No HIV transmissions with undetectable viral load: interim PARTNER study results show need for longer follow-up

Simon Collins, HIV i-Base

CROI 2014 logo-190The PARTNER study is an international observational study that estimates the risk of HIV transmission within HIV serodifferent couples who do not use condoms, when the HIV positive person is on ART and has an undetectable viral load.

This is not a question of merely academic interest – it is central to defining the safety of programmes that already emphasise the impact of treatment as prevention. It is also essential for people making their own decisions about levels of risk.

Results presented at CROI 2014, by Alison Rodger from University College London, from a planned interim analysis, reported that no linked transmissions have so far occurred after almost 900 couple years of follow-up. These results come from 586 heterosexual and 308 gay male couples. [1]

An entry criterion for the study is to not routinely use condoms. Sexual behaviour questionnaires are included in order to have greater certainty over the range for the residual risk of transmission for different types of sex. Although HIV treatment dramatically reduces this risk, this knowledge is based on very limited data, largely from heterosexual studies, and often when condoms continued to be used.

By November 2013, PARTNER had enrolled 1,110 couples. Median time on ART was almost five years (IQR: 1.9-11.4) and couples reported having sex without condoms for a median two years (IQR: 0.5-6.3).  Follow-up results included almost 44,500 times when sex was without condoms and over 21,000 times when this was from anal sex, (see Table 1). Couples had sex a median of 45 times a year without using condoms (IQR: 16-90).

For inclusion in this analysis, the positive partner needed to have a viral load <200 copies/mL. Although some HIV negative people became positive during the study, the phylogenetic analyses key to the results, showed these were not linked to their HIV positive partner. Sex with partners outside the partnership during follow-up period, also not using condoms, was reported by 34% of HIV negative gay men and 3% of heterosexual HIV negative partners.

Estimating risks is dependent not simply on the number of transmissions that may or may not occur, but in the statistical calculations for how confident the researchers are that the same results might not just occur by chance. This is the upper limit of the 95% confidence interval (95% CI), based on the standard assumption that a 5% possibility of the same results occurring by chance is acceptable.

Based on current analysis, the rate of within-couple HIV transmissions during eligible couple-years for the study as a whole was zero. However, the upper limits of the 95%CI were 0.40 per 100 couple years of follow-up (CYFU) for the study as a whole, 0.96/100 CYFU for anal sex (in gay and straight couples combined) and 1.97/100 CYFU for receptive anal sex with or without ejaculation (for gay couples).

The estimated transmission risk across the study per act, ranges from around 1 in 5,000 for an HIV negative man having vaginal sex to 1 in 2,000 for an HIV negative man having receptive anal sex.

A non-technical explanation of these risks, based on the PARTNER results so far, is that the risk of transmission occurring for one couple over ten years (based on having sex 45 times a year) could be as high as 4% for the average participant, and that the risk from anal sex could be as high as 9%. For receptive anal sex this reaches 32% risk over ten years. There is also a 2.5% chance that these risks could be higher.

If participants in the study had not been on ART, approximately 15 transmissions would have been expected in the straight couples and 86 transmissions in the gay couples, based on two meta-analyses of per act risk.

The data from PARTNER is already more substantial than the four other previous studies that have looked at this question. However, these upper estimates are dependent on the current number of years of follow-up and the study is only partly completed. If similar results continue, then the upper estimates will reduce further, but this is dependent on having this data.

In addition to the final results from the participants currently enrolled in PARTNER, an extension of the study called PARTNER 2 is looking to enroll an additional 450 gay male couples to achieve approximately 2000 patient years of follow up for anal sex. This will generate similar levels of confidence to the risk from vaginal sex.

The researchers have also produced online resources that help explain the implications of these results. [2, 3]

Table 1: Risk behaviour by the HIV negative partner and approximate estimated rates of HIV transmissions within couples

HIV status and sexual orientation of couples Type of sex without a condom by HIV negative partner Linked trans-missions (n) Couple-years of follow up (CYFU) Approx. no. of sex acts without condoms Risk per contact (95% CI)* Rate per 100 CYFU (95% CI) 10 year risk (95% CI)
Study overall All types of sex (VL <200) 0 894 44,450 0 (0 – 0.00008) 0 (0-0.40) 0 (0 – 3.9%)
All types of sex (VL < 50) 0 836 41,480 0 (0 – 0.00009) 0 (0-0.43) 0 (0 – 4.2%)
Anal sex 0 374 21,030 0 (0 – 0.00017) 0 (0-0.96) 0 (0 – 9.2%)
Straight couples (man positive) Sex 0 288 13,730 0 (0 – 0.00028) 0 (0-1.25) 0 (0 – 11.7%)
Vaginal sex with ejaculation 0 191 8,910 0 (0 – 0.00043) 0 (0-1.88) 0 (0 – 17.1%)
Vaginal sex without ejaculation 0 174 6,380 0 (0 – 0.00060) 0 (0-2.07) 0 (0 – 18.7%)
Straight couples (woman positive) Sex 0 298 14,300 0 (0 – 0.00027) 0 (0-1.21) 0 (0 – 11.4%)
Vaginal sex 0 272 14,150 0 (0 – 0.00027) 0 (0-1.32) 0 (0 – 12.4%)
Gay male couples Anal sex 0 308 16,420 0 (0 – 0.00023) 0 (0-1.17) 0 (0 – 11.0%)
Receptive anal sex (with or without ejaculation) 0 182 7,750 0 (0 – 0.00050) 0 (0-1.97) 0 (0- 17.9%)
Insertive anal sex 0 262 11,750 0 (0 – 0.00033) 0 (0-1.37) 0 (0 – 12.8%)

 Simon Collins is a community representative on the Steering Committee of the PARTNER Study.


It is important to be able to explain the concept of confidence intervals related to risk, even though this is not always easy in non-technical terms. However, without substantive data from observed rates in prospective studies such as PARTNER, it is nearly impossible.

These results do not prove the safety of having sex without a condom when viral load is undetectable: they provide the most reliable evidence to date on the level of risk for people who have already been having sex without condoms, sometimes for many years. Other factors affecting risk, include genetic predisposition to HIV infection and STIs could both make risks higher on an individual rather than population-based level.

These results are nevertheless extremely positive. They support modelling data indicating that new infections in the UK, especially among gay men, are likely to be driven by people who are not yet diagnosed, and  are often due to high viraemia in primary infection.  


  1. 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. (abstract)  (webcast)
  2. Q&A fact sheet and resource on the interim results from the PARTNER Study.  (PDF)
  3. An online interview with the principal investigator of the PARTNER study is posted on the community IFARI youtube channel.

Links to other websites are current at date of posting but not maintained.