Nick Zamiska had an interesting piece in yesterday's Wall Street Journal about the difficulty WHO is having encouraging a sense of urgency about a possible pandemic of avian influenza and the many other programs and problems with which they are daily engaged in a (literally) life and death struggle in many places in the world most of us have never heard of. The problem is made more difficult because of the roulette-wheel like nature of an emerging pandemic. We don't know when or if our number will come up or how much we should invest in the event it does. So far it hasn't happened, which is a far cry from saying it won't happen. It could be the next spin of the wheel, the hundredth spin, or the wheel could break before it comes up, ever. Meanwhile, TB, malaria, HIV/AIDS, schistosomiasis and many others exact a frightful toll around the world. It is a dilemma, because if the flu pandemic number does come up, the consequences could be frightful in ways none of these diseases are because of the effects on the global economy and social and international relations, not to mention the mortality and morbidity.
A typical reaction was that of Malaysia's minister of health, Dr. Chua Soi Lek, who said "the WHO has overemphasized the threat from bird flu at the expense of other, more pressing diseases such as AIDS, TB and malaria, which he says "don't have any glamour anymore -- they're not sexy. I'm not saying avian flu is not important, but we should have a balance....I have colleagues who have reservations, but they don't want to say it because they don't want to offend the WHO." (Nick Zamiska, Wall Street Journal)
But the threat of a pandemic has not receded. It is as real as it ever was, if not more so. Here is what a bubbling pre-pandemic looks like:
These graphs, covering the period from November 2003 to almost the present, show both the wider and broader case loads each season and the geographic spread, as seen in more and more colors. This is not the picture of a disease that is fading away, but one that is bubbling away.
Obviously a balance is needed, and striking that balance isn't easy. Neither WHO, nor nation states, nor municipalities, nor individuals should put all their eggs in one basket. But there are some kinds of eggs and some kinds of baskets that can do more than one thing, and shoring up our public health and social service infrastructures is one of the best examples:
The flurry of dire bird-flu warnings from scientists and public health officials alike, combined with the fact that a pandemic hasn't yet occurred, has already prompted a subtle shift in the rhetoric. Yesterday at the conference here, for instance, a senior Australian health official told her counterparts from across Asia that the work that countries were doing to combat and prepare for avian influenza would pay dividends down the road -- with other diseases.
I know I say this all the time. But it's something I've learned from the Bush administration: stay on message and keep repeating it. Over, and over, and over.
Zamiska calls arguing for the multiple uses of flu preparation a subtle shift in the rhetoric. I would put it differently. There is a time when the right message can be heard, and this is it. The existence of dire needs in many other areas of public health means that WHO has to keep pressing the urgency of the pandemic message, because without that kind of high level signal, regional and national health authorities will turn their attention to fighting fires and applying band-aids. Preparing for a pandemic involves longer term investment, making the house fire-resistant so fighting a fire in it is easier.
The bottom line, of course, is that there is much to do. There is more to global public health than bird flu. But pandemic preparation is deservedly near the core of global public health regardless of the many other needs. It is not exclusionary of those needs but complementary to them.
If we do it right. Which, alas, we don't always do.
- Log in to post comments
I received the following message a few days ago:
K A from Italy
it's A tragic situation like that in Iraq, I have a contact with Iraqi civilians in particular children and young boys, what they said that the hospitals more dangers than out side because of un related infections to the original disease, they comment, even a small burn injury will finish with an absolute death due to infections where usually un defined because of lack of diagnostic tools.
So, what I want to say to the Americans (the best people in the word) to look after the other humans because civilians died every day because of stupid decisions.
This morning I read this article:
Sep 19, 2006 ? GENEVA, Sept 19 (Retuers) - A three-year-old Iraqi boy in Baghdad has been confirmed as having survived a mild case of bird flu last March, the first confirmed human infection in the capital, the World Health Organization (WHO) said on Tuesday.
"The Ministry of Health in Iraq has retrospectively confirmed the country's third case of human infection with the H5N1 avian influenza virus," the WHO said in a statement.
The rest of this story is at
http://abcnews.go.com/International/wireStory?id=2462305
This cesspool of bugs brewing in Iraq under ideal conditions (for the bugs) should raise some hairs. We do not have proper precautions in place to prevent the spread of any of them by the heavy traffic in and out of there. Soldiers and civilians exposed to leishmaniasis are donating blood. Casualties are picking up MDR acinetobacter baumannii and its being spread from the military medical system to the community hospitals.
How could we even know how many cases of this flu there have been Iraq?
>The problem is made more difficult
>because of the roulette-wheel like nature of an emerging pandemic.
if it were roulette-wheel like, then insurances and experts
would have no problem to assign probabilities to it.
> We don't know when or if our number will come up or how much
>we should invest in the event it does. So far it hasn't happened,
> which is a far cry from saying it won't happen. It could be the
>next spin of the wheel, the hundredth spin, or the wheel could
>break before it comes up, ever.
but not all of these are equally likely.
We don't know, how the next superbowl will end, it could be 17:13
or 47:33, but it should be mentioned that the latter can be
considered much less likely. Indeed, the bookmakers will
give you exact quotes.
> Meanwhile, TB, malaria, HIV/AIDS,
> schistosomiasis and many others exact a frightful toll around
>the world. It is a dilemma, because if the flu pandemic number
>does come up, the consequences could be frightful in ways none
>of these diseases are because of the effects on the global economy
>and social and international relations, not to mention the mortality
>and morbidity.
>A typical reaction was that of Malaysia's minister
>of health, Dr. Chua Soi Lek, who said "the WHO has overemphasized
>the threat from bird flu at the expense of other, more pressing
>diseases such as AIDS, TB and malaria, which he says "don't have
>any glamour anymore -- they're not sexy. I'm not saying avian
>flu is not important, but we should have a balance....I have colleagues
>who have reservations, but they don't want to say it because they
>don't want to offend the WHO." (Nick Zamiska, Wall Street Journal)
the reason of this is, of course that the experts refuse to give
probability estimates and to discuss them.
How could we respect the warnings which come from people
who also say "no one knows about the magnitude of the risk" ?
If they refuse to quantify the risk, then why should we believe
them when they say that the risk is large or significant or severe -
they are creative in finding new words here just to avoid giving more
informative numbers.
>But the threat of a pandemic has not receded. It is as real
>as it ever was, if not more so.
but you just said some lines above that the risk can't be specified
and now you claim that it is "as real as it ever was" .
How can you know this, when you neither know how big the risk
is now nor how big it was earlier ? How are you comparing unknown risks ?
How are you comparing any risks if not by assigning numbers
to their estimated probabilities ?
>There is a
>time when the right message can be heard, and this is it. The
>existence of dire needs in many other areas of public health means
>that WHO has to keep pressing the urgency of the pandemic message,
>because without that kind of high level signal, regional and
>national health authorities will turn their attention to fighting
>fires and applying band-aids.
now, you justify the WHO warnings by the need to get the authorities
doing something. So, by some political timing not by news from H5N1 itself
and by a scientific analysis of the magnitude of the threat.
The authorities might see through this strategy and that further undermines
the credibility of WHO.
As the WHO supposedly has earlier based the intensity of its warnings
on the timing of some pending WHO-funding decisions.
Those warnings which for some obscure reason "can't" be quantified.
Honestly, even some of the "flu" sites have a terrible time staying on topic, so how can we expect the MSM, which has no particular focus to press, to find the topic of bird flu endlessly fascinating when even the flu sites make constant digressions into other topics? As a co-founder of one of the sites that includes the word "flu" in its name, I can testify that I fought endlessly against the dilution of the main message to no avail. It was impossible to keep a "flu" site focused on flu. It may be that the daily, unrelenting, and plodding work of H5N1-watching is just not exciting enough to some when there are "sexy" diseases like ebola to run with. People with decades of experience watching diseases in the third world do not become over-excited with news of a tropical disease outbreak somewhere, but such all too common events can completely derail a flu board and dilute its message. For some flu boards, the truth just may be that H5N1 is not exciting enough as a daily mission, and the broad variety of distractions out there may serve to alleviate that problem. For the MSM, the disease-of-the day is whatever sells papers, and that is their primary goal, one their shareholders remind them constantly to uphold.
WHO must, on a shoestring budget, deal quite literally with every disease known to man and we would rightly be angry if they lost that catholic focus. But no other disease, no matter how much compassion it may move in us, or excitement it may generate in those and uneducated in the subject of the world's diseases, has the potential to be as unstopable, as virulent, and as serious on a global scale as H5N1. There simply is nothing else like it. Nabarro and the other WHO officers tasked with preparing the world for a possible pandemic are doing yoeman's work, the hard daily work of public health officials who stay on topic, keep the lights on late, seek little glory, and are rarely thanked for their diligence.
anon: I would respectfully suggest that you consider taking another look at the "Cases by Onset Date and Country" graph above. The horizontal line can be fairly interpreted as the equivalent of a stock market analysis, of the "moving daily average" variety. Notice, for example, that for nearly the last three years there has been no significant dip, at any point (and there is no reason to assume that we will see any in the future), in this line.
You can translate this to a reasonable linear analogy: The "moving daily average's" inexorable horizontal increase with time can be imagined as a "burning fuse." History says -- convincingly, in my opinion -- that at the other end of this fuse there is a very large bomb. The fuse is finite, and each bit of ash is a reminder that we have a diminishing amount of time, until the bomb goes off. The progress of the fuse can be interrupted, and the "moving daily average" can take a precipitous, significant dive. But Nature, and history, strongly suggest that this will not be the outcome, here.
This is not a human-mediated, manageable event. It's not a market, and it's not a material object, that we can simply "sever" at some conceivable point. It is a relentless, observable, intractable, emerging natural phenomenon. As far as a "proactive" response is concerned, I believe that this lies almost entirely outside of our purview and present authority (we squandered any opportunity we might have had, long ago, through collective hubris, arrogance, and simple stupidity).
We have witnessed nothing but steady, quantifiable, irrefutable, forward progress and expansion -- on all fronts -- since the very inception of this incipient catastrophe. I've observed it from the very beginning. Not for a moment have I doubted what I'm seeing, here. H5N1 has spoken very convincingly, and repeatedly, on this subject, as far as I am concerned. As I mentioned on another forum, months ago, in the form of an anecdote. As a child I watched a house cat play with a mouse it had captured. The terrified mouse tried desperately to escape the cat. The cat continued with the game until it became bored...then it bit the mouse's head off. Cats do what they do. And influenza viruses do what they do.
We are clearly not ready for this thing. Not on any level that I can conceive of. And now, those who should prepare are collectively complaining of "fatigue." The cat's out there.
Pixie: MSM may mean mainstream media in blogospher, but it also means men having sex with men in the public health severe. I read your sentence as such, LOL
"Honestly, even some of the 'flu' sites have a terrible time staying on topic, so how can we expect the men having sex with men, which has no particular focus to press, to find the"
Was a bit jarring to say the least
Anon and Dylan: Dylan, while I do understand and agree with your perspective, you don't answer Anon's point. And Anon, you also miss Revere. Revere points out that the trend is going up and also, in so many words, passed performance is not an indicator of future events. Anon, from my reading, would like to know how you(Revere) know that the risk of one thing is greater than other, but Anon the probability of an event depends on the knowledge and certain factors. Moreover, the relationship between these factors is important as well. Your sports analogy has a finite set of known events with known relationships, 53 players with extremely specific stats (I still can't believe I lost on Monday night in the last 30 seconds). Fudge factors for home field advantage, coaches' passed performance can be statistically added as well.
We do know some things, how to assay, where it started, how far it has travelled, presentation, latency, transmission (for the most part), mortality, etc.
Anon, your point is well made, but do not forget that there are by far more unknowns about an H5N1 pandemic, if there is one, than knowns. No actuary, sports or not-related, would be able to create a credible *specific* probablity. Besides, bookies shift the line as the betting comes in so that there is even money on both sides. =)
For me, the important takeaway I think from Anon's post is that although masses suck at statitics and probabilities, it's hard for the masses to chew on trends.
Darin
Dylan - your version would probably garner more interest in the media than my original intended meaning!
Could you put a link to the original source of your graphic? It's quite difficult to read on the blog version.
Thanks!
bc: Inserted it in the post. And here it is again:
http://www.wpro.who.int/sites/csr/data/data_Graphs.htm
Dylan, there was also a chart about this recently by
Mingus on flutrackers showing that the increase
is only linear and not exponential. So there were ample
room for decades before it starts to become comparable
with other diseases. I can't find the link now,
maybe someone with access to flutrackers can find it.
Darin, the level of uncertainety or the large number
of "factors" does not prevent bookmakers from giving
quotes or football-experts from giving forecasts.
There is no principal difference. The question for
probability estimates _is_ being answered by people with
much less knowledge in polls and also by some few
experts, although most refuse.
Why do they refuse to participate these polls and
discuss the results ?
Good for you Pixie. I run an email list that pulls down all of the articles from the world every day and send them out to now over 1000 emails. Yep, you are right about the apathetic approach that everyone is taking. So I deliberately send pictures of peoples lungs, the clinical information of this shit as it takes a young childs life and I quoteth Revere often because if anything he does care like you.
It does matter what you do and you shouldnt give up because I always learned that its only a matter of hitting the right buttons. That includes the crap with our soldiers over there keep pushing the buttons. If you are there, grab a news guy from any country, take him into a room with one of those infections. As for the WHO, regardless of what you believe its as bad as our own CDC, before, during and what it will be after GWB leaves. It sucked in the past, and both will suck when the pandemic arrives. H5N1 if it comes will clean the slate, survival of the fittest. All I hear is about bonuses. Are these guys doctors or stock brokers? Justified in their departures? Yep, private sector pays better anyway.
here is the Mingus-link:
http://www.flutrackers.com/forum/showthread.php?t=10613
Thanks so much Randolph. Here's an example of H5N1 fatigue. Last year's "International Partnership on Avian Influenza" meeting was met with a room full of eager reporters. This year's first anniversary meeting was attended by a single-digit number of reporters, fewer in number than the presenters, in spite of that chart featured by Revere which shows no diminution of the threat. A description of the meeting:
http://blog.sciam.com/index.php?title=pandemic_flu_preparations_little_…
The fundamental problem is quantifying the risk that we will have an H5N1 pandemic within a certain timeframe. I'm wondering if any math modeling could help since this involves a threshold event. We had a long conversation a few months ago about tipping point math as I recall. Do we have enough data to plug into a tipping point model to evaluate a threshold date? Revere, remind us about the paper you I think you highlighted for us? (I don't think this was the paper by mills et al).
As a shrink I'd like to stress the process of habituation: one constant factor which is neglected at last. It has to do, paradoxically enough, with surviving.
When I lived in Indonesia, I was frightened every time when witnessing a car accident. You would certainly die when blood supply was an issue; they couldn't tell one blood group from another in the region where we lived. I won't go in details here. After a car accident, a driver could be blamed for it and bystanders killed him anyway if he survived. There were two things we did to adapt to this constant risky and stressfull life: 1. we integrated the risks in our way of living, by always having a professional driver thus reducing the risk of any car accident, 2. not thinking (and feeling) about this mortal danger nearly all the time (only remembered about it when confronted by another accident when we were passing it).
Doing so we enjoyed life pretty well and healty relaxed.
Maybe it's the same for a possible pandemic, prepare and enjoy life. Only when it's your professional responsability to reach as much readers as possible to get them starting or keep them going on with prepping, I think that's a completely different story, like you observed so well, Revere. I think it's like the mantra thing that we see every time when those sentences are repeated:
"Bird flu could trigger a pandemic ............. if it mutates ....into a form of the disease that can be passed between humans....." and so on. Curiously enough, in the real new action publications I don't read it anymore. Like in news releases someone is going to speak about the possible risks and how to prepare, or one is doing research on vaccines, and the like. They appear psychologically integrated and not habituated.
It's the good way of not causing panic or chronic stress, but instead give all opportunities to prepare.
Marissa, your idea of mathematical modeling could be used without making assumptions (human guesses about the variables) that might out-of-hand be accused of invalidating the results. For example, the Los Alamos modeling is often criticized for its assumtions and the numbers the humans have plugged into their formulas.
When a sophisticated financial person looks at the kind of chart Revere has positioned above, they see not necessarily a linear pattern, but instead what sticks out to them is the pattern of volatility. They trade the volatility. In financial models, the volatility can be and is measured. Maybe the volatility of this chart (and this situation) could be measured on its own, using the pure data above, with no other assumptions made - just the actual numbers. Option traders do this all the time, and it is in their interest to be able to measure the volatility itself within their models because that is where they make their cash, not on the linear trend, or the fundamentals. In the financial analysis of markets, they call this volatility "beta." I'd like to see that chart above run through a program that measures beta and volatility from a top outfit like Goldman Sachs. They understand volatility. The results, of course, will not be a predictor of "when" a pandemic will happen. But it might be a very interesting excercise to run such a model to see what the heights of those bars, and the intervals inbetween, may indicate and get some idea of the general level of fallout that could result. Just an idea, but one which may be a purer mathematical model than the kind done by Los Alamos which utilizes human assumptions (such as a 2.5% CFR, etc.). This kind of model, on the other hand, is very much blind to anything but the data. And yes, to the naysayers, I know its not a market - I know, I know. It's just an idea.
Pixie: Your comment contains some interesting ideas. Let me offer a brief response. These are not criticisms, just observations and clarifications.
I think you don't mean mathematical modeling but statistical modeling. They are a bit different. You are right that financial futures modelers do this fairly routinely, but it's not quite correct to say that financial folks "trade volatility". Volatility is important to them so that they can accurately measure risk. What they do trade is risk-reward, and so modeling volatility (better to say "variability" outside of a financial context) is important.
The typical way to handle this problem with surveillance data is to assume some kind of probablistic ("stochastic") process that changes over time, that is variability and mean change over time. Most of the methods used in statistical modeling of time series comes from the econometrics folks. Because of the differences in main interests, epidemiologists are several decades behind them, but we also have data issues and aspects of our problems that are very different from controlled financial markets, and so it's not clear that sophisticated methods designed for financial time series are going to improve our understanding of these data.
There are several models built to identify "changepoints in time series data," either changes in mean or changes in variability. Industrial people who work with control processes (quality control/industrial engineering etc) have been doing this for a very long time. Epidemiologists and bioinformatics scientists are now doing this with something called Hidden Markov Models to detect when a significant change occurs in a series of measurements taken over time. In essence, it looks at past patterns to predict future ones and when the prediction is very different it sends a signal to this effect. This is different than the financial approach of modeling variability but it is meant to do the same thing and it isn't clear which is more useful for epidemiological data. I consulted an expert in doing this and his response was, "If the mean is the same but the variability is greater, how do I interpret what's going on? I would be more concerned about inconsistent data collection and reporting as being an explanation for any changes I see in the time series."
Hope that is useful in some way.
Revere yes, statistical modeling is really what I mean vs. mathematical modeling.
But on the subject of volatility and variance, I think the financial modelers may have an edge in getting a grasp on the chart you show above, so heavily populated as it is with peaks and valleys. In finance, they draw fine distinctions between "volatility," "variance," and CV or "coefficient of variance" and each has a very interrelated but specific meaning. Epidemiologists, as you point out, are looking for a significant change to alert them that something new is happening. The models you spoke about do that, and provide that measure, since they alert them to new areas of concern. But once the kind of change we see in the chart above is detected, what then? I still think that financial models that parse levels of change, speed, duration, risk, etc., to a very fine degree may be helpful in taking analysis of the data in that chart above one step futher.
I'd argue that financial markets are not controlled, and are in fact wild open territory where it is likely that a new product or player or event will be introduced at any moment. Responsive - yes, but not controlled. They wish.
The biggest flaw in my adoption of a financial model here is that those models need a standard to derive their standard deviation from, so to speak, so what would our standard be? Perhaps the "norm" then would have to be a flu-free environment, or a "regular flu" environment. Still, I think that would be less subjective than human injected variables.
Your expert's concern that the data collection itself presents an issue is, of course, always a valid concern. But the spikey chart you highlight above contains data that most of us would take little issue with (although some would argue that its spikes are understated due to incomplete data collection). It's pretty clear a change has already taken place. Now the problem is how to measure the velocity and range of the change. I think that takes us beyond simple time series data analysis. The chart above, just examining its patterns and apart from its content, is a familiar chart to an options trader, but probably one that most manufacturers or epidemiologists (and even econometricians) would have nightmares about. I'm frankly glad that our economists and epidemiologists don't encounter this kind of pattern frequently! But the finance guys do, and maybe they can shed some light on it. What your expert asked, if "the variability is greater, how do I interpret what's going on?" is the crux of the problem. But to begin to get answers to that question, let's run the numbers first, see just what the variability is (even if it is "greater"), and then figure out what it means.
Pixie: Thanks for your observations. The data quality issues are a problem because it isn't so much that they understate by a fixed amount through time (that we could handle) but that the probability of ascertainment is also fluctuating in ways we cannot model.
The other issue I see as problematic for the use of financial models has to do with numbers. We are dealing with extremely tiny datasets here, compared to a financial modeler, where tens of thousands of datapoints (or more) are involved. That engenders some pretty severe power constraints.
However I am very much in your camp in the sense that we need to find new and effective tools to analyze these problems. Epidemiologists are very backward in that regard. They have hammers and make everything look like a nail. My own research involves other kind of mathematical tools than Markov Models or financial analyses, and although I know a bit about the mathematics of option pricing, not much at all. There are surprisingly few people working on this. There just aren't enough experts and too much to do and much of the work is highly specialized and one can't move from one technique to another easily.
You have good ideas, though. Maybe someone else is working on it. Or you?