Wednesday, June 22, 2005

I received these replies about the US Army's HIV stats and share them with you because I believe they shed important light on the stats and how the military gathers them, who is getting tested and why.

The first message is from Bob Roehr, a syndicated health care and gay issues reporter based in Washington, DC. The second one is from Sean Strub, veteran AIDS activist and publisher of POZ.
^^^



From: Bob Roehr

June 22, 2005



With regard to the military’s HIV stats, once you look at the details, there is less here than initially meets the eye. There are two principle flaws in the paper (by a candidate for a masters' degree) that would at least have been acknowledged in the discussion section of a peer reviewed publication, but are not here.



The first is that the high starting point of infections represents the backlog of infections that occurred over the course of activity (sexual, injection drug use, transfusions) over more than a decade, prior to the initiation of testing. As the paper points out, active duty personnel undergo periodic and regular HIV-1 testing every 2-5 years. So the real starting point for analysis should be when that backlog had been worked through—somewhere between 1988 and 1991—in which case the curve would look very different indeed. The chart that demonstrates this best is the one comparing incidence rates of those service members with more/less than three years on active duty (page 15).



The second major flaw is that it assumes that the pool of recruits entering the military, and hence being tested, remained exactly the same throughout the period of analysis. That is clearly not the case. Potential recruits are informed that they will be given a complete pre-induction physical and that being HIV-positive is one of the grounds for exclusion from induction. This dataset does not include testing numbers from pre-induction physicals, which would be a better measure of the true rate of infections occurring within the community.



Furthermore, there is anecdotal evidence (there may even be studies, I haven’t looked) that potential recruits, either on their own or on advice of recruiters, take their own HIV test before the pre-induction physical if they believe they might be at risk for testing positive. It is better to find out such information under conditions that they have some control over than at the physical. Bottom line, the pool of those being sworn into the service has changed over time, and this data reflects that.



If one uses a starting point of about 1990 (once the backlog has been worked through), then what one gets is a picture of those who seroconvert while they are in the military. One would be tempted to use this as a marker of gays, however, it also includes injection drug use, and I suspect given the population and popularity of tattoos, a fair number of folks who were infected through improperly cleaned tattoo needles.



The data showing health care personnel as those most likely to become infected with HIV would reinforce the view that gays tend to serve disproportionately in that field. But again, there is a statistical reason to be wary and not read too much into that—those career identifications have the largest confidence intervals and hence are the most suspect for accuracy.



- -



From: Sean Strub

June 22, 2005



This is great news of course, but I wonder how much of it is reverse self-selection as more people know their HIV status or their potential risk for HIV and check before entering the military to make sure they are negative, knowing they will get tested once in service. I suspect the military stats are not as reflective of the broader population today as they were earlier in the epidemic.

No comments: