Throughout sub-Saharan Africa young women have substantially higher rates of HIV infection compared to young men of the same age [1–3]. For example, it is estimated that 26% of females aged 15–24 years in Zimbabwe are infected with HIV, compared to 10% of males in the same age range . Biological factors may partially explain the gender difference in HIV prevalence as transmission of HIV from male to female is estimated to be two to three times higher than female to male transmission due to genital anatomy, more inoculum from men to women and co-morbidity from other sexually transmitted infections (STI), more often asymptomatic in women [5–8]. Nevertheless, some studies have found similar transmission probabilities from men to women in studies of sero-discordant couples [7,9,10]. Young women may be especially prone to HIV infection in comparison to older woman due to larger areas of cervical ectopy and trauma to the immature genital tract during sex [11–13]. Additionally, behavioral factors, including early age of coital debut, contribute to young women's heightened susceptibility to HIV. Other behavioral factors that are thought to increase a young woman's risk of HIV infection are older partners, inconsistent condom use, forced sex, and transactional sex.
In the USA, early age of coital debut has been shown to be associated with a number of negative subsequent outcomes such as STI, decreased condom and contraceptive use, and increased numbers of sexual partners [14–17]. Data on the role of early coital debut as a risk factor among women in sub-Saharan Africa are more limited . A recent study of sexual behavior and HIV infection in four African cities reported that in the two cities with high HIV prevalence, young women had their sexual debut earlier than in the two cities with a low HIV prevalence . In addition, research conducted among young people in Rwanda demonstrated that having sex at an early age (defined as 17 years of age or younger) was significantly associated with incident HIV infection . Reducing exposure to early age at first sexual intercourse has been advocated as a means to reduce the incidence of HIV infection in young women . In fact, delay in the age of first sexual intercourse is one of the changes in behavior credited with the decline in HIV infection in Uganda. In 1989 it was reported that approximately 50% of males and females in Uganda reported having had sex by the age of 15 years, while in 1995 only 20% of those in this age group reported having had sex . However, these studies do not take into account confounders such as condom use, amount of time sexually active, number of sexual partners and the current age of the woman. As most countries in sub-Saharan Africa are experiencing generalized HIV epidemics it is important to identify modifiable risk factors for HIV transmission in highly vulnerable groups to better tailor risk reduction messages and programs.
The primary aim of the research presented in this paper was to explore the independent relationship between early age of coital debut and risk of HIV infection, in a sample of sexually active urbanized Zimbabwean women, controlling for other behavioral factors.
Data used for this research came from an ongoing cohort study to examine the effect of hormonal contraception on HIV acquisition (HC-HIV Study). For the purpose of this analysis, screening data were used for women recruited between November 1999 and September 2002. The screening visit included a brief interview concerning socio-demographic factors and sexual history; HIV and pregnancy testing were also conducted. Ethics approval for this analysis was granted by the Committee for the Protection of Human Subjects, University of California Berkeley and for the larger study from the Committee on Human Research, University of California San Francisco, the Biomedical Research and Training Institute, Zimbabwe and the Medical Research Council of Zimbabwe. Informed consent was obtained from all participants.
Sexually active adult women 18–35 years of age were recruited from public sector family planning and reproductive health clinics in Harare, Epworth and Chitungwiza, Zimbabwe. The study was carried out in four clinics, chosen on the basis of similarity of the client population: low-income women in urban and peri-urban settings. A total of 4675 women were screened, of whom 4393 (94%) had complete data on sexual behaviors, contraceptive use and HIV sero-status.
HIV serostatus was determined at the screening visit using Abbott ELISA (Abbott Diagnostics, Abbott Park, Illinois, USA); positive results were confirmed by two consecutive rapid tests. In the case of discrepant test results, results were confirmed by Western blot and, if the Western blot was indeterminate, DNA PCR. The final HIV status from the screening visit was used as the primary outcome in these analyses.
Age of coital debut
During the screening interview, women were asked ‘At what age did you first have sex?'. Age of coital debut was defined as a binary outcome, (≤ 15 years, > 15 years) based on definitions used in the literature [17,18] and on sensitivity analyses we set with cut-off points at age 16, 17 (25th percentile) and 18 years (mean and median).
Other measures that were examined in the analyses included time since coital debut (current age minus age of coital debut); the log of the time since coital debut, as this transformation fit the data better in generalized linear regression models for HIV infection; current age (ages 17–24 years and 25–35 years); ever having used a male or female condom (the only measure of condom use collected at screening); lifetime number of sexual partners (one or more than one partner); education (completed high school or not); currently living with partner; marital status (currently married or not); ever engaged in commercial sex work; and current hormonal contraceptive use (having had a Depo-Provera injection in the last 4 months or using combined oral contraceptive pills regularly for the last 3 months).
Exploratory analyses of the association between HIV serostatus, early age of coital debut and other potential confounding variables were based on frequency distributions and graphical methods. Early age of coital debut was represented both as a continuous variable and as categorical, using a number of different cut-off points. Chi-square and Wilcoxon rank-sum statistics were used to assess significance of observed associations, and effect modification between the primary exposure and potential interaction variables were evaluated using the Breslow–Day test for homogeneity. We also investigated multivariate associations between early coital debut and significant factors from univariate analyses using logistic regression.
Multivariate analyses of factors associated with HIV infection risk were based on generalized linear regression models, taking a binary indicator of positive HIV serostatus as the outcome. Because of our interest in investigating the effect of early age of debut on infection risk, and simultaneously controlling for duration of exposure to infection, models were based on the complementary log–log rather than the logit link familiar from logistic regression. When log of the duration of sexual activity is included, such models can be interpreted as proportional hazards models for the unobserved time from sexual debut to infection, with regression coefficients representing log relative hazards . Related methods  were used to produce smooth semiparametric estimates of the infection distributions for women reporting early and late sexual debut. Results were used to evaluate the validity of the proportional hazards assumption, and suggest appropriate parametric models for infection distributions to use in regression models for additional risk factors. Results from additional modeling indicated that parametric Weibull proportional hazards (i.e., linear in log of duration of activity) provided acceptable fits to these data. Therefore, further regression models investigating associations with additional risk factors were based on this assumption. All analyses were conducted using Stata, version 7.0  and R, version 1.8.0 .
The prevalence of HIV infection among the women in this sample was 40.1%. The median age of coital debut was 18 years (range, 10–32 years), the 25th percentile was at age 17 years and the 75th was at age 20 years; 11.8% of women reporting having experienced coital debut at age 15 years or younger. The median age of this study population was 26 years (range, 18–35 years). Forty-nine percent of women completed high school. The majority of women were in monogamous marriages (83.5%), and reported currently living with their husband/partner (89%). Close to 65% of women were currently using either Depo-Provera or combined oral contraceptive pills. Almost 65% reported having had only one lifetime sexual partner and nearly 70% reported ever having had sex when a male or female condom was used. Just under 2% reported having ever engaged in commercial sex.
In bivariate analysis women who reported having experienced sexual debut at age 15 years or younger were significantly more likely to have had more than one lifetime sexual partner; to have been sexually active for longer; to not have completed high school; to not currently be living with their husband/partner; to not currently be using hormonal contraceptives; to have ever used a condom; and to have ever engaged in commercial sex work (Table 1).
In a multivariate logistic regression model examining the relationship between early debut and behavioral factors, early debut was associated with having more than one lifetime sexual partner [odds ratio (OR), 2.25; 95% confidence interval (CI), 1.82–2.78], not having completed high school (OR, 8.83; 95% CI, 6.63–11.77) and to have ever engaged in commercial sex work (OR, 2.23; 95% CI, 1.37–3.63), after controlling for age, education, living with partner, hormonal contraceptive use, condom use and ever engaging in sex work (Table 1).
Women who had their sexual debut at age 15 years or younger were also significantly more likely to be infected with HIV (P < 0.001). Of those women who reported having had sex at age 15 years or younger, 54.6% were infected with HIV compared to 38.2% among women who reported older age at coital debut (Table 2). Other factors associated with HIV infection included mean years since sexual debut (9.4 years versus 7.4 years; P < 0.001), being older (25–35 years), having had more than one lifetime sexual partner, not completing high school, not currently living with a husband/partner, not currently using hormonal contraceptives, ever having used a condom, and ever engaging in commercial sex work (Table 2).
Fig. 1 presents separate smoothed nonparametric estimates of the probability of remaining HIV uninfected as a function of duration of sexual activity. The curves for early debut and later debut and their 95% CI overlap in the first years of sexual activity and only start to diverge as the length of time the woman has been sexually active increases to more than 5 years. The 95% CI indicate significantly increased cumulative infection risk for women reporting early debut. However, significant differences are only evident among women reporting more than 12 years of sexual activity. Greater uncertainty in the estimate for women reporting early debut reflects the smaller number of women in this group.
Table 3 presents three proportional hazards regression models investigating the relationship between HIV infection risk and early age of coital debut; all of the models include the logarithm of duration of sexual activity (i.e., number of years since coital debut) and current age. The models move from the most basic (model 1 explores the association between early coital debut and HIV infection controlling for log of time since debut and current age) to more complex models that control for more variables. The three models demonstrate how the risk of HIV among women with early coital debut diminishes as other risk factors that are correlated with early debut are controlled for in the progressive models. In the first model (model 1), the estimated hazard ratio (HR) for early debut was 1.30 (95% CI, 1.13–1.50). When the binary indicator of greater than one lifetime partner was added to the model (model 2), the risk of infection due to early debut decreased (HR, 1.19; 95% CI, 1.03–1.37). This observation is consistent with the association between these two variables reported above. When other behavioral and socioeconomic variables are controlled for (model 3), the hazard ratio for early debut decreased to 1.15, and the lower bound of the 95% CI indicates marginal significance at the 5% level. Older age, not currently using a hormonal method of contraception, and not currently living with a husband/partner remained significant risk factors HIV infection, while never having used a condom and not completing high school were not significant in this model.
The results of this study support the hypothesis that early age of coital debut, defined here as having experienced first sexual intercourse at age 15 years or younger, is a risk factor for HIV infection among young women living in sub-Saharan Africa. However, it appears that much of the HIV infection risk attributable to early coital debut can be explained by other factors among women who have sex at an early age such as the length of time that a woman is sexually active and multiple lifetime partners [adjusted odds ratio (AOR), 2.25; 95% CI, 1.82–2.78; for more than one lifetime partner].
The prevalence of HIV infection among women in this study population is very high at 40%. This estimate is similar to the UNAIDS high estimate for HIV prevalence among women age 15–24 years in Zimbabwe which was estimated at 39.6% in 2001 . The 2000 ANC survey found that HIV prevalence ranged from 27.8% among women age 15–19 years to a high of 43.5% among those age 30–34 years. Nevertheless, despite the high prevalence, this population reported a relatively low prevalence of risk behaviors associated with HIV infection. The majority of women reported only ever having had one sexual partner and being in a monogamous marriage, and more than half reported coital debut at age 18 years or older. This profile is similar to that seen in other female populations in Zimbabwe [25,26]. A study recently conducted in Harare found that the mean age of sexual debut among women attending primary health clinics was 18 years .
Women who reported having had more than one lifetime sexual partner had the greatest risk of being infected with HIV. Only 35% of women, however, had more than one lifetime partner; 22% had two partners, and 13% had more than two (range, 3–200 lifetime partners). Nevertheless, young women that do begin sex at a younger age are more likely to have multiple partners. Supporting the message to reduce the number of sexual partners and to delay the age of first sex should remain an important emphasis of prevention campaigns. However, it must be recognized that many women are put at risk of HIV infection by the behaviors of their male partners, not their own.
For many young women, the choice to delay first sex is not a decision over which they have control. Gender-power imbalances, pervasive in many African countries, mean that many women are unable to refuse unwanted sexual advances; and when they have sex, have limited control over condom use. Moreover, the younger the woman is, the greater the power imbalance is likely to be . Gender inequality also leads to young women often having less access to HIV/STI prevention information or to health care services, including access to condoms, increasing their risk of infection . Also, coercion and rape are common in many sub-Saharan African countries  meaning that many young women are unable to negotiate condom use or to refuse unwanted sexual advances. Further, if trauma to the genital tract was sustained at coital debut due to rape, it may increase the risk of HIV transmission . Finally, economic hardship and disparity, which tend to affect young women more than men , mean that in many countries the only way a young woman can bring in food or income to her home is through transactional sex, another manifestation of the gender-power imbalance which again reduces the ability of women to negotiate condom use . In this study HIV prevalence was 40% despite the fact that women had a low HIV risk profile, emphasizing that women are at high risk due to the gender inequalities highlighted above and the risk behavior of their male partners.
Hormonal contraceptive use was also associated with HIV status, although because all analyses conducted here were cross-sectional, we cannot infer anything about directionality or causal relationships. Hormonal contraceptive use may be a proxy marker for socio-economic status, including access to health care services and education. Women who were using hormonal contraceptives were more likely to have completed high school compared to women not using hormonal contraceptives (57% versus 48%; P < 0.001). Also, they were more likely to have only one lifetime sexual partner compared to women not using hormonal contraceptives (46% versus 30%; P < 0.001).
Never having used a male or female condom was not a significant risk factor for HIV infection in the multivariate model. In bivariate analyses, never having used a male condom was actually associated with a lower risk for HIV infection, but this association disappeared when controlling for other factors. In this case ever having used a condom was likely a proxy for more sexual experience, and higher risk sex. In fact, women with more than one lifetime partner were more likely to report having ever used condoms compared to women with only one lifetime partner (82% versus 63%; P < 0.001). Additionally, ever condom use is not a sensitive measure of use as it does not capture the frequency or consistency of use, which are both essential to condom efficacy in HIV prevention.
Exploring the risk of early coital debut on HIV infection is complicated by the many interrelated factors that may put a woman at risk for HIV infection. These factors include: the age of the woman when she first had sex; the period in historical time when she became sexually active, which correlates with the stage of the HIV epidemic in the population; the length of time she has been sexually active; and her current age.
In order to best untangle these effects, we chose to use proportional hazards regression over conventional logistic models. Although both approaches yielded similar results, the former has the advantage that duration of sexual activity is naturally represented as the time scale over which infections occur, and resulting regression coefficients afford a hazard ratio interpretation. These models indicate that early debut is a risk factor for HIV infection, but this risk is dependent on a woman's sexual activity/HIV exposure level (partner number) and other variables (education, living with a partner, etc.). In Fig. 1, the estimated cumulative probabilities of remaining HIV negative for women with early and later coital debut are very similar for the first few years of activity, but diverge for greater than 5 years of activity. This implies that the risks operating in the early debut group are higher overall, and do not approach those for the later group even after many years of activity. Thus, even if there are early events that increase risk, the differences seem to imply a different pattern of risk overall, and in fact, later differences seem to drive the significant difference in HIV risk. Whether something about the event of early debut increase a woman's future risk or whether her behavior patterns that follow that event put her at increased risk cannot be determined from this study; cohort studies would be needed to answer this question. Limitations of these data include the fact that the sample is cross-sectional, and HIV status is consequently based on prevalent rather than incident infection. In addition, measures of duration of sexual activity, number of partners, and age of sexual debut are based on self-report taken at the time of recruitment.
Although we chose to categorize early debut at age 15 years or younger, the risk of HIV infection remained statistically significant in model 1 when first sex occurred at age 16 years, 17 years or younger. One potentially troubling aspect of these results is that women in the younger age group (age 17–24 years) appear to have experienced coital debut at a younger age than older women (25–35 years) (17.6 years versus. 18.6 years; P < 0.001). This effect may point to a trend towards earlier age of coital debut and is consistent with patterns seen in other countries . Alternatively, it is possible that women in the older age group who experienced early debut may have indeed been infected with HIV but were too sick or may have died from AIDS, and thus were not able to participate in the study. Therefore the age of debut may be remaining constant but appear different due to a ‘survivor bias'.
This study was cross-sectional in nature and the ability to clearly distinguish the time sequence of events leading up to HIV infection is limited. However, the age of first sex is a singular event that must come prior to heterosexual infection with HIV. In spite of this, we cannot definitively conclude that the factors described here increase the risk of HIV acquisition nor can we accurately assess ‘importance’ of early age of coital debut relative to other identified risk factors. Nevertheless, our results show that this variable is a significant predictor of prevalent HIV infection independent of other identified factors in this population, and indicate that controlling for age of debut may be important in investigations of transmission risk in other studies. Prevention programs that aim to delay the age of first sex are important in Africa and they should target both young men and women. Developing an understanding of the modifiable risk factors that precede early first sex is an important area for further research to address the barriers that prevent young women from opting to delay sex.
We thank all the women who participated in the study and the study staff from Zimbabwe. We acknowledge the multisite study's Principal Investigator, C. Morrison, and the sponsor and monitor of the study, Family Health International. Data management for this study was performed by SCHARP at the Fred Hutchinson Cancer Research Center in Seattle, Washington, USA.
Sponsorship: This project has been funded with Federal funds from the National Institute of Child Health and Human Development, National Institutes of Health Department of Health and Human Services, through a contract with Family Health International (FHI) (contract number N01-HD-0-3310).
The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services or FHI, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government.
1. Laga M, Schwartlander B, Pisani E, Sow P, Carael M. To stem HIV in Africa prevent transmission to young women.AIDS
2. Berkley S, Naamara W, Okware S, Downing R, Konde-Lule J, Wawer M, et al.AIDS and HIV infection in Uganda – are more women infected than men?AIDS
3. Gregson S, Garnett G. Contrasting gender differentials in HIV-1 prevalence and associated mortality increase in eastern and southern African: artifact of data or natural course of epidemics?AIDS
(Suppl 3): S85–S99.
4. Joint United Nations Programme on HIV/AIDS (UNAIDS). Report on the Global HIV/AIDS Epidemic 2002.
Geneva: UNAIDS; 2002.
5. Nicolosi A, Leite M, Musicco M, Arici C, Gavazzeni G, Lazzarin A. The efficiency of male-to-female and female-to-male sexual transmission of the human immunodeficiency virus: a study of 730 stable couples.Epidemiology
6. European study group on heterosexual transmission of HIV. Comparison of female to male and male to female transmission of HIV in 563 stable couples.BMJ
7. Fideli US, Allen SA, Musonda R, Trask S, Hahn BH, Weiss H, et al.Virologic and immunologic determinants of heterosexual transmission of Human Immunodeficiency Virus Type 1 in Africa.AIDS Res Hum Retroviruses
8. Carpenter LM, Kamali A, Ruberantwari A, Malamba SS, Whitworth JAG. Rates of HIV-1 transmission within marriage in rural Uganda in relation to the HIV sero-status of the partners.AIDS
9. Gray R, Wawer MJ, Brookmeyer R, Sewankambo NK, Serwadda D, Wabwire-Mangen F, et al.
for the Rakai Project Team. Probability of HIV-1 transmission per coital act in monogamous, heterosexual, HIV-1 discordant couples in Rakai, Uganda.Lancet
10. Quinn TC, Wawer MJ, Sewankambo N, Serwadda D, Li C, Wabwire-Mangen F, et al.
for the Rakai Project Study Group. Viral load and heterosexual transmission of human immunodeficiency virus type 1.N Engl J Med
11. Moss G, Clemetson D, D'Costa L, Plummer FA, Ndinya-Achola JO, Reilly M, et al.Association of cervical ectopy with heterosexual transmission of human immunodeficiency virus: results of a stud of couples in Nairobi, Kenya.J Infect Dis
12. Coombs R, Reichelderfer P, Landay A. Recent observations on HIV type-1 infectionin the genital tract on men and women.AIDS
13. Moscicki A, Ma Y, Holland C, Vermund S. Cervical ectopy in adolescent girls with and without human immunodeficiency virus infection.J Infect Dis
14. Manning W, Longmore M, Giordano P. The relationship context of contraceptive use at first intercourse.Family Planning Perspectives
15. Greenberg J, Magder L, Aral S. Age at first coitus a marker for risky sexual behavior in women.Sexually Trans Dis
16. Duncan E, Tibaux G, Pelzer A, Reimann K, Peutherer JF, Simmonds P, et al.First coitus before menarche and risk of sexually transmitted disease.Lancet
17. Dickson N, Paul C, Herbison P, Silva P. First sexual intercourse: age, coercion, and later regrets reported by a birth cohort.BMJ
18. 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
(Suppl 4): S41–S50.
19. Bulterys M, Chao A, Habimana P, Dushimimana A, Nawrocki P, Saah A. Incident HIV-1 infection in a cohort of young women in Butare, Rwanda.AIDS
20. Asiimwe-Okiror G, Opio A, Musinguzi J, Madraa E, Tembo G, Carael M. Change in sexual behaviour and decline in HIV infection among young pregnant women in urban Uganda.AIDS
21. Jewell NP, Shiboski SC. Statistical analysis of HIV infectivity based on partner studies.Biometrics
22. Shiboski SC. Generalized Additive Models For Current Status Data.Lifetime Data Anal
23. Stata Statistical Software: Release 7.0, Reference H-P. College Station, TX, Stata Corporation, 2001.
24. Ihaka R, Gentleman R. R: a language for data analysis and graphics.J Computational Graph Stat
25. Centers for Disease Control and Prevention. Sexual Activity and Contraceptive Practices among teenagers in the United States, 1988 and 1995. Vital and Health Statistics.
Atlanta: CDC; 2002.
27. Mbizvo E, Msuya S, Hussain A, Chirenje M, Stray-Pederson B. HIV prevalence in Zimbabwean women: 54–67% knowledge and perceived risk.Int J STD AIDS
28. Luke N. Cross Generational and Transactional Sexual Relations in sub-Saharan Africa: A Review of the Evidence on Prevalence and Implications for Negotiation of Safe Sexual Practices for Adolescent Girls.
Philadelphia: International Centre for Research on Women; 2001.
29. Jewkes R Abrahams N. The epidemiology of rape and sexual coercion in South Africa: an overview.Soc Sci Med
30. Davison J, Strickland R. Leveling the Playing Field: Promoting Women's Economic Capabilities and Human Rights.
International Center for Research on Women and the Centre for Development and Population Activities; 2000. Available online at http://www.icrw.org/does/SWECsynthesispaper.pdf