This is an interesting point about sampling, but I think it's also somewhat moderated by the fact that gay men also get tested much more frequently for HIV and other STIs.

Put another way: half of all new HIV diagnoses are in heterosexuals, who as a population are significantly less likely to be tested for HIV to begin with.

I'm the US and not the UK, but I think the last time I was tested for HIV was when I was in college. When I told doctors that I was straight, they ordered tests for other STIs, but never HIV.

But at most there would be a delay of a few years in diagnosis if they are more prompt about it. No one goes too long with an untreated HIV infection before symptoms become noticeable

> No one goes too long with an untreated HIV infection before symptoms become noticeable

It's not that simple: The chronic phase of HIV is frequently asymptomatic, and frequently lasts over a decade[1]. Even the AIDS phase can last for years before an opportunistic infection causes severe enough illness.

The initial (acute) phase has flu-like symptoms that people typically ignore.

[1]: https://hivinfo.nih.gov/understanding-hiv/fact-sheets/stages...

Granting a decade long delay in diagnosis of all heterosexual HIV+ cases vs instantaneous diagnosis of homosexual men, the disparity still doesn't explain much - unless there is orders of magnitude shifts in the infection rates among heterosexual people over that time period.

> unless there is orders of magnitude shifts in the infection rates among heterosexual people over that time period.

There are two forces at play: decreasing infection rates among gay men, and increasing infection rates among straight men. Both aspects of the curve are "doing work" to change the statistics.

But yes, there are significant shifts that explain these changes: increases in IV drug use, a decline in condom use among straight sex partners due to readily available alternative prophylactics, and an overall increased emphasis on testing the entire population for HIV/AIDS.

> There are two forces at play: decreasing infection rates among gay men, and increasing infection rates among straight men. Both aspects of the curve are "doing work" to change the statistics.

You may have missed the point — only one of those is modulated by (relative) delay in heterosexual diagnoses post-infection.

The article makes this point and the headline which several have objected to does not mention rates of infection in the two populations. New infection volumes are also affected by the population infection rate in the respective groups and the time to diagnose and treat. If you wanted to be a bore, you could model the number of expected encounters with the virus in each population and figure out the relative transmission rates too.

The take away from the article is that your risk as a man who does not have sex with men is nonzero and that regular testing is important to personal health outcomes and to keeping infection rates low, and that cheap and effective preventative measures are available, including PReP and PEP. After thirty-odd years of dealing with it, MSM be aware.

I think you main point stands, but way more men have gay sex than are willing to self-report as gay or bi in a government survey

That just increases the incidence relative to straight men.

The bigger question is whether they're reporting it as they join infection statistics.

Is that actually true? The last data I saw on this indicated that there are a huge number of people falsely claiming to be bisexual. That is, relationships amongst the population claiming to be bi are overwhelmingly hetero. The young especially think there are (real or perceived) social benefits from claiming to be gay or bisexual even when they're not.

Do you have the data for this? My experiences have been precisely the opposite. I have spoke to many bisexual men married to women on Grindr who were looking to cheat on their wives (or fantasised about doing it).

My experiences point to a lot of bisexual men living in the closet and are ashamed of what they are. I think guys in that situation typically want the heterosexual lifestyle of kids and marriage, but the thrill of sex with a guy on the side. There have been very, very few guys I've seen on there who are married to a woman who are able to openly explore sex with men with their wife's consent.

I can't really comment on women who are bisexual however.

Were those men genuinely bisexual, I wonder? Gay men pretending to be straight and ending up in family marriages is a very long standing issue, or was at least, I thought the "lavender marriage" was dying out.

https://slate.com/human-interest/2016/05/over-80-percent-of-...

"The massive 2013 Pew Research LGBT Survey found 84 percent of self-identified bisexuals in committed relationships have a partner of the opposite sex, while only 9 percent are in same-sex relationships"

101 times?

> "Around 1.7% of males identified themselves as gay or lesbian in 2017"

Another page on the same website says that "Between 2018 and 2019, the number of men identifying as LGB increased from 2.5% to 2.9%", so maybe 0.0145 is a better number than 0.0085?

https://www.ons.gov.uk/peoplepopulationandcommunity/cultural...

Yes, the ones you found are more up-to-date, and the more recent numbers probably suffer less from the closeted people/stigma bias. Also, as cuteboy19 pointed out, my number excludes the bisexuals. Unfortunately, I can't edit my comment.

I still think the overall point remains valid though.

I'm confused, why you are both halving that statistic?

Because what we have is:

  # of gays / # of males

but what we want is:

  # of gays / the whole population

It's so disproportional that a small number "wrongly" associated with the larger cohort would mean in reality it's far more skewed.

Your comment is quite misleading, I think.

It reads as if you expose the article to make a wrong statement about incident rates. The article however however is purely right in this regard. It does not talk about comparison of incident rates at all. Its first sentence already clearly talks about numbers of HIV diagnoses.

If your comment would be like "Don't read the title in a wrong way. HIV is still much more common among lgb men than among heterosexual men and women. Reason is that lgb men are around 2.3% of all men compared to heterosexual mean and women who are 93.2% of the population" it might be helpful.

> It reads as if you expose the article to make a wrong statement about incident rates.

No? But you could read it as "exposing" the fact that it mentions incident rates only vaguely [1], and that the title is chosen strategically. (The article looks seems to be part of a public health campaign to encourage heterosexuals to test for HIV, and that's probably part of the reason.)

There were a couple of comments which raised valid points about my rough estimate though.

[1]: It only says "Gay and bisexual men are still more impacted by HIV relative to population size", but the question of by how much seems to be left an exercise for the reader.

The concept of ratios/proportionality is like a grade 9 math concept. I haven't seen the article, but if this is the case, the title is grossly misleading.

True, the title does feel misleading, but I don't know how they could make it less misleading while conveying the same information. At least they mention in the article that "gay and bisexual men are still more impacted by HIV relative to population size".

I didn't see the headline as misleading because it did not mention individual risk.

Would appending a clause mentioning that gay men remain at greater risk improve the headline? Or is it possible that this story should be presented in a very different way or not at all?

It's misleading because the first thing that comes to mind is "wow, the non-gay community is worse than the gay community when it comes to HIV!", which is completely wrong. It would be right if the proportions of both communities were the same, but they are 100x different.

Even if the estimates are off by a lot, your point is still pretty clear. This is that bias talked about with covid, right? What’s the name again? Where “more vaccinated people have covid” doesn’t mean what some suggest when most people are vaccinated.

Base rate neglect

Those covid stats[0] are age standardized and per 100k people. There are still other differences between the populations that could explain it (for example the vaccinated are probably getting tested more often), but it's not as clear as this.

0 - https://publichealthscotland.scot/publications/covid-19-stat... page 33 of the full report

Simpson’s Paradox?

I don't understand why you would use .85% if 1.7% is the correct (but possibly underestimate) number. Additionally, gay/bi is included in the same bracket so the real number would probably be 1-.932=0.068 instead of 0.0085

.932/.068= 13.7

Even though this is still pretty higher, it's probably much more realistic

> I don't understand why you would use .85% if 1.7% is the correct (but possibly underestimate) number.

I think 1.7% is out of the males, and males are about half the general population.

> Additionally, gay/bi is included in the same bracket so the real number would probably be 1-.932=0.068 instead of 0.0085

I think the 1.7% number also includes bisexual males, but I'm not completely sure. (The title of Figure 1 of the section in the source is "Lesbian, gay or bisexual")

My overall point isn't about the exact numbers though, and I agree that the government survey number is probably an underestimate.

Sorry, your point about exclusion of bisexuals was actually totally correct. Also, creata found more up-to-date numbers.

Wow, HIV spreads poorly through penile-vaginal intercourse.

…or heterosexuals are a hundred times less likely to engage in risky behaviour.