Couples study in Uganda finds no HIV infections from partners on antiretroviral therapy

This article is more than 9 years old. Click here for more recent articles on this topic

A longitudinal cohort study of heterosexual couples in Uganda has found more evidence of the efficacy of antiretroviral therapy (ART) in curbing HIV infection within the community.

The study found 119 new HIV infections in 2334 couples over the course of the study, and 62 infections among the 254 couples that were initially or who became HIV serodiscordant. But it did not find a single example of transmission from a partner who was on antiretroviral therapy.

It also found that the rate of HIV infection between couples declined over time and that transmission likelihood was related to the HIV-positive partner’s viral load, although because widespread ART has only been available in Uganda relatively recently, neither of these findings quite reached statistical significance.



A serodiscordant couple is one in which one partner has HIV and the other has not. Many people dislike this word as it implies disagreement or conflict. Alternative terms include mixed status, magnetic or serodifferent.

prospective study

A type of longitudinal study in which people join the study and information is then collected on them for several weeks, months or years. 

voluntary male medical circumcision (VMMC)

The surgical removal of the foreskin of the penis (the retractable fold of tissue that covers the head of the penis) to reduce the risk of HIV infection in men.

person years

In a study “100 person years of follow-up” could mean that information was collected on 100 people for one year, or on 50 people for two years each, or on ten people over ten years. In practice, each person’s duration of follow-up is likely to be different.


The surgical removal of the foreskin of the penis (the retractable fold of tissue that covers the head of the penis) to reduce the risk of HIV infection in men.

It also found that men in the study were significantly more likely to transmit HIV, and women more likely to acquire it; that Muslims were considerably less likely to acquire or transmit HIV, in part probably because of the influence of circumcision; and it found that transmission and acquisition were both much more likely where the male partner was more than 15 years older than the female partner.


In 2011, the HPTN 052 study found that the HIV-positive partner in a positive/negative (serodiscordant) couple was at least 20 times less likely to transmit HIV if they were on ART than if untreated. While no one has disputed the efficacy found in this result, it was achieved within the highly controlled environment of a randomised controlled trial, where participants receive a great deal of monitoring, counselling and support, as well as free condoms; it was notable, for instance, that condom use in HPTN 052 was far higher than that seen in typical population surveys in Africa.

It is therefore important to find out whether ART is actually effective as a means of preventing HIV transmission within a community. Factors that may make it ineffective may include poor ART adherence and poor support for it, irregular drug access or supply, a high rate of extramarital or casual sex within the population, high rates of other sexually transmitted infections, and stigma or domestic violence that make it difficult for individuals to access or adhere to ART.

There have been a number of studies from San Francisco, British Columbia, South Africa and France that all show a correlation between HIV treatment, average viral load in the community and falling or stabilised HIV infection rates; but correlation is not causation, and these population-level studies cannot show a direct link between individuals starting ART and a reduction in the likelihood of their transmitting HIV. A Chinese study last December which carried out a retrospective analysis of HIV positive/negative couples and did find that the rate was lower in couples where the positive partner took ART. This found a relatively modest reduction in transmission of 26%, but this may have been modified by the fact that retrospective cohort studies cannot take account of factors that were not recorded at the time.

The study

A prospective cohort study, which takes a group of people and follows their medical history over time, is a better tool for measuring HIV infections within a community, as it can pre-set frequencies for tests and standards for record-keeping, and can better determine the cause/effect time sequence.

This Ugandan study was such a prospective study and has the additional advantage that it started back in 1989. The cohort studied comprises 20,000 people in 25 villages in southwest Uganda. Researchers visit each household annually and conduct demographic and behavioural surveys, and test participants for HIV and genital herpes (HSV-2). In this study, they also asked adults who their main partner was, if any, and collected whatever data were available on antiretroviral treatment, viral load and CD4 count in HIV-positive people.

The study looked at 2465 male/female couples whose HIV status was known, 2334 of whom were not already both HIV-positive and who were tested for HIV at least twice. The median follow-up time was 2.83 years.

The average age of participants was 33 in men and 26 in women. Unusually, in a continent where more women than men have HIV, 54% of the HIV-positive partners in this study were male.

Results: HIV transmission and acquisition

Of the couples, 2113 were both HIV negative at the first study visit, 131 (5.3%) were both HIV positive, and 221 (9%) were an HIV positive/negative couple.

During the period of the study, 53 of the HIV-negative partners in positive/negative couples caught HIV; 53 people in a negative/negative couple became HIV positive, and so the couple became serodiscordant; and amongst these, the other partner subsequently became HIV positive in nine cases. In addition, four negative/negative couples both caught HIV and became positive/positive between study visits. There were therefore 53 plus 9 = 62 infections within couples known to be serodiscordant at some point in the study.

Overall HIV incidence in negative partners was just over 7% a year, but the rate of infections declined over time, from 11% a year in 1990-1994 to 4.3% a year in 2005-2007. In multivariate analysis, compared with 1990-1995, the chance of the HIV-negative partner acquiring HIV was about 40% lower between 1995 and 2004, and was about 60% lower in 2005-2007. The same applied to the chance of the positive partner transmitting HIV. These associations were not quite statistically significant (p = 0.09 and 0.06 respectively) partly because the decline in risk occurred in two phases separated by a plateau; transmission risk fell in the mid-90s, an effect ascribed by the researchers to better community awareness and education, and then again in the mid-2000s. ART becoming more widely available may have contributed to this latter fall, but with the current study’s data only going up to the end of 2007, and relatively few people starting ART, antiretrovirals had probably not been used enough at that point to prove this.

Results: demographic factors

A number of demographic characteristics were associated with reduced or increased HIV acquisition/transmission. In the HIV-negative partners, women were 83% more likely to acquire HIV than men, and in the HIV-positive partners, men were 81% more likely to transmit HIV than women. Muslims were four times less likely than Christians to acquire HIV if negative and three times less likely to transmit it if positive. HIV-negative women were 3.7 times more likely than otherwise to acquire HIV if their male partner was more than 15 years older than them, and HIV-positive men 3.1 times more likely to transmit it if they were more than 15 years older than their female partner.

In the case of Muslims, clearly high circumcision rates explain part of the difference, but the effect was amplified by the fact that half the female partners were immigrants to the area, many of them refugees from nearby Rwanda, compared to only one in five men, and these women were also more likely to be Christian.

Results: viral load and antiretroviral therapy

Viral load in the HIV-positive partner did appear to affect transmission. Compared with people with viral loads under 10,000 copies/ml, people with viral loads between 10,000 and 50,000 copies/ml were over twice as likely to transmit HIV and people with viral loads over 50,000 copies/ml nearly six times as likely. However, because viral load results were only available for a minority of participants, this effect did not reach statistical significance (p = 0.25).

During the study period, some of the HIV-positive partners started ART. None of them transmitted HIV in 29 person-years on ART, compared with 62 transmissions in 843 person-years from HIV-positive people not on ART (transmission rate 7.35% a year). This difference was highly significant (p < 0.001) but more people will have to start on ART before a meaningful comparison of risks can be made.

One point worth noting is that, in a study like this, it is impossible to establish whether people caught HIV from their primary partner or from someone else. In HPTN 052, which DNA-tested everyone’s HIV, 28% of infections ‘in couples’ were in fact acquired from someone else. Although no infections have so far been seen where the positive partner is on ART, such ‘extramarital’ infections could make it look like people on ART were transmitting HIV when they were not.

A study in the Lancet this week estimated that 30% of all new HIV infections in men and 10% in women within stable relationships in Africa were due to sex outside the relationship.


Biraro S et al. HIV-1 transmission within marriage in rural Uganda: a longitudinal study. PLoS One 8(2):e55060. doi: 10.1371/journal.pone.0055060. 2013.