Monday, May 11

Swine flu/H1N1: Your life, riding on the CDC's slow boat from China


I wanted to plunge straight into my argument with the Centers for Disease Control and Prevention (CDC) but I decided to first outline my basic position.

There have been several developments on the swine flu front since I last posted, which produced reactions in many Americans that range from, "Whew! We dodged the bullet!" to "I knew from the start the government was making a mountain out a molehill because, after all, 36,000 Americans die every year from regular flu."

From where I sit both reactions miss the point, which is that the CDC's approach to crisis management is on par with the state of Louisiana's before Hurricane Katrina struck. And the CDC's approach to intelligence-gathering about emerging disease threats makes the pre-9/11 CIA's intelligence-gathering efforts look cutting edge.

From top to bottom, the CDC's management of the swine flu outbreak has been an unmitigated disaster. They are so far behind the times, and their crisis management is so completely disorganized, that you don't want to think about how many Americans would be dead today if the Piggy Flu, as one British wit dubbed it, had been an evil genius.

As for the 36,000 influenza deaths per year, which has been quoted ad nauseam by commentators since the swine flu outbreak in the effort to put the outbreak in context: it turns out that the number is a great illustration of the dictum, "Lies, damned lies, and statistics."

The CDC has no idea how many deaths there are each year from the flu because the number is a computer-generated guess based on mathematical modeling -- a model that's been used for more than 40 years, and which needs serious updating.

As the CDC's spokesman, Curtis Allen admitted a few years ago to medical journalist Kelly O'Meara, it's not a "real" number. He told her, "There are a couple problems with determining the number of deaths related to the flu because most people don't die from the influenza...We don't know exactly how many people get the flu each year because it's not a reportable disease and most physicians don't do the test [nasal swab] to indicate whether it's influenza."

However, using the CDC's own data, O'Meara managed to turn up that, "The greatest number of actual influenza deaths recorded since 1979 were 3,006 in 1981." (1, 2)

(A big hat tip to Deirdre Imus, writing for The Huffington Post, for digging up O'Meara's reports.)

Yet the same pesky computer modeling that's greatly overestimated the number of regular flu deaths in the USA is now greatly underestimating the number of swine flu cases and for the same basic reason -- hospitals and physicians are not swabbing every patient who presents with the symptoms of swine flu -- even when the patient is in a region that has shown swine flu infection.

The CDC is making a guess based on the number of swabs that have been tested and found positive. Yet even with all the will in the world to obtain a fairly accurate count, the CDC and public health officials at the state and local levels would be unable to do so because they were caught flat-footed by the outbreak.

They simply didn't have the protocols in place for dealing with that kind of emergency. It wasn't until the weekend of May 2-3 that the CDC finally got a testing kit to several localities, including New York City -- which had been a region hard-hit by the new swine flu virus. (3)

What did New York health officials do before that weekend? They did what health officials in every other U.S. state did: twiddled their thumbs waiting for the CDC laboratory to plow through the mountain of swabs for swine flu cases. That lab was the only place in the country that was equipped to test for the new virus. And it wasn't until May 5 that the CDC got test kits to all 50 states.

To put this all another way: While the CDC does the noblesse oblige thing for other nations and tests for diseases all over the world, such as hoof-and-mouth disease in Pakistan, citizens of the United States of America and their tax dollars have to make do with a horse-and-buggy version of infectious disease management.

But all that outrage pales next to the antiquated reasoning that informs the CDC's view of how best to handle a deadly epidemic that's jumped the US border from another country.

Following the swine flu story has not been easy for me -- and most others, for that matter. Often, news reports seemingly contradict each other, sometimes on an hourly basis or supersede each other at the same rate; there are significant gaps in the timeline of events; and parts of the story are incomprehensible to someone without a degree in a biomedical field.

Because I was out of my depth in the scientific/medical areas I chose to study the situation from the viewpoint of national defense, which I'd gotten familiar with during my years of closely following the Iraq theater of war and the war on terror in general.

I focused on news reports about how the Centers for Disease Control and Prevention and other U.S. government agencies were handling an emergency response to the swine flu outbreak.

I wasn't long into the task before I noticed a glaring contradiction in how officials at the CDC thought about emergency measures for dealing with a highly infectious disease outbreak. It doesn't take a degree in a scientific field, or even extensive knowledge of facts on the ground, to notice a sloppy premise and to argue against it. That's what I began to do in the last post, and that's where I pick up from in this one.
"One reason for the delay in stronger guidelines is that swine flu caught [CDC] planners off guard; they had anticipated being able to recognize a pandemic overseas, weeks or at least days before it hit the United States." (4)

In the old days -- the really old days -- a virus that started a pandemic far from home had tremendous character. Clinging to the wings of migrating fowl for days and weeks on end, getting buffeted by high winds and pelted by driving rain and snow, the virus that wanted to see the world had true grit.

Then, in the eras of globalized ship travel, an adventurous virus had to endure life in the bowels of a floating shack while cooped up for months with a bunch of smelly ill-tempered rats.

Travel conditions greatly improved for the intrepid viral world traveler with the advent of commercial air travel. These days the pampered virus takes an air-conditioned taxi to the international airport, flounces to a first-class seat, then disembarks a few hours later on the other side of the world, fresh as a daisy.

Yet planners at the CDC are laboring under the somewhat romantic notion that they have as long as it takes a clipper ship to sail from Old Shanghai before they must leap to defend the USA against a deadly viral outbreak on the far side of the world.

How did the planners arrive at such an antiquated notion? From the same report I quoted above:
[T]he official Pandemic Influenza Operation Plan, or O-Plan, of the U.S. Centers for Disease Control and Prevention, is based in large part on a history lesson -- research organized by pediatrician and medical historian Dr. Howard Markel of the University of Michigan. Markel was tapped by the CDC to study what worked and what didn't during the 1918 flu disaster>
In other words, the CDC's O-Plan is based in large measure on lessons learned from the 1918 outbreak.

If you say, "B-b-but this isn't 1918" -- Exactement!

While it is true that those who do not know history are doomed to repeat it, it's also true that those who don't factor present circumstances into their historical analyses can get us all killed.

Markel's research on the 1918 pandemic actually turned up some excellent advice. He stressed the importance of non-pharmacological interventions or "NPIs," as they're called, such as isolation of infected patients. However, from the viewpoint of dealing with a highly infectious disease, here is the critical difference between 1918 and 2009:

Today, a superkiller virus from the other side of the world can alight in large numbers on the same day at several U.S. international airports scattered all around the USA -- even before the lethality of the virus has been discovered by a government.

The key concept here is "appearing at several points at roughly the same time."

This is a very different concept from "spreading from one point."

Several days ago CNN showed a computer model of how ten isolated disease cases in California can, through exponential growth of infections, infect the entire country. This happens because each one of those ten people is assumed to infect an average of say, three people, and those infect three more, and pretty soon you need higher math to keep count.

That's not what happened in the USA with the swine flu virus. Infected people got off airline flights from Mexico at different points around the USA within a short period of each other.

New York City caught a break because for reasons known only to itself, the swine flu virus that arrived from Mexico with a bunch of private school students seemed to fizzle after clobbering 1,000 New Yorker within a few days. Medical detectives are all over the case. (5)

But the flu's behavior in New York, and the fact that it seemed relatively benign, tended to mask the critical need to update emergency planning for an era in which a serious disease outbreak can suddenly appear at several different points -- near large population hubs -- within a matter of hours.

Instead of updating the CDC stodgily clung to a fallacy, which they foisted on every U.S. official who had to comment publicly on the outbreak. On April 27 Janet Napolitano told a reporter:
"You would close the border if you thought you could contain the spread of disease, but the disease already is in a number of states within the United States."
Mr Obama echoed her statement. When asked at his April 29 press conference whether he was considering shutting the border with Mexico, he replied:
"I've consulted with our public health officials extensively on a day-to-day basis, in some cases, an hour-to-hour basis. At this point they have not recommended a border closing. From their perspective it would be akin to closing the barn door after the horses are out, because we already have cases here in the United States."
Ms Napolitano and Mr Obama were accurately outlining the reasoning that dominates at the CDC.

The problem is that the reasoning is based on the fallacy that stopping a viral outbreak is the primary goal once a highly infectious disease has arrived in a country. That thinking is dangerously wrong because at that point it's critical that the primary goal be shifted from containment to slowing the rate of infection. Why? I'll let a top CDC consultant, Dr Ira Longini, explain:
"The name of the game is to slow transmission until a well-matched vaccine can be made and distributed." (6)
A more precise statement of the goal is simply to buy time:

Yes, time until a vaccine can be developed, tested and distributed. But also time until health facilities can gear up to deal with an epidemic. And time so that the health facilities, and the society in general, won't be overwhelmed by the lightning-fast spread of an infectious disease.

This time-buying strategy is particularly vital today in the USA, given that the average of "3" (the "R" number or number of people that one infected person can in turn infect) is based on analysis of the 1918 pandemic in the USA not on the way things are today. (4)

Not only was the US population much smaller in 1918 and much less densely packed, but the volume of interstate travel and even travel within a city was much smaller. In 1918, many city people in America rarely if ever left their immediate neighborhood.

And there was no real suburbia, as we have today, from which millions of Americans in the workforce commute round-trip to a major population hub during the work week.

Also, you didn't have 4,000 flights a day between Mexico and airports all around the USA, and God Knows how many flights between many points in the USA and other areas of the world that are incubators for lethal exotic virus combinations.

So it's a safe assumption that the average number of people who can be infected by one disease carrier is higher today than it was during 1918.

More importantly, statistical averages go out the window when the disease enters the USA at several points within a short period. The R number might initially be 3 in Atlanta but in the densely packed mega-city of New York, R could be initially higher -- or vice versa, depending on the number of infected people who alight daily from the nearest airport hub near both cities.

The planners seem to have discounted the fact that the United States of America takes up much of a big continent. So the thinking about how to slow the speed of disease transmission should have been directed to seeing where the rate of infection could be slowed from region to region of the country.

In short, the United States isn't a barn and a virus is not a horse.

So of course it slows the rate of infection to suspend air flights and order a border and ports blockade when the country next door is in the middle of a deadly epidemic. Critically, in the early stage of the epidemic, the rate could vary wildly from region to region of the USA for several days or even weeks or months. The variables would translate into many lives saved if time-buying measures were immediately deployed.

If it still doesn't make sense to you to deploy draconian measures when a government knows that the measures have no hope of actually stopping the virus, think it through:

If 5 infected people slip through a border blockade or get past a temperature monitor every day during the outbreak -- if the measures halt 20 infected people every day, that's 20 people a day who've been stopped from infecting an average of 3 additional people each, per day. That's stopping those people who in turn would infect 3 more, and those who would infect 3 more -- and pretty soon you need higher math to keep count.

In short you're racing to brake the speed of the infection transmission in many regions around a vast tract of land, and for this you do everything you can think of:

  • Temporary border blockade

  • Quarantine of a suspected cluster of infected people

  • Temporary suspension of airline flights to and from a country that has a widespread outbreak of the highly infectious disease

  • Installment of "temperature-taking" thermal imaging machines at all ports -- I repeat "ports" -- in conjunction with tactics to offset the limitations of the machine's camera

  • Temporarily stopping all but essential off-loading at ports from a country with a highly infectious disease outbreak

  • "Social distancing;" e.g., urging the population to stay away from situations that have crowds of people, urging them to avoid close quarters with non-family members, etc.

  • PSAs hurled within hours onto every major media outlet to warn people to pay scrupulous attention to personal hygiene, such as hand-washing.

  • Again, no one of these tactics or a combination will stop a highly infectious disease from spreading once it arrives in a country. But if they can just slow the rate of infection by a fraction, while that won't stop deaths from a killer disease epidemic it would translate into thousands or even millions of human lives saved -- and a nation's social fabric hanging together during a deadly pandemic.

    Realize that the most drastic measures; e.g. blockade, would not need to be lengthy. How long? Until officials got a handle on the estimated length of the virus' incubation time.

    Yet what's most striking about the pattern of swine flu infection is that it didn't spread upward and outward from the U.S.-Mexico border region. The early pattern of the swine flu outbreak in the USA points to airline travel as the fast disseminator of the virus, not the Mexico-U.S. cross-border vehicular and foot traffic.

    It could turn out that some infected truckers who were heading from Mexico to various points in the USA or Canada helped spread the disease at pit stops across the country. But from what the numbers were saying up to May 5 (the date of the report I'm referencing) there's not much infection in the U.S. border states in relation to other states that have seen outbreaks. (7) And it's the early data on this phenomenon that count most.

    With three possible exceptions, which I'll discuss in my next post, virtually all the swine flu cases in the USA represent infected people who recently arrived from Mexico by plane, who were infected by contact with disembarked infected passengers from Mexico, and who in turn passed the infection to others. (6)

    May 12, 12:40 PM ET Update
    I'm going to interrupt myself here. I've just realized there's a glaring contradiction in my reasoning. I note in this post that swine flu test kits didn't arrive in all the states until May 5. And even if the border states were among those that received kits as early as the May 2-3 weekend, that doesn't necessarily mean that the test results show up in the numbers quoted in the report I reference.

    And the numbers in the report don't necessarily mean that the CDC had been able to test their own backlog of swabs from the border states by May 5.

    Yet despite these considerations I assert "it's the early data on this phenomenon that count most."

    Yeah, well, that's true in principle. But in this case the "early data" aren't necessarily a good picture, or even a good estimate, of the early number of swine flu infections in the border states. Duh, Pundita.

    So, with regard to my assertion in the third paragraph above, I think I should hedge by substituting "a great many" for "virtually all."

    This chart, dated May 11, 11:00 AM, shows the number of confirmed swine flu infections by state. Here are the numbers for the four border states:

    AZ 182, CA 191, NM 30, TX 179

    The question is whether those numbers reflect recent cases or clearing of a backlog.

    That's why I've inserted this correction rather than simply editing the paragraphs; the original writing illustrates the difficulty in this situation of arriving at valid conclusions when reasoning from very incomplete and imprecise data.

    Now to continue:

    Although it was known early in the U.S. outbreak that initially the biggest known cluster of infections came from a school trip to Mexico by New York students, there was nothing more than a passive effort at U.S. airports to screen incoming passengers from Mexico. (E.g., airline personnel eyeballing arrivals and asking them how they were feeling if they didn't look well.)

    And there was no action to suspend incoming or outbound flights to Mexico. This, despite the fact that officials knew that a highly infectious disease had reached epidemic stage in Mexico's capital city.

    When did they know? The answer is still speculative; I've touched on the issue in earlier posts and in the next post I'll grapple with it again. However, from all I've read, it can be argued that CDC officials were tardy by several days about alerting the U.S. public, and that their refusal to recommend suspending flights to and from Mexico explains how swine flu appeared in the majority of U.S. states within less than two weeks.

    Yet again and again the negligence was rationalized with the argument that because the disease had already entered the USA, there was no use taking measures to attempt to prevent its fast encroachment from Mexico.

    What if the disease had been a superkiller and stayed as highly infectious as it did when it first struck in Queens, New York? How many American dead would we be looking at by this time if we multiply the number of inbound infected airline passengers by the number of people they infected and keep multiplying from there?

    If Ms Napolitano would tell me that of course the U.S. government would have suspended flights from Mexico if the virus was highly lethal -- according to CDC's guidelines that would be a useless gesture. Ms Napolitano and the CDC can't have it both ways: talk gibberish when it suits them then fall back on reason when it doesn't.

    It's not the lethality of the virus that's at issue here. Yes or no: is it good practice to attempt to slow the rate of infection from a highly infectious disease?

    If yes, then: Does suspending air flights from an infected country help slow the rate of infection?

    If yes, I believe this is where I get to say "Checkmate."

    Given that my argument is hard to answer why do officials at the CDC refuse to consider it? Because they've continued to listen to Dr Longini and other members of his biostatistician tribe who advise the CDC.

    Longini and his colleagues have recently made public declarations that back them so far away from their assertion that an infection rate should be slowed at the start that they're making as much sense as Wonderland's Red Queen.

    Why would several mathematicians start talking gibberish about such a serious issue? I don't know. But TIME magazine's April 30 report, Why Border Controls Can't Keep Out the Flu Virus could be subtitled, "Lies, damned lies, and economics."
    "Once the virus has spread beyond its initial focus, travel restrictions just aren't effective," says Ira Longini ... With 4,000 flights a day between the U.S. and Mexico, "it's not worth the social disruption it would cause."

    That's not to say that very strict restrictions wouldn't have some effect on slowing the virus. In a 2006 study, Harvard epidemiologists John Brownstein and Kenneth Mandl examined the effect of the sharp reduction in air travel after the Sept. 11 attacks on that year's flu season. They found that the initial flight ban and general decline in air travel in the weeks after delayed the onset of the flu season but did little to reduce the overall number of infections and deaths that year.

    The data matches computer models run by biostatisticians like Longini, who found that even the strictest limits on air travel would only slow the start of a flu pandemic, not stop its spread. But, again, while that strategy may benefit countries that have not yet been infected with swine flu, there's still no way to know when it would be safe to lift those restrictions.

    "There's no question that air travel spreads the flu," says Mandl, a physician and researcher at the informatics program at Children's Hospital Boston and an associate professor at Harvard Medical School. "But the impact of limiting flights at this point is difficult compared to the downside of the economic impact."
    The observations overlook that no special measures were needed or taken in the USA against the 2001-2002 flu, which was seen as common. Yet a "delay in onset" would give health professionals a huge edge in preparing to deal with a superkiller virus.

    What I find particularly outrageous about the economic and "social disruption" arguments is their almost inhuman cruelty. I am haunted by a news photograph of an exhausted Mexican mother and her two daughters, slumped in exhaustion against her shoulders as they waited to be examined at a health clinic in Mexico. They were all wearing face masks.

    Here's a news flash: all decent people everywhere want to live. Parents don't want to see their children die before their eyes of an infectious disease.

    What the CDC and their scientific and medical advisors refuse to consider is that a failure to suspend cross-border and air travel traffic means that the infections are not one-way. Americans infected with swine flu who travel to that country can infect Mexicans -- in a country that is more poorly equipped than the USA to face a killer virus.

    So I don't want to hear about social and economic disruption. Memo to all concerned U.S. and Mexican agencies: Coordinate and do something called contingency planning. This, so that disruptions from a temporary border blockade and suspension of flights between Mexico and the USA are not catastrophic for people on either side of the border.

    If you don't know how to do contingency planning, if the concept is foreign to you, talk to major corporations and defense departments to find people who are experts in business and military logistics.

    Face it; you were lucky. If the death count from swine flu would have been high, all of you would have been straight on your way to Hell, where I sincerely hope you would have spent the rest of eternity sharing a slow boat from China with a particularly mean virus.

    1) INSIGHT MAGAZINE via Sci-Tech Archive: Flu Secrets You Should Know; Kelly Patricia O Meara; June, 2004

    2) Moms Against Mercury: A Shot in the Dark - Part 1; Kelly O'Meara; 2004?

    3) The New York Times: New York’s Lab (Don’t Ask Where It Is) Readies Its Own Tests for the New Strain; Anemona Hartocollis; May 3, 2009

    4) CNN: Scientists dig for lessons from past pandemics; Caleb Hellerman; April 30, 2009

    5) The Associated Press: Medical detectives probe flu virus spread in NYC; David B. Caruso; May 4, 2009

    6) CNN: Expert on flu's spread says new strain here to stay; Patrick Oppmann; April 28, 2009

    7) Phoenix [Arizona] Business Journal: Border states show no swine flu spike; Mike Sunnucks; May 5, 2009)
    This entry is crossposted at RBO. Thanks much to Procrustes for proofreading.


    bullmoosegal said...

    Pundita, while I agree that the CDC is woefully behind in monitoring and testing, I'd like to point out two mitigating factors. 1. Their funding for the kind of surveillance you're talking about is horrible. There are way to many specific antigens and serotypes of interest to provide surveillance against, and very few resources with which to do so. Their travel budgets alone don't allow for some of the surveillance, they don't have enough personnel, etc. etc. 2. The CDC doesn't have the authority from Congress to direct some of the activities you're requesting (e.g., mandating a nasal swap or antigen test for every suspected case). Congress has repeatedly refused that authority due to privacy concerns, tying the hands of CDC.

    Pundita said...

    Bullmoosegal: I agree with you but only up to a certain point. It was for the CDC to make a case to Congress that they were underfunded,understaffed, etc., which they'd had since the SARS and H5N1 outbreaks earlier in this decade to do.

    The agency seems to be a mess. In any case I intended in a future post to touch on the issues you mention. But for this post I was focused on arguing against what I believe is a dangerous fallacy.

    FYI, a few minutes ago I added strong and what I hope are very cutting concluding statements to this post.

    It's not my objective to be fair with regard to the CDC's problems. My objective is to sound an urgent warning.

    CGardner said...

    Another excellent article, Pundita!

    Perhaps your reference to Hurricane Katrina needs to be expanded to fit this situation. A virus is very much like a hurricane. . . .No guarantees on how it will behave over time, but preparedness saves lives. And, people have the right to accurate information to make the best decisions.

    As a matter of national defense, I think people will just need to learn to deal with uncertainty, as the coastal residents do every hurricane season.

    I was very disappointed in how the press dealt with this. After it appeared that the "worried well" might overwhelm the health system, new stories quickly dwindled, unless there was a confirmed case or a death. I'm sure most people don't know how the "confirmed cases" were calculated and certainly don't understand how these figures "low ball" the outbreak.

    I'm also disappointed with many fellow conservative acquaintances on the web, who placed this potential crisis into the tired, old "Obama power grab" category. They would be the first to cry "conspiracy" if this virus erupted into something more deadly.

    Finally, on a more technical note, I hope you can look at this table from CDC and perhaps run it by any lab people you know:

    What's very striking to me is the large proportion of PCRs which are "unsubtyped" or "could not be subtyped." I wonder whether this is a problem with the test kits, a nonspecific definition of when disease was confirmed or for the training of off-site CDC labs who were inadequately trained to read the results.

    DocJim said...

    Outstanding commentary, Pundita.

    I wish it were published in the New York Times and New England Journal of Medicine to get some real, honest discussion going.

    CGardner has some excellent comments. We MUST learn what these "unsubtyped" tests mean.

    I wish I had not lost faith in the CDC in 1980s with the AIDS problem. They became obviously politicized then and have continued in that mode.

    Pundita said...

    Thanks very much to all for your insightful comments. With regard to the question that CGardner raised about the "unsubtyped" tests, that's out of my league.

    Perhaps reader Bullmoosegal, who blogs at ModCon and who's a biologist, can throw light on whether the large number of unsubtyped cases are an unusual matter or a fairly common occurrence when processing large batches of virus samples. And whether this might indeed point to the kind of problems that CGardner outlined.

    bullmoosegal said...

    Re/the unsubtyped cases questions - it's very normal especially with a new viral strain. There can literally be hundreds or even thousands of variants to a primary strain. These occur through slight mutations in the virus within infected or carrier individuals (either through interaction with the host DNA or through 'swapping' bits of DNA with other viruses in the host). That's one of the reasons some people experience much more robust infections than others with the 'same' virus. As they just finished genotyping the primary strain circulating for this one, it'll be awhile before they identify new subtypes, and meanwhile, others will continue to develop. You're always behind the curve in this kind of game. Frequently, tests will work across the board for many subtypes because the sequence is similar enough that an antibody test or PCR will recognize the serotype, but that's not guaranteed. One of the reasons it took so long to identify all of the HIV strains is that tests did not fit across all strains. As a result, each new subtype had to be sequenced, and a test developed. Tests can't be readily developed unless you can get a good handle on the sequence (not always easy - took years in that case), and can grow up enough virus in tissue to figure out what antibody or other test works. Again, that part can sometimes take years to work properly. Training isn't the issue here - some sequences are just much much harder than others.

    Pundita said...

    Wow. That sets that to rest. Thank you very much, Bullmoosegal!

    CGardner said...

    Thanks for explaining that, Bullmoosegal!

    Anonymous said...

    Lots of good points. Agree that underfunding and staffing are big problems with CDC. Plus they are so enmeshed with the ways and logistics of the Feds, they are too dinosaurish to move any faster. CDC, did u read this post?