Estimating HIV Prevalence and Risk Behaviors Among High-Risk Heterosexual Men With Multiple Sex Partners: Use of Respondent-Driven Sampling : JAIDS Journal of Acquired Immune Deficiency Syndromes

Secondary Logo

Journal Logo

Epidemiology and Social Science

Estimating HIV Prevalence and Risk Behaviors Among High-Risk Heterosexual Men With Multiple Sex Partners: Use of Respondent-Driven Sampling

Chopra, Mickey MSc*; Townsend, Loraine MSc*; Johnston, Lisa PhD; Mathews, Cathy PhD*‡; Tomlinson, Mark PhD*‖; O'Bra, Heidi MPH§; Kendall, Carl PhD

Author Information
JAIDS Journal of Acquired Immune Deficiency Syndromes 51(1):p 72-77, May 2009. | DOI: 10.1097/QAI.0b013e31819907de
  • Free



Southern Africa continues to be highly affected by the HIV epidemic with an adult population prevalence that is substantially higher than other parts of Africa and Asia.1 Currently, the national HIV prevalence in South Africa is estimated to be 10.8% but with a great deal of subnational and gender variation. Higher aggregated prevalence rates are found for women between the age of 15 and 49 years (20.2%) compared with men of the same age (11.7%), for Africans (13.3%) compared with other racial groups (less than 2%), and for those who reside in urban informal localities (17.6%) compared with other locality types (around 10%).2

Distinct patterns of HIV distribution, along with close analysis of factors related to declines in HIV incidence in Uganda3-5 and Zimbabwe,6 strongly suggest that the sexual behaviors of men in their mid-20s and older residing in periurban settings are vital to understanding why HIV prevalence is so high. Infection rates are particularly high among young women under 24 years, and having an older male sexual partner is consistently found to be an HIV risk factor for younger women. High levels of concurrent sexual partners (ie, more than 1 overlapping sexual partner in the same period) among key groups of usually older men and younger women along with inconsistent condom use and low levels of male circumcision7,8 are being highlighted as critical factors driving the epidemic in the region.

The use of sexual networks9 to understand the spread of infectious diseases10,11 is also clarifying the specific context of the epidemic. For instance, a recent study that carefully mapped the sexual networks of heterosexual relationships on an island in Lake Malawi found that one fifth of the population was in mutually faithful relationships but that two thirds of the population could be linked by sexual relations over the last 3 years. These chains were held together by decentralized, robust, linkages of “ordinary” (ie, not sex workers, truck drivers, etc.) men and women in sexual networks marked by high levels of long-term concurrent partnerships.12 Studies suggest that such networks are widespread across Southern Africa especially in periurban settings.13-15

In response to the urgent need to monitor and understand high risk groups of men,16 it is necessary to have more information on male social and sexual networks in periurban settings. Traditional sampling methods such as those used at antenatal clinics and in households are either inappropriate or very inefficient in recruiting these men. Easy to use sampling methods, such as general snowball sampling, provide inaccurate estimates; more robust sampling methods, such as time-location sampling, are often logistically difficult to prepare and implement, and expensive to conduct.17 Respondent-driven sampling (RDS) is an approach that has been successfully used to sample specific high-risk groups such as injecting drug users,18,19 men who have sex with men,20,21 and sex workers.22 RDS, a chain referral sampling method, begins with a nonrandomly selected set of seeds (those who initiate the recruitment process) who, through the use of coupons, recruit members of their peer group with whom they share certain characteristics. The central features that distinguish RDS from snowball sampling are that all participants recruit only a set number of respondents thereby limiting their influence on the final sample composition, and that social network sizes of each participant are used to account for difference recruitment. A recent systematic review23 reported 123 successful RDS studies conducted in countries outside of the United States, but it has never been used in a population of heterosexual men at high risk for sexual transmission of HIV.

The aim of this study was to use RDS to measure HIV prevalence and describe key characteristics among periurban heterosexual men engaging high HIV risk behaviors.


Study Design

An HIV prevalence and sexual risk behaviors survey, using RDS,24,25 was conducted from September to December 2006 among a subset of sexually active males in a black African periurban settlement on the outskirts of Cape Town, South Africa (population: 330,000).26 In this survey, eligible persons received a recruitment coupon from a peer, used it to enroll in the survey at a fixed location, completed an interview, provided a biological specimen for HIV testing, and received an incentive and a new set number of coupons with which to recruit their peers. Recruiters received another incentive for each person they recruited into the survey. This process continued for multiple rounds or waves, creating long chains of recruits until the sample size was reached.

Eligible males were 18 years and older, who reported more than 1 female sexual partner in the previous 3 months who was either more than 3 years younger than the participant or below the age of 24. Formative research was conducted prior to the survey, and the results used to nonrandomly select 8 males who fitted the eligibility criteria and who stated that they were able to recruit into the survey other similar males. The initial recruiters, and each participant who completed the survey, received up to 3 recruitment coupons with which to recruit other eligible males. The coupons provided a brief description of the survey, the interview location address, and a phone number with which to make an appointment. The coupons also included a unique number that was used to track who recruited whom and to match the questionnaire and the biological specimens to the participant.

The sample size of 430 was calculated based on an estimated HIV prevalence of 25% among pregnant women in the geographic area27 with a precision of +/−5% and a design effect of 1.5.

Data Collection

Eligible participants received an explanation of the study process and provided written consent before being interviewed by a trained interviewer. Interviews consisted of questions about participants' current socioeconomic status, sexually transmitted infections history, current and past sexual risk behaviors with different types of sexual partners, and whether sex involved any payment of money or goods (food, cosmetics, clothes, transportation, items for children or family, school fees, or cash). Formative research identified the different categories of sexual partners as main partners, defined as steady partners or wives; casual partners, defined as regular partners (outside of steady partner or marital relationships); and 1-time partners, defined as partners with whom men had just a single sexual encounter. Concurrency was assumed if men reported a main sexual partner and casual or one-time partners in the same time period.

Participants were also asked the number of men they personally knew that they could recruit into the survey to estimate their network size. A dried blood spot was collected by a trained nurse according to standard operating procedures, and participants were offered free voluntary counseling and testing and HIV test results on site. Participants who completed the interview received a telephone voucher worth R60 (±US $8), and participants who recruited other eligible participants received telephone vouchers valued at R20 (±US $2.70) for each of up to 3 successful recruits. Successful recruits were required to present a coupon, fulfill eligibility criteria, and complete the survey. Ethical approval was granted by the University of Cape Town Ethics Committee.

Data Analysis

Estimates of proportions and 95% confidence intervals (CIs) were calculated using the Respondent-Driven Sampling Analysis Tool 5.6 (RDSAT) ( This software package generates sample weights to take into account the variations in participants' network sizes (degree weight) and differential recruitment and homophily (recruitment weight).28 In the univariate analyses, we estimated crude risk ratios of HIV status by all covariates separately. The dependent variable was weighted with population weights generated by RDSAT 5.6. All odds ratios and corresponding P values were calculated using STATA, version 9.0.


In total, 468 men presented coupons but 47 were ineligible providing a final sample of 421 men (excluding the initial recruiters) and obtaining a maximum of 13 waves. The median age of the sample was 28.7 years (interquartile range 23.0-32.5 years), and the average network size was 5.64 (range 1-45). Most (94.7%) of the men were never married. Men in this survey were more highly educated and more likely to be employed than those of a similar age group in the last population census in 2001 conducted in the same area (Table 1). The men reported a range of 2-39 sexual partners in the 3 months before the survey with a mean of 6 and median of 5. The median number of sexual partners in the last 3 months classified as main, casual, and one off was 1, 3, and 1, respectively. During the 3 months before the survey, 98% of men had concurrent sexual relationships.

Comparison of Education and Employment Status Between the RDS and 2001 Census Data Among Men in a Periurban Settlement Outside Cape Town, South Africa, 2006

HIV prevalence was 12.3% (CI: 8.3 to 16.9). Being older than 24 years and not using a condom during the last sexual intercourse with a one-off sexual partner were significantly associated with HIV infection (Table 2). More than one third of men reported symptoms of a sexually transmitted infection in the 12 months before the survey.

Sociodemographic and HIV Risk Factors Associated With HIV Status Among Men in a Periurban Settlement Outside Cape Town, South Africa, 2006

Just under two thirds of men (64.0%; CI: 57.4 to 71.5) did not use a condom during last sexual intercourse with their steady partner. A lower proportion did not use a condom at last sex with their casual and one-time sexual partners (35.9%; CI: 29.8 to 44.1 and 42.4%; CI: 33.9 to 50.2, respectively).

Most men (83.3%, CI: 77.4 to 88.5) indicated that their friends would approve if they had sex with a woman who was not their steady partner or wife, and 86.1% (CI: 80.1, 90.5) indicated that their friends would approve if they changed girlfriends often.

More men reported that their casual partners and one-time partners were 5 or more years younger than they were (24.3%; CI: 17.6 to 31.6 and 25.2%; CI 17.1 to 32.4, respectively) compared with the number of men who reported that their main partners were 5 or more years younger than they were (9.2%; CI: 5.8 to 14.4). Participants were asked whether they thought their sexual partners had had sex with them because they expected or had received any form of material goods. Forty-six percent of respondents (CI: 38.0 to 54.3) thought this was so for their main partners; 82.8% (CI: 75.4 to 87.0) for their casual partners; and 90.6% (CI: 84.8 to 95.0) for their one-time partners.

Shebeens (local drinking bars) or taverns were most commonly cited as places where men met with friends for recreation (31.2% shebeens; CI: 25.0 to 38.6 and 54.4% taverns; CI: 46.5 to 59.6, respectively). Most men (81.8%; CI: 76.2 to 87.5) reported drinking more than 5 beers/ciders or tots of alcohol during these visits. Seventy-four percent (CI: 67.1 to 80.7) indicated that they had met a new sexual partner at either of these venues in the past 30 days.


This study has identified and conducted detailed surveillance among a group of heterosexual men who are engaging in extremely risky sexual behaviors in a high HIV transmission setting. Evidently, this group of single relatively well-off men pursuing multiple female partners and concurrent relationships are connected through social networks. They are better educated and more likely to be employed than the wider population of similar aged men residing in the same area. The HIV prevalence of 12.3% (CI. 8.3 to 16.9) among this study population is much higher than that found in general household surveys in the same region.2

The men in this sample reported a large number of sexual partners, most of whom were concurrent. Comparisons between low-level and high-level concurrency populations using mathematical modeling of HIV spread suggest that although the total number of sexual relationships can be similar in both populations, HIV transmission can be as much as 10 times greater in a high concurrency population.29 Ethnographical and historical research suggest that the high cost of marriage has led men to remain single and, in turn, to show their social status through having multiple, concurrent sexual partners, and unprotected sex.30 Most men in our survey also reported that male peers were supportive of behaviors associated with having multiple sexual partners. Peer opinion leader type interventions that aim to shift social norms and have led to decreases in HIV and associated risky behaviors in other populations31,32may be appropriate for this population.33

Our finding that more men reported being 5 or more years older than their casual or one-time partners than their main partners has not been reported elsewhere in the literature. In a recent South African national survey, about one third of female adolescents had partners who were at least 5 years older and significantly more likely to be infected with HIV.34 Twenty-five percent of adolescents from a similar periurban community near Cape Town reported experiencing their sexual debut at an average age of 14.6 years, with someone who was 5 or more years older than them.35 The men in the study reported low levels of condom use especially with their main sexual partners. These types of relationships may result in sexual power imbalances whereby young girls in relationships with older men are less likely than women in relationships with similarly aged men to negotiate safe sex practices.36

Power imbalances in sexual relationships are also exacerbated by the common practice of giving some sort of material goods or cash to sexual partners. The majority of sexual acts, especially with nonregular partners, were accompanied by some sort of exchange of material goods according to our respondents. Also known as transactional sex, this practice is linked to intimate partner violence and risk of HIV through low condom use, and among women, low relationship control, and forced sex.37,38 Our findings indicate that transactional sex was more common with casual and 1-time partners, who were also more likely to be 5 or more years younger than the men, compared with main partners.

These data also suggest that there was a widespread occurrence of dissortative mixing between the men and their female partners. Dissortative mixing describes situations where people with differential risks for HIV (eg, younger low-risk females and older high-risk males) form sexual partnerships that substantially increases the risk of infection in those populations with lower risk.11 Mathematical modeling suggests that dissortative networks with highly active nodes (eg, sexually active heterosexual males who do not use condoms and have multiple sexual partners) are more easily disrupted than assortative (sexual mixing between people with similar HIV risk behaviors) networks. Interventions that dissuade people from entering into dissortative networks or that modify behavior of the core groups will reduce the spread of HIV.39

Possible Limitations

The RDS methodology provides representative estimates of men with a particular set of characteristics based upon our eligibility criteria, and it is therefore difficult to estimate what proportion of the total population of men this group represents. However, the local drinking venues described by the men in our sample as the most common places for meeting each other and acquiring new sexual partners are numerous in the townships. A recent study of men who frequent taverns has found similar attitudes and behaviors as reported here, and it identified more than 380 shebeens and taverns in the same geographical area of approximately 200 km2.40 This would suggest that men with these characteristics and behaviors are widespread.

The age differences used to determine eligibility were reported by respondents. This required that they knew their sex partners' ages with some accuracy. There is also a possibility that partners' ages were deliberately reported inaccurately in order to gain entry into the study.

The analysis software (RDSAT) generated the sample weights taking into account the variations in participants' social network sizes (degree weight) and differential recruitment and homophily (recruitment weight). However, it may be the case that this still underestimates the CIs in the multivariate analysis.


Recent ethnographic work demonstrates the behaviors reported here are not limited to this specific setting of Cape Town but are widespread in many other periurban situations.41,42 We have identified and described an important group of men who are actively engaged in exploiting the intersection of cultural sexual role expectations, differences in socioeconomic status, and gender power differentials to engage in high HIV risk behaviors. This study highlights how this neglected group is fueling the epidemic and bridging high-risk and low-risk populations. For this reason, UNAIDS/World Health Organization have strongly encouraged the development of surveillance systems for this type of population group.16 The use of the RDS approach has allowed us to establish an important baseline to describe this group's HIV risk behaviors and prevalence. The challenge is now to institutionalize such a survey as part of the broader HIV surveillance process and to mobilize resources and political will to impact behavioral changes in this group of men.


M.C., L.T., C.K., and C.M. conceived and designed the study, L.T., C.M., and M.T. managed the fieldwork, M.C., L.T., L.J., C.M., and H.B. conducted the main analysis and M.C., L.T., and L.J. wrote the first draft of the article. All authors contributed to the final article. M.C. is the guarantor of the data and study.


1. UNAIDS/WHO. UNAIDS/WHO AIDS Update:December 2006. Geneva, Switzerland: UNAIDS/WHO; 2006.
2. Rehle T, Shisana O, Pillay V, et al. South African National HIV Prevalence, HIV Incidence, Behaviour and Communication Survey, 2005. Cape Town, South Africa: HSRC; 2006.
3. Low-Beer D. HIV-1 incidence and prevalence trends in Uganda. Lancet. 2002;360:1788. Author reply 1788-1789.
4. Stoneburner RL, Low-Beer D. Sexual partner reductions explain human immunodeficiency virus declines in Uganda: comparative analyses of HIV and behavioural data in Uganda, Kenya, Malawi, and Zambia. Int J Epidemiol. 2004;33:624.
5. Stoneburner RL, Low-Beer D. Population-level HIV declines and behavioral risk avoidance in Uganda. Science. 2004;304:714-718.
6. Gregson S, Garnett GP, Nyamukapa CA, et al. HIV decline associated with behavior change in eastern Zimbabwe. Science. 2006;311:664-666.
7. Bailey RC, Moses S, Parker CB, et al. Male circumcision for HIV prevention in young men in Kisumu, Kenya: a randomised controlled trial. Lancet. 2007;369:643-656.
8. Gray RH, Kigozi G, Serwadda D, et al. Male circumcision for HIV prevention in men in Rakai, Uganda: a randomised trial. Lancet. 2007;369:657-666.
9. Liljeros F, Edling CR, Nunes Amaral LA. Sexual networks: implications for the transmission of sexually transmitted infections. Microbes Infect. 2003;5:189-196.
10. Adimora AA, Schoenbach VJ. Social context, sexual networks, and racial disparities in rates of sexually transmitted infections. J Infect Dis. 2005;191(Suppl 1):S115-S122.
11. Doherty IA, Padian NS, Marlow C, et al. Determinants and consequences of sexual networks as they affect the spread of sexually transmitted infections. J Infect Dis. 2005;191(Suppl 1):S42-S54.
12. Kohler H, Helleringer S. The Structure of Sexual Networks and the Spread of HIV in Sub-Saharan Africa. AIDS. 2007;21:2323-2332.
13. Helleringer S, Kohler HP. Social networks, perceptions of risk, and changing attitudes towards HIV/AIDS: new evidence from a longitudinal study using fixed-effects analysis. Popul Stud (Camb). 2005;59:265-282.
14. Leclerc-Madlala S. Infect one, infect all: Zulu youth response to the AIDS epidemic in South Africa. Med Anthropol. 1997;17:363-380.
15. Wood K, Jewkes R. Dangerous'' love: Reflections on violence among Xhosa township youth. In: Morrell R, ed. Changing Men in South Africa. Pietermaritzburg, south Africa: University of Natal Press; 2001.
16. UNAIDS/WHO. Third Generation HIV Behavioural Surveillance. Geneva, Switzerland: UNAIDS/WHO; 2005.
17. Kendall CK, Kerr LR, Gondim RC, et al. An Empirical comparison of respondent-driven sampling, time location sampling, and snowball sampling for behavioral surveillance in men who have sex with men, Fortaleza, Brazil. AIDS Behav. April 4, 2008 [epub ahead of print].
18. Abdul-Quader AS, Heckathorn DD, McKnight C, et al. Effectiveness of respondent-driven sampling for recruiting drug users in New York city: findings from a pilot study. J Urban Health. 2006;83:459-476.
19. Stormer A, Tun W, Guli L, et al. An analysis of respondent driven sampling with injection drug users (IDU) in Albania and the Russian Federation. J Urban Health. 2006;83(6 Suppl):i73-i82.
20. Ramirez-Valles J, Heckathorn DD, Vazquez R, et al. From networks to populations: the development and application of respondent-driven sampling among IDUs and Latino gay men. AIDS Behav. 2005;9:387-402.
21. Johnston LG, Khanam R, Reza M, et al. The Effectiveness of respondent driven sampling for recruiting males who have sex with males in Dhaka, Bangladesh. AIDS Behav. 2008;12:294-304.
22. Johnston LG, Sabin K, Hien MT, Huong PT,. Assessment of Respondent Driven Sampling for recruiting female sex workers in two Vietnamese cities: Reaching the unseen sex worker. J Urban Health. 2006;83(Suppl 7):16-28.
23. Malekinejad M, Johnston LG, Kendall C, et al. Using respondent-driven sampling methodology for HIV biological and behavioral surveillance in International settings: a systematic review. AIDS Behav. 2008;12:S105-S130.
24. Heckathorn DD. Respondent-driven sampling: a new approach to the study of hidden populations. Social Problems. 1997;44(2):174-199.
25. Heckathorn DD, Broadhead RS, Sergeyev B. A methodology for reducing respondent duplication and impersonation in samples of hidden populations. J Drug Issues. 2001;31:543-564.
26. Statistics South Africa. Census 2001: Results of National Census South Africa. Cape Town, South Africa: Statistics South Africa;2005.
27. Shaikh N, Abdullah F, Lombard CJ, et al. Masking through averages-intraprovincial heterogeneity in HIV prevalence within the Western Cape. S Afr Med J. 2006;96:538-543.
28. Salganik MJ, Heckathorn DD. Sampling and estimation in hidden populations using respondent-driven sampling. Sociol Methodol. 2004;34:193-239.
29. Morris M, Kretzschmar M. Concurrent partnerships and the spread of HIV. AIDS. 1997;11:641-648.
30. Jewkes R, Dunkle K, Koss MP, et al. Rape perpetration by young, rural South African men: prevalence, patterns and risk factors. Soc Sci Med. 2006;63:2949-2961.
31. Kelly JA, Amirkhanian YA, Kabakchieva E, et al. Prevention of HIV and sexually transmitted diseases in high risk social networks of young Roma (Gypsy) men in Bulgaria: randomised controlled trial. BMJ. 2006;333:1098.
32. Kelly JA. Popular opinion leaders and HIV prevention peer education: resolving discrepant findings, and implications for the development of effective community programmes. AIDS Care. 2004;16:139-150.
33. Lewis JJ, Garnett GP, Mhlanga S, et al. Beer halls as a focus for HIV prevention activities in rural Zimbabwe. Sex Transm Dis. 2005;32:364-369.
34. Pettifor AE, Rees HV, Kleinschmidt I, et al. Young people's sexual health in South Africa: HIV prevalence and sexual behaviors from a nationally representative household survey. AIDS. 2005;19:1525-1534.
35. Jaspan HB, Berwick JR, Myer L, et al. Adolescent HIV prevalence, sexual risk, and willingness to participate in HIV vaccine trials. J Adolesc Health. 2006;39:642-648.
36. Pettifor AE, Measham DM, Rees HV, et al. Sexual power and HIV risk, South Africa. Emerg Infect Dis. Emerg Infect Dis. 2004;10:1996-2004.
37. Dunkle KL, Jewkes R, Nduna M, et al. Transactional sex with casual and main partners among young South African men in the rural Eastern Cape: prevalence, predictors, and associations with gender-based violence. Soc Sci Med. 2007;65:1235-1248.
38. Dunkle KL, Jewkes RK, Brown HC, et al. Transactional sex among women in Soweto, South Africa: prevalence, risk factors and association with HIV infection. Soc Sci Med. 2004;59:1581-1592.
39. Doherty IA, Shiboski S, Ellen JM, et al. Sexual bridging socially and over time: a simulation model exploring the relative effects of mixing and concurrency on viral sexually transmitted infection transmission. Sex Transm Dis. 2006;33:368-373.
40. Weir SS, Pailman C, Mahlalela X, et al. From people to places: focusing AIDS prevention efforts where it matters most. AIDS. 2003;17:895-903.
41. Smith DJ. Modern marriage, men's extramarital sex, and HIV risk in southeastern Nigeria. Am J Public Health. 2007;97:997-1005.
42. Wardlow H. Men's extramarital sexuality in rural Papua New Guinea. Am J Public Health. 2007;97:1006-1014.

heterosexual men; HIV/AIDS; respondent-driven sampling; South Africa; surveillance

© 2009 Lippincott Williams & Wilkins, Inc.