonorrhoea was much higher in unregistered sex workers (22.0%; CI, 15.0–29.0% versus 6.7%; CI, 4.9–8.8%) [²⁰] but the PCR-based method used for diagnosis in our study was more sensitive than microscopic examination and culture used in registered sex workers. The lower HIV seroprevalence could be due to lower exposure, with a shorter time in prostitution (means, 3.3 years versus 5.8 years) and fewer clients (means, 1.8 versus 5.0 per week), which were not counterbalanced by lower condom use (65.2% versus 84.0%) or slightly higher STI prevalence rates [²¹]. The women's mean age was similar in the two populations (30.1 years and 31.7 years) and thus does not explain the difference either.

Even this relatively low HIV seroprevalence in a population of female sex workers [^{5,9,15,18,22–26}] is more than five times higher than that in the general population. A recent survey of clients of female sex workers in Dakar showed the role of the former in spreading HIV infection, initially to their steady partners, with whom they had unprotected relationships [²⁷]. In this and our survey, condoms were not used in a substantial proportion of contacts. Regular partners of female sex workers are also highly susceptible to HIV infection through unprotected relationships.

Surprisingly, women under 21 years of age accounted for only 14.6% of the study population, and as few as four of these women mentioned the legal age limit for prostitution as the reason for non-registration. Thus, this legislation does not appear to be the main obstacle to registration, although it should not be overlooked. Besides age, the conditions for registration include presentation of identity card for Senegalese or passport for foreigners but welfare assistants help the self-willed women to get it if necessary. Our survey showed that the reasons for non-registration are multifactorial, ranging from a lack of knowledge of the legal system to simple refusal.

Given the high prevalence of HIV infection and other STI among these unregistered sex workers, and the fact that two-thirds have a frequent sexual work and one-third only rarely use condoms, these women should be targeted by a multidimensional public health campaign. First, information on Senegalese law and its procedures should be provided in the field, and women seeking to register should be helped to do so. Second, the legal age limit for registration should be abolished or at least revised downwards (the legal age of majority in Senegal was reduced from 21 to 18 in 1999). Third, as some women will nonetheless continue to refuse to be registered, they should be offered specific follow-up including education, condom promotion and management of STI by trained multidisciplinary teams of medical personnel and social workers. This could be conducted both at the dispensary used by registered female sex workers and at another dispensary located in the suburbs, independently of the registration. The acceptability and feasibility of such an intervention must be determined by specific operational research, despite the stated agreement by most of the women studied here.

Acknowledgements

The authors thank M. Cissé Thioye, M. Soumaré, K. Gueye, M. B. Basse, D. Sene, P. M. Ndiaye and the female sex workers’ peer leaders for their contribution to the field work.

Sponsorship: Supported by the French national agency for AIDS research (ANRS). Christian Laurent was the recipient of a doctoral fellowship from ANRS.