Advocating a treatment that may not help the treated?


This page in:

Last month, NIAID released news that treating HIV-infected partners in mostly heterosexual HIV-discordant couples at 13 sites around the world reduced HIV transmission to the HIV-negative partner by 96%. This is obviously great news and an important scientific advance. Its implications may well be far reaching in terms of HIV prevention in the future. But, I am not sure they justify ‘test and treat’ as prevention policy touted in some corners now.

Of the 1,763 couples who agreed to be in the study (not a random sample of those approached – for example, people in the U.S. and homosexual individuals were less likely to participate), one group, randomly chosen, started taking a combination of three drugs immediately, while the other group (deferred treatment group) was treated when their CD4+ counts fell below 250. Some highlights from the study are:

·         The study started in 2005 and aimed to continue until 2015, but after an interim review by an independent data and safety monitoring board saw the data, the trial was stopped in 2011. So, I presume that the reported infection rates are incidence over a 5-6 year period.

·         39 initially uninfected partners were infected by 2011. However, only 28 of these were infections by the partner (confirmed by genetic analysis).

·         Of the 28 infections from partners in the study sample, 27 happened in the deferred treatment group and one in the immediate treatment group. This is the 96% reduction result.

So, the evidence is as conclusive as it gets that immediate treatment is effective in preventing transmission to uninfected partners. However, I’d like to know the treatment regimen among this group, i.e. whether adherence was monitored and enforced or not (the news release does not have this level of detail and does not link to a published study).

The study, according to the news release, “…was designed to evaluate whether antiretroviral use by the HIV-infected individual reduced HIV transmission to the uninfected partner and potentially benefited the HIV-infected individual as well (emphasis added).” So, what about those outcomes for the HIV-infected individual that have not been discussed in the media? The immediate treatment reduced the incidence of extrapulmonary TB (i.e. TB in organs other than the lungs), but this incidence was quite small (17 in the deferred treatment group and 3 in the immediate treatment group over the six year study period). Furthermore, there was no difference in mortality between the two groups (13 vs. 10). So, it’s fair to conclude (for the health outcomes reported for the HIV+ individual) that the impact of immediate treatment was small to non-existent for the treated individual. There are no negative effects for the patient that we know of other than the side effects of the drugs, which are reportedly few for the latest ARVs. I am not a medical specialist, but I also worry about adherence problems stemming from the fact that otherwise healthy individuals living with HIV have lower incentives to take their drugs regularly. Would this lead to a problem of drug resistant strains of HIV emerging?

In the end, we are faced with a treatment that accrues no benefits to the individual herself/himself, but confers great benefits to their regular sexual partners. The NYT article (linked above) asked Dr. Anthony S. Fauci, head of the National Institute of Allergy and Infectious Diseases whether it could be considered immoral for a doctor to accede to a patient’s request to delay starting combination drugs: “I’m not going to go there. I’m not going to say it’s immoral. But there is more and more data showing the advantages of starting as early as you can.”

This is quite a statement: there is obviously no way to force people to take drugs that they don’t want to take. Providing the best available information and offering the immediate treatment option (when it is available to the country and/or the patient) is the best we can hope for. Of course, we first need to know who is infected (and conducting ‘test and treat’ for just couples is not going to have the full desired effect on the trajectory of the epidemic), so we need to test everyone first.

Of course, the cost-effectiveness of ‘test and treat’ is the biggest barrier to the immediate change in policy. It is going to be difficult to argue that we should treat healthy people with drugs that many others who actually need them now cannot get. Even if there were enough drugs available to all people with CD4+ counts below 350, it is still not clear that a policy of ‘test and treat’ would be cost-effective. The implied annual incidence rate that I calculated is 0.5% in the deferred treatment arm. Over a 10-year period (the difference in years of treatment I am guessing there would be between the average immediate and deferred treatment), treating 100 people would prevent 5 primary infections. At a current estimated treatment cost of $880 per year under PEPFAR, it would cost $176,000 to prevent a primary infection, which is much more expensive than treating an infected person for their lifetime.

It is great news that we now know ARVs can prevent transmission. This could lead to other discoveries that would contribute to winning the fight against HIV/AIDS for good. However, given the ethical issues associated with and the practical infeasibility of ‘test and treat’, we should not expect substantial changes to HIV prevention strategy based on this study for the time being.



Berk Ozler

Lead Economist, Development Research Group, World Bank

Join the Conversation

Berk Özler
June 02, 2011

One of the benefits of blogging is that people point you to other sources of information you had not yet read. Mead Over over at CGD had written about this on May 17, which generated a discussion among the big names in the HIV prevention field, such as Epstein, Halperin, Shelton among others. You can see it here:…



Brett Keller
June 02, 2011

I think you make a lot of good points. One small quibble (and also not having read the papers in question) regarding what you say about health impacts for the people on treatment: "The immediate treatment reduced the incidence of extrapulmonary TB (i.e. TB in organs other than the lungs), but this incidence was quite small (17 in the deferred treatment group and 3 in the immediate treatment group over the six year study period). Furthermore, there was no difference in mortality between the two groups (13 vs. 10)."

True, there was no mortality in this study, but it looks to me like it wasn't really powered for that. A difference of 13 vs. 10 deaths isn't significant in this study but given enough time (ie, if it weren't stopped early) maybe it would have been. A study designed to look at health benefit for the people on treatment would probably look quite different anyway. You wouldn't expect very high mortality for people at that stage of HIV infection anyway (and I wonder how many of those 23 deaths were directly attributable to HIV? probably some but not all.) so it's quite conceivable that -- had the study been designed to be powered to see that -- there might actually be health benefits too.

In other words, I agree the study doesn't show health benefits, but that doesn't mean they're not there -- it simply doesn't appear to have been designed to show them in this timeframe.

Berk Özler
June 02, 2011

Fair point, and one I considered as I wrote. But, the quote I cite from the NIH release says that the study WAS designed to show benefits for the HIV-infected individuals within the proposed 10-year time frame. Still, it's hard to derive conclusions with just a news release as opposed to a published study. There could also be productivity gains from treating people with high levels of CD4+ counts, but we simply don't yet know.

However, even if there was an effect, we then have to think about whether we need to spend $1 million (1000 people at $1,000/year) to prevent 3 TB infections. Presumably there are cheaper ways to prevent TB and deaths. The point that seems lost by the staunch 'TasP' advocates is the marginal effectiveness of the additional $1 reallocated to ARVs compared with all of its other uses in improving welfare.

Matt Williams
June 14, 2011

It is a great shame that the study didn't look at the high risk groups mostly likely to transmit to more than 1 person. If this was the men who have sex with men British Columbia cohort and "a "test and treat" approach was taken with the 40 odd % who transmitted HIV to a median 17 people (see posters at 2011 Retrovirus conference), and treatment worked to prevent these new cases, suddenly the cost effectiveness looks much more favourable.

But these kind of studies are never done withy gusto with high risk groups.

- matt