Ewald bird flu spat

Scientific American has an editors' blog, SciAm Observations. I point to it because a series of recent posts has included an interesting exchange among evolutionary biologist Paul Ewald, the editor John Rennie, and an anonymous "senior public health scientist". The exchange developed after an LA Times op-ed by Wendy Orent, which discussed Ewald's views.

Ewald believes on evolutionary grounds that any bird flu pandemic is likely to have a much lower mortality rate from the disease than either the recent cases of the H5N1 variant or the 1918 pandemic. From his letter:

The future that we are debating concerns the evolution of influenza viruses. Being an expert on the molecular biology or epidemiology of influenza viruses does not translate into expert assessments of the future evolution of influenza viruses. One need only peruse the writings of such influenza experts to recognize that they generally fail to incorporate the most essential component of the evolutionary process into their arguments--they talk as though mutation and reassortment were the only processes relevant to the evolution of influenza viruses. To discuss evolutionary processes with any degree of expertise one must focus on natural selection. Mutations and reassortments generate the variation on which natural selection acts. But natural selection molds viral evolution.
A second process critical to accurate predictions of the future threat posed by H5N1 is the evolution of virulence. The evolution of virulence is linked to the evolution of transmissibility because high virulence itself curtails transmission unless conditions allow transmission from people who are immobilized with illness. Evolution of increased transmissibility of H5N1 from human to human is bound to go hand in hand with drastic evolutionary reductions in virulence. The history of influenza provides strong evidence for this conclusion. Except for the 1918 pandemic, all the trustworthy evidence from all the years of well documented influenza epidemics and pandemics indicate that influenza viruses maintain themselves evolutionarily at low to moderate virulence when transmission depends on host mobility. Influenza experts who do not understand natural selection--or choose to ignore its implications--fail to realize how powerful this evidence is. The last century represents countless thousands of "natural experiments" involving countless trillions of mutation-prone viruses. If the highly virulent variants were able to spread in competition with strains of low or moderate virulence we would see some evidence of regional outbreaks. Instead we see only epidemics and pandemics of low to moderate virulence.

Essentially, the theory is that a virus that spreads by social contact requires its host to be socially active in order to spread. Make him too sick, and he won't be seeing very many people. Some diseases, like malaria and cholera, don't fit this model because they can spread from inactive hosts --- mosquitoes carry malaria away from a prostrate body, cholera spreads in water fouled by the sick. Ewald argues that the 1918 flu was likewise an exception, because situations like the trenches in Western Europe and the open sick wards of hospitals allowed the virus to spread effectively from incapacitated hosts to many new ones.

A later post carries a reply by the anonymous critic. Here's an excerpt:

But it should be obvious, even to him, that if a very virulent disease becomes transmissible before virulent effects immobilize the host, very virulent diseases can propagate with ease. SARS is not very transmissible until the host is seriously ill. Influenza, on the other hand, is transmissible before and during the early phases of illness. Moreover, Ewald objects that a 2% mortality shows the virus is very virulent. Why? Yes, 2% is high mortality for seasonal flu, whose case-fatality rate is usually on the order of 0.1%, the very figure Ewald predicts will hold for a pandemic strain. What epidemiologists know about pandemic strains, however, is that they tend to have higher mortality than the seasonal ones. Ewald assumes that there will be no difference in mortality from a subtype against which there is no population immunity compared to one where there is substantial population immunity. We know 2% mortality is attainable by influenza. I guess being an expert in evolutionary biology doesn't make you an expert in immunology, public health or infectious disease epidemiology.

If you're interested in the bird flu, the exchange is worth reading in full, along with Rennie's earlier posts. None of the participants is saying (as far as I can tell) that a flu pandemic wouldn't be a terrible thing. They differ on the likely mortality rate in the event a pandemic should occur -- Ewald predicts it won't be greatly above the yearly flu strains that we already see, others predict it will either be much higher or even catastrophically higher.

I don't have a dog in this hunt, but I notice several points:

1. Almost no mainstream press accounts of the bird flu threat discuss anything about the evolution of influenza. This is probably the most important public impact of evolutionary theory today, but we hear almost nothing of the evolutionary modeling of how the virus may change.

2. Ewald is very well known for studying the evolutionary dynamics of disease. He is making an argument that is sound, as far as the dynamics of selection are concerned. Thus, there are good reasons to think that the worst will not happen, and this is a perspective that has been underplayed.

3. So far, the theory has only been tested by a relatively small number of instances -- there just haven't been so many pandemics that we can infer accurately from past events what the future will be like. It could certainly happen that some new influenza strain could violate the model in some unexpected way, and for this reason governments should play it safe rather than assume that no high-virulence pandemic will emerge.

4. A lot of public health scientists are going to be well-employed for as long as the bird flu remains in the public perception. This doesn't mean that they are wrong to convey alarm, but it does mean that they don't benefit by playing down the threat. It's sort of like NASA and the asteroid impact threat --- partly they are more concerned because they know more about the threat and its terrible effects, partly because it's their job to be concerned.

5. There are a lot of biologists who don't use or understand selection.