Prude women, more HIV?
December 12, 2007 11:07 PM   Subscribe

I read online that AIDS spreads faster in societies where the women are less promiscuous. What is the best way of creating a model which will show me the development and spread of the virus in the case where women are less vs the case where women are more promiscuous?

Short of sitting down to program this, is there any tool that will quickly let me graph the development of the virus? I need something that shows it like a video, with spread from one person to the other.

The constraints are as follow:

Assuming you have a nightclub visited by 10 men and 10 women, if all 10 women are promiscuous, how long would it be till all men are infected? If however only 5 are promiscuous, how long would it take then?

I also need to do things like find out the exact promiscurity level to spread the disease as fast as possible.
posted by markovich to Health & Fitness (28 answers total) 3 users marked this as a favorite
 
I read online that AIDS spreads faster in societies where the women are less promiscuous.

1) I've never heard of such a study, but even if the two are correlated, it doesn't necessarily mean anything. Repeat after me: Correlation does not equal causation

The only way I can envision this being true is if there are other factors that are more important, such as rate of contraceptive use, or number of homosexual relationships (so that the women aren't as big of a factor). If this is the case, then the headline you're referring to is misleading at best.

Modeling the spread of disease is a quite complex task, but you might start by looking for papers on epidemiology.
posted by chrisamiller at 11:14 PM on December 12, 2007


I also need to do things like find out the exact promiscurity level to spread the disease as fast as possible.


That answer is easy. If you want a sexually transmitted disease to spread through the population as quickly as possible, then everyone should be having sex with a wide circle of people almost constantly.

If all other factors are equal, then increasing promiscuity cannot possibly slow the spread of an STD.
posted by chrisamiller at 11:16 PM on December 12, 2007


Response by poster: By the way, I'm not interested in if this study is true or not, I'm only interested in the tool that will allow me graph it. I'm not doing a big scale study, it's just a simple model with 10 people and no external factors.
posted by markovich at 11:22 PM on December 12, 2007


Before you can graph anything you have to have data. You don't even have the data, so why are you worried about building a graph?

Er... You can build graphs using a ton of different programs. Notably Microsoft Office will let you do it in many different fashions, but without the data, you don't have anything to graph.

Further, I'm not sure how you would graph something like individual promiscuity. It doesn't seem like the type of thing even first tier researchers would present in graph form simply because it is so subjective.

So, I fear that your question needs revising, and I think you need to actually sit down with whatever data you're interested in and think about how you'd like to present it.

Lastly, chances are there are already pre-existing graphs out there which you can use (with citation)... I suggest this because it sounds like you're not actually doing this data sampling yourself.
posted by wfrgms at 11:36 PM on December 12, 2007


Response by poster: I think you are misunderstanding the question a bit. I'm trying to create a graphical model showing the spread of a disease within a fictional night club with the players having only the ability to infect other players. This is not a real world question about diseases.

I'm not trying to create a bar chart or a pie chart. I want a graphical model, a video-like simulation showing the spread of the disease from one person to another.
posted by markovich at 11:41 PM on December 12, 2007


I think you are misunderstanding the question a bit.

I think your question was unclear and has been somewhat clarified by your updates. Your title doesn't help matters.

This isn't a direct answer but the model of this that I am most familiar with is a dumb social software game called Breedster. You play an insect, you mate with other insects and have offspring. At some point in the game a virus was introduced which, while not fatal, renders the insect sterile and unable to have more offspring. It is always transmitted through mating. Since there is nothing else to do in the game besides eat and defecate and move around, it created some interesting observable patterns. You might want to talk to the people who made it.

Since your model is (supposedly) about HIV, you'd have to account for rates of transmittion between F/M, F/F and M/F partners since they differ. Women who were promiscuous but almost exclusively with women would not have teh same transmission rates as women who were either bisexual or related to men as well. If you just want to account for some random disease then you can ditch the "promiscuity" angle and just talk about infection rates and interaction rates. Or you could land in the middle with some sort of STD non-fatal thing.

Otherwise, fi the disease itself and how virulent it is is not a factor, just google around for "disease transmission modelling" and there's a lot there.
posted by jessamyn at 11:54 PM on December 12, 2007 [1 favorite]


OP, I think you may have misunderstood the study. I bet that what they found was that when the majority of women were not promiscuous, more men spent more time having sex with prostitutes -- and that was the reason for the increased rate of spread.

If that's the case, then your entire simulation you're trying to do would be bogus. You haven't included any prostitutes in your model.
posted by Steven C. Den Beste at 12:02 AM on December 13, 2007 [1 favorite]


What you read on line was probably the title essay from Steven E. Landsburg's book "More sex is safer sex", or discussion inspired by it. Landsburg describes how under some assumptions (and yes he admits they are unrealistic ones) greater promiscuity on the part of the otherwise least promiscuous women reduces the rate of spread of AIDS because it puts limits on the promiscuity of the most promiscuous women, ie: it reduces the demand for the services of prostitutes.
posted by Canard de Vasco at 12:05 AM on December 13, 2007 [4 favorites]


Here's a discussion on Metafilter about the book you're probably thinking of.
posted by painquale at 12:33 AM on December 13, 2007


Recently I've heard that HIV transmission is higher in societies where individuals are likely to have a couple of long term sexual partners compared to societies where individuals have many serially monogamous relationships.

It was probably on Quirks and Quarks, but I can't find the story..
posted by Chuckles at 2:11 AM on December 13, 2007


Metatalk
posted by Blazecock Pileon at 2:51 AM on December 13, 2007


Although several others in this thread have alluded to it, I don't think anyone has spelled it out. Imagine the extreme case: a population where no woman yet has AIDS and every woman except one (Ms. A) refuses to have sex except with a single, lifelong committed partner, while the remaining woman accepts sex with any man at any time. Clearly, if some of the males in the population have AIDS, and if most of the men have sex with Ms. A, AIDS will spread quickly. Change the value of "one" to "a small number" and you have pretty much the same situation, especially if the men are not picky about their partner. But this is clearly not the same as saying widespread promiscuity reduces the spread of AIDS because:
  • If all women hold out for committed males, AIDS will not spread;
  • If most of the men also hold out for committed females, AIDS will not spread;
  • If a high number of women adopt Ms. A's attitude, and the men choose their partners randomly for each sex event, AIDS will also spread (though perhaps not as quickly).
So although you could demonstrate what you want by concentrating on modeling the original scenario, the argument is still specious.
posted by ubiquity at 4:23 AM on December 13, 2007


Ubiquity... Here's the problem I see with your argument. You posit a world in which every one has a single monogamous partner and then you hypothesize about what would happen if men and happen held out for "committed" partners. There's no way for an individual to know with certainty another person's sexual history or current activities. Your model only works if everyone is completely honest about their behaviour. It seems unlikely to me that people would be honest about their history in a society where having one partner doomed them to never have sex with anyone else. The only attribute that men or women can select for is the appearance of monogamy not actual monogamy and the appearance of monogamy is not an effective protection against H.I.V.
posted by rdr at 5:13 AM on December 13, 2007


What's this study you're talking about? Everything I've seen would indicate to me that women's sexual empowerment, both in terms of reducing rape, increasing the safety of prostitution and preventing trafficking, in terms of ability to insist on condoms, in terms of ability to leave an 'unfaithful' husband and support herself economically correlates to lower rates of AIDS. For the purpose of your model, anyway, does it really matter whether men or women are the ones that are 'promiscuous'? (which by the way is a loaded and meaningless word...)

Anyway, your model doesn't work because even if all ten of the women were willing to sleep with any of the men without condoms, if the men weren't 'promiscuous' but committed to only sleeping with their one steady partner or with no one at all, AIDS wouldn't anywhere.
posted by Salamandrous at 5:22 AM on December 13, 2007


rdr, it's not my argument. I do not believe that more sex is safer sex. I was using an extreme case to make the results obvious. I totally agree it's not realistic.
posted by ubiquity at 5:37 AM on December 13, 2007


You have to model this with several differential equations
Take this as a starting point:
http://www.esam.northwestern.edu/~chopp/ESAM252-3/lab6.pdf
posted by yoyo_nyc at 5:56 AM on December 13, 2007


By the way, I'm not interested in if this study is true or not, I'm only interested in the tool that will allow me graph it. I'm not doing a big scale study, it's just a simple model with 10 people and no external factors.

But you don't want it to be completely meaningless, right? And disease just doesn't...work like this. Having sex with an infected person is not a 100% guarantee that the uninfected person will get HIV, let alone AIDS.
posted by desuetude at 6:15 AM on December 13, 2007


One possibility is some probability package where you define two groups, X and Y, and then set up probabilities for them interacting with each other. Assign one person to be the carrier. Rate both groups at some value from prude to promiscuous (0 to 1) and then if there is sexual contact assign a rate of transmission. If you make the men completely promiscuous then for any given rate of transmission you will get an answer of time (number of cycles, where each cycle is an opportunity for sex) as a function of women's prudery/promiscuity. After you get the data there are a number of Flash packages that will help you present it.

Of course this 'model' does not include homosexual contact but what does it matter? Any relation between it and reality is coincidental. It sounds like the whole subject of AIDS, and even disease, is pretty much irrelevant to the way you've expressed the problem. Doing the above won't give you an answer that matches the study you read. In that model, the higher the promiscuity, the quicker the spread of disease.

If you want the results to match the study you will have to use the algorithms the study uses. And models with populations of 10 men and 10 women, which ignore transmission by other means, probably won't do the trick. If you do include the relevant presumptions I suspect you'll find that the rate of female promiscuity which is most successful in spreading the disease will be dependent on the rate of male promiscuity. You've also got a hell of a problem in how to model promiscuity. If the men, or women, of one region have an average of 9 sexual partners over their lifetime, how can you convert that into the chance of their having sex at some random opportunity? Because that's what you're looking for in this model - the chance that a person from group A will have sex with someone new from group B on any given opportunity. And that statistic would be abstract to the point of meaningless, not to mention a little insulting.
posted by BigSky at 6:25 AM on December 13, 2007


Since your model is (supposedly) about HIV, you'd have to account for rates of transmittion between F/M, F/F and M/F partners since they differ.

A very, very important point, that deserves to be reinforced. Due to the way HIV transmission works, it's significantly more difficult to catch HIV from a woman than from a man. Herpes might be a better example for this kind of study.
posted by aeschenkarnos at 6:43 AM on December 13, 2007


It sounds like what you want is simulation software. And a graphing interface. I can think of two types of software you could use to do this: crystal ball (which is a vba add-on to excel) and vensim (which is a system dynamics software). If you want to keep your hypothetical sample small, use crystall ball. If you want to build a system, set rules and feedback loops, then vensim and system dynamics will do exactly what you want.
posted by zia at 6:57 AM on December 13, 2007


Mod note: A couple comments removed. There's a metatalk thread open, if you want to discuss language issues etc. instead of answering the question.
posted by cortex (staff) at 7:38 AM on December 13, 2007


There's significant differences in transmission patterns among individuals within cultures that favour parallel multiple relationships over serialised multiple relationships. One can have cultures with a median higher quantity of total sexual partners in a lifetime, yet have reduced transmission probabilities for slow viruses such as HIV, especially when you consider the average life expectancies and sexually active periods of the population. Also, the issue of allelic heterogeneity in the cellular transmission vectors also plays a big part in accelerating or retarding the transmission rates, and also the progression from HIV infection to AIDS (in later stages of AIDS individuals are rendered less active). Finally, sexual practices, prevalence of STDs that create lesions and patterns of cultural modification of the genital organs can also affect transmission rates.

Your simulation sounds more like a simple alife game, not a realistic HIV simulation.
posted by meehawl at 9:20 AM on December 13, 2007


Also, of course, different HIV clades and subtypes will exhibit different transmission patterns.
posted by meehawl at 9:24 AM on December 13, 2007


By the way, in the set of heterosexual persons, men and women must have the exact same average number of partners. That doesn't mean the median is the same, but the mean must be.
posted by delmoi at 9:41 AM on December 13, 2007


By the way, in the set of heterosexual persons, men and women must have the exact same average number of partners. That doesn't mean the median is the same, but the mean must be.

The variance is perhaps more important than the mean. Let's say you have five men (A through E) who each sleep with three of five women (1 through 5), and all are sexually active. You can easily have a situation where one woman sleeps with one man, while another woman sleeps with all five men. Scale that up to millions — that spread will have an effect on designing an epidemiological study.
posted by Blazecock Pileon at 1:02 PM on December 13, 2007


Blazecock Pileon: and that's the kind of distributions seen in real-world situations, though obviously not to the extreme extent shown in that 5 person example.
posted by Justinian at 1:04 PM on December 13, 2007


Repeat after me: Correlation does not equal causation

It is quite a conceit to think that such a trivial thing is somehow outside the knowledge of people dealing with studies and models every day of their lives. Proclaiming "correlation does not equal causation!" on the internet really translates to "I do not understand the methodology of the study, but its results challenge my prejudices and make me uncomfortable, so I will try to make it go away by assuming the study is flawed and so my prejudices need no examination.". As it happens, the study in question dealt purely with causation (it was based on quirky applications of hypothetical economics models).

As for a possible help with an answer, here is an online graphical / animated, and (somewhat) interactive simulation of a zombie infection spreading among a population. This gets you 90% there (or 100% if everyone is bisexual and highly promiscuous :-) ).
The source code is included, so modifying the code could get you the rest of the way, or simple serve as a example of how to approach the problem. (The page also include links to similar infection simulators.)
posted by -harlequin- at 8:24 PM on December 13, 2007


I am pretty sure this is the idea that OP is refering to: http://www.nytimes.com/2007/07/08/books/chapters/0708-1st-land.html?_r=1&pagewanted=1&oref=slogin

The book is called ‘More Sex Is Safer Sex: The Unconventional Wisdom of Economics’

Not actually research as much as a thought experiment.
posted by vegetableagony at 7:42 PM on December 25, 2007


« Older How much will I save in taxes by donating this...   |   What to do when your s-o doesn't come through in a... Newer »
This thread is closed to new comments.