Friday, April 21, 2006

Cambodian village seroprevalence study

A paper now being reviewed at The Lancet is reporting that the prevalence of antibodies against H5N1 is zero in 351 villagers in a Cambodian town that suffered a bird flu death in March of last year. This is surprising on several levels. First, the assumption many had (including us) that there was likely quite a lot of mild inapparent and undetected infection with this virus, thus artificially inflating the case fatality rate (CFR, which is technically not a rate but a proportion). The good side of this is that it implies the virus has very little transmissibility, either from bird to humans or human to human. The bad side is that it implies this is one of the most dangerous viruses known, with a CFR exceeding 50%.

The second surprise is that there seemed to be nothing unusual about the exposure of the 28 year old farmer who did become infected and died.
Dead birds were reported in the village in Cambodia’s southern Kampot province and the H5N1 virus was detected in poultry there. Many villagers surveyed said they had very close, daily contact with the birds – collecting dead or sick poultry, feeding them, cleaning up faeces, plucking and eating them.

“That supports data from all over the region suggesting that it’s actually very inefficient to transfer from birds to humans,” said Benjamin Coghlan, a WHO epidemiologist from Australia who participated in the study.

“So, what does (the transfer) require? Well, we’re not sure,” he said. “Certainly, this case in the village wasn’t doing anything unusual that everyone else in the village wasn’t doing.” (Ireland Online)
Host factors we don't understand are involved. It is not just the virus. We know that age is a significant host factor, because the virus seems to prefer the younger age group (see the age distribution of reported cases here and here). In addition we know that the environment is a factor, since there is a marked (but unexplained) seasonality to influenza, including H5N1.

This is a reminder of an important point. An infectious disease is a function of the host, the agent and the environment acting together in a particular combination. Virulence is not a property of a virus but of the combination of the virus and the host, modulated by the environment. So the talk of "mutation to a virulent" form is really a shorthand for a more complicated idea. It is also important to remember that transmissibility (the ability to pass on infection), and virulence (the ability of infection to cause serious disease), are separate things. At this point we have little or no idea of the genetic, host or environmental factors affecting either in human beings.

So far these data have not been published but only presented at a conference. We look forward to seeing the published paper with the supporting tables and description of methods. Since the prevalence was zero, this isn't a statistical issue, because while the results don't rule out a small probability of infection that couldn't be detected because of sample size, that in itself is a statement about transmissibility. More importantly, it doesn't rule out some kind of bias, for example a problem with the measurement or selection of subjects. Therefore it is important to see the published paper.

We can't assume too much about this virus. There is always the hope that the transition to transmissibility is more difficult for the virus than is a reversion to less virulence. This virus is continually changing and the result could go in either direction. We don't know the biology well enough to hazard a guess which. It is currently a very dangerous bug that is constantly evolving. It continually uses hosts to make copies of itself and if chance happens on a good recipe for a human host cases will come faster and more abundantly.

If that happens it won't matter whether you believe in evolution or not, nor does that kind of virus believe in abstinence.