Quantified Risks of Gay Male Sex

by pianoforte6115 min read18th Aug 201440 comments


MedicineSex & Gender

If you are a gay male then you’ve probably worried at one point about sexually transmitted diseases. Indeed men who have sex with men have some of the highest prevalence of many of these diseases. And if you’re not a gay male, you’ve probably still thought about STDs at one point. But how much should you worry? There are many organizations and resources that will tell you to wear a condom, but very few will tell you the relative risks of wearing a condom vs not. I’d like to provide a concise summary of the risks associated with gay male sex and the extent to which these risks can be reduced. (See Mark Manson’s guide for a similar resources for heterosexual sex.). I will do so by first giving some information about each disease, including its prevalence among gay men. Most of this data will come from the US, but the US actually has an unusually high prevalence for many diseases. Certainly HIV is much less common in many parts of Europe. I will end with a case study of HIV, which will include an analysis of the probabilities of transmission broken down by the nature of sex act and a discussion of risk reduction techniques.

When dealing with risks associated with sex, there are few relevant parameters. The most common is the prevalence – the proportion of people in the population that have the disease. Since you can only get a disease from someone who has it, the prevalence is arguably the most important statistic. There are two more relevant statistics – the per act infectivity (the chance of contracting the disease after having sex once) and the per partner infectivity (the chance of contracting the disease after having sex with one partner for the duration of the relationship). As it turns out the latter two probabilities are very difficult to calculate. I only obtained those values for for HIV. It is especially difficult to determine per act risks for specific types of sex acts since many MSM engage in a variety of acts with multiple partners. Nevertheless estimates do exist and will explored in detail in the HIV case study section.


Prevalence: Between 13 - 28%. My guess is about 13%.

The most infamous of the STDs. There is no cure but it can be managed with anti-retroviral therapy. A commonly reported statistic is that 19% of MSM (men who have sex with men) in the US are HIV positive (1). For black MSM, this number was 28% and for white MSM this number was 16%. This is likely an overestimate, however, since the sample used was gay men who frequent bars and clubs. My estimate of 13% comes from CDC's total HIV prevalence in gay men of 590,000 (2) and their data suggesting that MSM comprise 2.9% of men in the US (3).



Prevalence: Between 9% and 15% in the US

This disease affects the throat and the genitals but it is treatable with antibiotics. The CDC estimates 15.5% prevalence (4). However, this is likely an overestimate since the sample used was gay men in health clinics. Another sample (in San Francisco health clinics) had a pharyngeal gonorrhea prevalence of 9% (5).



Prevalence: 0.825% in the US

 My estimate was calculated in the same manner as my estimate for HIV. I used the CDC's data (6). Syphilis is transmittable by oral and anal sex (7) and causes genital sores that may look harmless at first (8). Syphilis is curable with penicillin however the presence of sores increases the infectivity of HIV.


Herpes (HSV-1 and HSV-2)

Prevalence: HSV-2 - 18.4% (9); HSV-1 - ~75% based on Australian data  (10)

This disease is mostly asymptomatic and can be transmitted through oral or anal sex. Sometimes sores will appear and they will usually go away with time. For the same reason as syphilis, herpes can increase the chance of transmitting HIV. The estimate for HSV-1 is probably too high. Snowball sampling was used and most of the men recruited were heavily involved in organizations for gay men and were sexually active in the past 6 months. Also half of them reported unprotected anal sex in the past six months. The HSV-2 sample came from a random sample of US households (11).



Prevalence: Rectal - 0.5% - 2.3% ; Pharyngeal - 3.0 - 10.5% (12)

 Like herpes, it is often asymptomatic - perhaps as low as 10% of infected men report symptoms. It is curable with antibiotics.



Prevalence: 47.2% (13)

 This disease is incurable (though a vaccine exists for men and women) but usually asymptomatic. It is capable of causing cancers of the penis, throat and anus. Oddly there are no common tests for HPV in part because there are many strains (over 100) most of which are relatively harmless. Sometimes it goes away on its own (14). The prevalence rate was oddly difficult to find, the number I cited came from a sample of men from Brazil, Mexico and the US.


Case Study of HIV transmission; risks and strategies for reducing risk

 IMPORTANT: None of the following figures should be generalized to other diseases. Many of these numbers are not even the same order of magnitude as the numbers for other diseases. For example, HIV is especially difficult to transmit via oral sex, but Herpes can very easily be transmitted.

Unprotected Oral Sex per-act risk (with a positive partner or partner of unknown serostatus):

Non-zero but very small. Best guess .03% without condom (15)

 Unprotected Anal sex per-act risk (with positive partner): 

Receptive: 0.82% - 1.4% (16) (17)

                          Insertive Circumcised: 0.11% (18)

         Insertive Uncircumcised: 0.62% (18)

 Protected Anal sex per-act risk (with positive partner):  

  Estimates range from 2 times lower to twenty times lower (16)  (19) and the risk is highly dependent on the slippage and   breakage rate.

Contracting HIV from oral sex is very rare. In one study, 67 men reported performing oral sex on at least one HIV positive partner and none were infected (20). However, transmission is possible (15). Because instances of oral transmission of HIV are so rare, the risk is hard to calculate so should be taken with a grain of salt. The number cited was obtained from a group of individuals that were either HIV positive or high risk for HIV. The per act-risk with a positive partner is therefore probably somewhat higher.

 Note that different HIV positive men have different levels of infectivity hence the wide range of values for per-act probability of transmission. Some men with high viral loads (the amount of HIV in the blood) may have an infectivity of greater than 10% per unprotected anal sex act (17).


Risk reducing strategies

 Choosing sex acts that have a lower transmission rate (oral sex, protected insertive anal sex, non-insertive) is one way to reduce risk. Monogamy, testing, antiretroviral therapy, PEP and PrEP are five other ways.


Testing Your partner/ Monogamy

 If your partner tests negative then they are very unlikely to have HIV. There is a 0.047% chance of being HIV positive if they tested negative using a blood test and a 0.29% chance of being HIV positive if they tested negative using an oral test. If they did further tests then the chance is even lower. (See the section after the next paragraph for how these numbers were calculated).

 So if your partner tests negative, the real danger is not the test giving an incorrect result. The danger is that your partner was exposed to HIV before the test, but his body had not started to make antibodies yet. Since this can take weeks or months, it is possible for your partner who tested negative to still have HIV even if you are both completely monogamous.


For tests, the sensitivity - the probability that an HIV positive person will test positive - is 99.68% for blood tests (21), 98.03% with oral tests. The specificity - the probability that an HIV negative person will test negative - is 99.74% for oral tests and 99.91% for blood tests. Hence the probability that a person who tested negative will actually be positive is:

 P(Positive | tested negative) = P(Positive)*(1-sensitivity)/(P(Negative)*specificity + P(Positive)*(1-sensitivity)) = 0.047% for blood test, 0.29% for oral test

 Where P(Positive) = Prevalence of HIV, I estimated this to be 13%.

 However, according to a writer for About.com (22) - a doctor who works with HIV - there are often multiple tests which drive the sensitivity up to 99.997%.


Home Testing

Oraquick is an HIV test that you can purchase online and do yourself at home. It costs $39.99 for one kit. The sensitivity is 93.64%, the specificity is 99.87% (23). The probability that someone who tested negative will actually be HIV positive is 0.94%. - assuming a 13% prevalence for HIV. The same danger mentioned above applies - if the infection occurred recently the test would not detect it.


 Anti-Retroviral therapy

 Highly active anti-retroviral therapy (HAART), when successful, can reduce the viral load – the amount of HIV in the blood - to low or undetectable levels. Baggaley et. al (17) reports that in heterosexual couples, there have been some models relating viral load to infectivity. She applies these models to MSM and reports that the per-act risk for unprotected anal sex with a positive partner should be 0.061%. However, she notes that different models produce very different results thus this number should be taken with a grain of salt.


 Post-Exposure Prophylaxis (PEP)

 A last resort if you think you were exposed to HIV is to undergo post-exposure prophylaxis within 72 hours. Antiretroviral drugs are taken for about a month in the hopes of preventing the HIV from infecting any cells. In one case controlled study some health care workers who were exposed to HIV were given PEP and some were not, (this was not under the control of the experimenters). Workers that contracted HIV were less likely to have been given PEP with an odds ratio of 0.19 (24). I don’t know whether PEP is equally effective at mitigating risk from other sources of exposure.


 Pre-Exposure Prophylaxis (PrEP)

 This is a relatively new risk reduction strategy. Instead of taking anti-retroviral drugs after exposure, you take anti-retroviral drugs every day in order to prevent HIV infection. I could not find a per-act risk, but in a randomized controlled trial, MSM who took PrEP were less likely to become infected with HIV than men who did not (relative reduction  - 41%). The average number of sex partners was 18. For men who were more consistent and had a 90% adherence rate, the relative reduction was better - 73%. (25) (26).

1: http://www.cdc.gov/mmwr/preview/mmwrhtml/mm5937a2.htm?s_cid=mm5937a2_w

2: http://www.cdc.gov/hiv/statistics/basics/ataglance.html

3: http://www.cdc.gov/nchs/data/ad/ad362.pdf

4: http://www.cdc.gov/std/stats10/msm.htm

5: http://cid.oxfordjournals.org/content/41/1/67.short

6: http://www.cdc.gov/std/syphilis/STDFact-MSM-Syphilis.htm

7: http://www.cdc.gov/mmwr/preview/mmwrhtml/mm5341a2.htm

8: http://www.cdc.gov/std/syphilis/stdfact-syphilis.htm

9: http://journals.lww.com/stdjournal/Abstract/2010/06000/Men_Who_Have_Sex_With_Men_in_the_United_States_.13.aspx

10: http://jid.oxfordjournals.org/content/194/5/561.full

11: http://www.nber.org/nhanes/nhanes-III/docs/nchs/manuals/planop.pdf

12: http://www.cdc.gov/std/chlamydia/STDFact-Chlamydia-detailed.htm

13: http://jid.oxfordjournals.org/content/203/1/49.short

14: http://www.cdc.gov/std/hpv/stdfact-hpv-and-men.htm

15: http://journals.lww.com/aidsonline/pages/articleviewer.aspx?year=1998&issue=16000&article=00004&type=fulltext#P80

16: http://aje.oxfordjournals.org/content/150/3/306.short

17: http://ije.oxfordjournals.org/content/early/2010/04/20/ije.dyq057.full

18: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2852627/





21: http://www.thelancet.com/journals/laninf/article/PIIS1473-3099%2811%2970368-1/abstract



23: http://www.ncbi.nlm.nih.gov/pubmed/18824617

24: http://onlinelibrary.wiley.com/doi/10.1002/14651858.CD002835.pub3/abstract

25: http://www.nejm.org/doi/full/10.1056/Nejmoa1011205#t=articleResults

26: http://www.cmaj.ca/content/184/10/1153.short


40 comments, sorted by Highlighting new comments since Today at 8:46 AM
New Comment

While it's probably justified to correct for the sampling bias in prevalence statistics, it's worth pointing out that sexual partners are not sampled uniformly: the prevalence of a given STD will potentially be higher in the population of likely partners than in the general population.

That's a good point, and probably applies to Mark Manson's guide too. It's similar to the well-known point that your friends are probably more popular than you are, because popular people have more friends.

and of course this is another case of 'just because you hired the top 1% of the CVs you got, doesn't mean that those you hired are in the top 1% of programmers'. Less good programmers are more often looking for a job.

Is there a name for this pattern?

Adverse selection.

[-][anonymous]7y 0

I think it's called "selection bias", though most people don't realize just how pervasive it is. Maybe we need subcategories. Another example is that the the neighboring lane in a traffic jam often moves faster than the lane you're in, because higher speed => larger gaps between cars => fewer cars => higher chance of finding yourself in the slow lane. (I have no idea if that reasoning is correct, but it sounds fun!)

[This comment is no longer endorsed by its author]Reply

While teaching a game theory class at Smith College I used an example that made mention of how blood banks didn't want donations from gay men. A student asked me why blood banks would do this and I said it was probably due to not being able to completely tell from screening if a donor has AIDS. My students then actually began debating among themselves whether gay men were more likely to have AIDS than heterosexual men were.

I once debated with myself whether I should donate blood given that I'd had sex with men before, but whom I was sure were HIV-negative. I did a quick Fermi estimate looking at the amount of contaminated blood samples the blood bank could expect nationally, first with only heterosexual donors, and then with heterosexual + homosexual donors, given that each blood sample underwent the most accurate HIV tests. The results were pretty staggering (order of magnitude difference).

That convinced me that the proscription was there for a very good reason and that I shouldn't violate it.

Yeah, the propaganda on this issue seems to have been quite effective since before reading this I had no idea the problem was (still) so severe.

While I recognize the true HIV prevalence is probably higher than most people would guess, what propaganda are you referring to?

I was young in the 80's, but my impression is that HIV/AIDS was considered a pretty gay-specific thing at first. Later there was more media pushing the idea that it can affect anyone - for example, one of my schools had a straight woman with HIV visit to tell us about it. While this was presumably well-meaning and may have even had good effects in terms of encouraging safety, it did lead me to a quite skewed perspective of the relative risks (I was still aware that it was more prevalent in gays, but not by how much).

"HIV can affect anyone" is far better than "HIV is a gay thing" along several criteria.*

One: factual - it can indeed affect anyone. And the difference in prevalence varies form country to country. In parts of Southern Africa the prevalence is about the same as the general population.

Two: Instrumental. If those involved in public health used HIV as leverage to get everyone to wear condoms more often, then I applaud them. Many lives were saved, and many infections prevented.

Three: Political. I'd be very curious to know what proportion of people reporting that HIV affect gay men more are doing so out of genuine concern rather than as code for "Those gays are dirty and disgusting and deserve to die/be ridiculed".

  • "Some groups are at higher risk for HIV than others, and the transmission rates depend on the nature of the sex act" is the most correct, but it is unwieldy for a public health effort.

I'm female, but I had no idea until after I'd had sex with bisexual men that the HIV risk was much higher than from sleeping with straight men. I used condoms anyway, but I was pretty shocked to learn about it. I still date bi men*, but I'm much stricter about making sure they've had STI tests than I used to be.

*My main social group are the UK bi/poly community, so two out of three of the men I've dated in the last few years have been bi.


Prevalence: Between 13 - 28%. My guess is about 13%.

Wow. That's at least an order of magnitude higher than I would have guessed. I even thought you might have accidentally omitted a decimal point at first.

I had the same thought.

Indeed the prevalence in the general US population is 50x smaller: http://www.cdc.gov/hiv/statistics/basics/
I knew that MSM were a high risk group, but I didn't realize the risk was that high.

As a heterosexual I'm not your target audience, but I voted this up for being a well-compiled and useful (to its audience) bit of research.

Whenever you speak of the prevalence I think it would you got to speak about the exact demographic it comes from. The US might have slightly different values than Europe.

Not just slightly, the HIV prevalence in the UK among MSM is less than 5% using demographic data here and here. And the total HIV prevalence in the UK is relatively high for Europe see here.

I should probably mention this in the OP. I live in the US, and I couldn't exactly obtain a prevalence rate for every single country and every disease. But the US is an outlier with respect to many diseases.

Quick thanks to Omid who came up with the idea for this post and gave me several suggestions about the content of the post.

This is great data. Now, it would be great to add some interpretation, either in a conclusion paragraph, or in a comment.

My interpretation is: "The prevalence of STDs in MSM is high, including 13% for HIV, especially considering their substantial impact on quality of life. If you're having anal intercourse, even if you're using a condom consistently, the associated risk is orders of magnitude higher than the risk from unprotected oral sex. Choice of sex act and testing your partner seem like relatively reliable ways to reduce risk"

[-][anonymous]6y 2

I want to have sex with this girl I just started dating without a condom, and pleasure her orally. I haven't done either of those things before, but I have a plan... However Mansons guide doesn't seem very helpful for me cause I think she'd be a virgin or pretty inexperienced.

The risk of aquiring HIV from sexual intercourse is actually super low. It's effectively a non-issue, particularly if you are having hetereosexual sex that's not regularly with a seropositive partner. Effectively, you're safe, even if you bang hookers like me :) It's good to get checked up, and you don't even need to give your real name: When I attended my cities sexual health clinic I used the anglicisation of my first name and only the first name of my last name, as my full name. They are happy for people to do this.

a key factor in the spread of the disease is the viral load. During the chronic infection stage, the load is low enough that it requires several contacts between two individuals over a short period of time for the infection to spread.

But what of the acute stage of the disease when the viral load is greater and the chances of infection much higher? Again Rocha and co say the disease cannot spread through the network, even if the infection rate is 100 per cent.

The reason is the time between sexual contacts. Rocha and co say there is an epidemic threshold in which the disease spreads if it is infectious for more than about a month. However, the acute stage of HIV appears to last for only for a couple of weeks. This just isn’t long enough to spread given the rate of contact that occurs in this network.

Fairly sure it's confounding by drug use with poverty, being a man who does anal, or high risk personality that explains sex workers having high rates of HIV, rather than their occupation.

Regarding HIV, what about Truveda?

Its the same as pre exposure prophylaxis.

Ah, sorry, I must have missed that due to reading too quickly.