pageok
pageok
pageok
What should we do about avian flu?

Avian flu has been attracting blogosphere attention for much of the last year, including from Glenn Reynolds and Andrew Sullivan. I have had a hand in starting an avian flu blog. Recently I have finished work on a 40 pp. piece on what we should do about avian flu, here is the pdf link. Here is the Executive Summary. My core approach is to rely on decentralization and incentives. I am skeptical of quarantine or use of the army. Well-functioning, decentralized health systems will likely save the most lives. Stockpiling Tamiflu or vaccines is often overrated in potential effectiveness. We also should, when needed, purchase the relevant intellectual property rights for a fair price rather than simply confiscating those rights. We should ease liability for vacciner makers and institute prizes for good vaccines. Better-developed prediction markets would be useful for charting the progress of the flu. Your comments, of course, are welcome, either below or email me.

Related Posts (on one page):

  1. Avian Flu:
  2. What should we do about avian flu?
Defending the Indefensible:
Stop feeding the panic, there is not even a single documented case of human to human transmission of "avian flu" and when/if such ever occurs, it would be a mutated form and hence a different virus than the one we're "preparing" for.

Nor is it fair to say (as you didn't, but media reports have been saying for months) that we are "due" for a pandemic based on the time since the last major flu outbreak -- this is the gambler's fallacy, if you've rolled the dice 35 times and and not gotten "snake-eyes," your odds of getting snake-eyes on the next roll are still: 1/35.
11.15.2005 9:30am
Defending the Indefensible:
And to clarify: if you've rolled the dice 70 times, and still not gotten snake-eyes, the odds on the next roll are STILL: 1/35.
11.15.2005 9:50am
Zargon (mail):
I hardly see how publishing an academic paper on the economics of epidemic response to be "feeding panic".
11.15.2005 9:53am
billb:
DTI: While absolutely correct in the context of dice, your claim that Tyler is committing the gambler's fallacy leaves something to be desired. We don't actually know the probability of having a flu pandemic in any given year (whereas we do know, a priori, the odds of a particular throw of the dice). The best that we can do is plot a histogram of the gaps in time between major flu outbreaks and to begin to speak in terms of probability distributions. Then, we should derive expectation values for things we care about based on these distributions (which, BTW, is what gamblers should be doing as well).

I personally doubt there's enough data in say the 1000 years or so of useful history to make any claims about whether we're "due" for a pandemic or not.
11.15.2005 10:22am
Dave W:
There is no reason to panic, but there is a lot of good reason to be concerned:

1. The H5N1 virus has a mortality rate of approximately 50% in the human cases where it was contracted via direct contact with birds. That's roughly 2x as deadly as the 1918 Flu that killed more than 40 Million. (Src: Scientific American - Last Month)

2. The H5N1 virus is about 100x as transmissible (among susceptible mammals) as the 1918 Flu. Recent studies have shown that while it takes an initial infection of 1000 1918 viruses to kill a mouse, it only takes an initial infection of 10-15 H5N1 viruses to start a lethal infection. (Src: Science News, Week of Nov 1)

Yes the virus is not current transmissible between humans, but there is nothing to indicate that if or when it acquires the genes required for human transmission, either of these factors will be reduced. In my book that's more than enough reason to be concerned.

As for the other points: It's not a matter of odds, it's a matter of frequency. The historical record shows that for as long as humans have tracked such things, a flu pandemic has occurred every 30-35 years. By that measure, we are due as the last one was in 1968.
11.15.2005 10:25am
dew:
DI: "and when/if such ever occurs, it would be a mutated form and hence a different virus than the one we're "preparing" for. "

Are there any examples of anyone saying we should only prepare for the current strain of the virus? Tyler Cowen's paper doesn't.
11.15.2005 10:44am
Syd (mail):
I's say a 1 in 35 chance per year that this virus may mutate to human-to-human transmission and kill 100-200 million people is enough that we should be thinking about ways to stop it if it does. And the longer it spreads among bird populations and thence to humans, the more likely it is to mutate.
11.15.2005 11:50am
jab (mail):
Ummm... The odds of rolling snake eyes on a pair of FAIR dice
is 1/36... not 1/35...
11.15.2005 12:49pm
b.trotter (mail) (www):
I believe, personally, that the hype of avian flu is far greater than the threat itself; that being said, preparation for pandemics in general is a wise course of action.
Quarantine, of course, is one of the most effective methods of stopping any transmission of infectious disease. I think, however, that it's been well demonstrated that we no longer have the capability of restricting a population to a defined region (or out of one). Don't think for two seconds, however, that it won't be tried. I think we've all seen enough Hollywood to envision the stereotypical containment (example: Outbreak starring Dustin Hoffman, Dreamcatcher starring Morgan Freeman). If only things would be that easy. I'm predicting (thanks to the irrational overreporting by the Mainstread Media) that quarantines will be imposed in more of an anarchy model... During the recent SARS epidemic, a coworker of mine was quite vocal about his collection of 30 firearms, and quite insistant that if SARS got close to home, that he expected us to join him at the entrance to town in an armed posse to keep folks out of our small town... As extreme as that sounds, isn't that what happened in New Orleans on a bridge leading out of the city?
11.15.2005 2:13pm
Jimbino (mail):
A woman has two kids. One is a boy. What are the chances that both are boys?
11.15.2005 2:37pm
Visitor Again:
A woman has two kids. One is a boy. What are the chances that both are boys?

I am no mathematician, statistician or even logician (other than in some of my legal arguments), but this one is not too difficult for me:

There are four options for gender combinations in a two-child family:

Girl/Girl
Girl/Boy
Boy/Girl
Boy/Boy

The first option is out since we know one of the children is a boy. So that leaves a one in three chance that both children are boys.

If we knew that the first child was a boy or that the second child was a boy, two of the possibilities would be out and the odds of the other child being a boy would be only one in two.
11.15.2005 3:12pm
Chuck Simmins (mail) (www):
The H5N1 mortality rate has to do with two primary causes, neither directly related to influenza. The first is poor health care in the region. The second is that much of the information is coming from two communist governments which cannot be relied upon to tell the truth. WHO has explicitly stated that labs in the region do not have the ability to accurately test for H5N1.

H5N1 was first found in 1959, in Scotland. In the 46 years since, it has not developed human to human transmission.

The historical record shows three influenza pandemics, all within the last 87 years. I have lived through two of them without effect. Most of the "experts" talking about influenza pandemics are speaking with far more certainty than knowledge.

Extensive blogging found here.
11.15.2005 4:12pm
Brent Michael Krupp (mail):
Maybe he's rolling 1 7-sided die, and 1 5-sided die. Then snake eyes would be 1 in 35. =)

More seriously, I suspect that the deadliness of avian flu is overstated as they're not testing everyone under the sun, only those who turn up sick enough to see a doctor. They'll thus miss folks who are unaffected or mildly affected and yet did catch the virus from a bird. Not saying it's not plenty deadly, just probably not quite so bad as it might look.
11.15.2005 4:37pm
Dick King:
Actually, if we had 7000 years of data and noticed that flu pandemics occur every 35 years, and the last one was 40 years ago, we might well be "about due".

Yes, the gamblers fallacy is a fallacy because dice have no memory.

Human immune systems, however, do indeed have a memory, especially WRT viruses. That memory is about seventy years long, the lifetime of a human, and it may well be that we get a pandemic whenever more than half of the humans now alive were never exposed to the previous one.

-dk
11.15.2005 4:50pm
BladeDoc (mail):
Quick question. Today Tamiflu reduces the symptoms of "normal" flu by a few days if taken right away. Not many people buy it normally in the grand scheme of things. Tomorrow it's found that the drug saves the life of 80% of people who get the avian flu virus. What is the "fair market value" what it was when it was a moderately useless medication or what it is when it's a lifesaver?
11.15.2005 5:08pm
ajf (mail) (www):
Quarantine, of course, is one of the most effective methods of stopping any transmission of infectious disease.

there's no "of course." the efficacy of quarantine to stop further transmission of an infectious disease is dependent on a number of factors, including but not limited to:

* the infectivity/transmission rate of the disease
* the mode of transmission of the disease
* when in the epidemic curve the outbreak is discovered

in today's world (US, anyway), i tend to think that quarantine would be more likely to be met with anger and resistance than compliance.
11.15.2005 6:15pm
triticale (mail) (www):
Ummm... The odds of rolling snake eyes on a pair of FAIR dice is 1/36... not 1/35...

I'm a firm believer in corner-loading one's Fulhams, but it would take a really subtle offset to produce this small a shif in the odds. Probably never get caught, tho.
11.15.2005 6:29pm
triticale (mail) (www):
The most encouraging news regarding the possible pandemic was the finding of the cytokine flood effect. Whether or not M. Simon and I are correct that marijuana could be of value in relieving flu symptoms, the fact that symptoms are being researched should mean that mortality rates are reducible.
11.15.2005 6:35pm
karlnewman (mail):
I think the snake eyes/dice analogy is a bit off for avian flu. The dice example requires that each throw be INDEPENDENT of each other. Since the virus picks up mutations as it moves in populations and mixes with human flu in humans or pigs, a better example might be a volcano. Sure, one could look and see that a particular volcano erupts every 100 years and just think that each year you have a 1/100 chance of an eruption. Of course since we know with science that right after an eruption there is a release of pressure and then we might detect the pressure building over years. Then geologists will warn us that the pressure has built up so we are more likely to have an eruption. So each year's risk is not INDEPENDENT of the prior years risk, but rather related. With flu, there is a similar pressure. This virus has picked up enough mutations and is spreading so quickly that the "pressure" is building and the chances that a few mutations could lead to a pandemic strain are now much higher. Of course nothing is 100%. So each year that this strain spreads, the chances that this could happen are higher than they were 25 years ago when this strain was not spreading.
11.16.2005 12:42pm