HIV EPIDEMICS IN SUB-SAHARAN Africa had their origin in urban areas. Urbanization in the 1970s and 1980s could well have been a major factor determining the onset of population-wide spread of the (long preexisting) HIV virus. In the urbanization process, many young adult males went to towns to look for employment. In growing cities like Nairobi, the preponderance of men over women, the lack of formal employment for women, and the social disruption resulting from the large number of migrants resulted in commercial sex to an extent not known before.1,2 During this period, HIV started to spread within and between urban centers, through mobile “core-groups” such as truck drivers and commercial sex workers, along the main trucking routes.3,4 From urban centers, HIV slowly spread into rural areas, through the sexual mixing of urban and rural residents by return migration.1,5
HIV/sexually transmitted disease control programs in sub-Saharan Africa have largely focused on (peri-)urban areas, because HIV prevalences in urban areas are mostly higher than in neighboring rural areas,6–10 and HIV interventions are more cost-effective in densely populated sites.11 It has been questioned whether HIV prevalence in rural areas in time will level up to those of the urban sites. Some think that higher levels of HIV in urban areas simply reflect an earlier start of the epidemic.7,10 Others believe that rural prevalences will remain lower, because of less risky sexual behavior in rural areas as compared with urban areas with their commercial sex, nightlife, opportunities for casual sex, and their less strict cultural rules toward sexual relationships.12,13 However, not many studies have focused on the comparison of sexual behavior in urban and rural areas.10
Our study took place in Nyanza, the densely populated province in the west of Kenya which is most affected by HIV. The province is located on the border of Lake Victoria and is crossed by the highway that links Mombasa seaport with Uganda, Rwanda, Burundi, and the Democratic Republic of the Congo. The high mobility in the region could have contributed to the fast spread of HIV in Nyanza. In its capital Kisumu, the third largest city in Kenya, HIV among pregnant women has risen from 19% in 1990 to 35% in 1997.14 A recent population-based study showed that HIV prevalence among the general population in Kisumu was 20% for men and 30% for women; in the age group 15 to 19, it was 4% for boys and as high as 23% for girls.15 There are no population data on HIV prevalence in the rural areas of Nyanza.
In this study, we have investigated the sexual risk behavior of young adults in Nyanza province. The aims of the study were 2-fold: to measure sexual behavior that puts young adults aged 15 to 29 at risk of HIV infection and to compare urban with rural areas. To this end, we compared Kisumu town, with around 200,000 inhabitants,15 with the rural districts of Siaya and Bondo (recently subdivided from 1 district called Siaya), with 640,000 inhabitants.16 The latter 2 districts are located in the northwest of the province where Bondo borders Lake Victoria. The main highway passes through Kisumu as well as Siaya. A recent study investigated sexual behavior in Kisumu,17 but we are not aware of studies of sexual risk behavior in the rural areas of Nyanza.
In this cross-sectional study in Nyanza province, we compared Kisumu town with the rural districts of Siaya and Bondo, which were selected because they are inhabited by the Luo ethnic group that also predominates in Kisumu. Multistage random sampling was used to select households. In the rural districts, 6 divisions were randomly sampled (Ugunja, Ukwala, Yala, and Karemo in Siaya district and Bondo and Rarieda in Bondo district), whereas in Kisumu town, the division that encompasses the town (Winam) was the starting point. In each division, we subsequently sampled: locations, sublocations, villages and subvillages (in rural areas) or residential areas called estates (in urban areas), and households. All members between 15 and 29 years of age who usually live (eat/sleep) in the sampled households were considered eligible. Three repeat visits were made to find eligible household members who were initially absent; learners who board at school were revisited during holidays or school breaks. After obtaining verbal consent, face-to-face questionnaires were administered by same-sex research assistants who were slightly older than the respondents. The 8 research assistants, most of whom were social scientists, had received 5 days of training on interview techniques and on how to handle the questionnaire. The median duration of the interviews was 2 hours. For adolescents aged 15 to 18 years, consent was also asked from 1 of the parents/guardians.
The questionnaire instrument was adapted from the multicenter study on factors determining the differential spread of HIV in 4 African cities.15 After a set of questions on sociodemographic characteristics, questions were asked on sexual debut, marriage, number of premarital partners, and lifetime number of sex partners. A partner-by-partner approach was used to ask detailed questions on sexual behaviors during the last 12 months for all spousal and nonspousal partnerships. Questions included sociodemographic characteristics of partners, beginning and end of each relationship (enabling to calculate concurrency), condom use, and whether money was exchanged for sex.
After a 1-week pilot test of the questionnaire, data collection took place from February until May 1999. Data were entered in Epi-Info 6 and analyzed using SPSS version 9.0. Urban and rural sites were compared by calculating the chi-squared test for categorical and ordinal data and the Mann-Whitney U test for continuous data because they were not normally distributed. Median age at sexual debut and at first marriage was calculated using Kaplan-Meier survival techniques, with censoring at age of interview of individuals who had not had sex and who had not married, respectively, and their statistical difference was tested using the log rank test in Kaplan Meier. Differences in sexual behavior variables between urban and rural areas were adjusted for sociodemographic differences (age, education, religion, ethnicity, place of birth, and occupation) using logistic regression analysis for dichotomous outcome variables and Cox regression analysis for continuous/censored outcome variables.
Of the 648 eligible household members aged 15 to 29 years of age, 26 did not give consent and 38 were not found in 3 revisits. The response rate was therefore 90%. In total, 584 respondents were interviewed, 150 in Kisumu and 434 in Siaya/Bondo (290 in Siaya, 144 in Bondo). The number of men and women was almost equal (49% vs. 51%) and did not differ significantly between urban and rural sites.
Sociodemographic differences between urban and rural respondents are given in Table 1 for males and females separately. Urban men were slightly younger than rural men. Compared with respondents from rural areas, men and women from urban areas were better educated, more likely to have a religion other than Christian, more likely to be from non-Luo origin, and less likely to have been born in the town/district where they lived now. More men from rural areas were farmers or manual workers, whereas urban men were more likely to be sales/service worker or unemployed. More women in rural areas were housewives or farmers, whereas urban women were more likely to be unemployed.
In the age group 15 to 19, approximately three fourths of boys and girls had ever had sex (Table 2). Over one third had had sex before age 15. For women, the age of sexual debut was nearly a year lower for rural areas (15.7) than for urban areas (16.5, P = 0.02). Men and women in urban areas were less often married than in rural areas (P = 0.01 and P = 0.04). Men in Kisumu married in their late 20s, whereas men in Siaya/Bondo married around age 25. Because of late age at marriage, few men were virgins when they married and the time between sexual debut and first marriage was over 10 years. Women married around 20 years; one fourth were virgins at the time of marriage in Kisumu, but only 7% in Siaya/Bondo (P = 0.01). Almost one third of married women had a spouse who was 10 or more years older than themselves.
The median lifetime number of partners for men was 4 to 6 (Table 3). Over one fourth of the young men aged 15 to 29 reported to have had 10 or more partners. Women in Kisumu reported fewer lifetime partners than rural women (median 2 vs. 3, P <0.01). Over half of men in Siaya/Bondo reported multiple partnerships in the past year, versus 38% in Kisumu (P = 0.04). Less than 10% of women reported such relationships in any site. At the time of interview, over one fourth of nonmarried men had more than 1 ongoing relationship. Few men reported commercial sex in the past year, and 16% had a 1-time sex contact. Condom use was higher in urban than in rural areas: 24% of urban men mostly or always used a condom with nonspousal partners in the past year versus 13% of rural men (P = 0.08); for women, this was 36% versus 11% (P = 0.01).
We analyzed whether the differences in sexual behavior variables between urban and rural sites could be explained by sociodemographic differences (age composition, education, religion, ethnicity, place of birth, and occupation). Separate logistic and Cox regression analyses were conducted for each of the variables in Tables 2 and 3, for which significant (P <0.05) differences between urban and rural areas were found. Figure 1 shows for each variable the crude and adjusted odds ratios and their 95% confidence interval, with urban residence as the reference category (ie, odds ratios for urban residence are 1.0).
For none of the outcome variables, difference in sexual behavior between urban and rural areas could convincingly be explained by sociodemographic differences (Fig. 1). There were no substantial decreases in point estimates for odds ratios when comparing adjusted with crude odds ratios, although upper or lower limits of the confidence interval sometimes crossed the border of significance (which is partly the result of the higher number of degrees of freedom in multivariate analyses). For some sexual behavior variables, urban/rural differences even increased after adjusting for sociodemographic confounders: being ever married for men, and having more than 2 lifetime partners and consistent condom use in casual relations for women. After adjustments, the following differences clearly remained significant: rural men were approximately 4 times more likely to be married than urban men, whereas rural women were approximately 5 times less likely to be a virgin at marriage than urban women, 3 times more likely to have had more than 2 lifetime partners, and 10 times less likely to be a consistent condom user with nonspousal partners.
This study among young adults aged 15 to 29 in Nyanza province in Kenya shows high-risk sexual behavior in both urban and rural areas. Men and women started sex at an early age, the age difference between spouses was rather large (ie, dissortative age mixing), there was a high rate of partner change and concurrency in relationships, and consistent condom use with nonmarital partners was low. These behaviors have proven, both empirically3,7,18,19 and by modeling,20,21 to be associated with an increased risk of STD/HIV infection. Not many men reported commercial sex, but this can be related to the age range of our sample (15-29 years), which is lower than the age of most clients of female sex workers in Nyanza.22
Our findings are not only “high risk” in an absolute sense, but also when compared to the Demographic and Health Surveys (DHS) from 28 countries in sub-Saharan Africa.23 Our study belongs to the most risky quartile for both urban and rural young men and women for all 3 DHS sexual behavior indicators, namely, median age of first sex, having multiple partners in the past year (except for urban males), and condom use during last sex with a nonspousal partner in the last year (except for urban and rural females).
In our study, rural women reported a riskier sexual behavior than women who live in town: they had a younger age at sexual debut, were less often a virgin at marriage, had more lifetime partners, and less consistently used condoms with nonspousal partners. These behaviors are known to be risk factors in HIV/STD transmission.18,24,25 The riskier sexual behavior among rural women could not be explained by sociodemographic differences between urban and rural areas.
Rural men married at a younger age, and the interval between sexual debut and first marriage was shorter than for urban men. From observational studies, it is not very clear whether (young age at) marriage is associated with increased or decreased risk for HIV/STD for men.3,26 The only urban-rural difference for men that is clearly associated with HIV risk, namely that rural men more often had multiple partners last year than urban men, could be explained by sociodemographic differences (rural areas had more Luo inhabitants, and Luo men more often had multiple partners). Overall, we can conclude that the sexual behavior of men was as risky in urban as in rural Nyanza.
The equally high (men) or lower risk behavior (women) in Kisumu town cannot be explained by risk reduction resulting from STD/HIV interventions. A recent sexual behavior study in Kisumu showed that for men, median age at first sex and number of premarital partners was equal among all age groups.27 For women, there was a trend toward more, in stead of less, premarital partners in the younger age groups.27 There was a tendency toward later age at first sex among young girls, but this is a pattern that is also seen in rural Siaya,23 as well as all over Africa.28 Selective AIDS mortality among high-risk individuals in Kisumu is unlikely to play a major role in explaining the relatively low sexual risk behavior in town, because our study population is young (median age, 21 years). The high-risk sexual behavior that we have found in the rural areas cannot be the result of mingling with Kisumu town; there is little urban-rural migration (two thirds of rural respondents have lived in the rural district since birth), the sampled rural areas are quite far away from Kisumu town (between 60 and 130 km), and there is little sexual mixing between Kisumu and the rural areas (for the rural respondents, 75% of the 390 nonspousal partners in the past year were met in the same rural district).
Our findings contradict the general assumption that sexual behavior is more risky in towns, where individualism and anonymity facilitate casual and commercial sex, as compared with remote rural villages, where cultural rules toward sexual relationships are more strict, especially for women.2,12,13 Is Nyanza an exception, because of its dense population and high mobility? Studies in nearby areas around Lake Victoria show more risky sexual behavior in urban than rural sites, with higher numbers of partners,3,7,9,29–31 and a higher proportion of sexually active adolescents8,29 in urban areas. When reviewing DHS data on urban-rural differences for 28 sub-Saharan African countries,23 we see that the proportion of young men and women (15-24 years of age) engaging in high-risk sex and having multiple partners in the past year is higher in urban areas for all countries. However, age of first sex is lower in rural areas in 5 of 28 countries for men and in 19 of 28 countries for women. Thus, in sub-Saharan Africa, urban residence is generally associated with casual sex and higher partner change rates, but rural residence is for women often associated with early sexual debut. We know of only 2 studies (Nigeria, United States) that are in line with our findings that risk behavior can be equally high or even higher in rural areas.13,32 Our study conformed to most other studies in both East and in West Africa in that condom use is higher in urban environments.33
The high risk sexual behavior in rural Nyanza implies that, even if the HIV epidemic is not fueled by new infections from urban areas, the disease might spread within rural areas as fast as it has spread in urban areas. It is well possible that HIV prevalences in Siaya/Bondo will level up to those in Kisumu. Another study has found that in the advanced epidemic in South Africa, the direction of HIV spread is not only from urban areas to rural areas by migrant men, but also from rural women to their partners (migrant or not).34 A significant amount of transmission thus occurs within rural areas, and this role of local rural transmission has not been much acknowledged.35
What are the limitations of our results? The response rate of 90% is rather high (especially because our sample concerned young adults who are relatively often traveling or boarding in school), although the missed 10% might have been a high-risk group. We tried to overcome the validity problems that are associated with face-to-face questionnaires36–38; our questionnaire was thoroughly developed and pilot-tested, the interviewers were same sex and not much older than the respondents, they were well trained, the interviews were conducted in the local language, a lot of time was invested in building rapport with respondents before starting the interview, and much attention was paid to conducting the interviews in privacy. Nevertheless, like in other sexual behavior studies12,27 we found that men reported more relationships than women (ratio 2:3) when excluding relationships with partners outside the age or geographic range of our sample. Because the ratio of partnerships in the last year reported by men versus women did not differ substantially between urban and rural areas (2.1 vs. 2.3), there is no indication of different reporting biases between urban and rural sites.
We tried to crosscheck the respondents’ answers by holding detailed in-depth interviews with every second respondent by the same interviewer within a few days after the questionnaire. For both men and women, the results revealed some underreporting in the initial interview for urban as well as rural areas. The ratio of relationships reported by men and women thereby slightly increased instead of decreased. This could mean that cultural taboos are so strong that even in in-depth interviews women hide multiple relationships, or that men still exaggerate theirs (which implies that they had to invent full stories). It could also mean that we have missed a core group of high-risk women like mobile sex workers (although not many men reported sex with a sex worker).
Our study has implications for STD/HIV control in Nyanza. Interventions aiming at sexual behavior change should be expanded from Kisumu town to the surrounding rural districts. In particular, the low levels of condom use in rural areas should be addressed by improving condom acceptance as well as supply. Although lower than in Kisumu town, the population density in the rural areas is relatively high when compared with other rural provinces in Kenya, which is favorable for the cost-effectiveness of interventions. Expanding behavior change interventions to rural areas will also favor equity between rural and urban residents. Because adolescents begin risky behavior early in their sexual career (half of boys 15-19 years have had 3 sexual partners, and HIV prevalence in girls aged 15-19 is as high as 23%15), interventions should address children/young adolescents before they become sexually active, ie, in their early teens. After all, it will be easier to establish patterns of safe sex practices from the start than to change the high-risk behaviors as shown in this study.
1. Quinn TC. Population migration and the spread of types 1 and 2 human immunodeficiency viruses. Proc Natl Acad Sci U S A 1994; 91:2407–2414.
2. Caldwell JC, Anarfi JK, Caldwell P. Mobility, migration, sex, STDs, and AIDS: An essay on sub-Sahara Africa with other parallels. In: Herdt G, ed. Sexual Cultures and Migration in the Era of AIDS, Anthropological and Demographic Perspectives. Oxford: Clarendon Press, 1997:41–54.
3. Serwadda D, Wawer MJ, Musgrave SD, et al. HIV risk factors in three geographic strata of rural Rakai District, Uganda. AIDS 1992; 6:983–989.
4. Bwayo J, Plummer F, Omari M, et al. Human immunodeficiency virus infection in long-distance truck drivers in east Africa. Arch Intern Med 1994; 154:1391–1396.
5. Pison G, Le Guenno B, Lagarde E. Seasonal migration: a risk factor for HIV infection in rural Senegal. J Acquir Immune Defic Syndr 1993; 6:196–200.
6. Mnyika KS, Klepp KI, Kvale G, et al. Prevalence of HIV-1 infection in urban, semi-urban and rural areas in Arusha region, Tanzania. AIDS 1994; 8:1477–1481.
7. Barongo LR, Borgdorff MW, Mosha FF, et al. The epidemiology of HIV-1 infection in urban areas, roadside settlements and rural villages in Mwanza region, Tanzania. AIDS 1992; 6:1521–1528.
8. Konde-Lule JK, Wawer MJ, Sewankambo NK, et al. Adolescents, sexual behaviour and HIV-1 in rural Rakai district, Uganda. AIDS 1997; 11:791–799.
9. Wawer MJ, Serwadda D, Musgrave SD, et al. Dynamics of spread of HIV-I infection in a rural district of Uganda. BMJ 1991; 303:1303–1306.
10. Caraël M. Urban-rural differentials in HIV/STDs and sexual behaviour. In: Herdt G, ed. Sexual Cultures and Migration in the Era of AIDS, Anthropological and Demographic Perspectives. Oxford: Clarendon Press, 1997:107–126.
11. Soderlund N, Lavis J, Broomberg J, et al. The costs of HIV prevention strategies in developing countries. Bull World Health Organ 1993; 71:595–604.
12. Munguti K, Grosskurth H, Newell J, et al. Patterns of sexual behaviour in a rural population in North-Western Tanzania. Soc Sci Med 1997; 44:1553–1561.
13. DiClemente RJ, Brown LK, Beausoleil NI, et al. Comparison of AIDS knowledge and HIV-related sexual risk behaviors among adolescents in low and high AIDS prevalence communities. J Adolesc Health 1993; 14:231–236.
14. UNAIDS/WHO. Kenya Epidemiological Fact Sheet on HIV/AIDS and STI, 2000 update (revised).
15. Buvé A, Carael M, Hayes RJ, et al. Multicentre study on factors determining differences in rate of spread of HIV in sub-Sahara Africa: Methods and prevalence of HIV infection. AIDS 2001; 15(suppl 4):S5–14.
16. Central Bureau of Statistics, Office of the Vice-President, Ministry of Planning and National Development. Kenya Population Census 1989, Migration and Urbanisation. Nairobi, Kenya: Government Printer, 1996.
17. Ferry B, Carael M, Buve A, et al. Comparison of key parameters of sexual behaviour in four African urban populations with different levels of HIV infection. AIDS 2001; 15(suppl 4):S41–50.
18. Pinkerton SD, Abramson PR. Effectiveness of condoms in preventing HIV transmission. Soc Sci Med 1997; 44:1303–1312.
19. Gregson S, Nyamukapa CA, Garnett GP, et al. Sexual mixing patterns and sex-differentials in teenage exposure to HIV infection in rural Zimbabwe. Lancet 2002; 359:1896–1903.
20. Morris M, Kretzschmar M. Concurrent partnerships and the spread of HIV. AIDS 1997; 11:641–648.
21. Garnett GP, Anderson RM. Factors controlling the spread of HIV in heterosexual communities in developing countries: patterns of mixing between different age and sexual activity classes. Philos Trans R Soc Lond B Biol Sci 1993; 342:137–159.
22. Voeten HACM, Egesah OB, Ondiege MY, et al. Clients of female sex workers in Nyanza province, Kenya: A core group in STD/HIV transmission. Sex Transm Dis 2002; 29:444–452.
24. Malamba SS, Wagner HU, Maude G, et al. Risk factors for HIV-1 infection in adults in a rural Ugandan community: A case-control study. AIDS 1994; 8:253–257.
25. Nuko S, Chiduo B, Mwaluko G, et al. Pre-marital sexual behaviour among out-of-school adolescents: Motives, patterns and meaning attributed to sexual partnership in rural Tanzania. Afr J Reprod Health 2001; 5:162–174.
26. Nunn AJ, Kengeya-Kayondo JF, Malamba SS, et al. Risk factors for HIV-1 infection in adults in a rural Ugandan community: A population study. AIDS 1994; 8:81–86.
27. Buvé A, Lagarde E, Carael M, et al. Interpreting sexual behaviour data: Validity issues in the multicentre study on factors determining the differential spread of HIV in four African cities. AIDS 2001; 15(suppl 4):S117–126.
28. Meekers D. Sexual initiation and premarital childbearing in sub-Sahara Africa. Pop Stud J Dem 1994; 48:47–64.
29. Boerma JT, Urassa M, Senkoro K, et al. Spread of HIV infection in a rural area in Tanzania. AIDS 1999; 13:1233–1240.
30. Killewo J, Nyamuryekunge K, Sandstrom A, et al. Prevalence of HIV-1 infection in the Kagera region of Tanzania: A population-based study. AIDS 1990; 4:1081–1085.
31. Mnyika KS, Klepp KI, Kvale G, et al. Determinants of high-risk sexual behaviour and condom use among adults in Arusha region, Tanzania. Int J STD AIDS 1997; 8:176–183.
32. Orubuloye O, Caldwell JC, Caldwell P. Men’s sexual behaviour in urban and rural southwest Nigeria: Its cultural, social and attitudinal context. Health Transit Rev 1997; 7(suppl):315–328.
33. Kapiga SH. Determinants of multiple sexual partners and condom use among sexually active Tanzanians. East Afr Med J 1996; 73:435–442.
34. Lurie MN, Williams BG, Zuma K, et al. Who infects whom? HIV-1 concordance and discordance among migrant and non-migrant couples in South-Africa. AIDS 2003; 17:2245–2252.
35. Lurie MN, Williams BG, Zuma K, et al. The impact of migration on HIV-1 transmission in South-Africa: A study of migrant and non-migrant men and their partners. Sex Transm Dis 2003; 30:149–156.
36. Catania JA, Binson D, van der Straten A, et al. Methodological research on sexual behavior in the AIDS era. Annu Rev Sex Res 1995; 6:77–125.
37. Konings E, Bantebya G, Carael M, et al. Validating population surveys for the measurement of HIV/STD prevention indicators. AIDS 1995; 9:375–382.
38. Weinhardt LS, Forsyth AD, Carey MP, et al. Reliability and validity of self-report measures of HIV-related sexual behavior: progress since 1990 and recommendations for research and practice. Arch Sex Behav 1998; 27:155–180.