Article Text

Download PDFPDF

Original article
The impact of peer outreach on HIV knowledge and prevention behaviours of male sex workers in Mombasa, Kenya
  1. Scott Geibel1,
  2. Nzioki King'ola2,
  3. Marleen Temmerman2,3,
  4. Stanley Luchters3,4
  1. 1Population Council, Nairobi, Kenya
  2. 2International Centre for Reproductive Health (ICRH), Mombasa, Kenya
  3. 3Department of Obstetrics and Gynaecology, International Centre for Reproductive Health (ICRH), Ghent University, Ghent, Belgium
  4. 4Centre for International Health, Burnet Institute, Melbourne, Australia
  1. Correspondence to Scott Geibel, Population Council, Box 17643-00500, Nairobi, Kenya; sgeibel{at}


Introduction Targeting most at-risk populations is an essential component of HIV prevention strategies. Peer education programmes have been found to increase HIV knowledge, condom use and safer sex behaviours among female sex workers in Africa and men who have sex with men elsewhere. The authors aimed to evaluate the impact of a peer-driven intervention on male sex workers who sell sex to men in Mombasa, Kenya.

Methods Using time-venue sampling, a baseline survey of 425 male sex workers was conducted in late 2006, after which, 40 peer educators were trained in HIV prevention, basic counselling skills and distribution of condoms and lubricants. A follow-up time-venue survey of 442 male sex workers was conducted in early 2008, and pre- and post-intervention changes were examined. The impact of peer educator exposure on HIV knowledge and condom use was analysed.

Results Positive changes in HIV prevention behaviours were observed, including increases in consistent use of condoms with both male clients (35.9%–50.2%, p<0.001) and non-paying male partners (27.4%–39.5%, p=0.008). Exposure to peer educators (AOR=1.97, 95% CI 1.29 to 3.02) and ever having been counselled or tested for HIV (AOR=1.71, 95% CI 1.10 to 2.66) were associated with consistent condom use in multivariate analysis. Peer educator contact was also associated with improved HIV knowledge and use of water-based lubricants.

Conclusions Peer outreach programming reached highly stigmatised male sex workers in Mombasa, resulting in significant, but limited, improvements in HIV knowledge and prevention behaviours. Improved peer coverage and additional prevention initiatives are needed to sufficiently mitigate HIV transmission.

  • Sex workers
  • men who have sex with men
  • peer education
  • HIV
  • Africa
  • behavioural science
  • commercial sex
  • surveillance
  • substance misuse
  • women issues
  • vaginal infections
  • epidemiology
  • infectious disease
  • antiretroviral therapy
  • STD
  • AIDS
  • genital infection
  • trials

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Within the past decade, there has been increased focus on the HIV epidemic among men who have sex with men (MSM) in Africa. MSM have been found to be more vulnerable to HIV infection than respective general populations,1 and large MSM subgroups of male sex workers have been identified as populations most at risk in some large African cities.2–4 Targeting key vulnerable populations such as MSM and male sex workers is an essential component of HIV prevention strategies, which is becoming increasingly recognised by some national HIV programmes in Africa.5

A number of evidenced-based strategies for sex work and MSM programming have been identified, including promotion of safer sexual behaviours, improving the availability of sexually transmitted infection (STI) prevention and care services, provision of care for those living with HIV/AIDS, health services outreach and peer education programmes.6 ,7 Peer education programmes have been found to increase HIV knowledge, condom use and safer sex behaviours, as well as reduce incidence of HIV and STIs among female sex workers.8 More specifically, peer education has had positive impacts among female sex workers in Africa9 ,10 and MSM in Europe11 and China.12 Evaluations of HIV programming for MSM or male sex workers in African settings, however, are lacking.

Historically, popular opinion perceived that male sex workers in Kenya's coastal region primarily catered to female more than male clients and that the male sex trade was primarily driven by foreign tourism.13 In 2002, formative focus group meetings suggested that male sex workers were a sizeable population in Mombasa and that their risk behaviours included high numbers of male sexual partners, inconsistent condom use, poor knowledge of STI prevention and high alcohol and drug use.14 In response to these concerns, we conducted an initial enumeration, which estimated 739 male sex workers were active, or currently seeking clients, in the Mombasa District area.4

In a subsequent 2006 survey conducted after the enumeration, male sex workers who sell sex to men in Mombasa reported poor knowledge of anal sex as a risk behaviour and high levels of unprotected anal sex and self-reported STI symptoms.15 As a result, evidence-based programming for male sex workers was implemented in Mombasa, and we aimed to evaluate the impact of this programme after 1 year of implementation. Pre- and post-intervention characteristics of the male sex workers are examined, the impact of peer education on consistent condom use—controlling for other factors—is analysed and dose–response effects of peer educator exposure on health-related outcomes are evaluated.


Study design

Utilising time-venue sampling frames derived from the 2006 enumeration exercise,4 we conducted two independent cross-sectional surveys using a pre- and post-intervention design. The baseline survey was conducted from October to December 2006, and after 12 months of intervention implementation, a follow-up survey was conducted from February to April 2008. Due to concerns about confidentiality and stigma, no personal or biological identifiers were recorded during the surveys, and therefore, an indeterminate number of male sex workers participated in both surveys. Both surveys maintained consistent recruitment procedures to facilitate comparative analysis, which are separately described in previously published reports.15 ,16

To summarise, study participants—defined as men aged 16 years or older who had recently sold and were currently willing to sell sex in exchange for money and/or goods—were sampled probability proportional to size from the time-venue sampling frame before both baseline and endline surveys. Trained peer mobilisers were recruited by the study team based on their familiarity and knowledge of the male sex worker population and venues and provided daily compensation for their time and effort. The peer mobilisers approached and identified male sex workers seeking male clients at the community sites, upon which they were invited to participate in the study. If agreed, the peer mobilisers then escorted recruited male sex workers to a secure central location, where a person-to-person behavioural survey was administered by trained interviewers.

Where needed, study participants were provided condoms, water-based lubrication and information or referrals for HIV and STI testing and treatment at local clinics. Respondents were compensated 300 Kenya Shillings (about US$4.50) for their time and return transport.

Intervention activities

Previously existing Kenya Medical Research Institute outreach services and drop-in centre north of Mombasa

Prior to implementation of the baseline survey, the Kenya Medical Research Institute in Kilifi District (KEMRI-Kilifi) began recruitment of MSM for an HIV vaccine preparedness cohort in late 2005. The KEMRI study group trained 10–15 peer mobilisers, who conducted routine outreach activities to approach and identify MSM in and around the Mombasa area. MSM were invited by KEMRI to a drop-in centre located in Mtwapa, approximately 20 km north of Mombasa's city centre, where they were availed HIV information and pre-enrolment counselling, as well as outside medical referrals where necessary. MSM who fully enrolled in the KEMRI cohort were then engaged in routine behavioural and clinical assessments, which entailed frequent visits to the Mtwapa drop-in centre.17

During and following the capture–recapture enumeration in May 2006, which we implemented jointly with KEMRI, a leaflet was distributed to MSM in Mombasa that contained information on anal STIs and HIV prevention.4 KEMRI peer outreach workers continued to distribute this leaflet to MSM through implementation of the baseline survey in October 2006.

Interventions implemented by the International Centre for Reproductive Health after baseline survey

Immediately following completion of the baseline survey, the International Centre for Reproductive Health (ICRH) Kenya opened a drop-in centre in the Mombasa city centre in January 2007. Services provided at the drop-in centre included condom and lubricant distribution, HIV counselling and testing, HIV prevention information and space for relaxation and television viewing.

We recruited 40 peer educators—all either male sex workers or non-sex worker MSM familiar with the sex worker environment—and trained them in HIV prevention in January 2007. In April 2007, all the peer educators were given additional training in basic counselling skills, including six peer leaders who were additionally trained as certified HIV testing counsellors. Later in 2007, the peer educators also attended workshops on alcohol and drug harm reduction related to HIV prevention.

From February 2007 to April 2008, approximately 1900 male sex workers and non-sex worker MSM contacts were recorded by the peer educators. Activities during these contact sessions included brief counselling, health referrals and/or condom/lubricant distribution. In the same period, 823 men were tested at the ICRH drop-in centre, but counselling and testing were available to the general public and same-sex sexual behaviour of the tested men were not recorded. Over 100 000 condoms were distributed to male sex worker and MSM peers, and 8000 5 ml packets of water-based lubricants were distributed.

Measures and statistical analysis

The behavioural surveys elicited information on socio-demographic characteristics, sexual behaviours, prevention knowledge and practices, reported STI symptoms, victimisation to violence and exposure to health services. To assess differences between baseline and follow-up survey indicators, key variables from the data were first examined using frequencies and descriptive procedures. χ2 Tests were used to detect associations among categorical variables, and the Wilcoxon rank-sum test was used to detect differences between medians.

We analysed the 2008 follow-up survey to assess intervention impact on condom use, where the outcome variable of consistent condom use with male clients in the past 30 days was categorised as ‘always’ use of condoms versus ‘sometimes or never’ use. Independent variables were assessed for distribution, content and for initial univariate logistic regression association (p<0.10) with consistent condom use. These variables were then included in an initial multivariate logistic regression model. Variables not significant at the (p<0.05) level were systematically removed from the model using the backwards stepwise method, after which a final model was identified. Independent variable measures assessed for the final model included intervention-related variables: had at least one contact with an ICRH or KEMRI peer educator in the past 12 months (yes or no), ever having visited the ICRH or KEMRI drop-in centre in the past 12 months (yes or no) and ever having been counselled or tested for HIV (yes or no). Other factors assessed for association included: alcohol use as measured by the Alcohol Use Disorders Identification Test (score of 0–19 or score of 20–40 reflecting possible alcohol dependence), number of male clients in the past 7 days (zero to two clients or three or more clients), knowledge of correct and consistent condom use for HIV prevention (yes or no), self-reported burning urination in the past 12 months (yes or no) and victimisation to verbal insults in the past 12 months (yes or no).

To evaluate the dose–response effects of peer educator contact, an independent variable of peer educator exposure was assessed for distribution and recoded into three categories (zero times, one to four times and five or more times). Dependent variables included in the analysis were health-related indicators in the 2008 follow-up survey, based on outcomes that changed significantly since the 2006 baseline survey (p<0.05). These measures included sexual practices with male clients and male non-paying partners (use of condom with last partner, use of water-based lubricant with last partner and ‘always’ use of condoms in the past 30 days), knowledge (knows that HIV can be transmitted via anal sex and knows that only a water-based lubricant should be used with a latex condom) and access to health services (ever been counselled or tested for HIV and ever attended an ICRH or KEMRI drop-in centre). We tested these measures for non-parametric ordered trend18 and also conducted univariate logistic regression to assess ORs across the three categories of peer educator exposure.

All analyses were conducted using Stata V.10.1 (StataCorp). For all univariate and multivariate logistic regression analyses, SEs were adjusted using Stata's svy procedures to compensate for cluster effects among participants contacted at same venues.


A total of 510 and 516 male sex workers were identified and contacted in the 2006 and 2008 surveys, respectively. A total of 425 and 442 interviews were conducted at baseline and follow-up, resulting in 83.3% and 85.7% acceptance rates. As shown in table 1, the age of the follow-up participants (median 23, IQR 21–27) were significantly lower (p<0.001) than male sex workers interviewed at baseline (median 26, IQR 22–31). A smaller but significant (p=0.026) difference in education levels was also detected, as higher percentages of follow-up survey participants had completed primary school. The distribution of religious affiliation among Muslim, Catholic and Protestant participants, however, remained similar.

Table 1

Baseline and follow-up survey comparisons of basic characteristics, sexual practices, knowledge, self-reported STI symptoms and exposure to HIV interventions among male sex workers in Mombasa

Participants who self-identified as bashas, kings or bisexuals were significantly more likely than other participants to be insertive partners during anal sex. Alternatively, other identities—most commonly cited as shogas, queens, gays and homosexuals—were more likely to report being receptive anal sex partners. The distribution of these two groupings, as well as anal sex roles, did not differ significantly between 2006 and 2008.

Some positive changes in HIV prevention behaviours were observed from 2006 to 2008 (see table 1), including increased condom use at last sex with both male clients (57.9%–67.6%, p=0.003) and non-paying male partners (50.9%–61.5%, p=0.027). Increases in ‘always’, or consistent, use of condoms were also detected with male clients (35.9%–50.2%, p<0.001), non-paying male partners (27.4%–39.5%, p=0.008) and non-paying female partners (25.4%–33.3%, p=0.011). Use of water-based lubricants greatly increased with male clients (21.7%–40.9%, p<0.001) and non-paying male partners (25.8%–47.8%, p<0.001).

Significant positive increases were reported in knowing that HIV can be transmitted during anal sex (64.7%–73.3%, p=0.006) and knowing that only a water-based lubricant should be used with a latex condom (19.8%–40.7%, p<0.001). No significant changes were noted in regards to numbers of male clients in the past 7 days, money received from clients, having female clients, knowledge of consistent condom use for HIV prevention or self-reported STI symptoms. A small decrease in having non-paying male partners in the past 30 days was detected (53.2%–46.4%, p=0.045), as well as an increase in female non-paying partners (25.4%–33.3%, p=0.011).

For intervention-related measures, respondents reported significant increases in exposure to HIV counselling and testing (55.8%–67.7%, p<0.001), visitation of an MSM-friendly drop-in centre at ICRH or KEMRI (22.1%–30.3%, p<0.001) and contact with peer educators (22.1%–32.6%, p=0.001).

Intervention effects on consistent condom use

Both intervention-related and other factors were found to be associated with consistent condom use with male clients in the past 30 days (table 2). Non-intervention factors associated with decreased consistent condom use included: having more paying male clients in the past 7 days (AOR=0.38, 95% CI 0.24 to 0.62), possible alcohol dependence as measured by Alcohol Use Disorders Identification Test (AOR=0.50, 95% CI 0.28 to 0.91), self-reported burning urination in the past 12 months (AOR=0.36, 95% CI 0.22 to 0.60) and victimisation to verbal abuse (AOR=0.62, 95% CI 0.41 to 0.94). Men who knew of correct and consistent condom use as an HIV prevention strategy were more likely to report consistent condom use (AOR=3.04, 95% CI 1.23 to 7.49).

Table 2

Intervention and other factors associated with consistent condom use with male clients in the past 30 days (n=442)

Among intervention-related factors, both exposure to peer educators (AOR=1.97, 95% CI 1.29 to 3.02) and ever having been counselled or tested for HIV (AOR=1.71, 95% CI 1.10 to 2.66) were associated with consistent condom use when controlling for the other variables in the final model. Having ever visited the MSM-friendly drop-in centres of ICRH or KEMRI was initially associated with condom use in univariate analysis (OR=1.52, 95% CI 1.09 to 2.12), but drop-in centre attendance became non-significant in multivariate analysis and was dropped from the final model. Only 25.4% of male sex workers who had ever been counselled or tested (n=299, data not shown) received their last test at a drop-in centre. The final multivariate model was statistically significant (F=8.91, p<0.001).

Peer educator dose effects

Increased levels of peer educator exposure were associated with some sexual practices, HIV knowledge and accessing of health services (table 3). Male sex workers who reported one to four exposures (78.0%, AOR=2.13, 95% CI 1.04 to 4.38) and five or more exposures (78.6%, AOR=2.21, 95% CI 1.21 to 4.06) were more likely to have used a condom with their last male clients than those with no exposure (62.4%). ‘Always’, or consistent, condom use with male clients was associated with increased peer educator exposure (44.4%, no exposure; 59.3%, one to four times; 64.0%, five or more times), especially when five or more exposures was reported (AOR=2.22, 95% CI 1.30 to 3.78). Peer educator exposure did not have any significant trend impact on condom use with non-paying male partners.

Table 3

Peer educator dose effects on male sex worker sexual practices, knowledge and counselling and testing (n=442)

Water-based lubricant use with male clients was more likely among respondents reporting one to four peer contacts (44.1%, AOR=2.96, 95% CI 1.80 to 4.86) and five or more contacts (51.2%, AOR=3.94, 95% CI 2.62 to 5.93) than among those with no exposure (21.0%). For water-based lubricant use with non-paying male partners, a significant increase in use was detected (28.0%, no exposure; 42.9%, one to four times; 57.8%, five or more times), especially among those reporting five or more contacts (AOR=3.52, 95% CI 1.83 to 6.75). This coincided with increased associations with correct knowledge about latex condoms from those with no exposure (31.0%) to those with one to four contacts (54.2%, AOR=2.64, 95% CI 1.35 to 5.17) and five or more contacts (65.1%, AOR=4.16, 95% CI 2.25 to 7.69). Respondents were also increasingly likely to correctly know that HIV can be transmitted through anal sex (67.3%, no exposure; 79.7%, one to four times; 89.5%, five or more times), especially among those reporting five or more contacts (AOR=4.15, 95% CI 1.93 to 8.90).

Finally, respondents with five or more peer educator contacts were more likely to have ever been tested for HIV (87.2%, AOR=4.37, 95% CI 2.04 to 9.36) than participants with no exposures (60.9%). Attendance at the ICRH or KEMRI drop-in centres—the base of operations for peer educators—was also highly associated with increased exposure from zero contacts (18.9%) to both one to four contacts (47.5%, AOR=3.89, 95% CI 2.27 to 6.65) and five or more contacts (58.1%, AOR=5.98, 95% CI 3.51 to 10.98).


This study aimed to determine if targeted and sensitive evidence-based interventions could reach male sex workers and improve levels of HIV knowledge and decrease reported sexual risk behaviours. Analysis indicates that levels of correct knowledge about HIV transmission through anal sex, lubricant use and self-reported condom use increased over the 12-month period of evaluation, although overall impact on HIV prevention may have been limited.

For consistent condom use, significant intervention-related predictors were direct contact with peer educators, as well as ever having been counselled or tested for HIV. Drop-in centre attendance was not associated with consistent condom use in multivariate analysis. Indeed, while provision of a drop-in centre as a ‘safe space’ is often perceived as important for stigmatised populations, such drop-in centres may also become a focus of community aggression. Since completion of this study, the KEMRI drop-in centre in Mtwapa and an ICRH drop-in centre in Malindi were targeted by separate community protests. These uprisings necessitated temporary shutdowns of the centres and affected the ability of peer educators to conduct outreach services.

We found that increased peer education exposure positively affected condom use with male clients but not with non-paying male sexual partners. The interpersonal relationship and condom negotiation dynamics among non-paying and cohabitating male and female partners require further examination. Correct knowledge and use of water-based lubricants with all male partners, however, was positively associated with peer educator exposure, as was improved knowledge of the risk of unprotected anal sex.

Given the lack of MSM-related initiatives in African HIV programming, it is not surprising that this study revealed increased access to information, condoms and lubricants. However, only 144 of 442 (32.6%) respondents at follow-up had knowingly interacted with a peer educator—a statistically significant but smaller than expected increase from baseline coverage (22.1%). This coverage would have reached less than half of the 739 male sex workers estimated to be active in 2006.4 This also implies that of the 1900 ‘contacted’ male sex workers/MSM recorded in service statistics, either same male sex workers were recorded multiple times and/or a sizeable proportion of the contacts reached by peer educators were non-sex worker MSM. Reasons for low coverage of male sex workers by the intervention were unclear, and it has been recommended to implementers to better monitor peer educator activities, increase coverage and improve documentation of unique individual contacts.

This study was conducted at time of rapid advancement in knowledge of effective HIV prevention strategies for MSM. Recent results from the international Pre-exposure Prophylaxis Initiative (iPrEx) trial have given credibility to pre-exposure prophylaxis (PrEP) as a potentially effective prevention strategy among MSM.19 Other strategies, such as testing and treating positive MSM to reduce transmission20 and development of rectal microbicides,21 are currently being evaluated. In Kenya, it will be necessary in the future to implement and test these new prevention initiatives to maximise impact towards reducing HIV incidence among male sex workers.

This study and the analysis assumed independence of the cross-sectional baseline and follow-up surveys, when in reality some participants participated in both surveys. The follow-up population was younger and more educated, which may have resulted in some biases in pre- and post-comparisons. All results were self-reported, which may have resulted in some social desirability bias in responses to interviewers. The study design was also quasi-experimental, and the lack of a comparison group makes it difficult to assess the confounding effects of other HIV programmes in the Mombasa area. Additionally, the baseline survey was conducted after implementation of enumeration and KEMRI activities, so this analysis may have underestimated the impacts of the interventions.

In summary, the use of peer outreach programming in Mombasa was a feasible and effective way to reach male sex workers, and significant increases in condom use for anal sex with male clients, use of water-based lubricants and HIV knowledge were observed. However, intervention impacts measured here were marginal and likely not enough to strongly mitigate the HIV epidemic among this population. We recommend that HIV programmes in African cities adopt these interventions, but additional evidence-based prevention and structural interventions are likely needed to meaningfully reduce HIV incidence among male sex workers.

Key messages

  • There is a need to evaluate HIV peer outreach programming for male sex workers who sell sex to men in highly stigmatised and discriminatory African environments.

  • Correct knowledge of HIV transmission through anal sex, lubricant use and self-reported condom use increased over a 12-month period of evaluation.

  • Contact with peer educators and ever having been counselled for tested for HIV were significantly, but marginally, associated with consistent condom use.

  • While behavioural interventions are important, complimentary efforts (eg, pre-exposure prophylaxis, test and treat and others) will be needed to maximise impact and reduce HIV incidence.


The authors would like to thank Dr A Kahindi, Coast Provincial Medical Officer, for his advice and support during implementation of the study and Dr Eduard Sanders, KEMRI-Kilifi, for his support during survey activities, as well as advising on the accuracy of the KEMRI intervention and service descriptions in the manuscript. Special thanks to Agnes Rinyiru, Masila Syengo, Daniel Lang'o and Jerry Okal for supervising various stages of implementation and to the men who participated in the study.



  • Funding This study was funded by the President's Emergency Plan for AIDS Relief through the Office of HIV/AIDS, Bureau of Global Health, US Agency for International Development (USAID), through the Population Council's Horizons Program cooperative agreement of Award No. HRN-A-00-97-00012-00. The opinions expressed herein are those of the authors and do not necessarily reflect the views of USAID.

  • Competing interests None.

  • Ethics approval This study was conducted with the approvals of the Kenyatta National Hospital Ethics and Review Committee and by the Population Council's Institutional Review Board.

  • Provenance and peer review Not commissioned; externally peer reviewed.