1Homosexuality and bisexuality in sub-Saharan Africa, socially condemned and sometimes punishable under law, have rarely been the focus of scientific research. Senegal has been a pioneer in this area, with several sociological and epidemiological studies conducted in the early 2000s as part of the country’s effort to combat AIDS. In this paper, Joseph Larmarange, Annabel Desgrées du Loû, Catherine Enel and Abdoulaye Wade present the results of a new survey carried out in 2007 on 501 men aged 18 years and over, both married or single, but with experience of sexual relations with other men. Using a detailed description of the many types of bisexuality in Senegal, the authors identifed six major “modes of sexual activity” based on the characteristics of the respondents’ sexual partners during the preceding year and at the time of the survey. This typology demonstrates the diversity and complexity of behaviours in terms of age at first sexual intercourse, sex of the first partner, number of partners and type of sexual practice. The differences in HIV prevalence and exposure to violence between the various modes of sexual activity open up new scope for preventive action.

2In Northern countries, many epidemiological and socio-behavioural studies have been conducted on male homosexuals in relation to HIV/AIDS. In sub-Saharan Africa, the subject was ignored for many years because HIV transmission was perceived as being mainly a heterosexual issue and, to a lesser extent, a perinatal one (Chin and Mann, 1988). Although the existence of homosexual practices in Africa has been documented for many years (Crowder, 1959), [1] it was only after 2000 that epidemiological and behavioural surveys were carried out on this population (Baral et al., 2007).

3Senegal is an African pioneer in this field and, starting in the late 1990s, several sociological surveys (Teunis, 2001; Sappe, 2003; Niang et al., 2002) revealed the particular vulnerability of men having sex with men (MSM) [2] in the country. Homosexual practices are strongly condemned and punishable by law in Senegal. Although homosexuality is not mentioned explicitly in any law, an amendment to the Penal code of 21 July 1965 stipulates that “unnatural sexual acts” may incur a five-year prison sentence (Loi de base n° 65-60 du 21 juillet 1965 portant Code penal). Homosexual men live in extreme insecurity and are subjected to domestic, community and institutional violence. They are stigmatized by Senegalese society as a whole and especially by healthcare professionals, whom they are reluctant to consult when symptoms occur.

4Against this backdrop, an epidemiological survey was carried out in 2004 on 462 men selected by snowball recruiting from the homosexual networks of five major cities including Dakar, with the agreement of the Senegal Ministry of Health. The survey showed a 21.5% prevalence of HIV, which is 30 times higher than in the population as a whole (Wade et al., 2005). The Senegal health authorities, working with NGOs, launched a number of programmes specifically targeting men having sex with men. They included provision of medical assistance to people with sexually transmissible infections (STI) and HIV, using specialized healthcare professionals trained in working with this understandably wary (because stigmatized) population. They also took the form of campaigns to increase awareness of sexual risks in the MSM networks and an appeal addressed to all players in public life highlighting the need to take the specific risks of MSM into account.

5In a society where homosexuality is strongly condemned, one of the main political arguments used to justify these programmes was the protection of the population as a whole: Senegal’s efforts in the fight against AIDS would come to nothing if the government failed to address this particular source of the epidemic in the homosexual population, since these men’s heterosexual practices could spread the epidemic to the entire population. The hypothesis whereby bisexual people form a “bridge” from MSM to the population as a whole, has frequently been used, and in a range of contexts. For instance, it came up in several chapters of the collective work edited by Peter Aggleton on bisexuality and AIDS (Aggleton, 1996). In Senegal, it is a public health argument that allows politicians to take charge of a socially stigmatized group while minimizing negative social reactions. However, this hypothesis has not been formally documented nor empirically proven, so must be advanced with caution. The relative weight of bisexuals in the total population is hard to measure and the epidemic can spread simultaneously in different ways (Kahn et al., 1997; O’Leary and Jones, 2006). Nevertheless, the risk to these men’s female partners is higher because of the unprotected heterosexual practices of some bisexuals (Crawford et al., 1996, p. 56).

6Research in Senegal has rarely examined in detail the heterosexual practices of MSM, however, even though they have been mentioned on occasion. Teunis (2001, p. 177) revealed that some MSM are married or have “girlfriends”. Sappe also mentioned bisexual behaviour, but restricted this to heterosexual unions used as a “social cover” (Sappe, 2003, p. 17). The 2000-2001 survey by Teunis in Dakar revealed that 88% of the 250 men interviewed reported having had vaginal sex with a woman at least once in their lifetime (Niang et al., 2003, p. 505). The 2004 epidemiological study confirmed the high frequency of bisexual practices: 94% of respondents had had sex with a woman at some time in their life and 74% in the past 12 months (Wade et al., 2005).
These indicators are insufficient. The concept of bisexuality refers to a whole range of very different realities, both in the practices and in the relationships they imply. Research in other contexts has shown the importance of taking sexual event histories into account, along with changes in sexual orientation and modes of sexual activity (Messiah and Mouret-Fourme, 1993). The term “sexual orientation” may designate both the way the individual defines him/herself (identity), his/her sexual preferences (desire) or the sex of his/her sexual partners (practices). The term “mode of sexual activity”, as used by Messiah and Mouret-Fourme (1993), refers to a more factual definition, namely the sex of the sexual partners over a given period, independently of the identities or desires of these persons.
In 2007 we performed a second epidemiological and behavioural survey as a follow-up to the 2004 one. We treated in detail the question of sexual practices with both men and women, and applied an approach that was both quantitative and qualitative. In this paper we identify the multiple forms of bisexuality observed in Senegal, based on this survey, and we produce an indicator of the current mode of sexual activity. We explore the ways in which various aspects of the respondents’ sexual and social life event histories vary according to the current mode of sexual activity in order to determine whether sexual risk-taking, HIV infection and exposure to violence vary by type of homo- or bisexuality.

I – Study population and survey method

7The main purpose of the 2007 ELIHoS Survey [3] was to measure the prevalence of HIV and STIs and the sexual behaviour of MSM, in order to observe changes since the previous survey in 2004 and evaluate the effects of current programmes targeting this population. The survey was carried out on the initiative of the AIDS/STI Division of the Institut d’hygiène sociale (Institute of social hygiene) in Dakar, [4] with the full approval of the Senegal Ministry of Health, at a time when the country’s health authorities were becoming aware of the need to take all vulnerable populations into account, especially MSM, [5] in the fight against AIDS. The survey was carried out in partnership with two Senegalese teams (the AIDS/STI division of the Institut d’hygiène sociale in Dakar under Dr. A. S. Wade, and the virology laboratory of the Le Dantec Hospital in Dakar under Professor S. Mboup), together with a French team of social science researchers under A. Desgrées du Loû (CEPED). A special MSM consultation clinic was set up in Dakar in 2003 [6] by Dr. Wade’s AIDS/STI division, and then extended to other major cities in the country, with doctors and social workers specially trained in dealing with this population.

8For the 2007 survey, we used the same snowball recruiting method as in 2004. Obviously this did not provide a representative sample of the MSM population over the entire country. However, it was the only possible recruiting method for a survey on a very stigmatized population for which no sampling frame exists. Our analyses of changes in the prevalence of HIV and STIs and of sexual practices between 2004 and 2007 have been published elsewhere (Wade et al., 2010).

9Respondents were recruited by leaders of the MSM community [7] in three locations (Dakar, the capital; Mbour and Thiès, two towns close to the coastal tourist resorts; and Saint-Louis, a medium-sized town in northern Senegal). Recruitment was carried out in different types of places (such as MSM bars and associations) and by word of mouth in various social networks. No financial incentives were offered to respondents apart from the payment of their travel expenses. [8] The MSM leaders, on the other hand, were paid for taking part in the research project.

10Respondents were required to be at least 18 years of age and to have experience of sexual relations with other men. The survey was described as being on the specific healthcare needs of MSM (including HIV and STI). It took place in the MSM health centre, where these men knew they would be well received and could speak in total confidentiality. The men who agreed to take part in the survey already used these health clinics. [9] They were received by a doctor and a social worker who explained the aims and procedures of the survey. After signing a consent form, they replied to a socio-behavioural questionnaire administered by the social worker or the doctor, after which the doctor carried out a clinical examination and took blood and urine samples for biological tests. All cases of sexually transmissible infections received symptomatic treatment. The respondents were asked to return a fortnight later to get their test results. When they returned, those with infections were treated. All treatment was free of charge. Respondents who tested HIV-positive were informed of their status by a specially trained social worker. They were then referred to the HIV department where full biological tests were performed to decide whether or not antiretroviral treatment should be administered. [10] The entire survey was carried out with a guarantee of total anonymity for respondents. The questionnaires and the clinical and biological forms were linked to each other by numbers (that were given to the respondents to obtain their biological test results). The interviewers were trained to abide by rules of strict confidentiality.

11The socio-behavioural questionnaire was administered to 501 men (306 in Dakar, 100 in Mbour/Thiès and 95 in Saint-Louis). It was a young sample (at least 80% were aged under 30, with the mean age being 24.7 years), and fairly well educated (47% had reached secondary level or above). Among respondents, 90% lived with their families and 41% reported that they belonged to an MSM organization (Table 4). Like all snowball samples, ours was biased, with the more “visible” / “accessible” respondents having agreed most readily to take part in the survey. [11] Only 20% of the men in the survey were aged 30 or over, although 51% of respondents reported that their sexual partners were usually aged 30 or above. [12] The young age of the sample was not specific to this survey. For example, in a multi-site study carried out in 2008, mean ages were 25.6 years in Malawi, 24.4 in Namibia and 25.8 in Botswana (Baral et al., 2009).

12Data from the questionnaires was entered into Microsoft Access and the statistical analysis was carried out using SPSS 16.0 software. Initial analysis of epidemiological data showed that HIV prevalence remained stable between 2004 and 2007, but that systematic use of condoms to protect against STIs had become more frequent, doubtless as a result of the prevention campaigns targeting MSN since 2004 (Wade et al., 2010).
The quantitative survey was associated with a complementary qualitative survey on bisexual behaviour. Twelve interviews were carried on men who had reported having one or several female partners, and twelve others on women who were cited by MSM as being female sexual partners. The respondents were recruited on the basis of the quantitative survey. When men stated in this survey that they had sex with both men and women, the interviewer asked their permission to arrange a more in-depth interview at a later date, and asked if the team could contact one of the declared female partners for an interview. [13] These female partners were informed that the interview was part of a reproductive health study. The homosexual practices of the referent man were never mentioned. The interviews were carried out by the French anthropologist, Catherine Enel, co-author of this paper, in a location selected by the respondent. Eleven interviews were held with men in the health centres, in a private room that ensured full confidentiality, and one was held in the respondent’s home. Four of the women’s interviews were carried out in the health centre, one in a bar (which was empty at the time of the interview) and seven in the homes of MSM leaders (who were not present at the interviews).
The women interviewed were aged between 18 and 30. One was married, two were divorced, and the nine others were single. Their educational levels varied from “none” to “two years of higher education”. One was still at school, one was a student in higher education, and two were in training. Three had a job (hairdresser, maid, shopkeeper) and the remaining five were inactive. The men interviewed were aged between 18 and 45, [14] nine were single, two divorced and one married. Two of them were students, six were in regular paid employment (health mediator, small shopkeeper, luggage handler, photographer, qualified apprentice) and one did occasional modelling work. [15]
In the next part of this article we will present the analysis of the bisexuality indicators provided by the ELIHoS study. They are used to construct a composite indicator that can be matched against various elements in the sexual and social life event histories of the men we interviewed. We will then compare these results, where possible, with data from other studies carried out in France and in sub-Saharan Africa.

II – How can bisexuality be measured?

13Bisexuality or MSMW (men who have sex with men and women) is not a uniform phenomenon with a clear definition. The fact of having had sexual relations with a person of the same sex during a given period (practices) does not necessarily imply that the individuals concerned will report being sexually attracted to persons of the same sex and vice-versa. Similarly, desire and practices are distinct from identity, i.e. the fact of defining oneself as a homosexual, bisexual or heterosexual (Bajos and Beltzer, 2008).

14In an African context, the terms “homosexual” or “bisexual” are not necessarily appropriate for capturing sexual identities. In the literature on homosexuality in Senegal, sexual identity is classified into two broad types: the passive homosexual, identified by the term ubbi (also ibbi), who recognizes that he is homosexual, and the active partner who identifies himself by the term yoos, and who does not see himself as a homosexual. The men interviewed told us that yoos means “loose woman”. During the anthropological interviews, men spontaneously used different terms of social identity to define themselves as men with homosexual practices. Some claimed to be “gay” or “branché” (trendy) while others used the term “MSM”, with ubbi and yoos only being used in reply to a question about their specific roles in homosexual relations (Enel et al., 2009). The interviews revealed this absence of systematic links between practice and identity, already observed in other contexts, notably in Africa. In a study in Kampala, Uganda (Kajubi et al., 2008), 64% of the 224 male respondents reported being “gay” and 36% “bisexual”, although 54% had had only male partners over the past five years, 13% mostly male partners but also women, 6% as many male as female partners, and 23% mostly female partners but also male ones. A total of 46% had therefore had at least one female partner in the past five years. The study by Lane et al. (2008) on 199 MSM in South Africa between 2004 and 2005 showed that 12% of the male respondents identified themselves as “bisexuals”, while 25% reported having had sexual relations with a woman, and only 4% reported having a female partner at the time of the survey.

15In the ELIHoS project, the quantitative questionnaire focused on high-risk sexual practices and consequently no questions were asked about desire or identity. The available indicators mainly concerned sexual partners, including those the respondent reported as “regular” [16] during the survey, and the concept of a couple. Table 1 shows these various indicators.

Table 1

The different measures of bisexual practices in 2007

Percentage Number Sexual partner(s) in a lifetime Man/men only 12.6 63 Man/men and woman/women 87.4 438 Sexual partner(s) over the year None 1.2 6 Man/men only 26.1 131 Woman/women only 1.2 6 Man/men and woman/women 71.5 358 Sexual partner(s) over the past month None 22.0 110 Man/men only 47.3 237 Woman/women only 9.0 45 Man/men and woman/women 21.8 109 Regular reported partner(s) No regular partner 18.2 91 Man/men only 29.1 146 Woman/women only 11.4 57 Man/men and woman/women 41.3 207 In a couple No 75.0 376 With a man 18.4 92 With a woman 6.6 33 Overall 100.0 501 Source: ELIHoS survey 2007.

The different measures of bisexual practices in 2007

16The numbers vary considerably according to the period over which bisexuality was measured; 88% of the men interviewed had had sexual relations with a man and with a woman in their lifetimes, and 72% in the year preceding the survey. A very small proportion (1%) reported only having had female sexual partners. However, in the month preceding the survey the proportion of respondents engaging in bisexual practices was quite different, with only 22% having sexual relations with both male and female partners. A further 22% reported having no sexual activity at all during this period, and among those who were sexually active, 39% reported at least one female partner, while 88% reported at least one male partner.

17At the time of the survey, 53% reported having at least one regular female partner and four out of five also had at least one regular male partner. But only 53% of the men who reported having a regular female partner also claimed to have had sexual relations with a woman in the course of the month. [17]

18In reply to the question “Are you in a couple?” 18% of respondents reported being in a couple with a man. This proportion was relatively stable with age (18% of the under-25 age group, 19% of the 25-29 age group, and 15% of those aged 30 and above). Only 33 men (7%) reported being in a couple with a woman and most of these were married (31 out of 33). [18] Indeed, although our sample contained 36 married men, they did not all report being in a couple with a woman, with 3 preferring to say they were in a couple with a man and 2 claiming not to be in a couple at all. The proportion of men in a couple with a woman increased sharply with age (1% before age 25, 6% at ages 25-29 and 22% at age 30 and over), and especially above age 30, the age at which social and family pressure to marry is the strongest.

19This first comparison of the various types of indicators shows that bisexuality is a multiform phenomenon, difficult to grasp using a single measure. Analysis of the in-depth individual interviews confirmed and increased this complexity by revealing that the women declared as “female partners” in the quantitative questionnaire were not necessarily sexual partners (Enel et al., 2009). The male respondents may describe as female partners some women with whom they have friendly or fraternal relations without any sexual relationship, [19] such as a jigéen (sister), an older woman with a “protective” role, or a “lesbian” friend, etc. Furthermore, wives are not necessarily perceived as female partners since their status as mothers commands respect. Prostitutes are also sexual partners who are not reported as female partners when the relationship is of a transactional nature. However, certain men also have non-transactional, friendly or sexual relationships with prostitutes, and may report them as female partners.
Because certain reported female partners are not sexual partners, the level of bisexual activity measured in the quantitative survey may be overestimated. However, this bias was minimized by drawing the interviewers’ attention to the matter and by matching the sexual partners’ reports against the detailed questions on sexual practices, which enabled us to correct a few discrepancies.
Despite these limitations, the rates observed remain very high, especially for a survey using convenience sampling that tends, at least in Western countries, to favour recruitment of exclusive homosexuals (Table 2). In Europe and the United States, the surveys carried out on representative samples of the population as a whole show that the vast majority of men who have had sexual relations with another man at least once in their life have also had sex with a woman. However, this proportion of bisexuals drops to a minority when measuring over the past year (except in the 1992 ACSF Survey in France). More detailed research on changes in the mode of sexual activity over a lifetime (Messiah and Mouret-Fourme, 1993) highlights a two-fold phenomenon. On the one hand, improved social acceptance of homosexuality makes exclusive homosexuality easier. On the other, “temporary” bisexuality appears to be more frequent at the beginning of people’s sexual lives, before individuals turn to either exclusive homosexuality or heterosexuality (Messiah and Mouret-Fourme, 1993; Pollak and Schiltz, 1987 and 1994).

Table 2

Proportions of bisexuals among MSM in various surveys

Country Year Sampling Percentage of bisexuals Source In a lifetime Over the past 5 years Over the past year Over the past 6 months Botswana 2008 Snowball – (a) – – 44 (Baral et al., 2009) Cameroon 2008 Snowball – – – 49 (Henry et al., 2009) Malawi 2008 Snowball – – – 63 (Baral et al., 2009) Namibia 2008 Snowball – – – 51 (Baral et al., 2009) Nigeria 2006 Word of mouth 86 80 65 44 (Allman et al., 2007) Uganda 2004 Respondent-driven sampling 73 46 – – (Kajubi et al., 2008) Senegal 2004 Snowball 94 – 74 – (Wade et al., 2005) Senegal 2007 Snowball 87 – 72 – ELIHoS Canada 1995 Gay clubs and bars, etc., press 58 – 15 – (Weber et al., 2001) Denmark 2006 Gay clubs and bars, etc., press – – 12 – (Cowan and Haff, 2008) France 1991 Gay press – – – 21 ( Pollak and Schiltz, 1994) France 1992 Students Ile-de-France(b) 70 – – – (Faure-Limouza, 1995) France 1995 Gay press – – 9 – (Schiltz, 1997) France 2004 Gay press, special interest websites 40–50 – 8 – (Velter, 2007) Belgium 1993 Representative 85 – 35 – (Sandfort, 1998) France (ACSF)(c) 1992 Representative 97 78 64 – (Messiah and Mouret-Fourme, 1993) France (CSF)(d) 2006 Representative 90 – 20 – (Bajos and Beltzer, 2008) Netherlands 1989 Representative 64 – 29 – (Sandfort, 1998) United States 1989-1990 Representative 86 – 25 – (Rogers and Turner, 1991) United States 1992 Representative 95 52 25 – (Laumann et al., 1994) (a) Data unavailable. (b) 90% of the students interviewed were aged between 19 and 24. (c) Analyse des comportements sexuels en France (Analysis of sexual behaviour in France). (d) Contexte de la sexualité en France (Context of sexuality in France).

Proportions of bisexuals among MSM in various surveys

20In the African context, the pressure to marry and have children combined with the strong social disapproval of homosexuality, may partly explain the high rate of bisexuality “over the past year” among MSM. In Table 2, the bisexuality rate measured in Senegal is higher than that observed in the other African countries. This discrepancy may be due as much to differences in the recruitment of respondents as to any Senegalese specificity. While the proportion of bisexuals “over the past year” remained stable between 2004 and 2007, the share of exclusive homosexuals “over a lifetime” rose from 6% to 13%. That factor, as well as the increase [20] from 11% to 41% (p < 0.01) in memberships of special interest organizations between the two surveys, may suggest that a homosexual community is “developing” in Senegal.
Because of its frequency (more than two-thirds of men), the issue of bisexuality is central to Senegalese MSM. Several researchers have already remarked on the limitations of a binary “homosexual” / “bisexual” categorization (Godelier, 2008). It is therefore important to build a more fine-tuned indicator to express the respondents’ current mode of sexual activity.

III – Expressing the diversity of homosexualities and bisexualities

21The sex of sexual partners over a lifetime does not reflect the current situation of individuals. The fact of already having had sexual relations with a woman did not necessarily imply that the individual was still actively bisexual at the time of the survey. In this survey we did not have detailed life event history data that would have allowed us to retrace the respondents’ various sexual and romantic trajectories. Nor did sexuality over the past month reflect the respondents’ current situation. The period was too short to give an accurate picture of all sexual partners, as illustrated by the fact that 22% of respondents reported no sexual intercourse at all over the period. It would therefore be more appropriate to use the sex of reported partners over the year to categorize the current mode of sexual activity.

22When the qualitative interviews was analysed (Enel et al., 2009), the question of romantic attachment to a man or a woman emerged as the main differentiating factor in understanding the respondents’ relational dynamics. When love is felt for another man, relations between the two men are almost exclusive and involve few other male partners. Relations with women are limited to the wife, fiancée, or betrothed even if the individual has occasional sexual relations with other types of female partners, notably when he feels that he is required to demonstrate his heterosexual “normality” and his “masculinity”, or when taking part in “folies”. [21] Men of this type, who mainly self-identify themselves as “gay” or “branché”, view marriage of convenience as a necessity, either to conceal their homosexuality – which often provokes strong feelings of guilt with regard to religion or the family – or else to put an end to family pressures to marry.

23However, when the romantic attachment is for a woman, sexual relations with other men are more likely to be marked by multi-partnerships involving a variety of practices in a search for maximum sexual pleasure, and/or transactional relations. Such men view heterosexual marriage, even a love marriage, as a way of escaping their homosexual practices which are a source of shame and dissatisfaction. This form of bisexuality is usually associated with a large number of multi-partnerships involving partners of both sexes, often simultaneously.

24The quantitative survey did not provide data about sexual identity, desire, or the respondents’ conception of their love lives since the questionnaire focused on sexual behaviour. However, we can use two indirect variables that are correlated with these notions. First, we knew if the individuals considered themselves to be in a couple with a man, with a woman or not in a couple. We also knew if they reported male or female partners to be “regular”. The fact of reporting regular partners did not imply the absence of other sexual partners.

25We selected five variables:

26• In a couple (no / yes with a man / yes with a woman)

27• Had sexual relations with a man at least one in the year (yes / no)

28• Had sexual relations with a woman at least one in the year (yes / no)

29• Reported at least one regular male partner at the time of the survey (yes / no)

30• Reported at least one regular female partner at the time of the survey (yes / no)
There are 14 combinations of these five variables. The details and the number of men in each are shown in Table 3.

Table 3

Modes of current sexual activity

Modes of current sexual activity Reports being in a couple Reports regular partner(s) Sexual partner(s) over the year Number Homo, no reg – – ♂ 35 Homo, reg M 96 Homo, reg M, not in a couple – ♂ ♂ 66 Homo, reg M, in a couple with a man ♂ ♂ ♂ 30 Bi, reg M 50 Bi, reg M, not in a couple – ♂ ♂♀ 26 Bi, reg M, in a couple with a man ♂ ♂ ♂♀ 24 Bi, reg MF 207 Bi, reg MF, not in a couple – ♂♀ ♂♀ 145 Bi, reg MF, in a couple with a woman ♀ ♂♀ ♂♀ 24 Bi, reg MF, in a couple with a man ♂ ♂♀ ♂♀ 38 Bi, no reg – – ♂♀ 50 Bi, reg F 51 Bi, reg F, not in a couple – ♀ ♂♀ 45 Bi, reg F, in a couple with a woman ♀ ♀ ♂♀ 6 Hetero, reg F, not in a couple – ♀ ♀ 3 Hetero, reg F, in a couple with a woman ♀ ♀ ♀ 3 No sexual relations (abstinent) – – – 6 Overall 501 Homo: exclusively homosexual over the year.Bi: bisexual over the year.Hetero: exclusively heterosexual over the year.Reg M: reported one or several regular exclusively male partners at the time of the survey.Reg F: reported one or several regular exclusively female partners at the time of the survey.Reg MF: reported regular partners of both sexes at the time of the survey.No reg: reported no regular partner at the time of the survey.♂ : Man/men; ♀ : Woman/women; ♂♀ : Man/men and woman/women; – : Neither men nor women. Source: ELIHoS survey 2007.

Modes of current sexual activity

31The dots in Figure 1 provide a visual representation of these 14 combinations in the factor plane formed by the two first axes of the multiple correspondence analysis (MCA) performed with these five variables. The horizontal axis of the factor plane represents sexual activity with women while the vertical axis represents sexual activity with men.

Figure 1

Representation of current modes of sexual activity in the factor plane (two first axes)

Representation of current modes of sexual activity in the factor plane (two first axes)

32The “couple” variable is not very discriminating, notably because three quarters of the sample did not report being in a couple. We therefore decided to group individuals into six categories based on the sex of their sexual partners over the year and their regular partners. For reasons of simplicity, we named these six categories using the terms “Homo” and “Bi” to designate homosexuals and bisexuals in the year, and the terms “reg M”, “reg F”, “reg MF” and “no reg” to designate the sex of regular partners reported by the respondents.
Six individuals had had no sexual intercourse in the past 12 months and six others only had female sexual partners. Being so few in number, these 12 individuals were excluded from the analyses below.

IV – Partners, practices, risks, HIV/STI and exposure to violence by mode of sexual activity

33Is this typology of current modes of sexual activity relevant to our understanding of the sexuality of MSM in Senegal? Do these various modes have distinct sociodemographic profiles? Although the typology refers to the individual’s current situation, is this situation marked by a specific first sexual experience? From a public health point of view, it is necessary to determine whether the current mode of sexual activity has an impact on the individual’s vulnerability to HIV and STI, be it in terms of behaviour, of practice and/or of exposure. In a context of considerable social stigmatization, it is also important to identify the persons most likely to be the victims of violence in order to reduce their exposure to risk.

Sociodemographic characteristics

34The sociodemographic profiles of the various groups are similar (Table 4). No difference is statistically significant, with the exception of marital status and the fact of having one or more children. Indeed, most married men and fathers were “Bi, reg MF” or “Bi, reg F”, in other words, men who reported having at least one regular female partner. Slightly more of the “Bi, reg F” group lived with their family.

Table 4

Sociodemographic characteristics of MSM by current mode of sexual activity

Sociodemographic characteristics Homo, no reg Homo, reg M Bi, reg M Bi, reg MF Bi, no reg Bi, reg F Overall Survey location+ Dakar 71.4 63.5 68.0 59.9 58.0 54.9 61.6 Saint-Louis 0.0 14.6 20.0 21.7 20.0 23.5 18.6 Mbour / Thiès 28.6 21.9 12.0 18.4 22.0 21.6 19.8 Age group+ 18-19 17.1 18.8 22.0 21.3 12.0 21.6 19.6 20-24 48.6 44.8 44.0 34.3 42.0 29.4 38.7 25-29 20.0 27.1 24.0 21.3 24.0 19.6 22.7 30-34 8.6 8.3 4.0 15.9 14.0 23.5 13.3 35+ 5.7 1.0 6.0 7.2 8.0 5.9 5.7 Educational level+ None 11.4 15.6 14.0 13.0 20.0 11.8 14.1 Primary school 45.7 42.7 38.0 40.1 34.0 33.3 39.5 Secondary school 31.4 34.4 42.0 42.5 32.0 47.1 39.5 Higher education 11.4 7.3 6.0 4.3 14.0 7.8 7.0 Activity status+ Inactive 11.4 3.1 4.0 0.8 8.0 7.8 5.5 Student / apprentice 25.7 28.1 30.0 23.7 26.0 25.5 25.8 Fashion / arts / tourism / dressmaking 22.9 30.2 26.0 20.3 24.0 7.8 22.1 Manual worker / shopkeeper / intermediate occupation 40.0 38.5 40.0 51.2 42.0 58.8 46.6 Lives with family+ 91.4 88.5 82.0 92.3 82.0 96.1 89.8 Marital status *** Single 100.0 97.9 100.0 86.0 98.0 88.2 92.2 Married 0.0 0.0 0.0 12.6 2.0 11.8 6.7 Divorced 0.0 2.1 0.0 1.4 0.0 0.0 1.0 Has one child or more*** 0.0 4.2 8.0 21.3 4.0 21.6 13.3 Member of an MSM organization+ 37.1 47.9 46.0 43.0 28.0 35.3 41.5 Has taken part in specific MSM preventative action+ 60.0 64.6 62.0 60.4 50.0 45.1 58.7 Took part in the 2004 survey+ 31.4 34.4 34.0 38.6 36.0 21.6 34.8 Overall 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Number 35 96 50 207 50 51 489 Chi-square test: + p > 0.05; * p ≤ 0.05; ** p ≤ 0.01; *** p ≤ 0.001. Source: ELIHoS survey 2007.

Sociodemographic characteristics of MSM by current mode of sexual activity

35The “Bi, reg M” and “Homo, reg M” categories were slightly younger and more likely to be members of an MSM organization. The younger age explains the slightly larger proportion of students in these two categories. However, there is no great difference in educational levels, unlike in France where exclusive homosexuals over the year had higher educational levels than bisexuals [22] (Messiah and Mouret-Fourme, 1993), or in Uganda, where 34% of the men defining themselves as “gay” had a higher level of education than the 11% who identified themselves as “bisexual” (Kajubi et al., 2008). [23]

First sexual intercourse

36Using French data from 1992 (ACSF Survey), Messiah and Mouret-Fourme (1993) showed that most of the individuals who were bisexual over a lifetime but exclusively homosexual for more than a year had their first sexual intercourse with a man (62%), whereas most of the bisexual individuals over the year had their first sexual intercourse with a woman (77%). Although statistically insignificant, we found similar results regarding the type of first sexual intercourse in Senegal (Table 5). Only a half of the “Homo, reg M” group had already had sexual intercourse with a woman, and their first sexual experience was, for the majority, with a man, whereas the reverse was true for the “Bi, reg F”. The other current modes of sexual activity were in intermediate positions.

Table 5

Type of first intercourse by current mode of sexual activity (among bisexuals over a lifetime)

Type of first sexual intercourse Homo, no reg Homo, reg M Bi, reg M Bi, reg MF Bi, no reg Bi, reg F Overall First intercourse with a man 26.3 52.0 36.0 36.4 30.0 27.5 35.9 At same age with both sexes(a) 26.3 8.0 14.0 14.1 20.0 11.8 14.3 First intercourse with a woman 47.4 40.0 50.0 49.5 50.0 60.8 49.8 Overall 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Number 19 50 50 206 50 51 426 Chi-square test: p = 0.252.(a) First intercourse is determined by comparing the age at first intercourse with a man and the age at first intercourse with a woman. If the two occurred at the same age, we are unable to determine the type of first intercourse. Note: 47% of the “Homo, reg M” group and 46% of the “Homo, no reg” group had never had sexual intercourse with a woman and were not included in this table. Source: ELIHoS survey 2007.

Type of first intercourse by current mode of sexual activity (among bisexuals over a lifetime)

37The Uganda study (Kajubi et al., 2008) found similar results concerning the type of first sexual intercourse, with 49% of the men who described themselves as “gay” having had their first sexual experience with a man, compared with only 9% of “bisexuals”.
Overall, the mean age at first sexual intercourse with a man or a woman was the same (17 years, Table 6). The age at first intercourse with a woman did not differ according to the current mode of sexual activity, as was observed in France. However, the age at first intercourse with a man differed slightly between bisexuals and homosexuals, with the latter having an earlier sexual debut. In France, in 1992, the reverse tendency was observed (Messiah and Mouret-Fourme, 1993).

Table 6

Age at first intercourse by current mode of sexual activity

Age at first sexual intercourse Homo, no reg Homo, reg M Bi, reg M Bi, reg MF Bi, no reg Bi, reg F Overall First sexual intercourse with a man *** Mean 14.9 15.6 17.3 17.1 17.4 18.6 16.9 Median 15.3 16.0 17.6 17.3 17.2 17.9 17.1 Minimum – Maximum 5–22 7–30 7–26 5–31 10–30 7–34 5–34 Standard deviation 4.7 4.1 3.8 4.9 3.4 5.5 4.6 Number 35 96 50 207 50 51 489 Already had sexual relations with a woman in lifetime Percentage 54.0 53.0 100.0 100.0 100.0 100.0 88.0 Number 35 96 50 207 50 51 489 First sexual intercourse with a woman (bisexuals in lifetime only)+ Mean 15.0 16.4 17.2 16.4 17.2 17.3 16.6 Median 14.8 16.5 17.6 16.8 17.2 16.6 16.9 Minimum – Maximum 20–20 6–27 10–23 5–32 10–27 6–34 5–34 Standard deviation 3.6 4.1 3.0 4.1 3.4 4.8 4.0 Number 19 50 50 206 50 51 426 One-way ANOVA (comparison of means): + p > 0.05; * p ≤ 0.05; ** p ≤ 0.01; *** p ≤ 0.001. Source: ELIHoS survey 2007.

Age at first intercourse by current mode of sexual activity

Sexual partners

38As expected, the number of male sexual partners in a lifetime differs according to the current mode of sexual activity (Table 7), the “Homo, reg M”, “Bi, reg M” and “Homo, no reg” groups having on average more than twice as many partners in their lifetimes as the “Bi, reg F” group.

Table 7

Sexual partners in a lifetime by current mode of sexual activity

Sexual partners in a lifetime Homo, no reg Homo, reg M Bi, reg M Bi, reg MF Bi, no reg Bi, reg F Overall Number of male partners (p = 0.076) Mean 14.1 15.4 13.9 11.2 9.7 6.6 11.9 Median 6.7 7.4 4.9 5.3 6.0 3.0 5.5 Minimum – Maximum 1–100 1–120 1–100 1–110 1–60 1–50 1–120 Standard deviation 22.4 21.5 21.5 18.1 10.5 9.2 18.3 Number 35 96 50 207 50 51 489 Already had sexual relations with a woman in lifetime Percentage 54.0 53.0 100.0 100.0 100.0 100.0 88.0 Number 35 96 50 207 50 51 489 Number of female partners (bisexuals in a lifetime) (p = 0.001) Mean 2.4 2.0 3.2 7.1 5.6 5.5 5.5 Median 1.7 1.6 2.4 4.2 3.8 3.5 3.1 Minimum – Maximum 1–10 1–10 1–15 1–100 1–50 1–30 1–10 Standard deviation 2.2 1.9 3.1 11.2 7.4 6.1 8.8 Number 19 51 50 207 50 51 428 p: One-way ANOVA (comparison of means). Source: ELIHoS survey 2007.

Sexual partners in a lifetime by current mode of sexual activity

39Only half the exclusive homosexuals in the past year had had sexual intercourse with a woman in their lifetimes. Like the “Bi, reg F” group, they had had half as many female partners on average.

40Quite logically, the frequency of sexual intercourse, whether with men or women, was correlated to the fact of having one or several regular partners of that sex (Table 8). Thus more respondents from the “Bi, reg MF”, “Bi, reg M” and “Homo, reg M” groups had had at least one male sexual partner over the month, and had more frequent intercourse on average.

Table 8

Sexual partners in the past month and frequency of intercourse by current mode of sexual activity

Sexual activity over the past month Homo, no reg Homo, reg M Bi, reg M Bi, reg MF Bi, no reg Bi, reg F Overall Sexual partners (%)(1) No sexual partner 54.3 27.1 14.0 9.7 34.0 27.5 21.1 Only a woman/women (a) 0.0 0.0 0.0 9.2 6.0 35.3 8.2 Only a man/or men (b) 45.7 62.9 76.0 39.1 48.0 15.7 48.5 Man/men and woman/women (c) 0.0 10.0 10.0 42.0 12.0 21.6 22.3 With a woman (a + c) 0.0 10.0 10.0 51.2 18.0 56.9 30.5 With a man (b + c) 45.7 72.9 86.0 81.2 60.0 37.3 70.8 Mean number of sexual partners Women 0.0 0.0 0.1 0.8 0.2 1.0 0.5 Men 1.1 1.8 1.7 1.5 1.0 0.5 1.4 Mean number of sexual acts With a woman 0.0 0.0 0.1 1.6 0.4 2.1 1.0 With a man 1.2 2.9 2.7 2.5 1.4 0.5 2.2 Number 35 96 50 207 50 51 489 (1) Chi-square: p < 0.001 Source: ELIHoS survey 2007.

Sexual partners in the past month and frequency of intercourse by current mode of sexual activity

Sexual practices over the past month and condom use

41Almost all the respondents who had had sexual intercourse over the past month reported practicing anal intercourse (Table 9). That does not necessarily signify that anal penetration is systematic in homosexual relations, but stems from the fact that respondents did not usually count non-penetrative sexual acts as sexual intercourse. This was also observed in the qualitative interviews (Enel et al., 2009) and is not specific to Senegal (Andro and Bajos, 2008).

Table 9

Sexual practices with a man in the past month by current mode of sexual activity (for 100 men with at least one male partner in the month)

Sexual practices with a man in the past month Homo, no reg Homo, reg M Bi, reg M Bi, reg MF Bi, no reg Bi, reg F Overall Insertive anal penetration*** 37.5 37.1 53.5 66.1 50.0 78.9 56.6 Receptive anal penetration*** 81.3 82.9 67.4 45.2 60.0 21.1 57.2 Anal penetration (nt)(a) 100.0 100.0 100.0 97.6 96.7 100.0 98.6 Active oral-penile sex* 50.0 44.3 46.5 32.1 26.7 15.8 35.8 Passive oral-penile sex+ 25.0 42.9 39.5 43.5 43.3 42.1 41.9 Oral-penile sex+ (a) 62.5 64.3 60.5 54.8 46.7 52.6 56.9 Active oral-anal sex (nt) 6.3 15.7 2.3 4.8 6.7 0.0 6.6 Passive oral-anal sex (nt) 12.5 21.4 7.0 12.5 3.3 5.3 12.4 Oral-anal sex*(a) 18.8 28.6 7.0 15.5 10.0 5.3 16.2 Masturbation+ 37.5 27.1 30.2 23.8 33.3 36.8 28.0 Partouze (nt) 6.3 8.6 7.0 8.3 6.7 0.0 7.5 Sex for money (money given) (nt) 0.0 10.0 4.7 9.5 3.3 0.0 7.5 Sex for money+ (money received) 31.3 41.4 39.5 41.1 30.0 21.1 38.4 Number 16 70 43 168 30 19 346 Chi-square test: + p > 0.05; * p ≤ 0.05; ** p ≤ 0.01; *** p ≤ 0.001. (a) These figures are not the sum of the two previous ones because multiple answers were possible. (nt): not tested (theoretical numbers too small). Source: ELIHoS survey 2007.

Sexual practices with a man in the past month by current mode of sexual activity (for 100 men with at least one male partner in the month)

42The type of anal penetration (insertive or receptive) differs significantly according to the current mode of sexual activity. Receptive penetration predominates among the “Homo, no reg”, “Homo, reg M” and “Bi, reg M” groups, whereas insertive penetration is most frequent among the “Bi, reg F” and “Bi, reg MF” groups, with a more marked contrast among the “Bi, reg F”.

43While the practice of oral-penile sex did not differ significantly from a statistical point of view, we did observe the same tendency, namely that it was more often given by the “Homo, no reg”, “Homo, reg M” and “Bi, reg M” groups and more often received by the “Bi, reg MF” group, and especially the “Bi, reg F” one.

44It therefore appears that the “Homo, no reg”, “Homo, reg M”, and “Bi, reg M” groups are more disposed to sexual practices that are socially associated with a “feminine and homosexual” role, while the “Bi, reg F” tend to engage in practices associated with a “male and heterosexual” role. The “Bi, reg MF” and “Bi, no reg” are in an intermediate situation. This is only a tendency however, since the various practices were observed in all categories.
The numbers were insufficient to reveal possible differences between heterosexual practices over the month (Table 10). Almost all respondents reported vaginal penetration.

Table 10

Sexual practices with a woman in the past month (for 100 men with at least one female partner in the month) by current mode of sexual activity

Sexual practices with a woman in the past month Bi, reg M Bi, reg MF Bi, no reg Bi, reg F Overall Vaginal penetration (nt) 100.0 98.1 100.0 100.0 98.7 Sex for money (money given) (nt) 60.0 13.2 22.2 6.9 14.1 Sex for money (money received) (nt) 0.0 3.8 0.0 0.0 2.7 Number 5 106 9 29 149 Chi-square test: + p > 0.05; * p ≤ 0.05; ** p ≤ 0.01; *** p ≤ 0.001. (nt): not tested (theoretical numbers too small). Source: ELIHoS survey 2007.

Sexual practices with a woman in the past month (for 100 men with at least one female partner in the month) by current mode of sexual activity

45Sex for money is less frequent with women than with men. Only four men reported having received money in exchange for heterosexual sex whereas almost 40% did so for homosexual sex. These monetary exchanges associated with sexual relations may refer to different realities. While they may correspond to a prostitutional relationship, the exchange of “gifts”, which may take the form of cash, can also occur in stable love relationships (Antoine and Nanitelamio, 1990; Thiriat, 1999).

46For each sexual practice reported over the past month, respondents were asked if they had “systematically”, “often”, “occasionally” or “never” used a condom. The detailed results per current mode of sexual activity are presented in the Appendix (Tables A.1 and A.2). The systematic use of a condom, for the same practice, does not vary significantly by current mode of sexual activity.
However, since the frequency of intercourse with a man does differ from one mode of sexual activity to another (Table 8), the proportion of men having had high-risk anal intercourse in the past month (Table 11) was significantly higher among those who had a regular male partner (18%-23% versus 7%-11%). Thus the “Homo, reg M”, the “Bi, reg M” and the “ Bi, reg MF” have more high-risk sexual relations with men than the others.

Table 11

High-risk anal intercourse with a man in the past month by current mode of sexual activity

High-risk anal intercourse with a man in the past month Homo, no reg Homo, reg M Bi, reg M Bi, reg MF Bi, no reg Bi, reg F Overall Percentage 11.4 22.9 22.0 18.8 10.0 07.8 17.4 Number 35 96 50 207 50 51 489 Chi-square test: p = 0.106. Note: The men who had had high-risk anal intercourse in the past month were those who reported at least one act of anal intercourse without systematically using a condom for this practice. Source: ELIHoS survey 2007.

High-risk anal intercourse with a man in the past month by current mode of sexual activity

47Overall, condoms were more frequently used in anal penetration with a man than in vaginal penetration with a woman (75% versus 62%) and oral-penile sex with a man was rarely protected (13%). In a study on bisexual men in France, R. Mendès-Leite and C. Deschamps (1997) showed that the management and perception of risk varied according to the partner’s gender, and that disparities were greater for regular partners than for occasional ones. In a survey in Uganda (Kajubi et al., 2008), 26% of the men who defined themselves as “gay” had had unprotected anal intercourse over the past six months, compared with 14% of those who defined themselves as “bisexual”.

Prevalence of HIV and various sexually transmitted infections

48Numbers were insufficient to highlight any differences for gonorrhoea, chlamydia or syphilis (Table 12). The prevalence of genital herpes (HSV-2) was the same (approximately 20%) for the various modes of sexual activity.

Table 12

Prevalence of HIV and various STIs by current mode of sexual activity

Sexually transmitted infections Homo, no reg Homo, reg M Bi, reg M Bi, reg MF Bi, no reg Bi, reg F Overall HIV* 40.0 25.0 18.0 20.4 26.0 9.8 21.9 Gonorrhoea+ 0.0 4.2 4.0 1.9 0.0 3.9 2.5 Chlamydia+ 0.0 2.1 4.0 4.4 4.0 2.0 3.3 Genital herpes (HSV 2)+ 20.0 20.8 16.0 20.9 22.0 19.6 20.3 Syphilis+ 2.9 4.2 4.0 4.9 0.0 0.0 3.5 Number 35 96 50 206(a) 50 51 488 Chi-square test: + p > 0.05; * p ≤ 0.05; ** p ≤ 0.01; *** p ≤ 0.001. (a) Results are missing for one person. Source: ELIHoS survey 2007.

Prevalence of HIV and various STIs by current mode of sexual activity

49However, the prevalence of HIV was significantly lower among respondents belonging to the “Bi, reg F” group (10%) who had had fewer sexual relations with men, whether in the past month (Table 8) or in a lifetime (Table 7). Furthermore, the majority practiced insertive, rather than receptive, anal penetration (Table 9), and the probability of HIV transmission per act for insertive anal penetration is between 4 and 25 times lower than for receptive anal penetration. [24] In a study conducted in Malawi, Namibia and Botswana, Baral et al. (2009) showed that defining oneself as “homosexual” or “bisexual” is associated with a greater risk of being HIV-positive (p = 0.06) than for men who define themselves as “heterosexual”.

50After controlling for age group, survey location and number of male partners over a lifetime (Appendix A.3), the “Bi, reg F” still had a lesser risk of being infected by HIV. The “Homo, no reg” group, however, had a greater risk. Although at the time of the survey they had less frequent sexual intercourse with a man (Table 8), they had as many sexual partners over a lifetime on average as the “Homo, reg M” or the “Bi, reg MF” groups (Table 7) and they practiced receptive anal penetration as often as the latter (Table 9). Further studies taking into account respondents’ entire sexual life histories would be required to identify the factors behind this higher prevalence of HIV.

Exposure to violence

51Against a background of stigmatization and criminalization of homosexuality, as is the case in Senegal, the modes of sexual activity we have defined do not appear to be associated with different levels of exposure to homophobic violence (Table 13).

Table 13

Violence sustained by current mode of sexual activity (%)

Type of violence sustained Homo, no reg Homo, reg M Bi, reg M Bi, reg MF Bi, no reg Bi, reg F Overall Violence at first sexual intercourse First intercourse with a man was forced+ 11.4 14.6 8.0 7.7 4.0 15.7 9.8 First intercourse with a woman was forced+ 0.0(a) 2.0(a) 0.0 2.4 0.0 0.0 1.4(b) Violence in a lifetime Forced sexual act(s)+ 34.3 35.4 38.0 34.8 28.0 27.5 33.7 Physical homophobic attack(s)+ 25.7 18.8 14.0 15.0 16.0 13.7 16.4 Number 35 96 50 207 50 51 489 Chi-square test: + p > 0.05; * p ≤ 0.05; ** p ≤ 0.01; *** p ≤ 0.001. (a): Calculated on the 51 “Homo, reg M” and the 19 “Homo, no reg” groups having had at least one sexual relation with a woman. (b): Calculated on the 428 men who had had sexual intercourse with a woman at least once. Source: ELIHoS survey 2007.

Violence sustained by current mode of sexual activity (%)

52The violence reported in a lifetime, whether forced sexual acts or physical attacks of a homophobic nature, does not differ significantly, possibly due to insufficient numbers. Nevertheless, there tend to be fewer victims of forced sexual intercourse in a lifetime among the “Bi, reg F” and “ Bi, no reg” groups. In the Ugandan study (Kajubi et al., 2008), the experience of violence or forced sexual acts did not differ according to the respondents’ identity (“gay” or “bisexual”).

53In Senegal, the differences observed in homophobic aggression were more related to membership of an MSM organization, since 25.4% of members of such organizations were victims of aggression compared with 10.1% of non-members (p < 0.001, exact Fisher test). Among members of MSM organizations, the “Homo, reg M” and “Homo, no reg” groups were the most exposed to violence, and especially the exclusive homosexuals over a lifetime (45.0% versus 23.2%, p = 0.036; Table not shown here). Among the MSM who were not members of such an organization, the “Homo, no reg” were the most exposed to violence [25] (22.7%), with the “Bi, reg F” reporting few attacks (3.0%), while the proportions ranged between 8% and 14% in the other four categories.
Although we do not have the necessary elements to explain this, it is probable that the men most exposed to violence are those with the greatest risk of being “identified” as homosexuals, whether by their clothing, gestures, and/or their militant commitment.

Discussion-Conclusion

54Bisexuality in sub-Saharan Africa is very different from that observed in Western countries. In the West, a majority of individuals who have experimented with same-sex sexual relations, have also, in the course of their lifetimes, had sexual relations with a person of the opposite sex. It would seem that after a period of “temporary bisexuality”, most settle into exclusive homosexuality or exclusive heterosexuality. This leaves a “bisexuality over the year” rate of between only 20%-35% among MSM. [26] This does not imply that bisexuals over the year form a homogenous group, since they include individuals in a sexual experimentation phase (temporary bisexuality), persons living a “double life” because they cannot or will not come to terms with exclusive homosexuality, or men and women who are attracted to both sexes and want to enjoy a bisexual lifestyle. Lastly, measuring bisexuality over the year may conceal a form of “sequential bisexuality” corresponding to individuals who successively have relations with a man and a woman.

55In the African context, exclusive homosexuality over the year is far rarer. In Senegal, it concerns just under one quarter of the men interviewed, despite the use of convenience sampling that would tend to over-represent exclusive homosexuals. This may be explained in part by the high prevalence of “permanent bisexuality” among MSM, in a context where homosexuality is condemned by law, where there is considerable social discrimination against homosexuality, strong social and religious pressures to marry and have children, and only embryonic special interest organizations.

56However we observe that the African situation is changing. Anthropological studies on homosexuality have been carried out for decades, but have been relatively discrete. The issue of male homosexuality [27] in Africa was only recently placed on the international agenda in the context of the fight against AIDS. The issue of African MSM arose during the 2004 International AIDS Conference in Bangkok, and was developed further at the Mexico Conference in August 2008 (Broqua, 2008) and at the African conference (CISMA) in Dakar in December 2008. Although associations dearly identified as MSM or presented as human rights organizations have existed for years, a francophone network of associations called Africagay [28] was not formed until 2007. In Senegal, several organizations were set up between 2004 and 2007. [29] The proportion of respondents who were members of an MSM organization rose from 11% in 2004 to 41% in 2007. A dozen such organizations were mentioned. The Senegalese government, via the AIDS/STI Division, has established specific action programmes and launched a number of preventive campaigns. Our study therefore took place against a rapidly changing backdrop, marked by a gradual increase in the visibility of homosexual practices and identities in Africa and in Senegal, often in the context of the fight against AIDS. Furthermore, we note that the proportion of men who reported themselves as exclusively homosexual over a lifetime rose from 6% to 13% between 2004 and 2007. Although these men are still a minority in Senegal, the proportion doubled in the space of three years.

57Nevertheless, the majority of the men interviewed in both MSM surveys (2004 and 2007) reported having had sexual relations with both men and women in the year prior to the survey. From a public health perspective, especially for the prevention of HIV and STIs, it is important to take into account the diversity of individuals’ trajectories, practices and relations in order to suggest appropriate preventive strategies. Most of the epidemiological studies on homosexual and bisexual men, especially in the field of HIV/AIDS, have focused on their sexual practices with other men, and consequently ignored the heterosexual practices of certain MSM.

58Our results highlight the diversity of bisexual situations among MSM in Senegal and show that these situations cannot merely be characterized by the partners’ sex. Using simple indicators collected by means of a closed questionnaire (sexual partners over the year and regular partners at the time of the survey), we were able to compose a six-group category of the current modes of sexual activity that proved to function effectively and may assist in devising risk-prevention messages adapted to each situation.

59The men who had only homosexual intercourse over the year and had no regular partner at the time of the survey, were the group most affected by HIV (40%). One-third of them reported having received money for sexual intercourse over the past month. Since they did not report having a regular partner, this probably signifies a large proportion of prostitutional sex, which constitutes a risk factor for HIV.

60Among the men who reported bisexual practices over the year, the prevalence of HIV infection varied according to the sub-groups defined by the mode of sexual activity at the time of the survey. Only 10% of bisexuals over the year with regular female partners at the time of the survey were infected by HIV, whereas this proportion rose to 20% or more for bisexuals with regular male partners at the time of the survey, or no regular partners.

61More of the bisexuals over the year with regular female partners at the time of the survey were married and fathers of a child or children, compared with other MSM. The majority had their sexual debut with a woman and their first intercourse with a man much later. Consequently they had had fewer male sexual partners in the course of their lifetimes and fewer had had homosexual intercourse over the past month. They were more likely than the other categories to engage in sexual practices associated with what might be qualified as “masculine” or “heterosexual” roles (insertive anal penetration, passive oral-penile sex, etc.). For equivalent sexual practices they used condoms as often as the other categories, but were less exposed to high-risk anal penetration because of the lower frequency of their homosexual relations, and were consequently less likely to be infected by HIV. However, this category should not be neglected in terms of prevention since these men have the most heterosexual sexual relations and, in 40% of cases, do not systematically use condoms during sexual intercourse with women.

62With regard to public health and HIV prevention, our results highlight the importance of taking heterosexual practices into account in preventive action specifically targeting MSM. This is generally not the case at present. Our results also show that the differences between the current modes of sexual activity, in terms of risk-taking, exposure to HIV and prevalence of HIV, are not due to differences in condom use from one category to another, but rather depend on the frequency of sexual activity and the individual’s participation in different sexual, partnership, and love networks, which need to be taken into account in action programmes and explored more fully in future surveys.

63Clearly, our categorization of current modes of sexual activity has its limitations. For one thing it does not integrate identity or sexual attraction, not covered in our study. Also, it only covers the period of the survey, whereas sexuality evolves over the course of a lifetime. Individuals may go through several periods in which their partners will be of one sex or another, or both. The transition from one situation to another may be a gradual process that takes the form of intermediate states, just as being in a couple may be a succession of “fuzzy states and complex trajectories”, to use the title of a collective work edited by P. Antoine and E. Lelièvre (2009). [30] The categories are therefore “porous”, with some individuals on the borderline between two modes of activity. The advantage of our typology, however, is that instead of placing individuals into separate boxes it highlights several distinct profiles according to data available from the ELIHoS Survey. It can therefore be improved or adapted to the needs of each survey and each context. Collection of more detailed data about the patterns of sexual activity and partners over a lifetime would enrich this typology with a life history dimension that we were unable to include here.
Nevertheless, this typological exercise has enabled us to show that while it is essential to take bisexuality into account in all studies about men having sex with men in Africa, bisexuality cannot simply be classified in terms of a homosexual-bisexual dichotomy. The types of partnerships with men and with women, the frequency of sexual intercourse, the affective, social and emotional aspects, as well as the sexual practices themselves, are multiform and we must be able to express that complexity.
Translated by Krystyna Horko.