That HIV study doesn't show what you think it does

1/03/2014 03:08:00 PM
During epidemics, the links within communities are usually more important than the links between them.
I see that several media outlets are unabashedly reporting that most HIV infections in women are transmitted to them by men-who-have-sex-with-men (MSM). These reports are a result of a this new study from New York City. Unfortunately, that's not actually what the study found.

The study was just a survey of a relatively small sample from a handful of gay bars in New York City. That right there suggests that we need to be cautious making conclusions based on this study, because the sample is both not really randomized and not necessarily representative of the population as a whole. Often cost constraints prevent us from collecting these kinds of data any other way, but realize that it is well established that using gay bars as the basis for study recruitment causes selection bias.

But, my point here is not to criticize the data, which is what it is, but to criticize how the media is reporting this particular study's conclusions. Here's the New York Post:
During the early years of the HIV/AIDS epidemic, many women contracted HIV from having sex with men who carried the deadly virus from drug use, particularly sharing needles that had contaminated blood. Or the women abused drugs themselves. But in recent years, the transmission HIV ... related to drug use has plunged as a factor in newly reported cases. That means women are contracting HIV from men who’ve had sex with other men
Is that what that means? Could have fooled me. The evidence they present is that intraveneous drug use accounts for a declining share of HIV infections in women, and their automatic conclusion is that women are getting HIV from the LGBT community. There is no warrant for this claim in the study at all.

The study itself does not do any kind of causal inference statistics at all. Instead, all that the study authors did was report descriptive statistics about the HIV population. Their key finding was that a possible link between the MSM population and straight women (WSM) exists. That is all. They did not find any evidence that straight women were getting HIV from gay and bisexual men, because they never even attempted to test this hypothesis.

I'm upset with this reporting because it leads to statements like these:
“However,” the CDC said, “these women may not have known of their male partners’ sexual activity.”
Let's compare what the CDC actually said to the context in which that the New York Post reported it. The CDC said that women don't necessarily know about their all of their partner's other sexual partners. That's a totally valid point that applies to everyone, everywhere. But the way the Post reports it, you'd think they are saying that women are getting HIV primarily from male partners living secret gay lives they don't know about. It is precisely this fallacy that the right wing has tried to use to tear apart the black community in attempts to pass homophobic laws like Prop 8. Listen: the incidence of HIV is higher among the black community than the white community, and it isn't because all black men are secretly having homosexual affairs. HIV follows much the same epidemiological patterns as other diseases, meaning that it hits the poorer and more marginalized communities the hardest.

I'm not trying to say that women can't get HIV from bisexual men. While this study does not show this is actually happening, it does show that the hypothesis is plausible because such a link does exist, at least in New York City. That's valuable information, I suppose. But it is important for the media to stick to what the science actually says, and not push their own pre-existing stereotypes and narratives. The fact is that even after this study, we have no idea what fraction of HIV infections in women came from gay and bisexual men. Do not pretend that this study answered that question.