AIDS and Behavior

, Volume 13, Issue 5, pp 860–865

Sexual Behavior of Female Sex Workers and Access to Condoms in Kenya and Uganda on the trans-Africa Highway

  • Chester N. Morris
  • Sheldon R. Morris
  • Alan G. Ferguson
Original Paper

DOI: 10.1007/s10461-008-9431-z

Cite this article as:
Morris, C.N., Morris, S.R. & Ferguson, A.G. AIDS Behav (2009) 13: 860. doi:10.1007/s10461-008-9431-z


Female sex workers and their clients remain a high risk core group for HIV in Africa. We measured sexual behavior of a snowball sample of female sex workers (FSW) along the Trans Africa highway from Mombasa, Kenya to Kampala, Uganda and surveyed the availability of male condoms at 1,007 bars and lodgings in Kenya along the highway trucking stops where transactional sex occurs. There were 578 FSW one month sex diaries analyzed, 403 from Kenya and 175 from Uganda. Kenyan FSW had a median of 45 sexual acts per 28 days compared to 39 sex acts per 28 days by Ugandan FSW (P < 0.05). Condom use by FSW for all sexual liaisons was 79% in Kenya compared to 74% in Uganda. In multivariate analysis, adjusting for repeated measures, Kenyan FSW were more likely to use a condom by an adjusted odds ratio of 2.54 (95% confidence interval 1.89–3.41) compared to Ugandan FSW. Condom use with regular clients was 50.8% in Uganda compared with 68.7% in Kenya (P < 0.01). The number of sex workers reporting 100% condom use was 26.8% in Kenya and 18.9% in Uganda (P < 0.01). Bars and lodges in Kenya compared to Uganda were more likely to: have condom dispensers, 25% versus 1%, respectively (P < 0.01); distribute or sell condoms, 73.9% versus 47.6% (P < 0.01); and have more weekly condom distribution, 4.92 versus 1.27 condoms per seating capacity (P < 0.01). Our data indicate that in both countries condom use for FSW is suboptimal, particularly with regular partners, and greater condom use by Trans African highway FSW in Kenya compared to Uganda may be related to availability. Targeted interventions are warranted for FSW and truck drivers to prevent transmission in this important core group.


HIV Sexual behavior Sex workers Transport workers Africa 


East Africa continues to provide some hopeful indications that the AIDS epidemic may be slowing suggested by a reduced HIV prevalence that was associated with a change in sexual behaviors in Uganda and Kenya (Hallett et al. 2006; Cheluget et al. 2006).

However the sources of this data are limited and poorly representative of higher risk subpopulations such as female sex workers (FSW) and their truck driver clients that from the beginning the AIDS epidemic in East Africa and elsewhere have been identified as a core transmitter group for HIV (Plummer et al. 1991; Orubuloye et al. 1993; Ramjee and Gouws 2002). Despite their importance there is no routine surveillance of FSW and their clients for HIV seroprevalence. The most recent published estimates of HIV prevalence for FSW and truck drivers were 27.6 and 14.8% in Kenya and 47 and 13.5% in Uganda, respectively (Kissling et al. 2005). Truck drivers are a mobile bridge group that connects to the general population, making this core especially suited to targeted interventions for control of the epidemic (Alary and Lowndes 2004; Morris and Ferguson 2006; Pisani et al. 2003).

In Uganda the national HIV prevalence peaked at over 15% in the early 1990s but appeared to steadily decrease thereafter (Stoneburner and Low-Beer 2004; Hallett et al. 2006). The decrease of HIV in Uganda has been attributed to a nationwide effort to curb the epidemic that broadly changed sexual behavior. In the most recent 2004–2005 Uganda HIV/AIDS Behavioral Survey that looked at 18,000 men and women the national adult HIV prevalence was estimated at 7% (Uganda Health Survey, 2005). While this prevalence is encouraging the sexual behaviors reported such as concurrency were concerning. Amongst the sexually active age group of 15–49, 18.2% of women and 39.4% of men reported engaging in extra marital sex in the past 12 months. When asked about their last extra marital sexual partnership only 40.5% of women and 49.5% of men said they used a condom. The survey found that women with more than 2 sexual partners over 12 months had a much higher HIV prevalence at 20.8% but did not specify FSW as a subgroup.

The HIV epidemic in Kenya peaked later than Uganda in 1990s with a lower overall HIV prevalence of 10% in adults and by 2003 was down to 7% (Cheluget et al. 2006), although HIV prevalence is not equally distributed across the country when divided by sub-populations and geographic location (Kenyan HIV/AIDS Data Booklet and (December 2005). Similar to Uganda, the overall decline may be the result of Kenyans adopting safer sexual behavior such as increased condom use with casual partners (Kenyan Demographic Health Survey 1998, 2003). In 2003 almost 24% of Kenyan women surveyed said they used a condom the last time they had casual sex, compared with 15% five years earlier. The proportion of men and women with more than one sexual partner reduced by more than half from 1993 to 2003, and more young men and women were also delaying their sexual debut.

To further understand how sexual behavior has been affected in the subgroup of FSW and their clients in this paper we describe sexual behavior among sex workers and condom access in places of transactional sex along the highway from Mombasa to Kampala in these two countries. The aim of this study is to contribute information that can be used in formulating targeted prevention measures for HIV.


This was a prospective observational study of sexual partnerships and condom use in female sex workers in the highway corridor between Kampala, Uganda and Mombasa, Kenya.


Female sex workers working on the Mombasa-Kampala transport route in East Africa and workers or managers at bars and lodgings along the highway corridor.


The 30 sites chosen for the FSW sample had the largest FSW populations on this transport corridor as determined by an estimation of sex worker numbers described elsewhere. (Ferguson and Morris 2007). Bars and lodgings at 47 sites were sampled representing all overnight stops for transport workers on this corridor (Fig. 1).
Fig. 1

Sites of transactional sex on Mombasa-Kampala highway: numbers of sex workers as enumerated at each site and number of overnight trucks. The figure shows the numbers of FSWs (bars) and the mean numbers of trucks parked overnight (circles). As can be seen there is the greatest number of FSWs at border crossings in Busia, Kenya and Malaba, Uganda


A sex worker coital diary was used to quantify condom use and behavior with clients at 30 major spots for transactional sex on the trans-Africa highway (Ferguson et al. 2006). The diaries were designed to run for a consecutive 28-day period so as to encompass a biological month that would negate significant temporal variation. The time of menses was also documented.

As most transactional sex on the highway occurs in bars and lodgings that line the corridor, these enterprises were surveyed for condom access. For each bar or lodging, a structured questionnaire targeted at managers or owners of these businesses was completed. Quantitative data on condom sales and distribution free was collected. For bars, and lodgings that included condom sales by brand, free condoms distributed and the presence of a condom dispenser.


Three staff trained in the methodology of snowball sampling and interviewing were responsible for data collection at all sites. The fieldwork covered the Nairobi-Mombasa segment first, and then proceeded sequentially to Kampala. The snowball sampling included recruitment of FSW at each highway site to complete a coital diary. FSWs who volunteered were asked to return 1 day after recruitment to be consented as per ethical review board approval in each country. To ensure data quality volunteers were instructed in the completion process of the diaries by the field team at each site using a sample diary. All FSW subjects were asked to return 2 days after commencing their diaries to have the team review how to complete the diary and discuss any barriers to completion. At the collection day, individual diaries were reviewed with each individual to validate the identity of the diary participant. At a designated time the participants were asked to return to meet at the same location for diary submission. Variables recorded by sex workers included daily client number, occupation, liaisons per day, number of condoms used each liaison, and sex acts for each liaison (Ferguson and Morris 2007). A liaison was a temporally isolated sexual encounter that could include a number of sex acts. Classification of clients into ‘casual’ and ‘regular’ was self-defined by the sex worker in filling out the diary. FSW also recorded whether the liaison occurred at the site of the site where they were trained or at another site along the highway or elsewhere. The same trained staff interviewed the managers and employees in the bars and lodgings premises and validated seating and condom data. As part of a broader study on transactional sex each establishment was also visited during evening hours of operation for patron counts and validation of questionnaire responses (Ferguson et al. 2006).

Data Analysis

For comparison between countries binomial distributed data was tested by the Chi-square test, normally distributed continuous data was compared by the Student t test and continuous data that was not normally distributed was compared with the Wilcoxon test. For analysis of condom use by FSW at the event level of the liaison a generalized estimating equation logistic regression was performed with each FSW as a repeated subject using an exchangeable correlation matrix for factors that were captured in the diaries that may confound the relationship of FSW country with condom use. Factors that had a P-value equal or less than 0.1 in univariate analysis were then entered in a multivariate regression model in a backwards elimination process with an elimination criterion for variables of P > 0.05.


Diaries were distributed to 857 sex workers in the 30 sites shown in Fig. 1 and 658 (76.8%) were returned. 79 diaries were incomplete, leaving an end sample of 578. A total of 1,007 bars were surveyed at 47 sites, 717 in Kenya and 290 in Uganda.

Table 1 shows the differences between Kenya and Uganda for key indicators on condom availability at bars and lodges. In total 39 sites in Kenya and 8 in Uganda had bars and lodgings surveyed. Of Kenyan bars and lodges along the highway 73.9% had on-site condom sales or free distribution compared to 47.6% in Uganda (P < 0.01). When condom dispensers were counted there was a large discrepancy between the countries with only 1% of bars and lodgings in Uganda having dispensers compared with 25% in Kenya (P < 0.01). A total of 97,933 condoms were sold and distributed free from bars and lodgings each week on the highway. When these condoms are divided by seating capacity the mean weekly condom distribution is much higher in Kenya compared to Uganda, 4.92 versus 1.27 condoms per seating capacity per week (P < 0.01).
Table 1

Condom access in bars and lodgings in Kenya and Uganda at hot spots of transactional sex along the highway between Mombasa and Kampala






No. hot spots where data collected



Total No. bars/lodgings surveyed



Total seating capacity of bars/lodging




%Bars/lodges selling/distributing condoms



χ = 21.3


%Bars/lodges having condom dispenser



χ = 65.4


Number of condoms per bar/lodging dispensed per week



t = 6.4


The FSW diary participants had a median age of 27 years old, with a range from 15 to 50 years and one-third were aged less than 25 years old. FSW in Uganda were significantly older than their counterparts in Kenya with a median of 27 years old compared to 25 years old (Table 2). By country, there was no significant difference in the number of different clients in 28 days. However, the countries had significant differences in total number of sex acts and number of liaisons with Kenyan FSW having a higher median number of both. A significantly higher number of Kenyan FSW reported 100% condom use at every liaison compared to Ugandan FSW (26.8% versus 18.9%). With analysis of condom use at the level of the liaison Kenyan FSW were more likely to use a condom than Ugandan FSW when adjusted for repeated measures in individuals (Table 3). Younger age FSW were less likely to use a condom during a liaison in those less than 25 years old compared to those 30 years old and over. Having a casual partner appeared to be strongly associated with condom use. Having a truck driver as partner compared to other occupations, having sex on-site compared to off-site and having menses at the time of the liaison did not seem to affect the use of a condom.
Table 2

Characteristics of Kenyan and Ugandan female sex workers at hot spots of transactional sex along the highway between Mombasa and Kampala




Wilcoxon Z statistic or chi-square


Diaries analyzed



Total number liaisons recorded



Median age (years)



Z = 3.3497


Median monthly different clients



Z = −0.7181


Median monthly liaisons



Z = −2.6542


Median monthly number of sex acts



Z = −2.1594


%FSWs with 100% condom use with every liaison

26.8% (108/403)

18.9% (33/175)

χ = 4.1727


Median number of condoms per 28 days



Z = −2.7894


Median number of condoms per liaison



Z = −4.2735


Table 3

Predictors of condom use by female sex workers at hot spots of transactional sex along the highway between Mombasa and Kampala


Total liaisons with condom use (%)

Univariate odds ratio

Multivariate adjusted odds ratio

FSW country


7962 (79.2)

1.25 (1.03–1.51)*

2.35 (1.76–3.15)*


2953 (73.9)




    Less than 25

3780 (75.8)

0.82 (0.65–1.04)*

0.65 (0.48–0.89)*


3238 (77.8)

0.96 (0.75–1.21)

0.81 (0.59–1.11)

    30 or older

3897 (79.5)



Casual client

6552 (89.9)

8.81 (6.46–12.01)*

5.65 (4.66–6.84)*

Regular client

4356 (64.5)



Place of liaison

    At hotspot

4486 (78.4)

1.04 (0.92–1.18)


    Not at hotspot

6429 (77.2)



Partner occupation

    Truck driver

3438 (79.3)

1.12 (0.85–1.49)



7477 (76.9)




300 (72.3)

0.96 (0.59–1.56)


No menses

10590 (77.8)


P ≤ 0.1

In the multivariate model adjusting for age and casual partners Kenyan FSW were more likely to use a condom compared to Ugandan FSW with an adjusted odds ratio of 2.35 (95% confidence interval 1.76–3.15). Having a casual partner has an adjusted odds ratio of 5.65 (4.66–6.84) for use of a condom during a liaison compared to a regular partner. Casual clients made up 82% of clients but participated in half of the liaisons (52%) therefore averaging fewer sexual acts per individual sex workers than regular clients (2.98 versus 13.20). Much of the lower condom use in Uganda was due to regular partners; condom use for casual partner was 90.1% in Kenya and 89.5% in Uganda (P = 0.4), but for regular partners it was 68.7% in Kenya and 50.8% in Uganda (P < 0.01).


In East Africa there is now an emerging consensus that the generalized HIV epidemic is declining in countries such as Kenya and Uganda. The reasons for this declining epidemic are debated but likely related to behavior change in the general population. The behavior of the most high risk members of these countries is less well studied as they are not easily reached by representative surveys. In this study, we measured the sexual behavior of FSW along the Trans Africa highway between Mombasa and Kampala and found FSW and their clients remain at risk for HIV transmission through unprotected sexual encounters suggested by only 26.8% of Kenyan FSW and 18.9% of Ugandan FSW reporting 100% condom use. FSW reported the lowest condom use with their regular partners. Our data suggests that less than desirable levels of condom use may be related to condom availability in the venues that these partners meet along the highway because, compared to Kenya, Ugandan FSW and their clients had more unprotected sex and had lower condom availability. We estimate that there were 97,933 condoms sold and distributed from bars and lodgings each week on the highway compared to a calculated 114,093 FSW sex acts based on our average reported number of sex acts per month and an estimate of 3,020 FSW along the highway in Uganda and 5,655 in the Kenya (Morris and Ferguson 2006).

Given the dynamics of the current HIV epidemic and the existing literature on core and bridge populations there is reason for concern that the current decline in the generalized epidemic could be eventually be reversed by a lack of targeted interventions (Lowndes et al. 2002). Our data would support that the geographic areas frequented by transport workers along the Trans Africa highway could be ideal locations where targeted prevention measures could be employed. Although the situation in Kenya is better than Uganda with respect to condom access, both countries could benefit from coordinated targeted programs for sex workers and transport drivers that would increase condom use and therefore reduce transmission HIV and other sexually transmitted infections. Prevention messages that are marketed to the general population to choose abstinence and be faithful are not effective for sex workers due to their economic needs. Even if safe sex messages linked to condom promotion aimed at the general population reach the sex workers of Kenya and Uganda, without access to condoms there will be no effect. Targeted interventions for the FSW should include increased condom availability, voluntary HIV counseling and testing sites along the highway, medical evaluation and treatment for sexual health and programs to help women transition into other employment.

A common finding to both countries is the differential condom use by casual and regular clients of sex workers. This is important as regular clients of sex workers have as many or more sexual interactions and use condoms less than causal clients. Therefore the risk of HIV transmission would be higher with regular clients. Further studies need to evaluate the HIV prevalence in these two groups and there needs to be a marketing campaign to make the FSW aware of the need to protect themselves from all clients.

Limitations of this study include bias in selecting participants for the sex worker diary. The sample was not random but generated through snowballing of known sex workers in highway communities. Alternative sampling of hard to reach populations such as respondent driven sampling may be useful in future studies. Another bias is the over-representation of more literate sex workers as the completion of the diary requires some basic skills of literacy and comprehension. Condom breakage or failure was not directly asked in the diary but would provide important information. Other limitations include the geographic representation of chosen sites as they only included 8 sites in Uganda from the Kenyan border to Kampala. However the differences in behavior and access would be tend to be less than rest of Uganda because the sites of this study were closer to Kenya. A larger sample of Ugandan sites of transactional sex would have been preferred but was not possible given funding constraints.

In conclusion, we have demonstrated suboptimal condom use and access along the trans-Africa highway in Kenya and Uganda. We discovered that the lowest condom use by FSW was found to be with their regular partners. Since sex workers and other high-risk groups, such as truckers, were catalysts in the spread of the HIV epidemic in these countries and that their behavior seemingly remains high risk it is important that targeted programs that enhance condom distribution and encourage condom use for all clients of FSW should be implemented.

Copyright information

© Springer Science+Business Media, LLC 2008

Authors and Affiliations

  • Chester N. Morris
    • 1
    • 2
  • Sheldon R. Morris
    • 3
  • Alan G. Ferguson
    • 1
  1. 1.Department of Medical MicrobiologyUniversity of ManitobaWinnipegCanada
  2. 2.Aurum Institute for Health ResearchJohannesburgSouth Africa
  3. 3.Division of Infectious Diseases, Department of MedicineUniversity of CaliforniaSan DiegoUSA

Personalised recommendations