Is treatment really reducing infections?

This article is more than 13 years old. Click here for more recent articles on this topic

This time last year, at the 17th Conference on Retroviruses and Opportunistic Infections (CROI), Dr Moupali Das of the San Francisco Department of Public Health presented evidence to show that the city’s intensive testing and treatment policy was beginning to result in a declining HIV infection rate there. Similar evidence was presented from the province of British Columbia in Canada.

The evidence presented still leaves some questions unanswered, however.

  • Is the reduction in viral load in the HIV-positive population (the 'community viral load' or CVL) really the cause of the decreased level of diagnoses seen in San Francisco in the last few years, or is it due to the success of prevention campaigns and reductions in risky behaviour?
  • Do reduced diagnoses really indicate reduced incidence of infection?
  • Is reducing the average viral load of diagnosed people a good indicator of the average infectiousness of people with HIV in the community – or do high viral loads in the minority who remain undiagnosed make this an unreliable indicator?

The answers to these questions are crucial as the future direction of HIV prevention policy may depend on them, in particular whether to concentrate on suppressing viral load or on behaviour change as the mainspring of prevention.

Update from San Francisco

Dr Das presented more evidence from San Francisco to show a strong correlation between reductions in community viral load and the falling number of HIV diagnoses. The most recent CVL, defined as the average viral load of all HIV positive people being seen for care in the city, went down from approximately 25,000 copies/ml in 2004 to 10,000 copies/ml in 2009. Over the same time, new diagnoses of HIV decreased from 820 in 2004 to 500 in 2009.


risky behaviour

In HIV, refers to any behaviour or action that increases an individual’s probability of acquiring or transmitting HIV, such as having unprotected sex, having multiple partners or sharing drug injection equipment.

mathematical models

A range of complex mathematical techniques which aim to simulate a sequence of likely future events, in order to estimate the impact of a health intervention or the spread of an infection.

statistical significance

Statistical tests are used to judge whether the results of a study could be due to chance and would not be confirmed if the study was repeated. If result is probably not due to chance, the results are ‘statistically significant’. 


A test used to measure something.

virological suppression

Halting of the function or replication of a virus. In HIV, optimal viral suppression is measured as the reduction of viral load (HIV RNA) to undetectable levels and is the goal of antiretroviral therapy.

For CVL to decrease, a number of other things have to happen first, in a ‘cascade’ of events. Firstly, the frequency of testing in the at-risk population must increase; this must result in a lower proportion of undiagnosed cases; those diagnosed must be linked to care, and the majority must start treatment; and the time between diagnosis and viral suppression must reduce.

Dr Das said all these indicators were strongly correlated. The average time from diagnosis to the start of treatment shrank from twelve months in 2004 to two months in 2009. The time between diagnosis and achieving an undetectable viral load went down from nearly three years in 2004 to eight months in 2009; and the percentage with an undetectable viral load within a year of diagnosis increased from 26% in 2004 to 82% in 2008.

The average CD4 count at the start of treatment increased from 357 to 445 in the two years between 2007 and 2009, but CD4 count  at diagnosis "remains stubbornly between 400 and 450", Dr Das commented.

Interestingly, therefore, the reduction in CVL appears not to be due to increasing the frequency of HIV testing – at least so far. Rather, it appears to have been mainly driven by San Francisco's decision to abandon CD4 count as a criterion for starting therapy and to offer treatment to everyone diagnosed.

All these trends were highly correlated and the correlation was highly statistically significant (p=<0.001). Nonetheless, it is notable that Dr Das did not include an estimate of trends in actual incidence (the number of people actually infected in a year). This is because without using a reliable assay that can actually estimate time of infection, the possibility remains that a current fall in diagnoses actually reflects falls in HIV transmissions, caused by improvements in risk behaviour, up to a decade ago.

Treatment-as-prevention models

Several modelling studies analysed the degree to which different interventions would result in a reduction in HIV infections. One based on the HIV epidemic in the Netherlands (van Sighem) calculated the result of three different interventions: a 40% reduction in risk behaviour, reducing the time from infection to diagnosis from the current 2.4 years to one year, or treating everyone immediately at diagnosis.

In the Netherlands, far from infections decreasing, the number of infections diagnosed per year increased from 382 in 2000 to 840 in 2008 and would increase to 1462 infections a year with no interventions.

A 40% reduction in risk behaviour would reduce the true number of infections (not diagnoses) in 2019 to 138. Reducing time to diagnosis to one year would reduce the number to 629. But immediate treatment, although it would result in a rapid decrease in infections, would not produce a sustained decline in infections and they would be back up to 914 a year by 2019.     

The decline in infections in British Columbia has, in recent years, been strongly tied to a successful initiative to reduce undiagnosed infections in injecting drug users. A study from Baltimore (Kirk) found that diagnosis and treatment was having a similar impact on injecting drug users there. In recent years, it found, HIV incidence decreased by 68% for each tenfold decline in the community viral load in this population. The risk of an HIV-negative member of this population becoming infected with HIV decreased by 5% for each 1% increase in the proportion of HIV-positive members of the population on HIV treatment.

A mathematical model (Lima) by the British Columbia team, however, finds that in the case of gay men, universal treatment and reducing community viral load will not reduce onward transmissions by the same extent. This is because gay men will still be infectious in the first six months on treatment, a minority will remain so and the impact of treatment on reducing new infections is extremely dependent on the number of sexual partners people have. This is especially the case if people are closely connected in sexual networks.

“While the power of HAART to decrease infection remains strong and substantial, these results show that other interventions coupled with full HAART coverage of all medically eligible HIV-infected individuals can provide a tremendous chance to lessen the epidemic among MSM,” say the researchers.

Abstracts and webcast

You can view the abstracts from this research on the official conference website:

Abstract 1022 and the PDF of the poster:

Abstract 483:

Abstract 484:

Abstract 486:

You can also view a webcast of the themed discussion Community Viral Load on the official conference website.

We rely on donations to continue our work to help people with HIV; support us today at:


Das M et al. Success of Test and Treat in San Francisco? Reduced Time to Virologic Suppression, Decreased Community Viral Load, and Fewer New HIV Infections, 2004 to 2009. Eighteenth Conference on Retroviruses and Opportunistic Infections, Boston, abstract 1022. See the abstract here and poster here.

Van Sighem A et al. Decreasing Community Infectiousness Is a Marker for Decreases in New HIV Infections among Dutch Homosexual Men. Eighteenth Conference on Retroviruses and Opportunistic Infections, Boston, abstract 483. See the abstract here.

Kirk G et al. Decline in Community Viral Load Strongly Associated with Declining HIV Incidence among IDU. Eighteenth Conference on Retroviruses and Opportunistic Infections, Boston, abstract 484. See the abstract here.

Lima V et al. Why the MSM-driven HIV Epidemic Is Not Slowing Down Even in the Presence of HAART. Eighteenth Conference on Retroviruses and Opportunistic Infections, Boston, abstract 486. See the abstract here.

We rely on donations to continue our work to help people with HIV, support us today at: