Estimating pandemic risk

We keep seeing these discussions about the probability of a pandemic next year. Sometimes they center on the "overdue" for a pandemic notion, sometimes on using available data to give an estimate of the rough chances of a pandemic. In the latter category is this little contretemps in the UK:

Contingency plans drawn up by the NHS are based on a 3 per cent chance in any given year that the virus will mutate into a form that infects humans. However, an international review at a summit of avian flu experts put the risk of a pandemic during the next year as between 5 and 20 per cent.

Leading scientists described the Government's current assessment as "dangerously optimistic" and called for more money to be made available for drugs to protect the public in the event of an outbreak.

A review carried out by the Department of Health, and described as a "comprehensive and state of the art" assessment of pandemic planning, based its risk rating on one statistic - the number of flu pandemics that took place in the 20th century. With three pandemics in the past 100 years, the probability is an annual risk of 3 per cent, it concluded.

However, the report also admits that medical experts who took part in the global study put the risk of a bird flu pandemic in Britain at 5 per cent, while non-medical experts concluded it was as high as 20 per cent. (The Telegraph [UK])

Since I don't know what the probability of a pandemic is for the next year or the next five years and don't have an opinion about it (other than it would be foolish to believe it couldn't happen at any time), I won't give my own estimate (since I don't have one). However I will make a few observations on the general question, as much to think it through myself as anything else.

What tools do we have for predicting a pandemic ahead of time (i.e., before it is actually underway; thus I am leaving out the use of surveillance systems that give us an early warning that one is already happening)? If we understood the dynamics of pandemic emergence well enough, we might be able to see the various elements as they came into play. For example, if we knew exactly what mutations were necessary for efficient person-to-person transmissibility and knew the rate at which mutations at those spots occur (or the probability that they would occur at those spots), then we could use this information for prognostic purposes. Or lots of other pieces of information we don't have at the moment. Maybe we will get them or figure out what we need. This is an active area of science and we can hope some of the work being done now will help us see when we are approaching a dangerous situation in a definable time period, i.e., we could use this information for prediction.

Since we don't have that kind of knowledge now, we have to fall back on the observed data about pandemics. That there have been three pandemics in 100 years, for an average of 3%/year, is the most simple minded way to do this. And since simplemindedness has its attractions, especially if there is no evidence that more complicated schemes are any better, that's what the UK authorities have relied upon.

But that's the rub, here: relied upon. As an estimate of the yearly probability this isn't unreasonable. It is mathematically the maximum likelihood estimator of the annual probability, assuming independent occurrence of a pandemic each year. This means that the estimate of 3%/year is more likely than any other estimate that assumes independent events. But it doesn't say how much more likely. Even for the (doubtful) assumption of independent events, there could be a wide range of estimates almost as likely as the 3% one. Here is one way to think about it. If you were betting on a number, 3% would be the one to use since it is the most likely one to come up (given the assumptions). But how much you should bet on it would also depend on how likely 1%, 2.5% 5%, 10% or 20% was to come up, too. If they were almost as likely as 3%, you wouldn't want to bet a lot on a 3% annual occurrence. But that is what the UK plan is doing. They are betting a lot on it. Probably not a good strategy.

Moreover the assumption that each year is independent of the other years (i.e., that the probability of a pandemic is the same each year) is almost certainly untrue. While we don't completely understand the dynamics of immunologic status of the population, the rate of emergence of different subtypes and the biologically competence of different subtypes to cause pandemics in human hosts, we understand enough to say the assumption of independent occurrence each year is doubtful at best, although it may still turn out to be a reasonable approximation.

Both the 3% estimate and the idea that we are "overdue" for a pandemic (these two views often held by entirely different people) have one thing in common: the belief that the future will be like the past. It is true that as far as we know human populations have been plagued with periodic pandemics with influenza A, although the data are scarce historically speaking. But the same could be said of smallpox, malaria, polio and many other diseases which we have learned or are learning to control or that have disappeared because the demographic and environmental conditions that supported them have changed. There is nothing to suggest any such conditions make the likelihood of a flu pandemic any less -- just the opposite. In fact pandemics may be more likely than in the past -- or not. Nor is there any evidence we can prevent them happening, although some believe we should try. In any event, whether the future will look like the past we don't know, even as we aren't quite sure what the past actually looked like.

If I had to bet money on whether a pandemic would happen within the next year or two, I'd probably say "no," based on past experience. But I wouldn't bet much on it. I certainly wouldn't bet my life on it.

More like this

Peter Doshi has a bone to pick with CDC . His particular idée fixe is that CDC is cooking the books on their estimates of excess mortality attributable to influenza and he aims to set the record straight. He's done it before. Doshi is not the kind of critic CDC is used to. He is a graduate student…
The spate of swine flu articles in The New England Journal of Medicine last week included an important "Perspective, The Signature Features of Influenza Pandemics — Implications for Policy," by Miller, Viboud, Baliska and Simonsen. These authors are familiar to flu watchers as experienced flu…
by revere, cross-posted from Effect Measure The spate of swine flu articles in The New England Journal of Medicine last week included an important "Perspective, The Signature Features of Influenza Pandemics â Implications for Policy," by Miller, Viboud, Baliska and Simonsen. These authors are…
The newswires are humming again with another story of the estimated toll a flu pandemic might exact, if it were as bad as 1918. This time the occasion is a paper just published (.pdf) in the British medical journal, The Lancet, which attempts the most careful estimate yet of the toll of the 1918…

Nearly the same thing applies for earthquakes. We know the New Madrid pops its willies about once every 200 years. So for every year we dont have "the Big One" our percentages go up about 1% until the inevitability line is crossed. Its been 200 years since the last one so we have a 200% chance of having one? No, only that statistically we have increased our chances of having one by one percent. Recently, the insurance companies in light of Hugo, Andrew, Katrina, WTC's quit writing earthquake policies in the central US right or left of the Mississippi because of that 200% number.

That 200% number is the probability of within the next 50 years that we will have a =>7.0. Far more likely is the chance of a 6.5 which happen with more regularity and we have nearly a 500% chance of that happening statisically. A 4.5 ? Right now we have a 1250% chance of that. In fact we are back to counting on that having had two of those two years ago. Having those two doesnt mean that we are now closer to having a 7.0, it just means we had the 4.5's.

It does mean though that we are sitting on a time bomb waiting to go off. We just dont know what time it is.

Four months ago, a fire broke out in downtown Memphis and within two hours had spread to almost 10 buildings. We awoke out east some 30 miles to the smell of smoke. On goes the tube to find out that a gentle 10 mph wind had spread the flames from a simple fire. It was getting out of control as it spread to the high rises. Only by bringing in help from surrounding areas was it put out.

So where are we on the panflu path? I dont know because Revere's numbers are like mine. I can only look at the general numbers for flu or a quake and think to myself that something is up. In the case of the flu its outside of Indonesia/Vietnam/China. Percentage wise, we are due and only a fool would ignore that in light of the fact that we do have a killer flu running about along with its "normal" flu friends.

By M. Randolph Kruger (not verified) on 21 Aug 2007 #permalink

thanks for posting about this, which I think, important problem.

>Estimating pandemic risk
>Category: Bird flu ⢠Pandemic preparedness
>Public health preparedness
>Posted on: August 21, 2007 7:42
>AM, by revere
>We keep seeing these discussions about the probability
>of a pandemic next year. Sometimes they center on the "overdue"
> for a pandemic notion, sometimes on using available data to give
>an estimate of the rough chances of a pandemic. In the latter
>category is this little contretemps in the UK:
>Contingency plans
>drawn up by the NHS are based on a 3 per cent chance in any given
>year that the virus will mutate into a form that infects humans.
> However, an international review at a summit of avian flu experts
>put the risk of a pandemic during the next year as between 5 and
>20 per cent.
>Leading scientists described the Government's current
>assessment as "dangerously optimistic" and called for more money
>to be made available for drugs to protect the public in the event
>of an outbreak.
>A review carried out by the Department of Health,
> and described as a "comprehensive and state of the art" assessment
>of pandemic planning, based its risk rating on one statistic -
> the number of flu pandemics that took place in the 20th century.
> With three pandemics in the past 100 years, the probability is
>an annual risk of 3 per cent, it concluded.
>However, the report
>also admits that medical experts who took part in the global study
>put the risk of a bird flu pandemic in Britain at 5 per cent,
>while non-medical experts concluded it was as high as 20 per cent.
> (The Telegraph [UK])
>Since I don't know what the probability
>of a pandemic is for the next year or the next five years and
>don't have an opinion about it (other than it would be foolish
>to believe it couldn't happen at any time), I won't give my own
>estimate (since I don't have one).

such an estimate has to be subjective, so in that sense there
is not "the" probability, but many individual estimates.
However, I'm convinced that,despite what you write,everyone
here "has" such an estimate.Just most only have it analoguous,
intuitive and are too lazy or too chicken to give it digitally.

>However I will make a few observations
>on the general question, as much to think it through myself as
>anything else.
>What tools do we have for predicting a pandemic
>ahead of time (i.e., before it is actually underway; thus I am
>leaving out the use of surveillance systems that give us an early
>warning that one is already happening)? If we understood the dynamics
>of pandemic emergence well enough, we might be able to see the
>various elements as they came into play. For example, if we knew
>exactly what mutations were necessary for efficient person-to-
>person transmissibility

that's unclear, but we have some data, e.g. for the RBD of HA.

>and knew the rate at which mutations at
>those spots occur (or the probability that they would occur at
>those spots),

that one we know quite well.

>then we could use this information for prognostic
>purposes. Or lots of other pieces of information we don't have
>at the moment.

yes, and lots of other pieces of information we do have
at the moment. E.g. from the effectmeasure-archive ...
that wasn't all invain, was it ?

>Maybe we will get them or figure out what we need.
> This is an active area of science and we can hope some of the
>work being done now will help us see when we are approaching a
>dangerous situation in a definable time period, i.e., we could
>use this information for prediction.
>Since we don't have that
>kind of knowledge now, we have to fall back on the observed data
>about pandemics. That there have been three pandemics in 100 years,
> for an average of 3%/year, is the most simple minded way to do
>this. And since simplemindedness has its attractions, especially
>if there is no evidence that more complicated schemes are any
>better, that's what the UK authorities have relied upon.
>But
>that's the rub, here: relied upon. As an estimate of the yearly
>probability this isn't unreasonable. It is mathematically the
>maximum likelihood estimator of the annual probability, assuming
>independent occurrence of a pandemic each year. This means that
>the estimate of 3%/year is more likely than any other estimate
>that assumes independent events.

well, when ignoring what else we know. E.g. progress with vaccine,
antivirals,hygiene, no pandemic since 40 years but then :
increased population-density, airtravel,

>But it doesn't say how much more
>likely. Even for the (doubtful) assumption of independent events,
> there could be a wide range of estimates almost as likely as
>the 3% one. Here is one way to think about it. If you were betting
>on a number, 3% would be the one to use since it is the most likely
>one to come up (given the assumptions). But how much you should
>bet on it would also depend on how likely 1%, 2.5% 5%, 10% or
>20% was to come up, too. If they were almost as likely as 3%,
>you wouldn't want to bet a lot on a 3% annual occurrence. But
>that is what the UK plan is doing. They are betting a lot on it.
> Probably not a good strategy.
>Moreover the assumption that each
>year is independent of the other years (i.e., that the probability
>of a pandemic is the same each year) is almost certainly untrue.
> While we don't completely understand the dynamics of immunologic
>status of the population, the rate of emergence of different subtypes
>and the biologically competence of different subtypes to cause
>pandemics in human hosts, we understand enough to say the assumption
>of independent occurrence each year is doubtful at best, although
>it may still turn out to be a reasonable approximation.
>Both
>the 3% estimate and the idea that we are "overdue" for a pandemic
>(these two views often held by entirely different people) have
>one thing in common: the belief that the future will be like the
>past. It is true that as far as we know human populations have
>been plagued with periodic pandemics with influenza A,

no periodicity can be seen. I assume you mean recurrent pandemics,
not periodic pandemics

>although
>the data are scarce historically speaking. But the same could
>be said of smallpox, malaria, polio and many other diseases which
>we have learned or are learning to control or that have disappeared
>because the demographic and environmental conditions that supported
>them have changed. There is nothing to suggest any such conditions
>make the likelihood of a flu pandemic any less -- just the opposite.
> In fact pandemics may be more likely than in the past -- or not.
> Nor is there any evidence we can prevent them happening, although
>some believe we should try.

there is some evidence. You just named how other diseases were beaten.
Why not also Influenza ? Much progress with antivirals
and vaccines in the last years. We have a good database to study
the viruses for the first time in human history now.
We have computers now and advanced communication.

>In any event, whether the future will
>look like the past we don't know, even as we aren't quite sure
>what the past actually looked like.
>If I had to bet money on
>whether a pandemic would happen within the next year or two, I'
>d probably say "no," based on past experience.

so less than 50% per year.

>But I wouldn't
>bet much on it. I certainly wouldn't bet my life on it.

if you could also win an extra life, you should bet.

See also gsgs' posts here:
http://www.flutrackers.com/forum/showthread.php?t=33085
and here:
https://www.singtomeohmuse.com/viewtopic.php?t=1483&sid=4ec31f7a44e673f…

Hmmm.... the blogsite cut a paragraph out.

Add in to the above just after para 4.

The fire spread inside of two hours to ten buildings and with four battalions downtown they couldnt get it under control. Having the fire department is a method of mitigation for both quake and panflu if a fire breaks out but not much more. We have healthcare facilities but they by their own admission state that they dont have the equipment, facilities or people to man a fully involved panflu event. Panflu is a fire waiting to start and we cant expect that there will be any real help after a week or two into it.

By M. Randolph Kruger (not verified) on 21 Aug 2007 #permalink

"Since I don't know what the probability of a pandemic is for the next year or the next five years and don't have an opinion about it (other than it would be foolish to believe it couldn't happen at any time), I won't give my own estimate (since I don't have one)."

Revere Thank you for this beautifully concise piece of scientific writing!!

Yogi Berra said it best:

"Making predictions is hard, especially about the future."

Recently, the insurance companies in light of Hugo, Andrew, Katrina, WTC's quit writing earthquake policies in the central US right or left of the Mississippi because of that 200% number.

If you look carefully, the insurance companies have also recently taken steps to limit their coverage (especially business coverage) to exclude losses resulting from 'viruses'.

We may not be able to predict when a pandemic will occur 'with certainty' but the international business people who get paid big amounts of money to calculate the probability of such risks so that their clients and employers can make 'ginormous' amounts of money have already placed their bets. They have not raised their premiums for such coverage, they have excluded it.

When it comes to deciding whether the odds of a pandemic are sufficiently high that the insurance companies should protect themselves, they are betting that the odds are high enough.

By Into the Woods (not verified) on 21 Aug 2007 #permalink

no, look at the prices of the mortality bonds.
Insurance companies transfer some of the risk to
the market, worth about $2 billion.

FWIW, recently I wrote a rebuttal to the notorious BMJ editorial FAFfing About (actually I wrote three, you can read them unedited in the rapid responses page here), and, to my surprise, they published the one entitled "evaluating pandemic risk" in their next print issue here. This is the one where I suggested, now that we have the ability to track the virology and epidemiology better than previous generations, we should make good use of the info and that

with regards to H5N1, we are in the same position as New Orleans 24 hours before Katrina

Now revere and others may disagree with this view, but I think it's a lesson for us to not give up on tptb and the media too easily, and to remember that even the most stuck-in-the-mud mainstream establishment skeptics can sometimes be persuaded to consider alternate opinions.

Ran across this today. It may reflect on the 2% vs 200% discussion. (Didn't something similar come down a few months ago?)JVI Accepts, published online ahead of print on 25 July 2007
This Article J. Virol. doi:10.1128/JVI.00921-07
http://jvi.asm.org/cgi/content/abstract/JVI.00921-07v1

Persistent Host Markers in Pandemic and H5N1 Influenza Viruses
David B. Finkelstein, Suraj Mukatira, Perdeep K. Mehta, John C. Obenauer, Xiaoping Su, Robert G. Webster, and Clayton W. Naeve*
Hartwell Center for Bioinformatics and Biotechnology, St. Jude Children's Research Hospital, 332 North Lauderdale Street Memphis, TN 38105-2794, USA; Department of Infectious Diseases, St. Jude Children's Research Hospital, 332 North Lauderdale Street Memphis, TN 38105-2794, USA; Department of Pathology, University of Tennessee Health Science Center, Memphis, TN 38105, USA

Avian influenza viruses have adapted to human hosts causing pandemics in humans. The key host-specific amino acid mutations required for an avian influenza virus to function in humans are unknown. Through multiple sequence alignment and statistical testing of each aligned amino acid we identified markers that discriminate human influenza viruses from avian influenza viruses. We applied strict thresholds to select only markers which are highly preserved in human influenza isolates over time. We found that a subset of these persistent host markers exist in all human pandemic influenza sequences from 1918, 1957 and 1968, while others are acquired as the virus becomes a seasonal influenza. We also show that human H5N1 influenza viruses are significantly more likely to contain the amino acid predominant in human strains for a few persistent host markers when compared to avian H5N1 influenza viruses. This sporadic enrichment of amino acids present in human-hosted viruses may indicate that some H5N1 viruses have made modest adaptations to their new hosts in the recent past. The markers reported here should be useful in monitoring potential pandemic influenza viruses.
Abstract (above) is free. Full article is by subscription only.In the popular press:
http://www.reuters.com/article/scienceNews/idUSN2136874820070821

MRK-
Your evaluation of earthquake probabilities in the mid-west is faulty. Perhaps you are confusing probabalistic seismic hazard risk:
http://earthquake.usgs.gov/research/hazmaps/haz101/haz101.php#why
with earthquake probability, but either way, probabilities are not additive. If you flip a quarter, you have a 50% chance of getting tails, if you flip it twice, you don't have a 100% chance of getting tails.
Reference earthquake recurrence intervals are averages, or means, and there may be individual time periods between earthquake that are significantly shorter, or longer, than the mean value. The USGS does not release probabilistic predictions of equal to or greater than 100% (meaning an earthquake is guaranteed to occur in given time frame) because that is not possible to predict.
In fields where there IS enough data to make probabilistic predictions based on specific models, it's important that people understand what those predictions mean.

Good post Revere. If you haven't already seen it, you should check out the study released by Swiss Re, the giant global re-insurance company -- presenting their assessment of the excess mortality risk posed by pandemic influenza http://www.swissre.com/INTERNET/pwswpspr.nsf/alldocbyidkeylu/SHAN-753GMT. For a number of reasons, I think it is a very interesting study, and is the only pandemic risk assessment of which I am aware that attempts to incorporate the differences between the social and technological environments associated with recent influenza pandemics (1918, 1957, 1968). Their result may be controversial, since the bottom line of their assessment will probably strike some who follow the issue as way too benign. But I appreciate their explicit reminder that while we are worrying about pandemic emergence, alot is happening that may be changing the character of the impact once it arrives. Who knows? Not me.

On the emergence prediction side of things, you should also be aware, as I'm sure you are, of the work being done by Dr. Daniel Janies, et. al., at Ohio State University on mapping the evolution of H5N1. http://bmi.osu.edu/news_detail.php?id=54 It's quite interesting...

I think we have been approaching the topic of pandemic prediction from the wrong direction...as pointed out to me by something Revere said the other day.

The question is can we make a historically and scientifically based sound argument that the pandemic could start at any point in the immediate future...and I believe the answer has become very clear in the last 8 months...and we will and are seeing a different approach, as a result, from regulators and governments.

Dr. Nabarro probably said it best...'We are on God's time'.

With every passing day, we are one day closer to the inevitability of a pandemic...lets hope it is a mild one and lets hope that it doesn't occur for years or decades...only time will tell!!

he swissre link above doesn't work

CC with all due respect you are looking at the 50 year projection maps I think. You should try this where it clearly shows a 40% chance right now of exceedance in the NMSZ. Use California as you know they have them there neraly every day, but forecasting the big one isnt close to being called science yet. .

http://gldims.cr.usgs.gov/nshmp2007/viewer.htm

Play around with the %.g on the right hand side. You will see we are at 320% in and around Memphis and St. Louis if you take it deep enough for 2%.

Its all based upon when the last one happened and WHERE it happened. They have it down now to the nitnoys of even a 2% exceedance. We are 1% a year since 1812-1813 based upon a 200 year event cycle as determined from the soil samplings in MO, TN, AR, MS. A 200 year give or take cycle was determined and for each year that it doesnt happen a 1 percent chance of it happening is added. Its not a percentage of prediction, its a percentage of probability. So if they are able to bring the two together they can narrow it down.

Each one of the known faults is mapped and assessed a number of percentage of probability. That is reversed to the easily understandable percentage of possibility. The possibility exists that it will happen today, right now and its about 1%, but because of probability the numbers say for a New Madrid quake we are at a 40% possibility within the next fifty years which turns that number back from the higher probability. In otherwords its not likely to happen today, but tommorow there's a better chance because it didnt happen today.

The possibility of an event within the next 50 years is quite high but the probability based on a single day is lowered or raised by the number of days we didnt or did have one, but it is a rising number. At the end of the 50 year phase the numbers will be astronomical that we would have one. That is if we hadnt had one already. Confused? Go with the concept that like bird flu, you cant predict it except to say its inevitable.

Potentialities are adjusted each year by the USGS and by their reseach. They were exploding charges downtown for a year two years ago and it produced a map. No problem they said. That map got dashed when they had a quake in the north side of town last year. Localized, small, but there. They then started chasing down thru streams, cutting channels in the ground and found that there is a huge fault under the city some 70 miles down... So much for the big ball game plan.

Refer to: Investigation of Quaternary Faulting in the City of Memphis and Shelby County, Tennessee.

There is a huge document there if you want to Google it and also you should look at what is known as the SG-4 paper as well. The raw data also CC is available and it shows the percentages for basically every lat/long in the area and its used in 2, 5, 10% probabilities. There are so many sites on the map that a single generation probabilities one was created. Overall... now 40%.

Lets hope that we get past either flu or the quake and dont have them back to back. Either one would hit us so hard that it might be mortal to the country.

By M. Randolph Kruger (not verified) on 24 Aug 2007 #permalink

CC with all due respect you are looking at the 50 year projection maps I think. You should try this where it clearly shows a 40% chance right now of exceedance in the NMSZ. Use California as you know they have them there neraly every day, but forecasting the big one isnt close to being called science yet. .

http://gldims.cr.usgs.gov/nshmp2007/viewer.htm

Play around with the %.g on the right hand side. You will see we are at 320% in and around Memphis and St. Louis if you take it deep enough for 2%.

Its all based upon when the last one happened and WHERE it happened. They have it down now to the nitnoys of even a 2% exceedance. We are 1% a year since 1812-1813 based upon a 200 year event cycle as determined from the soil samplings in MO, TN, AR, MS. A 200 year give or take cycle was determined and for each year that it doesnt happen a 1 percent chance of it happening is added. Its not a percentage of prediction, its a percentage of probability. So if they are able to bring the two together they can narrow it down.

Each one of the known faults is mapped and assessed a number of percentage of probability. That is reversed to the easily understandable percentage of possibility. The possibility exists that it will happen today, right now and its about 1%, but because of probability the numbers say for a New Madrid quake we are at a 40% possibility within the next fifty years which turns that number back from the higher probability. In otherwords its not likely to happen today, but tommorow there's a better chance because it didnt happen today.

The possibility of an event within the next 50 years is quite high but the probability based on a single day is lowered or raised by the number of days we didnt or did have one, but it is a rising number. At the end of the 50 year phase the numbers will be astronomical that we would have one. That is if we hadnt had one already. Confused? Go with the concept that like bird flu, you cant predict it except to say its inevitable.

Potentialities are adjusted each year by the USGS and by their reseach. They were exploding charges downtown for a year two years ago and it produced a map. No problem they said. That map got dashed when they had a quake in the north side of town last year. Localized, small, but there. They then started chasing down thru streams, cutting channels in the ground and found that there is a huge fault under the city some 70 miles down... So much for the big ball game plan.

Refer to: Investigation of Quaternary Faulting in the City of Memphis and Shelby County, Tennessee.

There is a huge document there if you want to Google it and also you should look at what is known as the SG-4 paper as well. The raw data also CC is available and it shows the percentages for basically every lat/long in the area and its used in 2, 5, 10% probabilities. There are so many sites on the map that a single generation probabilities one was created. Overall... now 40%.

Lets hope that we get past either flu or the quake and dont have them back to back. Either one would hit us so hard that it might be mortal to the country.

By M. Randolph Kruger (not verified) on 24 Aug 2007 #permalink

MRK-reread my post, and both of yours.

MRK: "Play around with the %.g on the right hand side. You will see we are at 320% in and around Memphis and St. Louis if you take it deep enough for 2%."

To be precise, that map that you link says that there is a 2% chance of having a strong ground motion event where the acceleration reaches 320% of the acceleration due to gravity within 50 years, i.e., over three g.

In your original posts you said "That 200% number is the probability of within the next 50 years that we will have a =>7.0."

I repeat, probabilities are not additive, and earthquake recurrence interval time periods are averages, individual time periods between large earthquakes can vary greatly (there can be a large standard deviation around the mean).

Recurrence intervals for an 1811-1812 style earthquake (Magnitude >7.6) are ~500 yrs +/- 300 years [Tuttle, 2005, Nature 435, 1037-1039; Tuttle, 2002, Bulletin of the Seismological Society of America, v. 92; no. 6; p. 2080-2089] Recurrence intervals for a M>6.0 are on the order of ~100 years.

I think we can both agree that the potential damage from even moderate earthquakes (Mag=5-6), much less strong earthquakes (Mag>6), is great, due to the relative lack of preparedness and the lack of earthquake resilient buildings in the mid-west, compared to areas such as the U.S west coast. (plus ground conditions which introduce less energy damping and thus damage is spread over a greater area)

However, when trying to educate the public, it does no good to make vague and misleading claims such as there is a 200% chance (meaning guaranteed twice over) of a severe earthquake in the next 50 years. Probabilities are not additive, something like that cannot be claimed. The New Madrid fault zone has the potential to create a great disaster for the U.S., but a calm, precise and logical statement of the potential is more important to disseminating that information than vague exaggerated claims. I would think the same would go for attempting to quantify the risk of pandemics.

BTW, I'm not confused, I'm a geologist. If I want to make authoritative statements like I have, I guess I should declare that.

CC that last statement runs counter to what I was told here. It is a cumulative aggregate prediction based upon the history here. In November of last year 05, I sat in on a regional meeting of everyone from Senators, Congressmen, State and Federal and I didnt miss any part of it. Arnsdale was one of the key speakers along with Represenative Ford, state majority leader Norris, half of the Memphis City Council, County Mayor Wharton, State EMA Director Bassham.

The statement was made that it was indeed a cumulative effect here and that it was based upon a best guess. Each day we dont have one is a march like the lemmings to the sea and we are not alone along the Mississippi. The calm approach as you put it CC hasnt carried any weight and now with the findings of the new faults in downtown Memphis, the Ellendale fault and one or two others and the information from the SG-4 put it squarely on the front burner. The SG-4 indicates an 850 cm uplift in and around Armorel Arkansas and more than that in many places. It would fail the levee system. Want to place bets on whether the river would flood all of Eastern Arkansas in a Katrina manner?

That little statement caused a stunned silence in the room when it was presented. Prior to this it was indeed the way you suggest it. Status quo. And they did go out on the limb to say it because as with all things scientific they can point to what they have today and say thats what we think, only the next day be hammered into the ground by
another study. Nothing vague about living in the NMSZ. If it happens it will destroy everything for the better part of 200 miles either side of the River from Jackson Ms. to Chicago.

But no matter, Memphis with its brick and mortar downtown old buildings would be gone very quickly. What doesnt fall down will burn down. That was stated too.

But I only repeat what was said. I have no idea where you are. You obviously cant make the statement that its not a cumulative effect as was asserted here either. I asked the same question and got that answer above. Application of percentages are nothing more than a guess and thats not vague especially if the lithofacies show a time line that indicates a big one about every two hundred years. I dont discount what you say though CC. I do though disagree with it. You could be right. But be advised those same models missed the nearly back to back 4.5's we were supposed to have and they were off by a mere 85% on their probabilities on that. That came up too in the meeting.

Jonesboro AR had a NMSZ meeting the night before the first one happened. The message, "Prepare but dont worry, it likely wont happen in your lifetime." The Army Corps rushed over to check the levees for damage and found a tad but not bad. The quakes were only 1 second RMSS out to 1.75.

As Revere states though about panflu, I wouldnt want to bet my or my families life on it that you are right.

By M. Randolph Kruger (not verified) on 24 Aug 2007 #permalink

so SwissRe thinks that a 1918-like pandemic
("1-in-200-years-event") nowadays would cause
300000 (=0.3M) excess deaths in USA.
They say Ferguson had reviewed the model, but don't
quote what he said about it.

apparantly H5N1 had been excluded from that model.As I got
the feeling that the actuaries since years seem to ignore
the H5N1-threat, while making the reader believe it were included in the analysis.

-----------------------
> The H5N1-virus that is currently infecting birds,
> and occasionally people,
> has been unable to transmit efficiently and
> sustainably between humans.
> Given that no H5-subtype has ever caused a
> human influenza pandemic,
> this virus may never be able to do so.
> However, the SwissRe model does make an allowance
> for such a development,
> but as a remote possibility.

how the article is done and several formulations
make me believe that SwissRe has an interest to present
the risk as lower than they think it is.
What could be their motivation ? Usually an insurance
company would want to present the risk as larger than they
think it is. But they are selling mortality bonds to the market. Worth about $2 billion.

MRK-
You are completely misunderstanding my point that you confuse hazard with risk and play loosely with probability math. I only belabor it here because one can compare and contrast the hazard/risk analysis for earthquakes to other natural disasters, including pandemics, and there are preparedness trade-offs that must be made with limited funds.

The issue is not whether the New Madrid seismic zone(NMSZ) is accumulating strain, the best paleoseismology and geodetic research shows that it is, and the Earth's crust can only accommodate a finite amount of strain. Eventually the stress exceeds the strength of the crust and earthquakes happen.

However, since we are still unable to predict the time, place and magnitude of future earthquakes, geologists instead use probabalistic estimates of the liklihood of an earthquake occuring in a given window of time, space and magnitude, specifically in order to "transmit earth science information on future earthquake hazards to officials in charge of emergency preparedness, earthquake engineers and the public. This trend is welcome because the probability conveys the information in a form which can be dealt with in response planning and cost-benefit analysis of mitigation" (Aki. U.S. National Report to IUGG, 1991-1994 Rev. Geophys. Vol. 33 Suppl., � 1995 American Geophysical Union)

For example, in the NMSZ, we have about 2000 years of (incomplete) records from paleoseismology, a few hundred years of historical records, and about 10 years of geodetic data. During the historical time period, there were two earthquakes (retroactively) determined to be between magnitude 6 and 7, the first in 1843 and the second in 1895. The USGS, scientistists at SIU, and the Central United States Earthquake Consortium agree on an average(note the average-and remember there can be a large standard deviation around the mean) recurrance interval of 80 years for earthquakes of that size. The estimated probability is that there is a 25-40% chance of an earthquake of that size on the NMSZ in the next 50 years.

Lets look at that. It was 50 years between those two quakes, but it's been 112 years since the last. According to your analysis of probabilities in your first post(although you contradict yourself) "we have nearly a 500% chance of that[a magnitude 6 earthquake] happening statisically". What if you took that to a policy maker and they made priorities based on that figure, raised taxes, forced hospitals to retrofit instead of hiring more staff, halved again the cost of construction to build everything up to higher codes, and there was no earthquake during the next 20 years, but there was another major hurricane, or catastrophic flooding, or a pandemic during that time frame? The probabilities aren't additive that way. And what would the failure for those predictions to occur have on the possibility of future funding for earthquake preparedness?

Will there be more earthquakes on the NMSZ? Yes, everything points to relatively long term stress application on the NMSZ, and accumulation of strain. Will they be devasting for a region unprepared? Yes. (Would I live in an unreinforced brick building in that region? Hell no). But, the short/medium term probabilities you calculate in your first post make no sense. A magnitude 6 earthquake may happen tomorrow, or before I can post this, but may not happen for another 40 years, a magnitude 8 earthquake (like 1811-1812) could happen next year, or may not happen for another 600 years (or longer). Realistic (based on the best knowledge at the time- which is why i quoted the author and journal in my prior post with the recurrance intervals for earthquakes M>7) estimates of relative probabilities are crucial for public planners. There are very severe hazards associated with a major earthquake in the eastern mid-west of the US, but a realistic assesment of the risk of an earthquake is necessary for decisions to be made. Do you retrofit all hospitals in the region to withstand an earthquake of M>6 within the next 10 years, at the expense of staffing? Or do you increase staff and equipment to handle more medical care in the event of a pandemic? Ideally you could fully fund both avenues of preparedness, but with limited funds decisions must be made.

Additionaly, the final report for the SG-4 grant from the Mid-America research center by Dr. Arsdale (not Arnsdale) is a great piece of work characterizing the long term tectonic evolution of the region and an attempt to reconstruct the deformation caused by the 1811-1812 earthquake sequence. One can use it to help determine the seismic hazard an earthquake would pose to the region: damage due to the rupture itself, direct building failure from the passage of seismic waves, liquifaction induced building failure and destruction of farmland, mass wasting events, reworking of river channels, and flooding due to subsidence &/or levee failure. The SG-4 final report did not attempt to quantify seismic risk- the earthquake probabilities.

Sorry for the length-

SwissRe is a 're-insurance' company, which can be thought of a wholesale insurance company. Retail insurance companies take part of their risk and pay SwissRe to backstop them. Insurance companies are very interested in managing risk. Re-insurance companies especially so. I would trust re-insurance analyses as long as it's what they're saying privately as well.

Another point is that Warren Buffett bought SwissRe. That's possibly the most authoritative recommendation you can find.

By Ground Zero Homeboy (not verified) on 24 Aug 2007 #permalink