Secondary Logo

Journal Logo

Infectious Disease

Heterosexual HIV-1 Infectiousness and Antiretroviral Use

Systematic Review of Prospective Studies of Discordant Couples

Baggaley, Rebecca F.a; White, Richard G.b; Hollingsworth, T. Déirdrea; Boily, Marie-Claudea

Author Information
doi: 10.1097/EDE.0b013e318276cad7


HIV-1 remains among the world’s leading causes of mortality, with an estimated 2.7 million new infections in 2010.1 Treatment-as-prevention has been proposed as a much-needed new tool to prevent HIV transmission.2 Although the efficacy of antiretroviral therapy (ART) for reducing the risk of HIV-1 transmission between stable partners has recently been estimated in the landmark HIV Prevention Trials Network (HPTN) 052 trial,3 such results do not necessarily reflect what would happen in less controlled settings (with less intensive risk-reduction and adherence counseling, for example) or if ART was scaled up more widely as treatment-as-prevention. The objective of this study is to comprehensively review all prospective studies of discordant couples to summarize HIV-1 transmission estimates by level of ART use by the infected partner (index case). These include many early and observational studies that involve study populations more similar to real-life populations.

We examined studies that followed couples who were discordant by HIV-1 status, from the time of diagnosis of the index case. This design has some limitations, notably selection bias. Couples in stable partnerships may therefore be at lower risk of transmitting than those in the general population. In addition, these studies are subject to left censoring, or frailty selection, whereby the most vulnerable couples of the so-called “high and fast transmitters” rapidly become seroconcordant4,5 and therefore ineligible for recruitment. Combined, these factors may result in oversampling of less infective index cases and less susceptible partners who remain uninfected longer, leading to underestimates of infectivity. Nonetheless, this study design is often judged preferable for estimation of per-partnership transmission because the enrollment of monogamous couples avoids having to account for partners with unknown serostatus. In addition, the time of seroconversion of the susceptible partner can be determined more accurately than from cross-sectional studies. Studies with frequent follow-up can also provide more reliable information with less recall bias on frequency and type of contact.6

The aim of this study is to summarize HIV-1 infectiousness for discordant couples by level of ART use, through a systematic review and meta-analysis. Our first “within-study” analysis systematically reviews all prospective discordant-couple studies directly comparing couples in which the index receives ART with those not receiving ART (ART-stratified studies). Our second, broader analysis includes all prospective heterosexual discordant-couple studies to provide a comprehensive summary of infectiousness estimates and to compare HIV-1 infectiousness estimates by level of ART use among each study sample. We also investigate the potential influence of direction of transmission (man-to-woman and woman-to-man) and setting (high- and low-income countries) on HIV-1 infectiousness and potential causes of heterogeneity across study estimates.


Search Strategy

The systematic review was undertaken according to the Meta-analysis of Observational Studies in Epidemiology (MOOSE) checklist for review of observational studies.7 Details of the search are published elsewhere,6 with a subsequent update to 31 July 2011. Our search terms were intentionally broad to capture publications estimating HIV-1 infectiousness of all types and modes of transmission, for use in other HIV-1 infectiousness reviews.6,8–10 We searched bibliographies of relevant articles to find additional publications not identified by electronic search. The search was not limited to English language articles, and we translated articles from other languages where required. In addition, we searched abstracts from the previous 2 years (2010, 2011) of International AIDS Society, Conference on Retroviruses and Opportunistic Infections, and International Society of Sexually Transmitted Research on “discordant.” Further search details are given in the eAppendix (

Selection Criteria and Data Extraction

We extracted data on two outcome measures of sexual HIV-1 infectiousness, cumulative incidence of infection over study follow-up periods and incidence rate after exposure to an HIV-1–infected partner; we focus our analysis on incidence rate. We included all empirical prospective studies reporting either type of estimate or studies from which such estimates could be derived. Infectiousness estimates for heterosexual sex (vaginal sex or vaginal and anal sex) are reported here; estimates for oral sex and for homosexual intercourse, with per-act transmission probabilities, are reported elsewhere.6,8,10

This analysis is reported in two parts. The first “within-study” meta-analysis included ART-stratified studies. The second analysis included studies in which infectiousness estimates were not stratified by the ART use of index cases (non-ART–stratified studies). We used these studies to compare estimates by level of ART use. Only the results from the control arms of randomized controlled trials (RCTs) of non-ART interventions were included (results from the intervention arm available on request). We excluded cross-sectional estimates, indirect estimates from transmission dynamics modeling studies, narrative or methodological reviews, abstracts pre-2000, and study estimates based on fewer than 10 subjects. Non-ART–stratified studies were categorized as no-ART-use if they: (1) reported <3% ART use by study participants; (2) did not report on ART use but follow-up was censored at 1996 or earlier11 for high-income countries or 2005 or earlier12 for low-income countries; or (3) did not state follow-up dates, and publication was before 1997 for high-income countries or before 2005 for low-income countries. All other estimates were classed as “any ART use.” We allowed for up to 3% ART use as a maximum level of contamination within the no-ART-use group.

Where publications have been superseded by more recent information, the later source was included, unless an earlier source provided a larger sample size. Each relevant publication was examined by two reviewers (R.F.B. and M.-C.B. or R.G.W.) who compiled information on estimates and on study and participant characteristics. Discrepancies were resolved by consensus. Where possible, couples reporting no sexual activity or 100% condom use were excluded. Nineteen authors were contacted to obtain additional information or to clarify potential overlap in study participants between publications; 10 responded. A list of all included and excluded studies is available on request.

Data Analysis and Statistical Methods

Forest plots were used to explore graphically the heterogeneity among study estimates.13 Heterogeneity was also explored using Cochran’s Q statistic14 and I2 statistic.15 If there was significant heterogeneity (using P < 0.05 as a guideline but taking into account sample size and without a strict cutoff), no overall pooled estimate was provided on forest plots. We explored potential causes of heterogeneity through subgroup analyses based on study-level characteristics: condom use, presence of sexually transmitted infections (STIs), HIV-1 infection stage of index (based on clinical stage or CD4 count data—see eAppendix, for classification methods), prevalence of male circumcision, study design (RCT/observational); and population characteristics: region (Americas and Europe/Africa/Asia) and country-level prevalence of male circumcision. There were insufficient data to investigate other risk factors such as HIV-1 viral load. We used forest plots to explore graphically the heterogeneity among random-effects summary estimates calculated for subgroups.13

Regarding condom use, STIs, and infection stage, we created aggregate study-level variables because of incomplete and noncomparable reporting of these risk factors among studies (see eAppendix, and figure legends for details). An ecologic analysis pairing estimates with country-level male circumcision prevalence data16 was used because of the lack of information on male circumcision in most studies.

Where relevant, we used random-effects Poisson regression models to obtain summary estimates for HIV-1 incidence rates and cumulative incidence, with 95% confidence intervals (95% CIs). For each study or stratum, the total number of events was considered to be Poisson distributed for a given sum of person-years. Poisson regression models were fitted with a logarithmic link function. Total exposure time (incidence rate) or initial number of couples at risk (cumulative incidence) per study was used as an offset variable, with γ-distributed random effects at the study level. Further details of the methods are provided in the eAppendix (


Summary of Search

Fifty prospective studies of discordant couples were included (eFigure 1, Nine studies fit the inclusion criteria for the first analysis, with HIV-1 transmission stratified by ART use of the index case3,17–25 (one study was reported in two abstracts20,21). Forty-one non-ART–stratified studies provided estimates for the second analysis, reporting either no (n = 29)26–54 or any (n= 12)55–66 ART use. Eighteen studies were excluded based on eligibility criteria (study details available on request). Several publications were excluded because they were superseded by more recent articles. Although Musicco et al17 and Saracco et al59 are from the same study, both were included (Musicco for the first analysis; Saracco for the second) because the former reports estimates stratified by ART status, whereas the latter does not; the same was true for Donnell et al24 and Celum et al.28 Most studies reported estimates of both HIV-1 incidence rate and cumulative incidence.

Among all publications included in these analyses, more were from low-income (n = 34) than high-income (n = 17) countries (Cohen et al3 provide results from nine countries, only one of which is high-income, and therefore, the study is classified here as low-income). Seven publications reported results from RCTs: one study testing ART use of the index (early vs. delayed ART),3 one testing pre-exposure prophylaxis,67 one testing herpes simplex virus suppressive therapy (Celum et al,28 non-ART–stratified; Donnell et al,24 ART-stratified), one testing male circumcision for prevention of man-to-woman HIV-1 transmission,53 and two cluster RCTs of a behavioral intervention55 and an STI treatment intervention.43 The remainder were observational studies.

ART-Stratified Studies (Within-Study Comparisons)

One of the nine studies reporting HIV-1 transmission stratified by ART use (Fig. 1)3,17–25 was excluded because the therapy was zidovudine monotherapy.17 An additional conference abstract reported stratification of ART use in the form of pre-exposure prophylaxis67; this study was included in the forest plot but not the pooled ART effect estimate. The included studies reported combined man-to-woman and woman-to-man transmission and were from low-income settings, except for one from Spain22 and the one combining results from nine study countries (of which one was high-income).3 All studies provided incidence rate estimates (or data from which such estimates could be derived), except Wang et al.23 Cohen et al3 was included, although it should be noted that a small number of homosexual couples were included in the analysis (2% men who have sex with men, <0.1% women who have sex with women). Four studies did not state cumulative incidence.19–21,25,67 Summary incidence rate was 0.2/100 person-years (95% CI = 0.07–0.7) among ART-receiving couples and 3.6/100 person-years (2.0–6.5) among non-ART–receiving couples (studies from all settings, n = 73,18–22,24,25; incidence rate ratio (IRR) = 0.09 [0.04–0.21], P < 0.001; Table), constituting a 91% (79–96%) reduction in per-partner HIV-1 incidence rate with ART use. Because only two included studies reported results stratified by direction of transmission,20,21,68 we did not perform stratified meta-analyses. Further details of studies are shown in eTable 2 (

Meta-Analysis of Incidence Rate Ratios for ART-Stratified Study Estimates, Comparing ART Use Among Index Cases with No ART Use
Forest plot summary of HIV-1 incidence rate estimates per heterosexual partnership for ART-stratified studies, with 95% confidence intervals. The first row for each study denotes couples where the index was receiving ART (forest plot boxes in blue); the second row denotes couples with no ART received (forest plot boxes in red). Size of boxes is proportional to number of couples, except for Watera et al19 and Sullivan et al,20 , 21 which do not provide these data. Baeten* et al67 refers to ART for the initially uninfected partner rather than the index (pre-exposure prophylaxis) and is shown in the figure but has not been included in the summary estimates. Inc indicates per-partnership HIV-1 incidence rate per 100 person-years; n, number of HIV-1–discordant couples; NR, not recorded in publication; x, number of HIV-1–transmitting couples.

Non-ART–Stratified Studies (Between-Study Comparisons)

Infectiousness by ART Use

HIV-1 incidence rate estimates ranged from 0.0 to 17.4/100 person-years for high-income settings (Fig. 2) and from 0.0 to 33.7/100 person-years for low-income settings (Fig. 3). Forest plots highlight the considerable heterogeneity among study estimates for low-income settings (Fig. 3, P < 0.001 test for heterogeneity for all no-ART-use estimates and combined man-to-woman and woman-to-man any ART use; insufficient data for any ART use by direction of transmission). High-income settings were more homogeneous, probably due to the small numbers of estimates per strata (Fig. 2; for combined transmission: no ART use, I2 = 33%, P = 0.110, n = 6; any ART use, I2 = 0%, P = 0.764, n = 2; for man-to-woman transmission: no ART use I2 = 0%, P = 0.388, n = 4; any ART use I2 = 60%, P = 0.032, n = 2; and for woman-to-man transmission: insufficient estimates). This heterogeneity and lack of sufficient estimates for some strata prevent straightforward interpretation of summary estimates. We therefore omitted these from the forest plots but provided them in eTable 3 ( However, although absolute values of summary transmission rate estimates by level of ART use may not be reliable, the IRR of 0.31 (95% CI = 0.14–0.69, P = 0.001) for difference in infectiousness, any versus no ART use (combined man-to-woman and woman-to-man, all settings), is compelling.

Forest plot summary of HIV-1 incidence rate estimates per heterosexual partnership, with 95% confidence intervals, for non-ART–stratified studies from high-income settings, by direction of transmission: (A) combined man-to-woman and woman-to-man transmission; (B) man-to-woman transmission; and (C) woman-to-man transmission. Estimates are classified as no ART use (up to 3% antiretroviral use by study participants—see Methods for classification criteria) and any ART use. Within these two groups, study estimates are plotted in order of increasing ART use and then chronologically. Size of boxes is proportional to number of couples. ART indicates reported percentage ART usage among index cases; Inc, per-partnership HIV-1 incidence rate per 100 person-years; NR, not recorded in publication.
(Continued) Forest plot summary of HIV-1 incidence rate estimates per heterosexual partnership, with 95% confidence intervals, for non-ART–stratified studies from low-income settings, by direction of transmission: (A) combined man-to-woman and woman-to-man transmission; (B) man-to-woman transmission; and (C) woman-to-man transmission. Estimates are classified as no ART use (up to 3% antiretroviral use by study participants—see Methods for classification criteria) and any ART use. Within these two groups, study estimates are plotted in order of increasing ART use and then chronologically. Size of boxes is proportional to number of couples. Hugonnet et al34 provides two per-partnership estimates: one risk for partners of infected individuals at baseline and one risk for partners of individuals who seroconverted during follow-up. ART indicates reported percentage ART usage among index cases; Inc, per-partnership HIV-1 incidence rate per 100 person-years; NR, not recorded in publication.

Study estimates in Figures 2 and 3 are presented in order of increasing ART coverage, but few non-ART–stratified studies reported the precise level of ART use among study participants (n = 12), and, thus, no strong dose-response relationship is observed (see eFigure 3, for additional plots). However, for the stratum with the most estimates (combined man-to-woman and woman-to-man transmission for low-income countries, Fig. 3A), there were sufficient data to analyze the IRR by ART level (Fig. 4). Changing the comparison from no versus any ART use to none/any versus 100% use decreased the IRR to 0.04 (95% CI = 0.005–0.36, P = 0.004).

Incidence rate ratios for HIV incidence rate estimates from low-income countries for combined man-to-woman and woman-to-man transmission, by ART coverage of index cases within each study. Estimates included in the analysis are shown in Figure 3A. Li et al64 and Guthrie et al62 are included in the no-versus-any-ART-use scenario but excluded from the other IRR calculations because their level of ART coverage was not reported. Error bars represent 95% confidence intervals.

Infectiousness by Setting

Incidence rates tended to be smaller in high-income countries (Fig. 2) than in low-income countries (Fig. 3). There were many estimates of zero transmission rates from high-income countries, even among the no-ART-use studies.37, 40,42,50,52 Of the 15 high-income estimates, only three35,41,57 were ≥5.0/100 person-years. Robertson et al57 reported on any ART use (prevalence of ART use not stated), but study subjects came from a population where index cases were current or former intravenous drug users, and, therefore, there may be HIV risk from needle contamination. There is no obvious explanation for the high estimate from Laurian et al35 (9.7/100 person-years for partners of hemophiliacs), but the sample size was small (n = 17). Thirty-three percent of couples followed by Operskalski et al41 reported anal intercourse.

In contrast to the low estimates from high-income settings, there were many high incidence rate estimates from low-income settings (Fig. 3). For combined man-to-woman and woman-to-man transmission, only 11 of 24 (46%) estimates were <5.0/100 person-years,28,39,46,54,58,60–65 four of which were from no-ART studies.28,39,46,54 Two large studies of no ART use reported particularly low infectiousness estimates,28,39 and three small studies produced particularly high estimates.33,34,38 The high estimate from Hugonnet et al34 was from partners of people who seroconverted during follow-up and therefore included the period of high infectiousness associated with acute infection. The index cases in the study by Hira et al33 were mostly symptomatic (14% AIDS, 82% AIDS-related complex) and therefore also likely to be more infectious. The paucity of information reported by Mao et al38 regarding risk factors makes interpretation of their high estimate difficult. Low-income studies were also more recent and tended to be far larger than those from high-income settings, particularly the RCTs. There were eight any-ART-use studies,56,58,60–65 which tended to be more recent (published 2006–2011) than the no-ART studies (1990–2010).

Infectiousness by Direction of Transmission

There was no clear pattern regarding HIV-1 infectiousness by direction of transmission, woman-to-man compared with man-to-woman. Of the publications providing man-to-woman and woman-to-man infectiousness estimates from the same study, seven26,29–31,33,34,41 reported no significant difference between them, with the eighth27 being significant before adjustment but nonsignificant after adjusting for age.

Subgroup Analyses

We undertook subgroup analyses to identify sources for the heterogeneity of infectiousness estimates. However, there was a lack of comparable information on HIV-1 risk factors. For example, of the 22 studies reporting no ART use for combined man-to-woman and woman-to-man transmission, only five provided data on the prevalence of male circumcision. Few differences in summary estimates by subgroup reached statistical significance, and those that did may be due to chance. Nevertheless, the data available do allow trends to be identified. Subgroup summary estimates are plotted for no-ART-use studies from high- and low-income settings for combined transmission and for low-income settings for man-to-woman and woman-to-man transmission (eFigure 2A–D, There were insufficient data to provide meaningful plots for any-ART-use estimates and woman-to-man and man-to-woman no-ART-use estimates from high-income countries.

Summary incidence rate estimates were lowest for studies from Europe and the Americas (combined transmission, no ART use 3.6/100 person-years [95% CI = 0.4–32.6]), then Africa (7.7/100 person-years [5.5–10.7]), and highest for Asia (10.9/100 person-years [1.7–69.5]; eFigures 2A and B, There were no substantial differences in summary estimates by level of condom use and STIs. HIV-1 infection stage classified studies as having index cases who were more (high proportion of late-stage or acute-stage HIV-1) or less infectious. For the majority of strata, the “more-infectious” subgroup had the higher summary incidence rates, although this was not the case for combined transmission high-income estimates (but only three studies reported relevant data40,41,52) and for woman-to-man low-income estimates (five studies28,30,39,43,69). In the first case, Operskalski et al41 was an outlier with high infectiousness, as discussed above. In the second case, the only two studies in the “more infectious” subgroup had small sample sizes.39,69

Although male circumcision status was not well reported, we found that studies with higher prevalence levels had lower summary incidence rates. The ecologic analysis using country-level circumcision prevalence data found no difference in summary estimates for woman-to-man transmission from low-income settings (5.8/100 person-years [95% CI = 3.5–9.5]) for high-prevalence settings; 5.8/100 person-years [3.7–9.1] for low-prevalence settings, eFigure 2D, Pooled estimates by study type were higher for trials than for observational studies for combined and man-to-woman transmission and roughly equal for woman-to-man transmission, but interpretation is limited because of the small number of trials included.

The inconsistent reporting of risk factors for HIV-1 transmission across studies means that only univariate analysis was possible for the subgroup analysis. The trends identified by each HIV-1 risk factor are generally as expected and therefore explain some of the heterogeneity in study estimates within each stratum. For most subgroup pooled estimates, substantial heterogeneity remained (eg, combined transmission no-ART-use all settings: 13 of 15 pooled estimates had I2 ≥ 50%, data not shown).


HIV-1 Infectiousness with ART

Our systematic review and meta-analysis of per-partner HIV-1 infectiousness includes all published prospective studies of discordant couples and suggests a 91% reduction (95% CI = 79–96%) in infectiousness with ART use by the index case from ART-stratified observational studies. Our results from non-ART–stratified studies support this observation for populations that more closely resemble real-world settings. The IRR of 0.04 for 100% versus <100% ART use among index cases does not mean a 96% reduction in infectiousness. We are not comparing consistent use with zero use, we cannot directly compare populations from different studies conducted in different settings at different time points, and we are relying on two small studies representing 100% ART use.60,63 However, the estimate does demonstrate a striking reduction in transmission within discordant couples with ART use for observational studies. Together with the stratified study analysis, there is compelling evidence that ART reduces HIV-1 infectiousness for sexual transmission.

The HPTN 052 RCT provides gold-standard evidence of the efficacy of ART for reducing the risk of HIV-1 transmission, finding a 96% reduction in linked HIV-1 transmission events within heterosexual stable couples.3 Our analysis summarizing findings from all ART-stratified studies found a 91% reduction. The other ART-stratified studies included in our review are observational and tended to involve ART-receiving index cases with more advanced HIV-1 disease than untreated index cases, probably with less intensive risk-reduction and adherence counseling; thus, a slightly lower effect of ART is unsurprising. This proof of concept that ART does indeed reduce sexual transmission of HIV-1 is not enough to predict the population-level impact of ART interventions on HIV-1 spread. Such predictions would require more data on the effectiveness of infectiousness reduction with ART when adherence-promotion counseling is set at the more modest intensity characteristic of existing ART programs and would need to look back at the incidence rates from the non-ART–stratified discordant-couple studies reviewed here to estimate transmission rates within stable partnerships in real-world settings—again, without risk-reduction counseling being as intensive as now justifiably required for prospective HIV-1 studies.

Incidence rate estimates from no-ART-use studies seem to have reduced over time (eg, Fig. 3A, combined man-to-woman and woman-to-man transmission, low-income settings). It is difficult to estimate a “typical” transmission risk to use for HIV-1 projections because (1) it is difficult to define a typical discordant couple, (2) levels of counseling differ between studies, and (3) awareness among the general population of HIV/AIDS and risks for transmission varies by time and location. However, the 5–10/100 person-years incidence rates reported between 1992 and 2001 by the majority of low-ART, low-income studies is a reasonable starting point from which to base predictions of the impact of ART, or indeed, pre-exposure prophylaxis or other HIV-1 interventions, within discordant couples from low-income settings.

In 2009, Attia et al70 reviewed five cohorts reporting HIV-1 transmission rates among heterosexual discordant couples with ART-treated index cases, finding that no transmission events occurred in patients who were treated with ART and who had viral loads <400 copies/ml. Although our findings were not able to stratify risk by viral load of the index, maintaining a lower or undetectable viral load on ART probably reduces transmission risk even further.70 Since the report by Attia et al, a reanalysis of the Partners in Prevention data (represented in our analysis by Celum et al28 and Donnell et al24) has supported the findings by Attia et al, with higher viral load increasing risk of HIV-1 transmission.71 No transmission events were observed from index cases with undetectable viral load in their blood, but there were a small number from people with undetectable viral load in genital fluids.72 In addition to the ART-stratified trials, we looked at all prospective discordant-couple studies and provided estimates of per-partner incidence rate from older studies, many of which did not involve such intensive risk-reduction counseling. These older studies may provide more real-world incidence rates, more accurately reflecting conditions for discordant couples in the general population.

Infectiousness by Setting

Rates of HIV-1 transmission to partners were generally higher in low-income than high-income countries. There may be several explanations, including differences in methodology, rates of monogamy, rates of circumcision, differences in the stage of the HIV-1 epidemic or the proportion of symptomatic cases, patterns of sexual behavior, prevalence of other STIs, and variations in viral strains. More recent studies have used more sophisticated phylogenetic analyses to exclude transmissions that originated outside the partnership (see, for example, Celum et al28 and Fideli et al31).

Infectiousness by Direction of Transmission

There was no clear difference in HIV-1 infectiousness by direction of transmission. For high-income settings, this may be due to the lack of studies (only one incidence rate for woman-to-man transmission was identified). In low-income countries, a higher prevalence of HIV-1 cofactors such as STIs may mask a true biologic difference in HIV-1 infectiousness. These findings are consistent with results from a previous meta-analysis of per-sexual-act HIV-1 infectiousness.6

Study Limitations

Transmission probabilities are affected by a combination of risk factors for the person transmitting the virus (infectiousness) and the partner acquiring it (susceptibility). Factors known to affect HIV-1 infectiousness include stage of infection,73 viral subtype,74 STIs,75 type of sexual act,76 and male circumcision.77 Therefore, average transmission probabilities will depend on the distribution of these measured and unmeasured risk factors, and it may not be appropriate to generalize across populations. Discordant-couple studies usually recruit monogamous couples, who may have lower prevalence and incidence of various risk factors than the general population (eg, lower incidence of STIs). Furthermore, self-reported data on sexual behavior are often inaccurate because of recall and social-desirability bias,78 leading to imprecise estimates of frequency of sexual contacts, condom use, and reports of monogamy. There is more potential for bias in our non-ART–stratified study comparison than our ART-stratified comparison because we are comparing different studies, where distributions of these risk factors will differ between populations by place and time.

Our HIV-1 cumulative incidence and incidence rates show considerable heterogeneity, likely due to different distributions in HIV-1 risk factors across study populations. Although a meta-analysis is presented, care must be taken with the interpretation of summary estimates from such heterogeneous data and from studies with such different patterns of risk factors. This meta-analysis is informative at a qualitative level, identifying trends in estimates between risk groups. However, with incomplete or noncomparable reporting of HIV-1 risk factors and cofactors, we cannot adequately control for differences among studies. For non-ART–stratified studies, our distinction of no and any ART use is based on inferences from the study dates. There was incomplete reporting on dates and on many other HIV-1 risk factors and cofactors, which we therefore could not adequately control for in our analysis. Our analysis by ART use is also limited by the ecologic nature of our measure: for studies with ART use between 0 and 100%, we do not know which of the transmitting and nontransmitting couples included index cases who received ART.

The quality and emphasis placed on risk-reduction counseling of participating discordant couples is likely to have varied substantially among studies and over time. For example, couples identified as discordant only retrospectively by Quinn et al43 may have different risk behavior patterns from those who were aware of their serostatus. Such study designs are unlikely to be replicated. In contrast, the RCTs had intensive levels of risk-reduction counseling.

It has been stated that “longitudinal studies of discordant couples are the preferred epidemiologic design for the investigation of heterosexual HIV transmission.”34 However, this design misses primary infection and involves left-censoring selection bias, selecting for couples with biologic or behavioral characteristics that impart a lower risk of HIV-1 transmission. Therefore, the rates derived are likely to underestimate the true average HIV-1 transmission risk. To avoid this bias would require prospective recruitment of concordant HIV-1–negative couples to measure transmission rates from the time of HIV-1 acquisition of the index case. Such a large trial would be so costly as to be unfeasible. The exceptions are the Rakai Project Study,43 which retrospectively identified those persons who were in stable partnerships and, as a result, has been able to provide HIV-1 transmission risk for the acute stage of infection73—although it seems unlikely that such a study design will be replicated.79 By overcoming left censoring, Quinn et al43 did indeed estimate a higher infectiousness than other studies: the third highest transmission rate of 16 man-to-woman low-ART resource-poor setting estimates (Fig. 3B) and the highest of 13 for woman-to-man estimates (Fig. 3C). The second exception is Operskalski et al.41 The extremely high incidence rate (17.4/100 person-years) would be even higher if we took into account the two couples who were concordant positive at recruitment and for whom time since sexual contact began was known, after a blood transfusion of the index (incidence rate 22.0/100 person-years).

Prospective discordant-couple studies are also limited either by the small number of participants or the small number of seroconversions resulting from dissolution of couples, intensive condom promotion, short follow-up periods, or falling rates of transmission over time because of behavioral change or exhaustion of susceptible persons. They cannot allow for partnerships that have broken up (which often happens after an HIV-1 diagnosis). Nonetheless, monogamous HIV-1–discordant couples provide the best means of estimating HIV-1 infectiousness for sexual transmission because they (in principle) eliminate contamination from outside sources. For the purposes of comparison (such as by ART use, setting, and direction of transmission), these limitations in study design are less important because they are applicable to virtually all studies. However, in the absence of better data, these transmission rates are also used to inform trials and modeling studies. Therefore, it is important to appreciate the biases, especially the likely underestimation in risk as a result of left censoring.


ART reduces viral loads and thus has been believed to reduce infectiousness. In the last 5 years, studies have demonstrated that this reduction in infectiousness can be marked.3,80 Therefore, widespread access to ART has the potential to have an impact on HIV-1 transmission, both at the individual level (reducing transmission risk and thus allowing a safer route to conception for many couples) and, potentially, at the population level, as an HIV-1 prevention tool.2,81–83 The use of ART has been broadened from a therapeutic role to prevention, not only by reducing infectiousness as shown here, but also by its use as pre-exposure prophylaxis for reduction of susceptibility.84 Our study reinforces the results of previous studies and meta-analyses70,85 that have shown how ART can reduce a person’s infectiousness through sexual transmission in trial or intervention scenarios. The conditions under which this potential has been demonstrated do not seamlessly translate to a prediction of the effects of ART on real-life situations, whereas our meta-analysis, with its inevitable limitations, shows that real-world studies of such interventions can provide useful estimates of their likely public health benefit.


We gratefully acknowledge James Goedert, Koya Ariyoshi, Steven Reynolds, Deborah Jones, Szonja Vamos, Breno Santos, and Maria Wawer for providing additional data. We also thank Elizabeth and Richard Baggaley, Dongmei Liu, Lorenzo Sabatelli, Kendra Wu, and Isaac Fung for translations.


1. UNAIDS/WHO. . Global HIV/AIDS response. Epidemic update and health sector progress towards Universal Access (Progress report 2011). Available at: Accessed 21 April 2012.
2. Granich RM, Gilks CF, Dye C, De Cock KM, Williams BG. Universal voluntary HIV testing with immediate antiretroviral therapy as a strategy for elimination of HIV transmission: a mathematical model. Lancet. 2009;373:48–57
3. Cohen MS, Chen YQ, McCauley M, et al.HPTN 052 Study Team. Prevention of HIV-1 infection with early antiretroviral therapy. N Engl J Med. 2011;365:493–505
4. Shiboski SC, Padian NS. Epidemiologic evidence for time variation in HIV infectivity. J Acquir Immune Defic Syndr Hum Retrovirol. 1998;19:527–535
5. Shiboski SC, Jewell NP. Statistical analysis of the time dependence of HIV infectivity based on partner study data. J Am Stat Assoc. 1992;87:360–372
6. Boily MC, Baggaley RF, Wang L, et al. Heterosexual risk of HIV-1 infection per sexual act: systematic review and meta-analysis of observational studies. Lancet Infect Dis. 2009;9:118–129
7. Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA. 2000;283:2008–2012
8. Baggaley RF, White RG, Boily MC. HIV transmission risk through anal intercourse: systematic review, meta-analysis and implications for HIV prevention. Int J Epidemiol. 2010;39:1048–1063
9. Baggaley RF, Boily MC, White RG, Alary M. Risk of HIV-1 transmission for parenteral exposure and blood transfusion: a systematic review and meta-analysis. AIDS. 2006;20:805–812
10. Baggaley RF, White RG, Boily MC. Systematic review of orogenital HIV-1 transmission probabilities. Int J Epidemiol. 2008;37:1255–1265
11. Palella FJ, Delaney KM, Moorman AC, et al. Declining morbidity and mortality among patients with advanced human immunodeficiency virus infection. HIV Outpatient Study Investigators. N Engl J Med. 1998;338:853–860
12. UNAIDS/WHO. . . Progress on global access to HIV antiretroviral therapy. Available at: Accessed 24 May 2011. 2006.
13. Thompson SG, Higgins JP. How should meta-regression analyses be undertaken and interpreted?. Stat Med. 2002;21:1559–1573
14. Cochran WG. The combination of estimates from different experiments. Biometrics. 1954;10:101–29
15. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ. 2003;327:557–560
16. Waskett JHGlobal circumcision rates (review).. Circumcision Independent Reference and Commentary Service. Available at: Accessed 9 March 2012.
17. Musicco M, Lazzarin A, Nicolosi A, et al.Antiretroviral treatment of men infected with human immunodeficiency virus type 1 reduces the incidence of heterosexual transmission. Italian Study Group on HIV Heterosexual Transmission. Arch Intern Med. 1994;154:1971–1976
18. Melo MG, Santos BR, De Cassia Lira R, et al. Sexual transmission of HIV-1 among serodiscordant couples in Porto Alegre, southern Brazil. Sex Transm Dis. 2008;35:912–915
19. Watera C, Mutonyi G, Levin J, Birabwa E, Bawulira F, Vudriko T, et al.Changes in reported sexual behaviour and incidence of HIV infection in a cohort of HIV discordant couples in Kampala, Uganda. 5th Conference on HIV Pathogenesis and Treatment and Prevention, Cape Town. Abstract MOPEC005. 2009
20. Sullivan P, Kayitenkore K, Chomba E, et al.Is the reduction of HIV transmission risk while prescribed antiretroviral therapy (ARVT) different for men and women? Results from discordant couples in Rwanda and Zambia. 5th IAS Conference on HIV Pathogenesis and Treatment. Abstract WEAC101. 2009
21. Sullivan P, Kayitenkore K, Chomba E, et al. Reduction of HIV transmission risk and high risk sex while prescribed ART: results from discordant couples in Rwanda and Zambia. 16th Conference on Retroviruses and Opportunistic Infections. Abstract 52bLB. February 2009Montreal8–11Montreal, Canada. 2009
22. Del Romero J, Castilla J, Hernando V, Rodríguez C, García S. Combined antiretroviral treatment and heterosexual transmission of HIV-1: cross sectional and prospective cohort study. BMJ. 2010;340:c2205
23. Wang L, Ge Z, Luo J, Shan D, Gao X, Ding G-W, et al. HIV transmission risk among serodiscordant couples: a retrospective study of former plasma donors in Henan, China. JAIDS. 2010;55:232–238
24. Donnell D, Baeten JM, Kiarie J, et al.Partners in Prevention HSV/HIV Transmission Study Team. Heterosexual HIV-1 transmission after initiation of antiretroviral therapy: a prospective cohort analysis. Lancet. 2010;375:2092–2098
25. Reynolds SJ, Makumbi F, Nakigozi G, et al. HIV-1 transmission among HIV-1 discordant couples before and after the introduction of antiretroviral therapy. AIDS. 2011;25:473–477
26. Allen S, Tice J, Van de Perre P, et al. Effect of serotesting with counselling on condom use and seroconversion among HIV discordant couples in Africa. BMJ. 1992;304:1605–1609
27. Carpenter LM, Kamali A, Ruberantwari A, Malamba SS, Whitworth JA. Rates of HIV-1 transmission within marriage in rural Uganda in relation to the HIV sero-status of the partners. AIDS. 1999;13:1083–1089
28. Celum C, Wald A, Lingappa JR, et al.Partners in Prevention HSV/HIV Transmission Study Team. Acyclovir and transmission of HIV-1 from persons infected with HIV-1 and HSV-2. N Engl J Med. 2010;362:427–439
29. de Vincenzi IA longitudinal study of human immunodeficiency virus transmission by heterosexual partners.. European Study Group on Heterosexual Transmission of HIV. N Engl J Med. 1994;331:341–346
30. Deschamps MM, Pape JW, Hafner A, Johnson WD Jr. Heterosexual transmission of HIV in Haiti. Ann Intern Med. 1996;125:324–330
31. Fideli US, Allen SA, Musonda R, et al. Virologic and immunologic determinants of heterosexual transmission of human immunodeficiency virus type 1 in Africa. AIDS Res Hum Retroviruses.. 2001;17:901–910
32. Fischl MA, Dickinson GM, Scott GB, Klimas N, Fletcher MA, Parks W. Evaluation of heterosexual partners, children, and household contacts of adults with AIDS. JAMA. 1987;257:640–644
33. Hira SK, Kamanga J, Macuacua R, Mwansa N, Cruess DF, Perine PL. Genital ulcers and male circumcision as risk factors for acquiring HIV-1 in Zambia. J Infect Dis. 1990;161:584–585
34. Hugonnet S, Mosha F, Todd J, et al. Incidence of HIV infection in stable sexual partnerships: a retrospective cohort study of 1802 couples in Mwanza Region, Tanzania. J Acquir Immune Defic Syndr. 2002;30:73–80
35. Laurian Y, Peynet J, Verroust F. HIV infection in sexual partners of HIV-seropositive patients with hemophilia. N Engl J Med. 1989;320:183
36. Lawrence DN, Jason JM, Holman RC, Heine P, Evatt BL. Sex practice correlates of human immunodeficiency virus transmission and acquired immunodeficiency syndrome incidence in heterosexual partners and offspring of U.S. hemophilic men. Am J Hematol. 1989;30:68–76
37. Lusher JM, Operskalski EA, Aledort LM, et al. Risk of human immunodeficiency virus type 1 infection among sexual and nonsexual household contacts of persons with congenital clotting disorders. Pediatrics. 1991;88:242–249
38. Mao YR, Zheng XW, Re ZY, et al. [An epidemiological study on sexual transmission of human immunodeficiency virus among pre-marital group in Yining city, Xinjiang]. Zhonghua Liu Xing Bing Xue Za Zhi. 2004;25:322–324
39. Mehendale SM, Ghate MV, Kishore Kumar B, et al. Low HIV-1 incidence among married serodiscordant couples in Pune, India. J Acquir Immune Defic Syndr. 2006;41:371–373
40. O’Brien TR, Busch MP, Donegan E, et al. Heterosexual transmission of human immunodeficiency virus type 1 from transfusion recipients to their sex partners. J Acquir Immune Defic Syndr. 1994;7:705–710
41. Operskalski EA, Stram DO, Busch MP, et al.Role of viral load in heterosexual transmission of human immunodeficiency virus type 1 by blood transfusion recipients. Transfusion Safety Study Group. Am J Epidemiol. 1997;146:655–661
42. Padian NS, Shiboski SC, Glass SO, Vittinghoff E. Heterosexual transmission of human immunodeficiency virus (HIV) in northern California: results from a ten-year study. Am J Epidemiol. 1997;146:350–357
43. Quinn TC, Wawer MJ, Sewankambo N, et al. Viral load and heterosexual transmission of human immunodeficiency virus type 1. Rakai Project Study Group. N Engl J Med. 2000;342:921–929
44. Ragni MV, Faruki H, Kingsley LA. Heterosexual HIV-1 transmission and viral load in hemophilic patients. J Acquir Immune Defic Syndr Hum Retrovirol. 1998;17:42–45
45. Rockstroh JK, Ewig S, Bauer T, et al. Male-to-female transmission of HIV in a cohort of hemophiliacs–frequency, risk factors and effect of sexual counseling. Infection. 1995;23:29–32
46. Roth DL, Stewart KE, Clay OJ, et al. Sexual practices of HIV discordant and concordant couples in Rwanda: effects of a testing and counselling programme for men. Int J STD AIDS. 2001;12:181–188
47. Ryder RW, Kamenga C, Jingu M, Mbuyi N, Mbu L, Behets F. Pregnancy and HIV-1 incidence in 178 married couples with discordant HIV-1 serostatus: additional experience at an HIV-1 counselling centre in the Democratic Republic of the Congo. Trop Med Int Health. 2000;5:482–487
48. Senkoro KP, Boerma JT, Klokke AH, et al. HIV incidence and HIV-associated mortality in a cohort of factory workers and their spouses in Tanzania, 1991 through 1996. J Acquir Immune Defic Syndr. 2000;23:194–202
49. Serwadda D, Gray RH, Wawer MJ, et al. The social dynamics of HIV transmission as reflected through discordant couples in rural Uganda. AIDS. 1995;9:745–750
50. Siddiqui NS, Brown LS Jr, Phillips RY, Vargas O, Makuch RW. No seroconversions among steady sex partners of methadone-maintained HIV-1-seropositive injecting drug users in New York City. AIDS. 1992;6:1529–1533
51. Tovanabutra S, Robison V, Wongtrakul J, et al. Male viral load and heterosexual transmission of HIV-1 subtype E in northern Thailand. J Acquir Immune Defic Syndr. 2002;29:275–283
52. van der Ende ME, Rothbarth P, Stibbe J. Heterosexual transmission of HIV by haemophiliacs. BMJ. 1988;297:1102–1103
53. Wawer MJ, Makumbi F, Kigozi G, et al. Circumcision in HIV-infected men and its effect on HIV transmission to female partners in Rakai, Uganda: a randomised controlled trial. Lancet. 2009;374:229–237
54. Zhang H, Wu H, Xu L. Behavior interventions and health education for HIV-infected people and their spouses. Chin J AIDS STD (Chin).. 2001;7:286–288
55. El-Bassel N, Jemmott JB, Landis JR, et al.NIMH Multisite HIV/STD Prevention Trial for African American Couples Group. National Institute of Mental Health Multisite Eban HIV/STD Prevention Intervention for African American HIV Serodiscordant Couples: a cluster randomized trial. Arch Intern Med. 2010;170:1594–1601
56. Kumarasamy N, Venkatesh KK, Srikrishnan AK, et al. Risk factors for HIV transmission among heterosexual discordant couples in South India. HIV Med. 2010;11:178–186
57. Robertson JR, Wyld R, Elton R, Brettle RP. Heterosexual transmission of HIV in men and women in a Scottish cohort. AIDS. 1998;12:823–824
58. Rojanawiwat A, Ariyoshi K, Pathipvanich P, Tsuchiya N, Auwanit W, Sawanpanylaert P. Substantially exposed but HIV-negative individuals are accumulated in HIV-serology-discordant couples diagnosed in a referral hospital in Thailand. Jpn J Infect Dis. 2009;62:32–36
59. Saracco A, Veglia F, Lazzarin ARisk of HIV-1 transmission in heterosexual stable and random couples.. The Italian Partner Study. J Biol Regul Homeost Agents. 1997;11:3–6
60. Glynn J, Price A, Floyd S, et al.Antiretroviral therapy reduces HIV transmission in discordant couples in Northern Malawi. 19th International Society of Sexually Transmitted Research Conference, Quebec, Canada. Abstract P1-S6.53. STI, 2011. July 2011;87(suppl 1):A218
61. Muwonge T, Namimbi F, Bakabulindi P, Wakabi L, Weguli S, Kambugu A. Low HIV sero-conversion rates among HIV discordant couples attending a large urban HIV clinic in Uganda. 6th International AIDS Society Conference on HIV Pathogenesis, Treatment and Prevention, Rome, Italy, July 2011. Abstract CDC335, 2011
62. Guthrie B, Choi R, Bosire R, et al.Low rates of HIV-1 transmission in a cohort of discordant couples in Nairobi, Kenya. 6th International AIDS Society Conference on HIV Pathogenesis, Treatment and Prevention, Rome, Italy, July 2011. Abstract MOPE330. 2011
63. Apondi R, Bunnell R, Ekwaru JP, et al. Sexual behavior and HIV transmission risk of Ugandan adults taking antiretroviral therapy: 3 year follow-up. AIDS. 2011;25:1317–1327
64. Li JY, Li L, Li H, et al. [Cohort study on human immunodeficiency virus discordant couples in the countryside of central China]. Zhonghua Liu Xing Bing Xue Za Zhi. 2006;27:192–195
65. Duan S, Yang YC, Xiang LF, et al. [Incidence and risk factors of HIV infection among sero-negative spouses of HIV patients in Dehong prefecture of Yunnan province]. Zhonghua Liu Xing Bing Xue Za Zhi. 2010;31:997–1000
66. Pan XH, Yang JZ, Chen L, Xu Y. [Analysis of epidemiological characteristics of HIV infections among immigrant marriage women in rural areas in Zhejiang province]. Zhonghua Yu Fang Yi Xue Za Zhi. 2010;44:1023–1026
67. Baeten J, Celum CTeam obotPPS. . Antiretroviral pre-exposure prophylaxis for HIV-1 prevention among heterosexual African men and women: the Partners PrEP study. 6th International AIDS Society Conference on HIV Pathogenesis, Treatment and Prevention, Rome, Italy,. Abstract MOAX0106.. July 2011 Available at: Accessed 1 August 2011.
68. Reynolds S, Makumbi F, Kagaayi J, et al.ART reduced the rate of sexual transmission of HIV among HIV-discordant couples in rural Rakai, Uganda. 16th Conference on Retroviruses and Opportunistic Infections Abstract 52a. February 2009 Montreal, 8–11 Montreal, Canada. 2009
69. Hira SK, Feldblum PJ, Kamanga J, Mukelabai G, Weir SS, Thomas JC. Condom and nonoxynol-9 use and the incidence of HIV infection in serodiscordant couples in Zambia. Int J STD AIDS. 1997;8:243–250
70. Attia S, Egger M, Müller M, Zwahlen M, Low N. Sexual transmission of HIV according to viral load and antiretroviral therapy: systematic review and meta-analysis. AIDS. 2009;23:1397–1404
71. Hughes J, Baeten J, Lingappa J, et al.Determinants of per-act infectivity of HIV-1 in the Partners in Prevention Study. Eighteenth Conference on Retroviruses and Opportunistic Infections (CROI), Boston. Abstract 135.. Available at: Accessed 26 May 2011.
72. Baeten J, Kahle E, Lingappa J, et al.Genital HIV-1 RNA levels predict risk of heterosexual HIV-1 transmission. Eighteenth Conference on Retroviruses and Opportunistic Infections (CROI), Boston. Abstract 154. Available at: Accessed 26 May 2011.
73. Wawer MJ, Gray RH, Sewankambo NK, 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
74. Cohen J. Differences in HIV strains may underlie disease patterns. Science. 1995;270:30–31
75. Cameron DW, Simonsen JN, D’Costa LJ, et al. Female to male transmission of human immunodeficiency virus type 1: risk factors for seroconversion in men. Lancet. 1989;2:403–407
76. Nicolosi A, Corrêa Leite ML, Musicco M, et al.The efficiency of male-to-female and female-to-male sexual transmission of the human immunodeficiency virus: a study of 730 stable couples. Italian Study Group on HIV Heterosexual Transmission. Epidemiology. 1994;5:570–575
77. Cameron DW, Simonsen JN, D’Costa LJ, et al. Genital ulcer disease and lack of male circumcision facilitate female to male sexual HIV transmission The Lancet. 1989;2:403–407
78. Catania JA, Gibson DR, Chitwood DD, Coates TJ. Methodological problems in AIDS behavioral research: influences on measurement error and participation bias in studies of sexual behavior. Psychol Bull. 1990;108:339–362
79. Angell M. Investigators’ responsibilities for human subjects in developing countries. N Engl J Med. 2000;342:967–969
80. Cohen MS, Gay C, Kashuba AD, Blower S, Paxton L. Narrative review: antiretroviral therapy to prevent the sexual transmission of HIV-1. Ann Intern Med. 2007;146:591–601
81. Cohen MS, Gay CL. Treatment to prevent transmission of HIV-1. Clin Infect Dis. 2010;50(suppl 3):S85–S95
82. Bezemer D, de Wolf F, Boerlijst MC, et al. A resurgent HIV-1 epidemic among men who have sex with men in the era of potent antiretroviral therapy. AIDS. 2008;22:1071–1077
83. Garnett GP, Anderson RM. Antiviral therapy and the transmission dynamics of HIV-1. J Antimicrob Chemother. 1996;37(suppl B):135–150
84. Abdool Karim Q, Abdool Karim SS, Frohlich JA, et al.CAPRISA 004 Trial Group. Effectiveness and safety of tenofovir gel, an antiretroviral microbicide, for the prevention of HIV infection in women. Science. 2010;329:1168–1174
85. Anglemyer A, Rutherford GW, Baggaley RC, Egger M, Siegfried N. Antiretroviral therapy for prevention of HIV transmission in HIV-discordant couples. Cochrane Database of Systematic Reviews.. 2011 CD009153. doi: 10.1002/14651858.CD009153.pub2.

Supplemental Digital Content

© 2013 Lippincott Williams & Wilkins, Inc.