STI Risks: Questions and Modeling (3/4)

In the last post, I explored the role of feelings in managing STI risks. I talked about why even getting testing can have complex feelings associated and shared some of the experiences that helped me learn what my feelings are.

This post is the most challenging to write because I have less to say here that's generally applicable. As I discussed at the beginning of this series, the process of thinking through things yourself and coming to grips with your own feelings is critical; I can't give someone else the answers. There are many assumptions that go into a model of risk probabilities; even more so than feelings such a model will be individual. I have shared my specific assumptions and specific analysis with those closest to me--those who need to understand what risks I'm taking. If you'd like to see my specific conclusions and model, I do have them written up, and I'd be happy to talk about whether I feel comfortable sharing. If I know you well enough to have confidence that you'd take it as input into your own process and that you can think about statistics enough to understand the many limitations, then I would almost certainly be comfortable with that.

Once I got to a point where I thought I understood how I'd react if I did have to face an infection, I started trying to model activities I might engage in. For me, the final result was a spreadsheet representing a numerical risk model. On one tab I can enter assumptions including things like how sexually active I expect to be and assumptions about the population I fuck in. Another tab gives a variety of statistics about risks under those assumptions. In my first post on this topic, I pointed to an article discussing why this sort of modeling is dangerous. Certainly, there are a lot of limitations to any model, and a model is only as good as its input. However, I think this helps with a number of questions I have.

First, I wanted to see if my practices were consistent. For example I'd hate to be spending a lot of effort (or giving up opportunities) to avoid a consequence in one way, but run a relatively high risk of the same consequence because of something else I was doing. I'd also be suspicious if I found myself spending a lot of effort to avoid a relatively low-cost consequence while vulnerable to a high cost consequence.

I wanted to be able to understand the relative risks of various activities related to the threats I had identified. For example I've always wanted to develop intuition for whether unprotected oral sex was more or less dangerous than protected vaginal sex with regard to threats that matter to me. Ultimately I wanted to develop a set of practices for my behavior. My hope is that if I follow these practices, I will feel that I've treated myself with care, regardless of the actual outcome. To do that I want to have an intuitive feel for how likely risks are.

Understanding risk numbers at an intuitive level is hard. One in ten thousand seems like a low probability. Sometimes it is. If there were advances so that our probability of death from cancer was one in ten thousand over an entire lifetime, the medical community would be filled with joy. Even so, tens of thousands of people alive today in the United States would still be expected to die of cancer. However, if my probability of getting shot on the way to the store was one in ten thousand, I'd know many people killed on the way to the store, and shopping death would be one of the more likely ways I'd die.


Here are some things I pondered when putting together assumptions for the model. As I mentioned in the first post, one thing I was looking for while researching was statistics on the chance of transferring an infection through a particular activity.

Infection Frequencies

The way I'd heard a lot of people talking about safe sex approaches, I'd assumed that the biggest factor in risk mitigation was what protective strategies you used. I've come to the conclusion that while safe sex approaches such as condoms are certainly valuable, I end up caring a lot more about how common infections are than I expected. Even perfect use of condoms doesn't always work, and other methods and drugs have failure rates high enough that I need to consider them. (I admit needing to look more at the Truvada studies, although I'd imagine a lot of participants combined Truvada with condoms.)

If I'm going to consider infection rates in the community, I need to consider the affects of testing. For myself, I'm comfortable insisting on recent tests before engaging in a lot of sexual activities. So, the question becomes how likely are infections in the set of people who claim to have recent enough tests and claim not to be infected? A lot of factors can influence this. People might lie about their testing status.

People might be infected since the last time they were tested.

With HSV there are additional complications. Absent an outbreak, tests can tell you what strain of HSV you have, but not where. As I've mentioned, I don't buy the presumption that HSV type 2 is genital and HSV type 1 is oral. Making that assumption is an example of one of those cases where I'd be throwing away a lot of opportunities while still exposed to the risk. So, I need to make assumptions about how I decide whether I am at risk of genital HSV when someone tests positive for some strain of HSV but has not had recent outbreaks.

I'll flag one unpleasant truth about population statistics. Race (and presumably economic class) matters in infection rates. I have no idea what to do about that; for myself I've decided to ignore the issue, but that's an area where I can see difficult decisions.

Unless I assume very high probabilities that someone will lie, testing significantly reduces infection rates in the population of people I'll interact with sexually. For me that had more of an effect than I expected.

Number of New Partners

I'll discuss the general question of independence of events in a moment. However, there's one specific case I'd like to call out. What fraction of my interactions are with new partners matters a lot. If I have a hundred interactions in a year with a hundred new partners, then it's reasonable to assume that the probability that my partner is infected is independent for each of those hundred interactions. However, if I have a hundred interactions spread across 5 partners, then my overall probability of getting infected will be different because the conditional probability of getting a particular infection given a partner who does not transmit that disease is zero, and my interactions are not independent with regard to that conditional probability.

Independence of Events

More generally, when looking at probabilities you need to consider how events are related. There are all sorts of ways things might be related:

Acute Infection Period, Viral Load, and Asymptomatic Shedding

There are a lot of reasons someone might be more or less infectious than average. At the beginning of a viral infection, someone is likely to be much more infectious.

With HIV, if someone's viral load is low, research suggests it is a lot harder to spread the virus.

With HSV, we believe it is much harder to spread the virus when it is in a dormant period and not being shed from the skin.

You'll be making some assumption or another about this. Perhaps you'll decide that this can all be factored into averages. Perhaps you'll try to explicitly account for it.

How long does it take to detect an Infection?

If you're like me and you would choose different behavior when facing a partner who had an infection, you probably want to think about how long it would take to detect an infection. Especially given acute infection periods, it probably matters what sort of testing schedule you're looking at for yourself and partners.

HSV and the Yearly Risk Rates

Studies I've seen discussing the risk of contracting genital HSV give results in your percentage chance per year of contracting the infection with an infected sexual partner. They don't give a chance of contracting the infection from a particular sexual interaction. That's problematic because it makes it hard to adjust for frequency of sexual activity or number of partners. I ended up having to make the assumption that the chance of getting genital HSV is independent of the number of partners I had. That assumption is clearly false.

Unfortunately it's quite difficult to do any better than that. HSV spreading is very dependent on asymptomatic shedding. That varies by person and throughout the year. So it seems like it would be quite difficult to get a per-fuck chance that meant anything. This is a great illustration that a lot of the assumptions you make will be necessitated by incomplete information and will decrease the accuracy of any model.

What Can I Use it For

After looking at all those assumptions, it's easy to see that the value of any model is going to be limited. However, I've found it very helpful for reasoning about order-of-magnitude risks. I was able to compare risks of getting diseases to other health challenges I will face as aging and based on how I felt about those potential challenges answer questions like "Would a 1% lifetime risk of that be acceptable? What about a 10% risk? Do I need to keep things under one in a thousand?" Thinking about how many people I'd know with a condition if a lot of people I knew had a one in a thousand chance or a 1% chance also helped. With that I was able to come up with lifetime risk tolerances I tend to think accurately represent my feelings.

I think the sort of work I'm doing is good enough to let me reason about how far my practices diverge from those lifetime risk tolerances. That's been amazingly useful to me.

It's also been helpful to think about what activities I care most about. I've also been able to reason about how much of my partner's sexual history I need to know and about how comfortable I am interacting with partners who use different practices.

I've also found it helpful to reason about how important I think condoms are and about how comfortable I'd be with multiple partners where I chose not to use protection. I'll say that I think safe sex is another area where we spread fear and shame. I've continued to use condoms for penetration, but I'm a lot less afraid of doing something else than I expected to be. I'm also very strongly defensive of people who choose not to employ safe sex measures. I do think it's worth investing the time to become comfortable having great sex even when you use protection, because that seems like a wonderful skill to have. However, in an attempt to protect those who did want to establish safe sex boundaries, I think we've chosen to shame those who choose differently. I don't think the science supports that shame (like any science possibly could). For myself this is a case where exploring STI risks has helped me be comfortable fighting that shame, while still fighting for people's write to establish their boundaries.

Beyond that, I don't think the modeling work I've done would have much use. I think it would be very dangerous to apply it to find the risk of specific situations rather than general trends. It certainly wouldn't work to apply something this simplistic to estimate how a community would change over time. Even so, it helped me understand my needs and factor medical information into what is a very important set of feelings.

That is of huge value to me.

Partners with Infections

I haven't talked about how I'd approach a partner who had an infection. I've faced that with regard to genital HSV. However, even that is fairly rough and raw. This is an area where I have more growth to do, and where I am uncertain of what to say. Empathy, respect and love are important. As far as the decisions or how to make them? I suspect others involved in this discussion have far more to say than I; I'll listen not speak.

Looking Forward

In the final article in this series, I'd like to talk about communicating about STI risk and precautions. That communication has been one of the most painful areas of interaction as a lover. None of the important relationships in my life have escaped pain from that communication. Even so, I've worked to improve, and learned some things I'd suggest considering along the way.