Today in extrapolating beyond your data
Oct. 30th, 2013 01:07 pm![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
pktechgirlbackupYou could be forgiven for thinking that getting "The HPV Vaccine", or worse, "The Cervical Cancer Vaccine" means you will never get HPV and/or cancer, because that's what a lot of news coverage has indicated, and because public health workers tend to overpromise and obscure details in order to motivate. The truth is that there are many strains of HPV (NIH says >100), there's no reason to believe our list is comprehensive, and a vaccine against one is not necessarily effective against another.
That's okay. HPV is incredibly common, to the point that some scientists think some strains be commensal. That makes it hard to prove what it's effects are: does increased prevalence of strain in renal patients mean it causes renal disease, that it's harmless but increases in prevalence in response to renal stress, or that it's harmless in healthy patients but harmful in the quantities seen in renal patients?
What we can agree on: some strains cause warts. Warts won't kill you, but they can hurt and create a vulnerable point other infections could use. How bad is that? For a modern American who wears shoes all day, plantar (foot) warts are more likely to harm you through bad ergonomics than an opportunistic infection.* If you are poor and shoeless in a sub-saharan Africa that vulnerability is a really big deal. Genital warts make it easier to catch another STD, but the exact probability HPV leads you to another STD you wouldn't otherwise have caught depends on the STD status of the people you have sex with.
We also agree that HPV can cause cancer. You can't prove it's impossible to get cervical cancer without it, but it's probably safe to say that if you do so you've either been storing nuclear waste in your vagina or severely pissed off a vengeful deity. Now that we're looking for it we're also finding certain strains associated with penile, rectal, and oral cancers.
Given that there are a large and growing number of identified HPV strains, some of which might even be beneficial**, and each strain must be separately cultured, increasing expense how do you decide which to vaccinate against ? When making Gardasil, Merck chose four strains, two of which caused 90% of genital warts, and two of which caused 70% of cervical cancer.
Or did they? New data is out suggesting that the vaccine is less useful in black women than white women because black women are more likely to have strains the vaccine doesn't cover. Some people are describing this as "less effective in black women", but that's misleading. As far as we know the vaccine is equally effective against the strains it claims to be effective against*** on a biological level. It's just not useful because black women are much more likely to be exposed to strains the vaccine doesn't protect against. By far the simplest explanation is that whatever study generated the prevalence estimates oversampled white women.
This demonstrates a couple of things. One, the importance of sampling across the entirety of the population you want the data to apply to even if you are really, really sure they're genetically identical. I would not be at all surprised to discover geographic differences in strain distribution. But if I'm correctly interpreting this newspaper article with no link to the underlying study, participants were recruited at the same site and so roughly the same geographic area. Assuming no racial influences on susceptibility or response, this suggests that white and black women, and their partners, are swimming in entirely separate sexual pools.
I'm not that naive. I knew people tended to have sex primarily with same-race partners. But my epidemiology intuition says it shouldn't take *that much* cross over for strains to reach prevalences much closer than what's being reported here, because once a strain has crossed over, it should rapidly colonize a wide open pool.
Alternate possibilities:
exposure to one strain makes you resistant but not immune to another, so which you strain you have is correlated much more heavily with early sexual partners than later ones. Without looking it up I'm pretty sure people's first partners are much more likely to be the same race as them. This suggests that the wrong-strain vaccines are still likely to be some helpful, but not as helpful as the right strain.
People who engage in interracial sex are atypical in their engagement with their own race. They might have fewer partners, observe a higher standard of sexual safety. or have sex nearly exclusively with members of that race, making them part of that cluster.
the true clustering is around a factor other than race but with a heavily non-random distribution, like location or SES.
*True story: the only time I've ever used crutches is after having a plantar wart burnt off.
**Commensal is defined as one side (HPV) benefiting and the other side (humans) receiving no benefit. However, if harmless HPV is taking up space on our skin that would otherwise be occupied by something damaging, that's a benefit, like those spiders that avoid humans and eat black widows. Or the HPV could be involved in some weird but ultimately beneficial cycle with the bacteria on our skin. We don't understand the ecosystems within our own bodies at all, and our overconfidence at what can be safely removed has caused a lot of trouble over the years.
***Never say never, but I'd be shocked if the per-strain effectiveness differed significantly between races, because even if there was a genetic component, race is a stupid categorization that tells you very little about an individual's genetics.
That's okay. HPV is incredibly common, to the point that some scientists think some strains be commensal. That makes it hard to prove what it's effects are: does increased prevalence of strain in renal patients mean it causes renal disease, that it's harmless but increases in prevalence in response to renal stress, or that it's harmless in healthy patients but harmful in the quantities seen in renal patients?
What we can agree on: some strains cause warts. Warts won't kill you, but they can hurt and create a vulnerable point other infections could use. How bad is that? For a modern American who wears shoes all day, plantar (foot) warts are more likely to harm you through bad ergonomics than an opportunistic infection.* If you are poor and shoeless in a sub-saharan Africa that vulnerability is a really big deal. Genital warts make it easier to catch another STD, but the exact probability HPV leads you to another STD you wouldn't otherwise have caught depends on the STD status of the people you have sex with.
We also agree that HPV can cause cancer. You can't prove it's impossible to get cervical cancer without it, but it's probably safe to say that if you do so you've either been storing nuclear waste in your vagina or severely pissed off a vengeful deity. Now that we're looking for it we're also finding certain strains associated with penile, rectal, and oral cancers.
Given that there are a large and growing number of identified HPV strains, some of which might even be beneficial**, and each strain must be separately cultured, increasing expense how do you decide which to vaccinate against ? When making Gardasil, Merck chose four strains, two of which caused 90% of genital warts, and two of which caused 70% of cervical cancer.
Or did they? New data is out suggesting that the vaccine is less useful in black women than white women because black women are more likely to have strains the vaccine doesn't cover. Some people are describing this as "less effective in black women", but that's misleading. As far as we know the vaccine is equally effective against the strains it claims to be effective against*** on a biological level. It's just not useful because black women are much more likely to be exposed to strains the vaccine doesn't protect against. By far the simplest explanation is that whatever study generated the prevalence estimates oversampled white women.
This demonstrates a couple of things. One, the importance of sampling across the entirety of the population you want the data to apply to even if you are really, really sure they're genetically identical. I would not be at all surprised to discover geographic differences in strain distribution. But if I'm correctly interpreting this newspaper article with no link to the underlying study, participants were recruited at the same site and so roughly the same geographic area. Assuming no racial influences on susceptibility or response, this suggests that white and black women, and their partners, are swimming in entirely separate sexual pools.
I'm not that naive. I knew people tended to have sex primarily with same-race partners. But my epidemiology intuition says it shouldn't take *that much* cross over for strains to reach prevalences much closer than what's being reported here, because once a strain has crossed over, it should rapidly colonize a wide open pool.
Alternate possibilities:
*True story: the only time I've ever used crutches is after having a plantar wart burnt off.
**Commensal is defined as one side (HPV) benefiting and the other side (humans) receiving no benefit. However, if harmless HPV is taking up space on our skin that would otherwise be occupied by something damaging, that's a benefit, like those spiders that avoid humans and eat black widows. Or the HPV could be involved in some weird but ultimately beneficial cycle with the bacteria on our skin. We don't understand the ecosystems within our own bodies at all, and our overconfidence at what can be safely removed has caused a lot of trouble over the years.
***Never say never, but I'd be shocked if the per-strain effectiveness differed significantly between races, because even if there was a genetic component, race is a stupid categorization that tells you very little about an individual's genetics.
