The proportion of heterosexual HIV transmission in sub-Saharan Africa that occurs within cohabiting partnerships, compared with that in single people or extra-couple relationships, is widely debated. We estimated the proportional contribution of different routes of transmission to new HIV infections. As plans to use antiretroviral drugs as a strategy for population-level prevention progress, understanding the importance of different transmission routes is crucial to target intervention efforts.
We built a mechanistic model of HIV transmission with data from Demographic and Health Surveys (DHS) for 2003—2011, of 27 201 cohabiting couples (men aged 15—59 years and women aged 15—49 years) from 18 sub-Saharan African countries with information about relationship duration, age at sexual debut, and HIV serostatus. We combined this model with estimates of HIV survival times and country-specific estimates of HIV prevalence and coverage of antiretroviral therapy (ART). We then estimated the proportion of recorded infections in surveyed cohabiting couples that occurred before couple formation, between couple members, and because of extra-couple intercourse.
In surveyed couples, we estimated that extra-couple transmission accounted for 27—61% of all HIV infections in men and 21—51% of all those in women, with ranges showing intercountry variation. We estimated that in 2011, extra-couple transmission accounted for 32—65% of new incident HIV infections in men in cohabiting couples, and 10—47% of new infections in women in such couples. Our findings suggest that transmission within couples occurs largely from men to women; however, the latter sex have a very high-risk period before couple formation.
Because of the large contribution of extra-couple transmission to new HIV infections, interventions for HIV prevention should target the general sexually active population and not only serodiscordant couples.
US National Institutes of Health, US National Science Foundation, and J S McDonnell Foundation.
In the past 2 years, major research advances have been made in anti-HIV interventions. Antiretroviral drugs can help prevent HIV transmission, either by reducing infectiousness when given as antiretroviral therapy (ART) to HIV-positive individuals (treatment as prevention [TasP]),1, 2 or by reducing the susceptibility of HIV-negative individuals when given as oral or topical pre-exposure prophylaxis (PrEP).3, 4 These advances have led to debate about how best to use ART to further reduce HIV incidence.5 An approach that combines several biomedical and behavioural interventions will be needed,6 and policy makers are debating the criteria used to target interventions, including TasP and PrEP.
A serodiscordant couple, defined as an HIV-positive and HIV-negative individual in an ongoing sexual relationship, is a clear example of a susceptible individual being at risk of HIV infection from an infectious individual.7, 8 Targeting of well defined, high-risk groups such as seronegative individuals in serodiscordant partnerships is expected to be resource-efficient. Thus, research of HIV transmission and intervention efficacy has tended to focus on cohorts of serodiscordant couples7 such that seronegative individuals in these partnerships are often the first group in which a new intervention is shown to work. For example, in response to the proven effectiveness of TasP in prevention of transmission in a cohort of serodiscordant couples,1 WHO has recommended this strategy to HIV-positive partners in serodiscordant couples, irrespective of immune status.9 However, not all transmission is within serodiscordant couples; routes also include infection of individuals who are single, and of those in couples by sexual partners outside their relationship (extra-couple relationships). Granich and colleagues10 propose a test-and-treat policy that would target all heterosexual routes of transmission. This approach consists of annual voluntary testing of the entire sexually active population, with immediate and sustained provision of ART to those who test HIV positive. This approach is more expensive and logistically difficult than are targeted approaches, and its value is strongly dependent on the proportion of new transmission events that occur between partners in serodiscordant couples versus those occurring by other routes.
We constructed a mathematical model to estimate rates of HIV transmission before couple formation, rates attributable to extra-couple intercourse, and rates within serodiscordant couples, to assess the proportional contribution of different routes of transmission to new HIV infections. Because the probability that an individual acquires HIV during any period is a function of the period’s duration,11 we disentangled routes of transmission by relating couple serostatus to information about couple duration, duration of sexual activity before couple formation, the population prevalence of HIV, and age-specific estimates of HIV survival.
Our findings show three major conclusions. First, extra-couple transmission has played and still plays a major part in driving HIV incidence for both sexes, but particularly for men; second, within couples, HIV seems to be propagated more from men to women than vice versa; third, women have a period of high infection risk before entering a cohabiting partnership. We emphasise that the fitted transmission coefficients aggregate several behavioural and physiological processes and thus should be interpreted cautiously. Because the hazard of infection is the product of transmission coefficients and prevalence in the opposite sex, comparisons between male and female transmission coefficients should be made with consideration of the differing HIV prevalences for each sex. For example, although we estimate that more men than women are infected through extra-couple transmission, the estimated transmission coefficients are roughly similar because female infectious HIV prevalence is greater than that of the opposite sex. The transmission coefficients will also partially absorb unmodelled mixing patterns. For example, young women tend to mix with older men, who have a greater probability of being seropositive than do younger men. This effect would tend to increase female incidence before partnership formation, thus needing greater fitted female before-partnership transmission coefficients to fit the recorded data, but not necessarily biasing the estimate of incidence through this route.
Investigators of previous studies17—23 using DHS and similar cross-sectional couple data have come to diverse conclusions (panel). Analyses of DHS couples data have noted that slightly less than half of serodiscordant couples had seropositive women rather than men;17, 20 others used mathematical models to estimate the proportion of transmission that took place outside serodiscordant partnerships versus between partners.21, 22 These studies all conclude that the high prevalence of female-positive and male-positive serodiscordant partnerships suggests that, contrary to mainstream beliefs,20 both women and men often have risky extra-couple intercourse, with the modelling studies estimating that much of the transmission to both sexes is from outside rather than within the couple. These studies have largely overlooked that routes of infection cannot be directly inferred from cross-sectional data, such as DHS. Estimations of transmission from outside a couple combine infections occurring from extra-couple intercourse with those acquired before that couple’s formation when the individual was either single or in another couple. Thus, the existence of serodiscordant couples does not necessarily suggest extra-couple transmission, and estimates of the proportion of transmission from outside existing partnerships do not measure extra-couple transmission.
A second important factor largely overlooked in analyses of cross-sectional couple data is survival bias—ie, only couples in which both partners survive to be sampled are recorded. Median survival time after seroconversion is about 6—13 years, dependent on the age at seroconversion.12 Many couples in which one or both partners become infected are thus removed from the population before the sample is taken. This effect will be different for serodiscordant and seroconcordant couples. Studies analysing cross-sectional couple data while ignoring mortality17,20—22,28 could therefore yield biased conclusions for the proportional contribution of extra-couple intercourse to incidence.
Our findings show that extra-couple and within-couple transmission are both important routes of HIV infection and both account for many recorded infections in men and women; however, results vary substantially by country. We obtained this result despite finding that fitted extra-couple transmission coefficients were by far the smallest of the three routes of infection. This result is consistent with Chemaitelly and colleagues’28 finding that most infections in serodiscordant couples are due to within-couple transmission. The large contribution of extra-couple transmission at the population level is because most cohabiting couples are concordant negative and, on average, the surveyed individuals had spent most time in a couple since their sexual debut. Thus, the large amount of person-time spent at risk from extra-couple transmission more than compensates for its small transmission coefficients. Results from our analysis, which was only of couples, greatly contrast those of Dunkle and colleagues,19 who concluded that within-couple transmission accounts for most of the HIV incidence in sexually active urban populations (ie, in single individuals and those in couples) in Zambia and Rwanda. This contrast is probably because of the reliance of Dunkle and colleagues on downwards-biased self-reported rates of intercourse with non-cohabiting partners, which could lead to substantial underestimation of the contribution of extra-couple intercourse.29
When available, molecular evidence shows the importance of extra-couple transmission. In several cohort studies of serodiscordant couples,1,2,25—27 13—32% of incident infections were from virus not linked to that isolated from the seroconverter’s partner and were presumably due to extra-couple intercourse. Compared with cohort studies, we attributed a smaller proportion of transmission within serodiscordant couples to extra-couple intercourse, which might be because individuals enrolled in cohort studies differ systematically from the general population, which is more representatively sampled by DHS. Furthermore, seronegative individuals in cohort studies might engage in more extra-couple and less within-couple intercourse upon finding that their partner is seropositive.27 This behavioural effect could explain why our estimated rates of within-couple transmission are generally greater than those from cohort studies (table 2).1, 2, 25
Our finding that, within couples, the directionality of HIV propagation is more from men to women than vice versa is because of the greater average duration of sexual activity in men before couple formation and additionally, for some countries, because of their greater hazard rate for extra-couple infection. Although the average duration of sexual activity before partnership formation is much shorter for women than for men, we noted that, as reported elsewhere,11 this difference is partly compensated by the greater risk of infection per unit time in women before partnership formation.
With use of relationship and serostatus data, country-specific trends for the prevalence of HIV, and estimates of HIV survival times to explicitly estimate the probability that infections were because of pre-couple, within-couple, or extra-couple transmission, our model addresses several limitations of previous studies, and advances estimations of transmission breakdown by behavioural routes from cross-sectional data. However, our model retains certain assumptions. We assumed homogeneous mixing between age groups for sexual partners chosen before couple formation or during extra-couple intercourse. Although this assumption might bias our results, to the extent that patterns of age mixing cause a consistent bias for overestimates or underestimates in the estimated prevalence that individuals are exposed to, this bias will be counteracted by underestimates or overestimates in transmission coefficients, with no effect on estimates of total hazard and per-route contributions to transmission.
We also assumed that the probability of infection via a particular transmission route is dependent on only the duration an individual is at risk by that route, the time-varying HIV prevalence in the population of the opposite sex (or partner seropositivity for within-partner transmission), and a transmission coefficient for each gender-route combination. In reality, the frequency of intercourse and the number and riskiness of partners also affect transmission. Other causes of heterogeneity not considered here include genetic and immunological factors, type of sexual exposure, sexually transmitted infections, viral loads, viral characteristics, tendency to seek care, male circumcision, and protected sex; many of these factors vary both between individuals and through time within individuals.7, 16 Although we assumed that individuals were homogeneous, our results were robust to this assumption. Our sensitivity analysis shows that even with a large individual-level heterogeneity in hazard rates, the association between relationship histories and serostatuses was substantial enough for the model to accurately infer the proportional breakdown of infections by transmission routes.
Hazards can vary over time for reasons other than changing prevalence. Declines in HIV prevalence in several countries have been attributed to behavioural changes in response to interventions or overall HIV awareness.30 Such changes would lead to decreasing transmission coefficients during the epidemic, but how this decrease might be divided among the routes of transmission we considered is unclear; therefore, we were unable to assess this possibility. We did not include effects of ART on HIV survival times or within-couple transmission in our main analysis because DHS surveys do not provide the drug status of individuals, and because we believe that the within-couple effects of therapy were small. On the basis of policies created before WHO’s 2012 TasP recommendations, most treated individuals would have already exposed their partners to infection for a long time before they become ill, get tested, have CD4 counts decrease to less than 200 counts per μL, and start ART. Furthermore, coverage of ART in the countries analysed was negligible for most of the period covered by the couples in our survey.13 This factor explains why our results were robust in sensitivity analyses allowing for ART to affect within-couple transmission or relaxing the assumption that all individuals on ART are non-infectious.
Finally, in view of the range of the DHS and the relatively narrow scope of our study, we necessarily excluded many couples because of missing or inconsistent data. However, these exclusions are unlikely to cause major selection bias and our results are roughly generalisable to the couples in the population as sampled by DHS. In particular, our results are likely to be more representative of the general population than are those from virological linkage cohort studies, which have more specific selection criteria and alter the behaviour of participants.27
We have shown that substantial HIV transmission occurs through all transmission routes: within serodiscordant couples and before couple formation and from extra-couple intercourse. We make no assumptions about the morality31 or potential for mitigation32 of extra-couple sex. Extra-couple sex does not necessarily constitute a choice and could be motivated by basic needs or indicate large social support structures.33 However, policy choices should be made in view of our finding that extra-couple transmission by both sexes has a major role in the HIV epidemic in sub-Saharan Africa.
Offering of TasP to only HIV-positive individuals in stable, serodiscordant couples is tempting because the partner is identifiable, and clearly at risk. However, the aggregate risk to partners not in stable relationships with positive individuals is also high. This finding does not mean that TasP and PrEP programmes have no place in targeted treatment of serodiscordant couples. These programmes have been effective and represent major advances in HIV prevention strategy. PrEP, in particular, could change the gender power dynamics in serodiscordant couples by empowering women to prevent HIV transmission. In view of this fairly small proportion of populations constituted by serodiscordant couples, these approaches could be a good starting point for HIV control efforts, especially in the context of resource limitations. However, our results do imply that behavioural and biomedical interventions focused on serodiscordant couples will not be sufficient to cause major reductions in HIV incidence at the population level. Interventions should address all transmission routes to fight the HIV epidemic. Despite its expense and logistical demands, the universal test-and-treat strategy could reduce all forms of heterosexual transmission.