Secondary Logo

Journal Logo

Epidemiology and Social

Only a fraction of new HIV infections occur within identifiable stable discordant couples in sub-Saharan Africa

Chemaitelly, Hiama; Shelton, James D.b; Hallett, Timothy B.c; Abu-Raddad, Laith J.a,d,e

Author Information
doi: 10.1097/QAD.0b013e32835ad459



Recent landmark randomized clinical trials have established the efficacy of antiretroviral drugs, as either treatment to the infected partner or prophylaxis to the uninfected partner, in reducing HIV heterosexual transmission among stable HIV sero-discordant couples (SDCs), in which one partner is infected whereas the other is not [1–3]. Moreover, multiple cross-sectional epidemiologic surveys revealed that over half of couples affected by HIV in sub-Saharan Africa (SSA) are discordant [4–6]. It is widely perceived that ‘most transmissions of HIV-1 infections in Africa are thought to occur within stable, cohabitating HIV-1 SDCs’ [1].

There is an intense debate about the relative contribution of HIV sero-conversions within SDCs to total HIV population-level incidence, and the priority of HIV prevention among SDCs compared with other prevention approaches such as among the general population, young people, unmarried sexually active men and women, and commercial sex networks [7–9]. Although one modeling study of two urban settings in Zambia and Rwanda attributed more than half of new heterosexually transmitted HIV infections to sero-conversions within married or cohabiting SDCs [8], empirical evidence from a cohort of stable couples in Rakai, Uganda, revealed that only a minority of HIV infections occur among established SDCs that could be identified [10]. Accordingly, we present an evidence-based quantitative assessment to assess the contribution of HIV-1 sero-conversions within SDCs to total HIV population-level incidence across 20 countries in SSA.


Our mathematical model estimates the annual number of HIV-1 transmissions from the infected partners to the uninfected partners among established SDCs, and compares its value to an estimate for the overall HIV population-level incidence in the country. Multiple data sources were used to obtain the model parameters (Fig. S1 of Supplemental Digital Content; SuppDC,, with the primary sources being the Demographic and Health Surveys (DHS), which are nationally representative household-based surveys. HIV population-level incidence estimates were calculated using the Joint United Nations Programme on HIV/AIDS (UNAIDS) SPECTRUM model predictions [11, Gouws E. Joint United Nations Programme on HIV/AIDS (UNAIDS), personal communication 2011], or derived using DHS measures of HIV-1 prevalence when those were not available (SuppDC,

Model structure

Our model sets out to ask the following question: if there is a national screening survey at a point in time which identifies SDCs, what proportion of all HIV infections that occur over the following year will be ones in which the infected individual in an identified SDC transmitted HIV to their partner? This is the quantity that should determine the relative priority of interventions among SDCs.

In that hypothetical screening survey (at Time 0), three different types of couples will be found (Fig. 1):

  1. Concordant HIV-1 positive couples (both partners are HIV-1 sero-positive).
  2. Discordant HIV-1 couples (one partner is HIV-1 sero-positive and the other partner is HIV-1 sero-negative).
  3. Concordant HIV-1 negative couples (both partners are HIV-1 sero-negative).

At the time of the second cross-sectional survey 1 year later (Time 1), several scenarios are possible in terms of HIV-1 sero-status for each of these types of partnerships (Fig. 1):

Fig. 1:
Model conceptualization for our methodology of estimating HIV-1 seroconversions among stable HIV-1 serodiscordant couples as identified through repeated cross-sectional surveys.The figure shows the possible outcome scenarios for the three types of couple serostatus at Time 1, one year after a baseline screening survey at Time 0. The highlighted part of the figures displays HIV incidence within identifiable stable discordant partnerships that is the focus of our analysis.
  1. Concordant HIV-1 positive couples

This type of partnership remains unchanged, as no new HIV sero-conversions are possible between the two HIV-1 sero-positive partners.

  • Discordant HIV-1 couples
  • Three different scenarios are possible:
    1. No sero-conversion occurs in the partnership during the course of the year and the couple remains discordant.
    2. The sero-positive index partner transmits the infection to the sero-negative partner over the course of the year before the next cross-sectional survey. This type of sero-conversion contributes to HIV incidence among SDCs, as it is from a source within the SDC. As these HIV transmissions occurred in couples that have already been identified as discordant at Time 0, these infections define the identifiable HIV incidence among SDCs (for mathematical derivations see SuppDC, Such transmission events are the only ones that have the potential to be averted by interventions that target both couples’ joint behaviors and factors that affect the rate of transmission from the infected partner (in particular, antiretroviral drugs). This category is of primary interest in determining the relative priority of interventions to identifiable SDCs.
    3. The HIV-1 negative partner sero-converts by acquiring the infection from an external source. This type of incidence is distinct from HIV incidence due to transmission from within the SDC.
  • Concordant HIV-1 negative couples
  • Three different scenarios are possible within a concordant-negative couple:
    1. No sero-conversion occurs in the partnership during the course of the year and the couple remains concordant negative.
    2. One or both partners are infected from external sources during the course of the year and the couple becomes either discordant or concordant positive. These new infections occur to individuals who are engaged in a stable couple, but cannot be attributed to the stable couple and would not be averted by interventions that aim to change couples’ joint risk of infection. For this reason, they are not counted as infections within SDCs.
    3. One of the partners (index partner) acquires the infection from an external source and then transmits the infection to the other partner within the year. This rapid transmission to the HIV-1 sero-negative partner through an infection transmitted from such a recently infected index partner is a transmission within an SDC. However, from an intervention perspective, these transmissions are difficult to prevent using interventions targeted for SDCs, as these couples would have been identified as concordant negative in the initial screening at Time 0. These infections therefore are not counted as transmissions occurring among identifiable SDCs.

Model parameterization

Socio-demographic characteristics, HIV-1 prevalence, and sero-discordancy prevalence in sub-Saharan Africa

We used the most recent DHS data for 20 countries in SSA [12], complemented by population size information from the United Nations Population Division Database [13], to calculate country-specific demographic, epidemiological, and behavioral indicators (Table 1). Countries were considered for analysis based on the availability of the DHS HIV serological biomarker survey (SuppDC,

Table 1:
Key demographic and HIV-1 related indicators across different countries in sub-Saharan Africa.

Following DHS methodology, we define a stable sexual couple as a man and a woman living in a consensual union within a household at the time of the cross-sectional DHS survey [14]. For details on DHS sampling methodology see SuppDC, We excluded from our analyses couples where one or both partners did not test for HIV. Missing HIV information among all couples ranged from 2.2 to 28.1% (mean of 13.0%) across countries. We followed DHS guidelines in applying the sampling weights retrieved from the DHS databases to our calculations [14,15].

HIV population-level incidence and incidence rate

Estimates for the country-specific HIV population-level incidence rate were obtained from the available UNAIDS SPECTRUM model predictions for each country for the specific year of the DHS survey [11, Gouws E. Joint United Nations Programme on HIV/AIDS (UNAIDS), personal communication 2011]. SPECTRUM is an integrated modular program used for the estimation and projection of HIV epidemiologic indicators based on up-to-date surveillance data [16,17]. For countries where SPECTRUM model predictions were not available or where the bounds of the 95% confidence interval (CI) were not precisely specified, estimates of HIV population-level incidence rate were derived using the DHS country-specific HIV-1 prevalence (SuppDC, The number of new HIV infections per year was then calculated using the relation:

HIV-1 transmission probability per coital act and coital frequency

We parameterized our model using the average of the empirical measures of HIV-1 transmission probability per heterosexual coital act (p) as reported by the Rakai Study [18] and the Partners in Prevention HSV/HIV Transmission Study (Partners in Prevention Study) [1,19]. Both studies provide the best available direct empirical data for p among SDCs. The Rakai estimate is 0.0012 (95% CI: 0.0009–0.0015) [18], and the Partners in Prevention Study estimate is 0.0011 [19, Hughes JP, personal communication, 2010]. Both estimates represent the average p for men and women, and neither study found a significant effect of sex on the transmission probability. For SDCs in which women were HIV infected and the susceptible male partners were circumcised, we reduced p by 58% to account for the efficacy of male circumcision in reducing HIV acquisition [20–23]. The frequency of coital acts among stable couples (n) was assumed at 8.3 coital acts per month based on the median measured for men and women in the Rakai cohort [18]. This assumption results in an incidence rate among SDCs (in absence of male circumcision and condom use) of 10.8 per 100 person-years, consistent with that reported in the Rakai study of 8.4 per 100 person-years [18]. Male circumcision and condom use reduce the incidence rate among SDCs according to their coverage in each specific country and whether the index partner in the SDC is a male or a female (SuppDC,

Uncertainty and sensitivity analyses

Uncertainty analyses were performed for the estimates of the contribution of identifiable HIV incidence among SDCs to total HIV incidence for each country using Monte Carlo sampling from uniform distributions for the relevance CIs or credible ranges for the demographic, biological, and epidemiological parameters (Fig. S2 of SuppDC, Country-specific distributions for the estimated contribution of SDCs to total HIV incidence were then generated, and used to calculate the mean and associated 95% CIs of this estimate. A sensitivity analysis was also conducted to assess the sensitivity of estimates to variations in two key model parameters that determine the annual risk of HIV-1 transmission within an SDC: p and n.


Key demographic and HIV-1 related indicators shown in Tables 1 and 2 reveal considerable heterogeneity in HIV-1 prevalence and numbers of SDCs across countries of SSA. The fraction of the population of reproductive age engaged in stable couples ranged from 35% in Swaziland to 76% in Niger with a mean of 59% across countries. HIV-1 prevalence ranged between 0.5% in Senegal and 23.0% in Lesotho, and the estimates of HIV incidence rate ranged from 0.05 per 100 person-years in Senegal to 2.9 per 100 person-years in Swaziland. The proportion of couples that were sero-discordant ranged between 0.4% in Senegal and 17.2% in Lesotho. On average, women were the HIV-1 infected partner in half of all SDCs, as previously reported [24]. The fraction of circumcised men among these SDCs varied between 9% in Zimbabwe and 100% in several countries. Very few individuals in SDCs used condoms at last sex with a median of 5%, though with some notable exceptions (e.g. 24% of couples reported condom use at last sex in Lesotho and Swaziland).

Table 2:
Key epidemiological indicators used in the model and estimates for the contribution of HIV incidence among stable HIV-1 serodiscordant couples to total HIV population-level incidence across sub-Saharan Africa.

Across the 20 countries, 545 046 new HIV infections are expected to occur each year, of which 162 207 infections (29.8%) would be among identifiable SDCs. The mean of the country-specific point estimates for the contribution of identifiable HIV incidence among SDCs to total HIV incidence was 29.3%, with a range of 10.2–52.2% across countries (Table 2). Figure 2 shows the mean and 95% CI for this measure across all 20 countries using the multivariate uncertainty analysis. Estimates for the various countries were largely close to the mean contribution of 29.3%, although countries with general population HIV epidemics and higher HIV-1 prevalence tended to have lower estimates for the contribution of identifiable HIV incidence among SDCs. The upper bound of the 95% CI seldom exceeded 50%. The only notable exception was Niger with a mean contribution of 57.1% (95% CI: 37.2–81.5%).

Fig. 2:
Mean and 95% confidence interval of estimates for the contribution of identifiable HIV incidence among stable HIV-1 serodiscordant couples to total HIV population-level incidence across sub-Saharan Africa.Estimates are calculated using the country-specific likelihood distributions of outcome for this measure. The distributions were generated using 200 000 runs of the model through Monte Carlo sampling from uniform distributions for the specified ranges of uncertainty of the demographic, biological, and epidemiological parameters of the model. Countries are shown in order of increasing HIV-1 prevalence.

The main drivers of variability for all countries were the fraction of the population in reproductive age engaged in stable couples, the prevalence of HIV-1 sero-discordancy among stable couples, HIV population-level incidence rate, and importantly the annual risk of HIV-1 transmission within an SDC. The latter is determined in our model by p and n, which are influenced by male circumcision and condom use coverage. For countries where HIV-1 prevalence is low (∼1%), the number of SDCs in the DHS sample was small contributing to wide CIs for some of the measures.

Figure 3 presents the findings of a sensitivity analysis assessing the sensitivity of our estimates for the contribution of identifiable HIV incidence among SDCs, as applied to Kenya as an example, to variations in the two key parameters: p and n. Although the analysis shows some variability underlying the estimates, it also shows that it is unlikely for the incidence within SDCs to contribute the majority of HIV incidence in SSA. In fact, with assumptions about lower coital frequencies, as observed in earlier DHS data [25] and recent clinical trials [1,2], the contribution of identifiable SDCs to overall HIV incidence would be lower than the estimates described above.

Fig. 3:
Sensitivity analysis assessing the sensitivity of estimates for the contribution of identifiable serodiscordant couples (SDCs) to overall HIV incidence to variations in two key model parameters that determine the annual risk of HIV-1 transmission within an SDC: HIV-1 transmission probability per coital act and the frequency of coital acts.The model presented is for Kenya. The ranges shown in the figure represent those used in the uncertainty analyses.


We found that identifiable HIV-1 sero-conversions among SDCs contribute slightly less than a third of the total number of new HIV infections in SSA. The uncertainty and sensitivity analyses further indicated that the contribution of SDCs to total HIV incidence is unlikely to exceed 50%. Perhaps counter to intuition, lower proportionate contributions from identifiable SDCs were found for countries with larger general population HIV epidemics, especially in the world's largest epidemics of Swaziland and Lesotho. This finding results from the fact that, in high HIV prevalence countries, the proportion of couples affected by HIV that are discordant is lower than in low prevalence countries [6]. Moreover, higher proportions of people are infected overall in general population HIV epidemics, thus increasing the chances for any of the other categories of HIV transmission besides the identifiable HIV incidence among SDCs to occur. Engagement in stable couples is also lower in high HIV prevalence countries, whereas condom use among couples is reported to be higher (Table 1).

Notably, our findings are corroborated by empirical evidence from the Rakai Study in Uganda [10]. In Rakai, 24% of infections occurred among SDCs after adjustment for couples with unknown sero-status, a value remarkably within the range of our estimates. As Uganda is in a state of a general population HIV epidemic, it aligns even better with our somewhat lower estimates for general population HIV epidemics.

Our findings also support an intuitive paradigm of HIV transmission within the context of the heterosexual HIV epidemic in SSA. Every HIV sero-conversion from the infected to the uninfected partner within an SDC was transmitted from an index partner who most likely acquired the infection from a source external to SDCs, rather than within the context of a previous SDC. Consequently, the contribution of SDCs to total HIV incidence is not likely to exceed 50% [7,10]. Of note that this argument assumes specific assumptions about the structure of partnerships and temporal evolution of the epidemic which may not be always strictly valid such as in relation to partnership turn-over (no partnership dissolution), concurrency of partnerships (no concurrent partnerships), and epidemic phase (endemic at equilibrium). On the contrary, the majority of the population in reproductive age in SSA engages in at least one stable sexual partnership over their reproductive lifespan (Table 1). This creates opportunities for HIV transmissions within the context of stable couples, particularly considering the high annual risk of transmission from the infected to the uninfected partner within an SDC [4]. Therefore, a pattern of every two infections outside of SDCs leading on average to one infection within the context of SDCs seems plausible.

Our findings are seemingly at odds with a quantitative assessment that attributed the majority of new heterosexually acquired HIV infections in Rwanda and Zambia to cohabitating couples [8]. However, that model assumes a large value for the annual risk of HIV transmission within an SDC (20 per 100 person-years) in contrast with the majority of available empirical estimates [4,18,19, Hughes JP, personal communication, 2010], and used an unrealistic simplifying assumption about partnering, that all couples were assumed to be stable and were only segregated by whether they were cohabitating or not cohabitating. The model does not account for the various other categories of HIV-1 transmission in the population. Lastly, it relies on a comparison between the risk of HIV acquisition through cohabiting couples versus the risk of HIV acquisition through noncohabiting partnerships, as opposed to comparing the incidence arising within SDCs to the actual total HIV incidence in the population. Our approach in contrast relies on a direct comparison between the estimated incidence among SDCs, derived based on nationally representative empirical measures of HIV sero-discordancy, and the total incidence in the population. Our approach does not depend on assumptions about the uncertain risk of HIV infection through noncohabiting partnerships.

Our estimates are possibly biased towards over-estimating the role of SDCs in driving HIV incidence across SSA, rather than underestimating it. We used a coital frequency value of about eight coital acts per month as per the median reported value in the Rakai Study [18], whereas other data suggest a lower coital frequency of four to six coital acts per month among SDCs [25]. Furthermore, HIV status disclosure and the resulting uptake of prevention interventions among SDCs will presumably lead to a lower unprotected coital frequency and hence a lower risk of transmission, as suggested by a review of available evidence for preventing HIV transmission among SDCs in SSA [9], and affirmed in intensively counseled cohorts [1,2,19,26,27].

Several study limitations might have affected our findings. First, calculating the contribution of identifiable HIV incidence among SDCs to total HIV incidence across SSA requires a vast volume of data to parameterize the quantitative assessment. Our analysis is limited by the quality and precision of available data. Some of the country-specific data are only available with small sample sizes and wide CIs affecting thereby the precision of our estimates. This is particularly true in the lowest HIV prevalence countries of Niger and Senegal. Reliance on multiple data sources that use different methodologies could also potentially lead to inconsistencies that may affect our predictions. Our results may have been also biased because of inherent limitations in the DHS data, such as recentness of the epidemiological indicators, variability in response rates to HIV testing [28], and selection-bias in restricting our analysis to couples with complete HIV sero-status information, for which we made no correction. Although a margin of error may affect the preciseness of our quantitative estimates, our findings are qualitatively robust that only a minority of HIV incidence occur among SDCs.

Furthermore, our analysis does not consider several factors that may affect HIV transmissibility in couples such as the presence of other sexually transmitted infections [29,30], tropical coinfections that increase HIV viral load [31], viral factors [32–34], and host genetics and immunology [32] which could potentially affect the contribution of identifiable HIV incidence among SDCs. The rates of HIV-1 transmission per coital act from the Rakai and Partners in Prevention studies used in our model were measured in settings in which such factors might have partially existed, but variability in those factors, and how they influence variability in the risk of within-couple transmission between countries, could not be included in our model because of lack of data. Our estimates are more likely to over-estimate the contribution of SDCs to HIV incidence in settings in which HIV-1 transmission probability per coital act is lower than that in the Rakai or Partners in Prevention studies (after controlling for male circumcision), and underestimate it in areas where HIV transmission probability is higher (Fig. 3). If HIV-1 transmission probability per coital act was higher in high HIV prevalence countries (as may be the case of viral sub-types in those countries leading to, for instance, longer period of elevated viral concentration earlier in infection) [34], this could affect the results for those countries.

We have assessed the contribution of SDC transmissions that are identifiable only in principle and under rather optimal conditions. Thus, our estimates delineate the maximum possible impact for interventions that target SDCs, rather than a projection of the range of impact that could be expected in realistic intervention scenarios. For interventions to be effective among SDCs, couples would need to agree to joint testing, undertake appropriate prevention interventions, and comply with these interventions over a sustained period of time.

Despite the established efficacy of several prevention interventions in reducing HIV heterosexual transmission [2,3,35–37], prioritizing prevention interventions among SDCs will likely not avert the majority of HIV incidence in SSA. Prevention interventions among SDCs can have indirect effects on averting HIV transmissions to casual partners as well as on reducing HIV onward transmission in the population. Nonetheless, as onward transmission of HIV following HIV acquisition among the susceptible partner in an SDC may be limited, a prevention approach focused on the major drivers of HIV transmission is more likely to yield substantial reduction in incidence. If incidence levels in SSA were to decline and more people initiated antiretroviral drugs, the contribution of identifiable SDCs to total HIV incidence could change. Prevention efforts among identified SDCs deserve focused attention, but overall, a balanced ‘combination’ approach to HIV prevention utilizing a variety of interventions is clearly called for [38].

In conclusion, we quantified the role of HIV incidence among identifiable SDCs to total HIV incidence across SSA using a mathematical model parameterized by nationally representative data. Our parsimonious quantitative approach minimizes the chance of structural modeling assumptions influencing the results, and incorporates uncertainty and sensitivity analyses that demonstrate a degree of robustness in the main findings to sources of potential error. We conclude that HIV-1 seroconversions among identifiable SDCs are only a fraction of the total number of new HIV infections each year in SSA. HIV policy initiatives focused on HIV prevention among SDCs need to be considered in light of this finding.


We thank Dr Eleanor Gouws for kindly providing us with the UNAIDS SPECTRUM model incidence estimates, and Professor Richard Hayes for insightful comments that inspired this work. H.C., T.B.H. and L.J.A. are grateful for the Qatar National Research Fund for supporting this work (NPRP 08–068–3–024), and the support provided by the Biostatistics, Epidemiology, and Biomathematics Research Core at the Weill Cornell Medical College in Qatar. L.J.A. and T.B.H. are also grateful to the National Institutes of Health (R01 AI083034), and T.B.H. to The Welcome Trust, for supporting this work. H.C. managed the DHS databases, conducted most of the statistical and mathematical modeling analyses, and wrote the first draft of the article. L.J.A. conceived, led the design of the study, mathematical modeling analyses, and drafting of the article. All authors contributed to the conduct of the study, the interpretation of the results, and the writing of the article.

Disclose funding received for this work: The Qatar National Research Fund (NPRP 08–068–3–024), the Biostatistics, Epidemiology, and Biomathematics Research Core at the Weill Cornell Medical College- Qatar, National Institutes of Health, and the Wellcome Trust.

Disclaimer: The views expressed in this article are not necessarily those of USAID.

Conflicts of interest

There are no conflicts of interest.


1. Celum C, Wald A, Lingappa JR, Magaret AS, Wang RS, Mugo N, et al. Acyclovir and transmission of HIV-1 from persons infected with HIV-1 and HSV-2. N Engl J Med 2010; 362:427–439.
2. Cohen MS, Chen YQ, McCauley M, Gamble T, Hosseinipour MC, Kumarasamy N, et al. Prevention of HIV-1 infection with early antiretroviral therapy. N Engl J Med 2011; 365:493–505.
3. World Health Organization. Guidance on couples HIV testing and counselling including antiretroviral therapy for treatment and prevention in serodiscordant couples. Geneva, Switzerland: Department of HIV/AIDS-World Health Organization; 2012. pp. 15–18.
4. Guthrie BL, De Bruyn G, Farquhar C. HIV-1-discordant couples in Sub-Saharan Africa: explanations and implications for high rates of discordancy. Curr HIV Res 2007; 5:416–429.
5. Lingappa JR, Lambdin B, Bukusi EA, Ngure K, Kavuma L, Inambao M, et al. Regional differences in prevalence of HIV-1 discordance in Africa and enrollment of HIV-1 discordant couples into an HIV-1 prevention trial. PLoS One 2008; 3:e1411.
6. Chemaitelly H, Cremin I, Shelton JD, Hallett TB, Abu-Raddad LJ. Distinct HIV discordancy patterns by epidemic size in stable sexual partnerships in sub-Saharan Africa. Sex Transm Infect 2012; 88:51–57.
7. Shelton JD. A tale of two-component generalised HIV epidemics. Lancet 2010; 375:964–966.
8. Dunkle K, Stephenson R, Karita E, Chomba E, Kayitenkore K, Vwalika C, et al. New heterosexually transmitted HIV infections in married or cohabiting couples in urban Zambia and Rwanda: an analysis of survey and clinical data. The Lancet 2008; 371:2183–2191.
9. Matovu JK. Preventing HIV transmission in married and cohabiting HIV-discordant couples in sub-Saharan Africa through combination prevention. Curr HIV Res 2010; 8:430–440.
10. Gray R, Ssempiija V, Shelton J, Serwadda D, Nalugoda F, Kagaayi J, et al. The contribution of HIV-discordant relationships to new HIV infections in Rakai, Uganda. AIDS 2011; 25:863–865.
11. UNAIDS. HIV estimates with uncertainty bounds 1990–2009. In: Recent reports and data: UNAIDS Report on the Global AIDS Epidemic – 2010; 2010. http://
12. MEASURE DHS. Demographic and health surveys. In: MEASURE DHS. Calverton: ICF Macro; 2012. http://
13. United Nations Population Division. World Population Prospects: the 2008 Revision Population Database. [Accessed August 2011].
14. Rutstein S., Rojas G. Guide to DHS statistics. In. Calverton, Maryland: ORC Macro; 2006. pp. 171. http://
15. MEASURE DHS. Demographic and Health Surveys. In: Frequently asked questions: sampling and weighting: MEASURE DHS; 2011. http://
16. UNAIDS. Epidemiological software and tools. 2010. [Accessed September 2011].
17. Stover J, Johnson P, Hallett T, Marston M, Becquet R, Timaeus IM. The Spectrum projection package: improvements in estimating incidence by age and sex, mother-to-child transmission, HIV progression in children and double orphans. Sex Transm Infect 2010; 86 (Suppl 2):ii16–ii21.
18. Wawer MJ, Gray RH, Sewankambo NK, Serwadda D, Li X, Laeyendecker O, et al. Rates of HIV-1 transmission per coital act, by stage of HIV-1 infection, in Rakai, Uganda. J Infect Dis 2005; 191:1403–1409.
19. Hughes JP, Baeten JM, Lingappa JR, Magaret AS, Wald A, de Bruyn G, et al. Determinants of per-coital-act HIV-1 infectivity among African HIV-1-serodiscordant couples. J Infect Dis 2012; 205:358–365.
20. Weiss HA, Halperin D, Bailey RC, Hayes RJ, Schmid G, Hankins CA. Male circumcision for HIV prevention: from evidence to action?. AIDS 2008; 22:567–574.
21. Auvert B, Taljaard D, Lagarde E, Sobngwi-Tambekou J, Sitta R, Puren A. Randomized, controlled intervention trial of male circumcision for reduction of HIV infection risk: the ANRS 1265 Trial. PLoS Med 2005; 2:e298.
22. Bailey RC, Moses S, Parker CB, Agot K, Maclean I, Krieger JN, et al. Male circumcision for HIV prevention in young men in Kisumu, Kenya: a randomised controlled trial. The Lancet 2007; 369:643–656.
23. Gray RH, Kigozi G, Serwadda D, Makumbi F, Watya S, Nalugoda F, et al. Male circumcision for HIV prevention in men in Rakai, Uganda: a randomised trial. The Lancet 2007; 369:657–666.
24. Eyawo O, de Walque D, Ford N, Gakii G, Lester RT, Mills EJ. HIV status in discordant couples in sub-Saharan Africa: a systematic review and meta-analysis. Lancet Infect Dis 2010; 10:770–777.
25. Brown MS. Coitus, the proximate determinant of conception: inter-country variance in sub-Saharan Africa. J Biosoc Sci 2000; 32:145–159.
26. Ruzagira E, Wandiembe S, Abaasa A, Bwanika AN, Bahemuka U, Amornkul P, et al. HIV incidence and risk factors for acquisition in HIV discordant couples in Masaka, Uganda: an HIV vaccine preparedness study. PLoS One 2011; 6:e24037.
27. Ndase P, Celum C, Thomas K, Donnell D, Fife KH, Bukusi E, et al. Outside sexual partnerships and risk of HIV acquisition for HIV uninfected partners in African HIV serodiscordant partnerships. J Acquir Immune Defic Syndr 2012; 59:65–71.
28. Mishra V, Barrere B, Hong R, Khan S. Evaluation of bias in HIV seroprevalence estimates from national household surveys. Sex Transm Infect 2008; 84 (Suppl 1):i63–i70.
29. Abu-Raddad LJ, Magaret AS, Celum C, Wald A, Longini IM Jr, Self SG, et al. Genital herpes has played a more important role than any other sexually transmitted infection in driving HIV prevalence in Africa. PLoS ONE 2008; 3:e2230.
30. Korenromp EL, de Vlass SJ, Nagelkerke NJ, Habbema JD. Estimating the magnitude of STD cofactor effects on HIV transmission: how well can it be done?. Sex Transm Dis 2001; 28:613–621.
31. Abu-Raddad LJ, Patnaik P, Kublin JG. Dual infection with HIV and malaria fuels the spread of both diseases in sub-Saharan Africa. Science 2006; 314:1603–1606.
32. Kaul R, Cohen CR, Chege D, Yi TJ, Tharao W, McKinnon LR, et al. Biological factors that may contribute to regional and racial disparities in HIV prevalence. Am J Reprod Immunol 2011; 65:317–324.
33. Novitsky V, Woldegabriel E, Kebaabetswe L, Rossenkhan R, Mlotshwa B, Bonney C, et al. Viral load and CD4+ T-cell dynamics in primary HIV-1 subtype C infection. J Acquir Immune Defic Syndr 2009; 50:65–76.
34. Novitsky V, Ndung’u T, Wang R, Bussmann H, Chonco F, Makhema J, et al. Extended high viremics: a substantial fraction of individuals maintain high plasma viral RNA levels after acute HIV-1 subtype C infection. AIDS 2011; 25:1515–1522.
35. Partners PrEP Study Team. Antiretroviral preexposure prophylaxis for HIV-1 prevention among heterosexual African men and women: the Partners PrEP Study. In: 6th IAS Conference on HIV Pathogenesis, Treatment and Prevention. Rome, Italy: International AIDS Society; 2011.
36. Abdool Karim Q, Abdool Karim SS, Frohlich JA, Grobler AC, Baxter C, Mansoor LE, et al. Effectiveness and safety of tenofovir gel, an antiretroviral microbicide, for the prevention of HIV infection in women. Science 2010; 329:1168–1174.
37. Thigpen MC, Kebaabetswe PM, Smith DK, Segolodi TM, Soud FA, Chillag K, et al. Daily oral antiretroviral use for the prevention of HIV infection in heterosexually active young adults in Botswana: results from the TDF2 study.6th International AIDS Society Conference on HIV Pathogenesis, Treatment, and Prevention. Rome, Italy: International AIDS Society; 2011.
38. Schwartlander B, Stover J, Hallett T, Atun R, Avila C, Gouws E, et al. Towards an improved investment approach for an effective response to HIV/AIDS. Lancet 2011; 377:2031–2041.

demographic and health surveys; HIV incidence; mathematical model; sero-discordant couples; sub-Saharan Africa

Supplemental Digital Content

© 2013 Lippincott Williams & Wilkins, Inc.