The evidence for U=U (Undetectable = Untransmittable): why negligible risk is zero risk
Note: See postscript for results from the PARTNER2 study that provided further evidence for U=U in gay men. Table updated May 2019.
Simon Collins, HIV i-Base
Over the last year, hundreds of HIV organisations, including the US Center for Disease Control (CDC), have supported the statement that HIV transmission does not occur when viral load is undetectable on ART.
And while the dramatic impact of ART on reducing HIV transmission has been known for a long time, saying ART stops this completely is new.
This change is especially important given that prejudice and discrimination against HIV positive people is still widespread. So while it is easy to simply answer “no” to the question of whether someone with an undetectable viral load is still infectious, it is more complicated to explain why.
This article summarises selected key studies from 20 years of accumulating evidence that should directly challenge the prejudice and fear of HIV that is still widespread.
U=U: Undetectable = Untransmittable (or Uninfectious)
Launched in 2016, the Undetectable = Untransmittable (U=U) campaign is based on the following statement: “A person living with HIV who has undetectable viral load does not transmit HIV to their partners”. [1, 2]
The campaign statement has been endorsed by more than 350 from 34 countries HIV organisations who joined the U=U campaign, including by leading scientific and medical organisations such as the International AIDS Society (IAS), UNAIDS, and the British HIV Association (BHIVA).
The support for the statement is also remarkable given that science is not able to prove a negative – ie that something will not happen.
Instead, people who claim that HIV is transmittable when viral load is undetectable, should be challenged to prove it.
20 years of accumulating evidence
The scientific approach to understanding the world usually involves three stages.
- Observing something.
- Deciding on one or more hypotheses that might explain it.
- Testing any theory in a suitable experiment.
The strength of this approach is that a good study, by definition, should be repeatable. If the results are true and not by accident, other researchers should be able to repeat the study and get similar and consistent results each time.
The evidence supporting U=U includes different types of research spanning observational studies, randomised trials, systematic reviews and expert opinion. See Table 1.
Table 1: Key selected evidence supporting U=U
|San Francisco cohort||Clinical results from small cohort of HIV positive women using triple ART during pregnancy.||Transmission from mother to baby was reduced to approaching zero.||1998||Beckerman K et al. |
|DHHS guidelines||Expert opinion included in evidence-based guidelines.||Theoretical plausibility of reducing transmission risk was used as a factor for early ART.||1998||DHHS guidelines. |
|Ugandan cohort (Rakai)||Prospective observational cohort in ~ 400 serodifferent couples.||Zero transmissions when viral load was less than 1500 copies/mL.||2000||Quinn TC et al. |
|Spanish cohort||Prospective observational study in 393 heterosexual discordant couples enrolled from 1991 to 2003 where the negative partner became HIV positive.||Zero transmissions in couples where the HIV positive partner was on ART with undetectable viral load. Cautions emphasised good adherence and no STIs.||2005||Castella A et al. |
|Swiss Statement||Expert opinion and evidence review of >25 smaller studies looking at impact of ART on risk factors for HIV transmission.||Concluded that transmission would not occur undetectable with viral load.||2008||Vernazza P et al. |
|HPTN 052||1763 serodifferent heterosexual couples randomised to immediate or deferred ART.
Although condom use was high the impact of ART was highly significant.
|All infections occurred in people with detectable viral load: n=17 in the deferred ART group and one early infection in the ART group before VL was undetectable. Follow-up reported out to four years.||2011||Cohen M et al. [8, 9]|
|PARTNER||Prospective observational European study in ~900 serodifferent couples who were not using condoms.||Final results reported zero transmissions after more than 58,000 times couples had sex without condoms when viral load was undetectable <200 copies/mL.||2014 (interim).
|Rodgers A et al. [10, 11]|
|Opposites Attract||Prospective observational study in 358 serodifferent gay male couples in Australia, Thailand and Brazil.||Zero transmissions when viral load was undetectable <200 copies/mL.||2017||Grulich A et al. |
|PARTNER2||Extension of PARTNER study to collect additional follow-up in gay male couples.||Zero transmissions after more than 77,000 times gay male couples had sex without condoms when viral load was undetectable <200 copies/mL.||Presented August 2018. Published May 2019||[12, 19, 20]|
Key stages in this timeline include:
- 1998: observations that triple therapy ART reduced transmission.
- 1998: expert opinion that risk would be reduced (including based on reviewing evidence related to the details of this protection).
- 2000 – 2005: prospective observational studies and related research (Rakai cohort and others).
- 2008: further expert opinion and evidence review (Swiss Statement).
- 2011: first evidence from a randomised clinical trial (HPTN 052).
- 2014 – 2017: further prospective observational studies (PARTNER and Opposites Attract) – the first studies to provide data about risks for gay men.
- 2016 – 2017: further expert opinion (U=U campaign).
Each of these studies is now explained in more detail.
Early evidence: mother-to-child and Ugandan heterosexual couples
A remarkable report in July 1998 provided some of the first clinical evidence for the impact of viral load on HIV transmission.
At the IAS conference held in Geneva, Dr Karen Beckerman reported on a small cohort of HIV positive women in San Francisco who had used triple therapy during pregnancy. Instead of the 30% mother-to-infant transmissions reported before ART, or the 10% seen with AZT monotherapy, triple therapy reduced transmissions to approaching zero. 
Although this study reported on vertical rather than sexual transmission it provided clinical results showing that an undetectable viral load stopped a much higher risk of transmission.
Then later that year, the December 1998 update to the US DHHS guidelines, included “possibly decreasing the risk of viral transmission” as an additional reason for starting early ART. 
These expert guidelines noted the lack of direct evidence supporting this statement and emphasised that condoms should still be used even with undetectable viral load – but this inclusion in the 100-page document from leading US doctors this was important.
One of the next key studies provided direct evidence linking viral load with risk of HIV sexual transmission. This was a prospective observational cohort study in 415 serodifferent heterosexual couples in Rakai, Uganda, where one partner was HIV positive and the other was HIV negative. The study, by Thomas Quinn and colleagues was published in the New England Journal of Medicine in 2000. 
After median follow-up of 22 months, the risk of HIV transmission was not only clearly linked to higher viral load. No transmissions were reported among the 51 couples where the HIV positive partner had viral load below 1500 copies/mL.
Several details of the Rakai study are important. It was before ART was available and condom use was low. It found that transmissions rates were similar for men and women and that other STIs didn’t affect HIV risk. It also reported highly significant impact from circumcision – all the men who became positive during the study were uncircumcised.
These results were 17 years ago.
Expert opinion and evidence review: the Swiss statement
From 2000 to 2008, many smaller studies reported reductions in other routes of transmission, or supplemented observational data with supportive research, such as reporting the impact of ART in genital fluids.
For example, in 2005, a Spanish cohort reported on almost 400 heterosexual serodifferent couples where the negative partner became HIV positive during the period 1991 to 2003. The results were presented for three time periods – pre-ART (1991–1993), early-ART (1996–1998) and late-ART (1999–2003) – and reported no transmissions when the positive partner was on ART. 
Cautions for these results were that other risks reduced over time, such as condoms being more widely used and people having less sex as they grew older, but zero transmissions was still significant.
In 2008, Pietro Vernazza and colleagues published the first high profile evidence review that concluded that ART stopped transmission. 
This paper, published in French but quickly translated into English, was a response to the laws in Switzerland that criminalised an HIV positive person if they had sex with a negative partner, even if condoms were used or if a couple wanted to conceive with full consent. This paper reviewed more than 25 studies and concluded that transmission did not occur. The estimated risk as a very rare event was less than 1 in 100,000 (0.001%) – and therefore effectively zero.
Important considerations for the Swiss Statement included that the HIV positive person should be adherent on effective ART (not missing doses), have an undetectable viral load, and not have sexual infections that might increase viral load.
The Swiss statement was not only widely publicised but it was also widely cricitised, generating a very high profile. As such, it set a challenge to other doctors and researchers to report any cases that disproved the statement. Given the competitive nature of academic research, it is notable that after almost ten years no cases have been published that refute the Swiss statement.
Randomised data: HPTN 052
Scientists grade evidence based on the design of studies to be able to prove a link between and intervention and outcome. For many questions, the best quality of evidence comes from a randomised clinical trial. The process of randomly assigning participants to two or more groups where only the intervention is different, is the best way to rule out the results having been due to chance.
Because there is always the potential for other factors to affect outcomes, randomised studies are usually credited as the gold standard for evidence.
In 2011, US researchers, led by Myron Cohen and colleagues at the HIV Prevention Treatment Network (HPTN) reported early results from the HPTN 052 study. 
HPTN 052 recruited more than 1700 serodifferent couples (mainly in southern Africa, Latin America and South-East Asia. These were almost entirely heterosexual couples, and the HIV positive partners were randomised to either start ART immediately or wait until their CD4 count dropped to 350 cells/mm3 (the then threshold in WHO guidelines for starting treatment).
All couples were supported with condoms and information on reducing the risk of HIV transmission, but it soon became clear that HIV transmissions were almost exclusively occurring in the group waiting for ART. Of the 39 transmissions, 28 were linked to HIV positive partner. Of these, 27/28 were in group waiting for ART. The single transmission in the immediate ART group occurred within weeks of starting treatment, when viral load would have still been high and certainly detectable.
This provided a very high level of evidence that ART was directly linked to protection against sexual transmission and as a result the HPTN 052 study was stopped early so that all HIV positive participants could receive immediate ART. Longer follow-up of HPTN continued for at least another four years and confirmed the durability of these early results. 
HPTN 052 produced evidence to enable HIV positive people to access ART earlier in order to protect their partners – called Treatment as Prevention (TasP). But limitations of the study meant that it could only report relative differences between the two study groups, rather than quantify any actual risk (even if the risk was theoretical).
Again, this was a heterosexual study, anal sex was rarely reported and condom use was relatively high. This meant that while ART could be proved to reduce infection, the study couldn’t estimate how low this risk became, or the likely risk for different types of sex.
Large observational cohorts: PARTNER study and Opposites Attract
In 2009, several years before the results from HPTN 052, a group of European researchers led by Jens Lundgren from the Centre of Excellence for Health, Immunity and Infections (CHIP) launched the prospective observational PARTNER study. [10, 11]
The PARTNER study was important for enrolling serodifferent couples where the HIV positive partner was on ART and where the couples were already not always using condoms (often for many years).
Importantly, approximately one-third of the almost 900 couples were gay men and the study included detailed questionnaires on sexual activity to estimate risk based on actual exposure.
As with all studies, information about reducing HIV transmission, including free condoms, were included for all participants. All couples were then followed over time, trying to see whether transmissions occurred.
In a planned early analysis, presented at a conference in February 2014, PARTNER reported zero linked (within-partner) transmissions after more than 44,000 times when condoms hadn’t been used and viral load was undetectable (defined as less than 200 copies/mL). 
PARTNER also provided reassurance for previous theoretical concerns from viral load blips or other STIs. No transmissions were seen in the 91 couples where the positive partner reported an STI (approximately one-third of gay couples had open relationships). The final results, presented and published in July 2016, reported zero transmissions after 58,000 times without condoms. 
The PARTNER results made headlines globally, but a less well-known aspect of this study was that the ground-breaking results took nearly two years to be published. This is likely linked to the implications the results would have on HIV prevention campaigns that were based on always using a condom, even when the limitation of condom-only prevention were clear from continued high rates of HIV transmission.
Because an important outcome of the PARTNER study is to quantify the theoretical range of risk (the upper limit of the 95% confidence interval), the PARTNER 2 study continued to collect results in gay couples to provide an equal balance of evidence compared to heterosexual data. 
Finally, at the IAS conference held in Paris in 2017, results from the Opposites Attract study in 358 gay male couples from Australia, Thailand and Brazil, also reported zero linked transmissions after almost 17,000 when condoms were not used. 
Again, STIs were not uncommon (present in around 1,000 of these occasions) and didn’t result in HIV transmission.
Zero to negligible: what is in a word?
HIV transmission, even without a condom and without ART, is generally an uncommon event.
For example, the average upper range of estimated per-exposure risk ranges from 0.014 for receptive anal sex (14 in 1000) to from 0.001 for receptive or insertive vaginal sex (1 in 1000) and the lower ranges are many fold lower. 
However, during the first 2 to 4 weeks after infection, when viral load can be millions of copies/mL and people still believe they are HIV negative, risk will be higher. This led to many health campaigns pointing out that someone who believes they are HIV negative based on their last HIV test is associated with a much higher relative risk than any HIV positive person with undetectable viral load on ART.
Nevertheless, the semantic difference between zero risk and negligible risk, even when this theoretical risk is increasingly tiny (as with the Swiss Statement), prevented some people saying that the risk was effectively zero.
The most significant change over the last year, driven by the U=U campaign, has been for leading HIV scientists to now assert that a negligible theoretical risk is effectively zero.
Reversing the challenge: to now prove whether transmission is possible
Under ideal circumstances, large prospective studies that were designed to find cases of transmission when viral load was undetectable have not been able to do so.
So the evidence gap in 2017 is now the lack of any proof showing that HIV transmission is possible when viral load is undetectable.
This reverses the scientific challenge from proving safety to proving risk. Purely theoretical risks are no longer a good enough level of evidence to sustain stigma and discrimination and certainly not criminalisation.
Instead, there is no evidence to show that HIV transmission occurs when viral load is undetectable. People who want to assert the theory that HIV transmission might be possible, now have to provide some level of proof.
A comprehensive body of evidence now supports the U=U statement. This ranges from early clinical and theoretical studies, through small observational studies, randomised trials and the large prospective cohorts.
In addition, no cases of HIV transmission have been reported, over nine years since the Swiss Statement set this challenge. This includes data for gay men, for couples that have anal sex, over periods when low-level viral blips are likely and even when STIs are present.
In reality, even if the actual risk is zero, it is not healthy to think about anything in life as being risk-free. Even if at some point in the future an unlucky and rare case of transmission is reported with undetectable viral load, the U=U campaign is still right for closing the gap between zero and the real-life meaning of negligible in real terms.
The article is based on a talk given to the Positive People’s Forum held in Glasgow on 1 July 2017. 
Simon Collins is on the Steering Committee for the PARTNER studies.
The US CDC endorsement in September 2017 is especially important. 
Currently, the HIV criminalisation laws in many US states are outdated and severe. As a result, hundreds of HIV positive people have been imprisoned for many years for non-disclosure and often in the absence of actual transmission.
Since this article was published further evidence has become available to support U=U.
In July 2018, results from the PARTNER 2 study reported zero linked HIV transmissions from nearly 1000 gay couples who had sex more than 77,000 times without condoms. [17, 18]
In May 2019, these results were published in the Lancet. 
- Undetectable = Untransmittible
- U=U consensus statement: Risk of sexual transmission of HIV from a person living with HIV who has an undetectable viral load.
- Beckerman K et al. Control of maternal HIV-1 disease during pregnancy. Int Conf AIDS 1998 Jun 28-Jul 3; 12:41. Poster abstract 459.
- U.S. Department of Health and Human Services (DHHS). Guidelines for the Use of Antiretroviral Agents in HIV-1-Infected Adults and Adolescents. December 1998.
- Quinn TC et al. Viral load and heterosexual transmission of HIV type 1. Rakai Project Study Group. N Engl J Med 2000; 342: 921-929. Free online access.
- Castilla J et al. Effectiveness of highly active antiretroviral therapy in reducing heterosexual transmission of HIV. J Acquir Immune Defic Syndr. 2005;40:96-101. Free full text.
- Vernazza P et al. HIV-positive individuals not suffering from any other STD and adhering to an effective antiretroviral treatment do not transmit HIV sexually. (Les personnes séropositives ne souffrant d’aucune autre MST et suivant un traitment antirétroviral efficace ne transmettent pas le VIH par voie sexuelle). Bulletin des médecins suisses 89 (5), 30 January 2008. Included with English translation.
- Cohen MS et al for the HPTN 052 Study Team. Prevention of HIV-1 infection with early antiretroviral therapy. Supplementary information. NEJM 2011; 365:493-505.
- Cohen MS et al. Final results of the HPTN 052 randomized controlled trial: antiretroviral therapy prevents HIV transmission. IAS 2015, 19 – 22 July 2015, Vancouver. MOAC0101LB.
- Rodger A et al. HIV transmission risk through condomless sex if HIV+ partner on suppressive ART: PARTNER Study. 21st CROI, 3-6 March 2014, Boston. Oral late breaker abstract 153LB.
- Rodger AJ et al for the PARTNER study group. Sexual activity without condoms and risk of HIV transmission in serodifferent couples when the HIV-positive partner is using suppressive antiretroviral therapy. JAMA, 2016;316(2):1-11. DOI: 10.1001/jama.2016.5148. (12 July 2016). Full free access.
- PARTNER2 Study (2014–2018).
- Grulich A et al. HIV treatment prevents HIV transmission in male serodiscordant couples in Australia, Thailand and Brazil. IAS 2017, Paris. Oral abstract TUAC0506LB.
- Fox J et al. Quantifying sexual exposure to HIV within an HIV-serodiscordant relationship: development of an algorithm. AIDS 2011, 25:1065–1082. DOI:10.1097/QAD.0b013e328344fe4a. Free online access.
- Collins S. Undetectable = Uninfectious. Positive Person’s Forum, 1 July 2017, Glasgow.
- US CDC News. Dear colleague: national gay men’s HIV/AIDS awareness day. (27 September 2017)
- Zero HIV transmissions in PARTNER study after gay couples had sex 77,000 times without condoms – an undetectable viral load stops HIV. HTB August 2018.
- Rodger A et al. Risk of HIV transmission through condomless sex in MSM couples with suppressive ART: The PARTNER2 Study extended results in gay men. AIDS 2018, 23-27 July 2018, Amsterdam. Late breaker oral abstract WEAX0104LB.
- Rodger AJ et al. Risk of HIV transmission through condomless sex in serodifferent gay couples with the HIV-positive partner taking suppressive antiretroviral therapy (PARTNER): final results of a multicentre, prospective, observational study. Lancet 2019; published online May 2. (Open access)
- Collins S. PARTNER 2 study published in the Lancet: global news coverage that ART stops HIV transmission. HTB, May 2019. (03 May 2019).
- Google news. Compilation from Google search “HIV transmission 3 may 2019”. (3 May 2019).