The efficacy of antiretroviral drugs in the prevention of mother-to-child transmission of HIV is well documented  and there may also be a role of antiretroviral therapy (ART) in the prevention of sexual transmission of HIV . Any reduction in the capacity of HIV to replicate is likely to reduce the risk of HIV transmission, unless the effect is offset by behavioural risk compensation [2,3]. HIV-infected men treated with zidovudine monotherapy in Italy were half as likely to transmit infection to their female partners than untreated men, after controlling for their more advanced disease . Highly active antiretroviral combination therapy (ART) that can suppress HIV viraemia sustainably should have greater impact [2,3].
In January 2008 the Swiss Federal AIDS Commission stated that HIV-infected people on effective antiretroviral therapy and without other sexually transmitted infections were sexually noninfectious . National public health bodies [6,7] have, however, reasserted existing guidance about the need for consistent condom use whereas some groups have supported the statement . Our objective was to review the literature and obtain summary estimates of the risk of HIV transmission according to viral load, treatment with ART and presence of other sexually transmitted infections.
We included original articles and conference abstracts reporting on longitudinal studies of couples with one HIV-infected partner and documenting the number of HIV infections in previously seronegative sexual partners, and information about viral load in the HIV-seropositive partner, use of ART, or both. We excluded studies of preexposure prophylaxis and case-reports.
We searched the Medline and EMBASE databases from January 1996 to May 2008 and updated searches in February 2009. We used subject-heading terms for ‘HIV infections’ and ‘disease transmission’ and combined these with terms for either ‘viral load’ or ‘antiretroviral therapy, highly active’ (full search strategies available on request). We examined the reference lists of full text reports. We also searched the abstracts of the International AIDS Society conferences from 2001–2008 and the Conference on Retroviruses and Opportunistic Infections from 1997–2009 using key words ‘HIV’ and ‘discordant’, or ‘discordant’ and ‘couple’. There was no restriction on the language of published articles.
Two reviewers independently assessed all titles and abstracts of published articles (S.A., M.M.) and conference abstracts (S.A., N.L.). If there was insufficient information in the title or abstract we retrieved the full text. We determined eligibility by consensus, with a third reviewer (M.E.) making a final decision in the case of disagreement. Two reviewers (S.A., M.M.) extracted the same information about each study. A third reviewer (N.L. or M.E.) resolved discrepancies. We extracted information about: study characteristics and population; the number of HIV transmission events and duration of follow-up; plasma viral load, use of ART and sexually transmitted infections in the seropositive partner; types and frequency of sexual intercourse; and condom use.
We contacted authors of potentially eligible studies identified in the first search to confirm eligibility and to request additional information. We asked about numbers of HIV transmissions and follow-up time according to the viral load of the HIV seropositive partner (<400, 400–499, 500–9999, 10 000–49 999, 50 000 and more copies/ml) . We defined an undetectable viral load as fewer than 400 copies/ml of blood viral load, according to the detection limit of tests used in most eligible studies. We asked for the lowest measured viral load at which transmission to a seronegative partner had occurred. We relied on published data if the authors were not contacted, could not be reached or additional data were not provided.
We also asked for data in the predefined viral load categories according to whether the seropositive partner had any other sexually transmitted infections. We defined any sexually transmitted infection as: positive serological tests or microscopy for syphilis, positive test results for Neisseria gonorrhoeae, or Chlamydia trachomatis from swabs or urine specimens, or positive genital ulcer swab results or serological tests for herpes simplex virus. We defined ulcerative sexually transmitted infections as syphilis and genital herpes. If this information was not available, or if diagnosis was based on self-report we categorized the status as unclear.
We aimed to estimate the risk of HIV transmission per unprotected act of sexual intercourse. In the absence of data about frequency of unprotected sexual intercourse we used the HIV transmission risk per 100 person years of follow-up. If the exact follow-up time was not available we estimated this from the reported mean or median. We used a random effects Poisson regression model to obtain a summary estimate of the transmission rate with 95% confidence intervals (95% CI). 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 and total exposure time per study as an offset variable, and included γ-distributed random effects on the study or stratum level. If there were no events observed, we assumed that the number of events was Poisson distributed and obtained an upper 97.5% confidence limit based on exact Poisson probabilities. All analyses were conducted using STATA version 10 (Stata Corporation, College Station, Texas, USA).
Our searches yielded 305 publications, including 56 conference abstracts. Figure 1 depicts the process of identifying studies. We contacted the authors of 21 of 26 potentially eligible studies: nine replied and four provided additional information [10–13]. We identified 11 eligible cohorts reporting on 5021 couples and 461 HIV transmission events in 16 publications or abstracts from eight countries [9–24] (Table 1). The largest number of serodiscordant couples was reported in five studies from sub-Saharan Africa [9,17,18,23,24]. All included studies were conducted amongst heterosexual couples. There were insufficient data to allow estimation of summary rates of transmission through sexual intercourse without condoms, or to separate female–male and male–female transmission.
We analysed the risk of HIV transmission per 100 person-years of follow up because we only identified one study reporting on HIV transmission per coital act and stratified by viral load [15,16]. Data about coital frequency and transmission events were collected from the same population of untreated HIV-infected individuals in Rakai, Uganda. The estimated probability of HIV transmission per coital act, after controlling for age, ranged from 0.0001 when viral load was below 1700 copies/ml (sexual intercourse 10.4 times per month) to 0.0023 when viral load was greater than 38 500 copies/ml (sexual intercourse 7.9 times per month) .
HIV transmission and highly active antiretroviral therapy
Five studies included couples in which the HIV-seropositive partners used antiretroviral therapy, with 1098 person years of follow-up [10,11,18,23,24] (Table 1). One study reported specific antiretroviral regimens . We did not identify any studies that reported on both viral load and all microbiologically diagnosed sexually transmitted infections.
The overall HIV transmission risk from antiretroviral-treated patients to heterosexual partners, irrespective of viral load and other sexually transmitted infections, was 0.46 (95% CI 0.19–1.09) per 100 person-years, based on five episodes of HIV seroconversion (Fig. 2) [10,11,18,23,24]. Information on the lowest measured viral load at which transmission had occurred while taking antiretroviral therapy was not available for either of the studies in which new HIV infections occurred [18,23]. In the two studies with information stratified according to viral load there were no reported episodes of HIV transmission from HIV seropositive people with undetectable viral load in 291 person years of follow-up (upper 97.5% confidence limit 1.27 per 100 person years) [10,11]. Three studies did not report on associations between HIV transmission and other sexually transmitted infections [18,23,24]. Melo et al. and Castilla et al. reported no association with the infections assessed in their studies (Table 1).
HIV transmission from people not on antiretroviral therapy
Ten studies included HIV seropositive people not receiving antiretroviral therapy with 9998 person years of follow up [9–14,17,21,23,24]. The overall HIV transmission rate, irrespective of viral load category and sexually transmitted infections, was 5.64 (95% CI 3.28–9.70) per 100 person years (Fig. 2). Amongst people with viral load below 400 copies/ml, irrespective of sexually transmitted infections, the transmission rate was 0.16 (95% CI 0.02–1.13) per 100 person years, based on one episode of HIV transmission in six studies [9,11–14,17]. The transmission rate increased with increasing viral load to 9.03 (95% CI 3.87–21.09) per 100 person years amongst individuals with viral load at least 50 000 copies/ml (Fig. 2).
The lowest measured viral load values around the time of HIV-transmission events were available for seven studies (Table 1). Three values, all in untreated individuals, were below 1500 copies/ml: at 362 copies/ml (Castilla J, 17 July 2008, personal communication), 600 copies/ml (Ragni M, 21 July 2008, personal communication) and 1497 copies/ml .
This systematic review did not identify any study from which the risk of HIV transmission per act of unprotected sexual intercourse amongst persons with suppressed viraemia following ART could be quantified directly. The available studies found no episodes of HIV transmission in discordant heterosexual couples if the HIV-infected partner was treated with ART and had a viral load below 400 copies/ml, but the data were also compatible with one transmission per 79 person-years. There were insufficient data to stratify rates according to the presence or absence of sexually transmitted infections, use of condoms, direction of transmission, or practise of vaginal or anal intercourse. The comparison of overall rates in patients on ART and not on ART nevertheless indicate that heterosexual transmission was reduced by 92%, from 5.64 to 0.46 per 100 person-years. Of note, our review did not identify any study with data on ART and transmission risk in homosexual men.
The main strengths of this study were that we searched systematically for published and unpublished literature and attempted to quantify statistical uncertainty around the transmission rate. Additional information from several authors allowed us to combine data in consistent viral load categories to increase the precision of estimated transmission risks [10–13] and to report the minimum viral load at which HIV transmission occurred. The main limitations of the study relate to the lack of data that could be combined statistically. Four included studies were only available as conference abstracts with limited details [18,21,23,24]. Precision was also limited by small or zero numbers of events in each viral load category and short follow-up times. There are recognized difficulties in obtaining confidence intervals when no events have been observed . The interpretation of the lower limit of zero and upper 97.5% limit obtained using exact Poisson distribution probabilities differs from the standard 95% confidence interval. However, they demonstrate the uncertainty about the true HIV transmission rate by describing a range of values for the true quantity of interest that are compatible with the observed data. Describing the likelihood function about the true value of the parameter is an alternative but the results obtained would not alter our conclusion.
We found no direct evidence that, as stated by the Swiss Federal AIDS commission , the HIV transmission risk through unprotected sexual intercourse from an infected individual taking ART consistently under medical supervision, with blood viral load below 40 copies/ml and without any other sexually transmitted infection was ‘much lower than one per 100 000 acts of sexual intercourse.’ We found that there is considerable uncertainty about this risk: first, although there were no observed episodes of HIV transmission from people with undetectable viral load on highly active antiretroviral therapy, data are compatible with one new HIV infection for every 79 person-years of follow-up (one per 7900 sex acts if the yearly average is 100 contacts  and transmission probability is constant). Second, episodes of HIV transmission were found to have occurred at viral load levels lower than reported in earlier studies . There might therefore be no transmission threshold or a lower threshold than previously believed [5,26].
Mathematical models have been developed to predict the effects of antiretroviral therapy on HIV transmission but variability in assumptions based on epidemiological and biological data makes them difficult to interpret . The reduction in HIV transmissibility due to antiretroviral therapy includes estimates from two to 100 times , two to 10 times , and 100 times . Our meta-analysis should be useful in this context and inform future modelling studies.
There is also uncertainty about the role of sexually transmitted infections. Focusing on ulcerative conditions and symptoms as a proxy for lower genital tract inflammation  is problematic. First, symptoms in women correlate poorly with clinical signs of inflammation or diagnosed infections . Second, HIV transmission appears to be enhanced by bacterial vaginosis , a vaginal infection characterized by an absence of inflammation . Third, sexual transmission of herpes simplex virus, the most common cause of genital ulcer disease in many countries, can occur during asymptomatic virus shedding . Suppressing clinical recurrences with acyclovir does not reduce the risk of HIV transmission [34,35]. Furthermore, adherence to recommendations for regular testing for sexually transmitted infections in HIV-infected people would have to improve from current levels .
The risk of HIV transmission from people on highly active antiretroviral therapy is likely to be very low but it is nevertheless important that statements on transmission risk are based on thorough evaluation of the available data [37,38]. The need for systematic searches and clear documentation about the design, quality and consistency of evidence, and the availability or absence of direct evidence to address important clinical and public health questions is well recognized . The users of recommendations can then distinguish between statements based on appraisal of evidence by experts and those based on systematic methods. The results of our systematic review show where there is a lack of direct evidence and where further research is required.
Greater precision about the HIV transmission rate per sexual act on highly active antiretroviral therapy can be obtained from empirical studies. An upper 95% confidence interval of one in 100 000 per unprotected sex acts would be obtained if the observed HIV transmission rate were one in 550 000 sex acts. This is equivalent to 5500 person years of observation with an average of 100 unprotected sex acts per year, or 1100 couples followed for 5 years having unprotected sex and free of sexually transmitted infections, assuming a constant transmission probability and each act as an independent event. An ongoing randomized trial to follow 1750 HIV serodiscordant heterosexual couples for a median of 5.75 years with the infected partner receiving highly active antiretroviral therapy will help to provide this information . Studies to determine HIV transmissibility through insertive and receptive anal intercourse when viraemia is fully suppressed are needed to provide direct evidence for men who have sex with men. The implications of differences between antiretroviral agents in drug levels in plasma and genital tract, and of intermittent viral ‘blips’ also need to be clarified .
In conclusion, our study supports the World Health Organization's view  that at present there is insufficient evidence to formulate guidance on the role of ART in HIV prevention, both at the level of the individual and the population. Further studies quantifying transmission risk in different patient groups and under different conditions are required to inform such recommendations.
We are grateful to Jesús Castilla, Jorge del Romero, Kenrad Nelson, Margaret Ragni, and Breno Santos who provided additional data from their studies that contributed to the results presented in this paper. S.A. was supported by a ThinkSwiss! Fellowship from the Swiss Government.
Author contributions: S.A. did the literature searches, study selection, data extraction, and wrote the first draft; M.E. obtained funding, designed and supervised the study, and revised the paper; M.M. contributed to study selection, data extraction, and revision of the paper, M.Z. provided statistical advice and revised the paper; N.L. designed and supervised the study, conducted the analysis, and revised the paper.
Conflicts of interest: N.L. became a member of the Swiss Federal AIDS Commission on 1 January 2008. M.Z. was employed at the Swiss Federal Office of Public Health from 1998–2000 and head of the office of the Swiss Federal AIDS Commission at the Swiss Federal Office of Public Health from 1988–1989. The views expressed here are those of the authors.
1. Dao H, Mofenson LM, Ekpini R, Gilks CF, Barnhart M, Bolu O, et al
. International recommendations on antiretroviral drugs for treatment of HIV-infected women and prevention of mother-to-child HIV transmission in resource-limited settings: 2006 update. Am J Obstet Gynecol 2007; 197(3 Suppl):S42–S55.
2. De Cock KM, Gilks CF, Lo YR, Guerma T. Can antiretroviral therapy eliminate HIV transmission? Lancet 2009; 373:7–9.
3. 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.
4. Musicco M, Lazzarin A, Nicolosi A, Gasparini M, Costigliola P, Arici C, 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.
5. Vernazza P, Hirschel B, Bernasconi E, Flepp M. HIV- infizierte Menschen ohne andere STD sind unter wirksamer antiretroviraler Therapie sexuell nicht infektiös [HIV-infected people free of other STDs are sexually not infectious on effective antiretroviral therapy]. Schweizerische Ärztezeitung 2008; 89:165–169.
6. Centers for Disease Control and Prevention. CDC underscores current recommendation for preventing HIV transmission.
1 February 2008. http://www.cdc.gov/hiv/resources/press/020108.htm
. [Accessed 1 February 2009]
7. Public Health Agency of Canada. PHAC continues to emphasize safer sex for preventing HIV transmission
. 17 April 2008. http://www.phac-aspc.gc.ca/aids-sida/new-nouv-eng.php
. [Accessed February 2009]
8. Mexico Manifesto: a call to action by people with HIV and AIDS. 3 August 2008. http://www.ondamaris.de/wp-content/uploads/2008/07/lhive-mexico-manifesto1.pdf
. [Accessed February 2009]
9. Quinn TC, Wawer MJ, Sewankambo N, Serwadda D, Li C, Wabwire-Mangen F, 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.
10. Melo MG, Santos BR, Lira Rd, Varella IS, Turella ML, Rocha TM, et al
. Sexual transmission of HIV-1 among serodiscordant couples in Porto Alegre, Southern Brazil. Sex Transm Dis 2008; 35:912–915.
11. Castilla J, del Romero J, Hernando V, Marincovich B, Garcia S, Rodriguez C. Effectiveness of highly active antiretroviral therapy in reducing heterosexual transmission of HIV. J Acquir Immune Defic Syndr 2005; 40:96–101.
12. 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.
13. Tovanabutra S, Robison V, Wongtrakul J, Sennum S, Suriyanon V, Kingkeow D, 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.
14. Operskalski EA, Stram DO, Busch MP, Huang W, Harris M, Dietrich SL, et al
. Role of viral load in heterosexual transmission of human immunodeficiency virus type 1 by blood transfusion recipients. Am J Epidemiol 1997; 146:655–661.
15. Gray RH, Wawer MJ, Brookmeyer R, Sewankambo NK, Serwadda D, Wabwire-Mangen F, et al
. Probability of HIV-1 transmission per coital act in monogamous, heterosexual, HIV-1-discordant couples in Rakai, Uganda. Lancet 2001; 357:1149–1153.
16. 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.
17. Fideli US, Allen SA, Musonda R, Trask S, Hahn BH, Weiss H, et al
. Virologic and immunologic determinants of heterosexual transmission of human immunodeficiency virus type 1 in Africa. AIDS Res Hum Retroviruses 2001; 17:901–910.
18. Bunnell R, Ekwaru JP, King R, Bechange S, Moore D, Khana K, et al.3-year follow-up of sexual behavior and HIV transmission risk of persons taking ART in rural Uganda.15th Conference on Retroviruses and Oppportunistic Infections
. 3–6 February 2008. Boston, USA.
19. Brill I, Macaluso M, the Rwanda/Zambia HIV Research Group. A SAS program for the computation of seroconversion rates in a prospective study of HIV discordant couples in Lusaka, Zambia
. 2nd IAS Conference on HIV Pathogenesis and Treatment: Poster Abstract no. 1130
. 13–16 July 2003. Paris, France.
20. Kayitenkore K, Bekan B, Rufagari J, Marion-Landais S, Karita E, Allen S. The impact of ART on HIV transmission among HIV serodiscordant couples.AIDS 2006 - XVI International AIDS Conference: Abstract no. MOKC101.
13–18 August 2006. Toronto, Canada.
21. Mehendale SM, Kishore Kumar B, Ghate MV, Sahay S, Gamble T, Godbole SV, et al.Low HIV incidence in HIV sero-discordant couples in Pune, India.AIDS 2004 - XV International AIDS Conference: Abstract no. MoPeC3462
. 11–16 July 2004. Bangkok, Thailand.
22. del Romero J, Hernando V, Castilla J, Garcia S, Gil S, Rodriquez C. Lack of HIV heterosexual transmission attributable to HAART in serodiscordant couples.AIDS 2008 - XVII International AIDS Conference 2008: Abstract no. THPE0543
. 3–8 August 2008. Mexico City, Mexico.
23. Sullivan P, Kayitenkore K, Chomba E, Karita E, Mwananyanda L, Vwalika C, 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.
Montreal, 8–11 February 2009; Montreal, Canada.
24. Reynolds S, Makumbi F, Kagaayi J, Nakigozi G, Galiwongo R, Quinn T, 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.
Montreal, 8–11 February 2009; Montreal, Canada.
25. Hanley JA, Lippman-Hand A. If nothing goes wrong, is everything all right? Interpreting zero numerators. JAMA 1983; 249:1743–1745.
26. Wilson DP, Law MG, Grulich AE, Cooper DA, Kaldor JM. Relation between HIV viral load and infectiousness: a model-based analysis. Lancet 2008; 372:314–320.
27. Baggaley RF, Ferguson NM, Garnett GP. The epidemiological impact of antiretroviral use predicted by mathematical models: a review. Emerg Themes Epidemiol 2005; 2:9.
28. Blower SM, Gershengorn HB, Grant RM. A tale of two futures: HIV and antiretroviral therapy in San Francisco. Science 2000; 287:650–654.
29. Law MG, Prestage G, Grulich A, Van d, V, Kippax S. Modelling the effect of combination antiretroviral treatments on HIV incidence
30. 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.
31. Holmes KK, Sparling PF, Stamm WE, Piot P, Wasserheit JN, Corey L, et al
. Sexually transmitted diseases. New York: McGraw-Hill; 2008.
32. Myer L, Kuhn L, Stein ZA, Wright TC Jr, Denny L. Intravaginal practices, bacterial vaginosis, and women's susceptibility to HIV infection: epidemiological evidence and biological mechanisms. Lancet Infect Dis 2005; 5:786–794.
33. Wald A, Zeh J, Selke S, Ashley RL, Corey L. Virologic characteristics of subclinical and symptomatic genital herpes infections. New Eng J Med 1995; 333:770–775.
34. Watson-Jones D, Weiss HA, Rusizoka M, Changalucha J, Baisley K, Mugeye K, et al
. Effect of herpes simplex suppression on incidence of HIV among women in Tanzania. N Engl J Med 2008; 358:1560–1571.
35. Celum C, Wald A, Hughes J, Sanchez J, Reid S, any-Moretlwe S, et al
. Effect of aciclovir on HIV-1 acquisition in herpes simplex virus 2 seropositive women and men who have sex with men: a randomised, double-blind, placebo-controlled trial. Lancet 2008; 371:2109–2119.
36. Nandwani R. 2006 United Kingdom national guideline on the sexual health of people with HIV: sexually transmitted infections. Int J STD AIDS 2006; 17:594–606.
37. Hayward RS, Wilson MC, Tunis SR, Bass EB, Guyatt G. Users' guides to the medical literature. VIII. How to use clinical practice guidelines. A. Are the recommendations valid? The Evidence-Based Medicine Working Group. JAMA 1995; 274:570–574.
38. Guyatt GH, Oxman AD, Kunz R, Vist GE, Falck-Ytter Y, Schunemann HJ. What is ‘quality of evidence’ and why is it important to clinicians? BMJ 2008; 336:995–998.
39. Cohen MS, Bollinger RC, Celentano D, Chariyalertsak S, Grinstejn B, Hakim J, et al.HPTN 052. A randomized trial to evaluate the effectiveness of antiretroviral therapy plus HIV primary care versus HIV primary care alone to prevent the sexual transmission of HIV-1 in serodiscordant couples.http://www.hptn.org/research_studies/HPTN052StudyDocuments.asp#Protocol
. [Accessed February 2009]