Search Peter’s website:

Archive of Swine Flu Pandemic
Communication Updates

June 29, 2010

The “Fake Pandemic” Charge Goes Mainstream
and WHO’s Credibility Nosedives

The last Swine Flu Pandemic Communication Update on this site was January 17 – five months ago. It made four points:

• We don’t know what’s coming next.
• It’s a real pandemic.
• It’s a mild pandemic, at least so far.
• It’s probably not over – but we don’t know what’s coming next.

That’s about it, really – still … except for the blame game, which is what we are going to analyze below. The extremely long assessment that follows advances an argument we can summarize in a single sentence:

That’s where we’re going.

Where is the pandemic going? In January, swine flu incidence in the northern hemisphere had peaked and was decreasing in many countries, but was still widespread. Now, as summer approaches, it’s quite low. It’s quite low in the southern hemisphere too, but with somewhat higher levels reported in some tropical countries. Seasonal H3N2, the most severe of the seasonal strains, is still circulating at low levels. At this point, unfortunately, it does not appear to have been replaced by the pandemic virus.

In other words, there is no clear evidence yet that pandemic H1N1 is or is not going to “act” like a seasonal flu strain from now on, which is what former pandemic flu strains have done after 1–3 years. That’s certainly one of the possibilities, but there are two others. It could cause additional pandemic waves more or less like the ones it has already caused. Or it could get a lot more virulent – which most experts don’t expect at this point but don’t rule out either. (Experts do remember that the mild 1968 pandemic returned to cause a more severe European wave in the winter of 1969–1970, a full year after the very mild first European wave. Almost anything is possible with influenza.)

The pandemic is like a hurricane that initially and briefly looked like it might be a Category 4 whopper, turned out to be just barely a Category 1 hurricane – but still a hurricane – and then dissipated. But unlike a dissipated hurricane, the pandemic isn’t completely gone.

In our daily lives, though, the pandemic feels like it’s over, and for some people it feels like it never happened. Other than ongoing vigilance (surveillance and assessment), precautions are in abeyance in most of the world. The most expensive precaution, a new vaccine, wasn’t ready until after the 2009 northern and southern hemisphere waves were over or receding. Still, the vaccine could have prevented a lot of morbidity and some mortality if it had been ready sooner. And if the pandemic comes roaring back, many people will wish they had availed themselves of the vaccine they so disdained as the last wave receded.

Today, the only individuals continuing to take pandemic precautions are obsessed, and the only individuals complaining that we should never have taken precautions in the first place are ignorant.

As we write this in late June 2010, there is now a huge gap between WHO’s pandemic risk communication and the public’s pandemic risk perception. That is, what most people think they just lived through – a mild pandemic that has virtually disappeared – is radically different from what WHO tells them they are living through: a “moderate” pandemic that is still ongoing.

The result: Widespread skepticism about WHO’s credibility, and thus widespread receptiveness to previously fringe allegations.

“Widespread” is a relative term here. Most people pay virtually no attention to the World Health Organization. They didn’t realize it had a lot of credibility, and they didn’t notice when its credibility collapsed. And WHO’s most devoted followers – many of them public health professionals – have tended to rally round the organization in its hour of need. (Some of them have supported WHO publicly while muttering to each other that it really should stand down from some of its earlier pronouncements.)

But there’s a super-important group in the middle: people who don’t follow WHO closely but who do pay enough attention to have learned first that it seemed to be overstating the seriousness of the swine flu pandemic and then that it was accused of doing so on purpose to help Big Pharma make big bucks. That group includes many of the world’s opinion leaders: corporate executives, government officials, medical reporters, foundation heads, etc. Losing ground in their eyes matters for WHO, and may continue to matter for decades to come.

The collapse of WHO’s credibility is important not just in general because the world needs a credible international health agency, but also, in particular, because WHO’s pandemic warnings about the future (though not its status reports about the present) are right on target. H1N1 could mutate into a much deadlier virus. That would be unusual after a year, but not unprecedented. The world could also face another pandemic in the near future, caused by another novel flu virus. There was apparently a flu pandemic in 1830–31 and another in 1833. The 1968 pandemic started less than ten years after the 1957 pandemic ended. And of course all flu-watchers know that the extremely deadly H5N1 bird flu virus is still Out There. If bird flu ever starts transmitting easily from humans to humans, it’s a whole new ball game.

It is very hard for WHO to convince people – governments, journalists, ordinary citizens – to take the near-term pandemic risk seriously as long as WHO keeps describing the past year’s pandemic experience in ways they simply cannot take seriously.

The one-year anniversary of the identification of the new swine-origin H1N1 influenza virus came in late April, 2010. The media ran a lot of anniversary stories. Some health agencies issued anniversary news releases, often featuring “lessons learned” about what had gone well and not so well. The Center for Infectious Disease Research & Policy (CIDRAP) at the University of Minnesota did a whole series of anniversary stories. In preparation for one on communication lessons learned, Lisa Schnirring of CIDRAP sent us a list of questions. We dutifully wrote answers and posted them on this site.

The one-year anniversary of the World Health Organization’s (WHO) declaration of a full-fledged “Phase 6” pandemic was June 11, 2010. There were only a few anniversary stories.

This site’s Swine Flu Pandemic Communication Update about the Phase 6 declaration, posted on June 17, 2009, was entitled “Would you like another wakeup call?” It noted that: “For those who were already awake to pandemic realities and possibilities, [the WHO pandemic declaration] was basically a nonevent – a welcome if belated confirmation of what we knew.”

But in retrospect, the WHO pandemic declaration a year ago looks to many (though not to us) like a big mistake, or even an intentional deception. It was already pretty clear by mid-June 2009 that the pandemic was mild so far, more like the last two pandemics than like the nightmare possibility experts had warned about. But it was early days yet. It was perfectly possible that the pandemic could become much more virulent. (The first wave of 1918’s horrific pandemic was also mild.) Now, a year later, that possibility looks much slimmer. It can’t yet be ruled out; flu is famously unpredictable. But at this point it would be a big surprise. And so, with 20-20 hindsight, a lot of people think the initial pandemic declaration was unjustified.

Four things happened in early June 2010 that make this a compelling risk communication story, a fit but sad ending to the saga of Swine Flu Pandemic Risk Communication (Volume One).

That’s the big swine flu risk communication story now, the biggest in months: In June 2010, the credibility of the World Health Organization crashed and burned. Charges that it had manufactured a fake pandemic in deference to the economic interests of Big Pharma gained mainstream attention.

These charges gained traction at this time, in our judgment, not because they are valid (they are not) but because WHO has badly mishandled certain aspects of its pandemic risk communications.

WHO has made three fundamental errors. In diminishing order of importance, they are:

• Failing to acknowledge that the pandemic has been mild overall, and that its incidence is now quite low. By “mild,” we mean similar to certain previous flu pandemics that WHO has long characterized as “mild” or “relatively mild.” By contrast, WHO has insisted on calling this pandemic “moderate” instead. And its tone has often left people feeling as if it were claiming “severe.”
• Failing to acknowledge that WHO changed some flu pandemic definitions and descriptions just as H1N1 was emerging. The technical meaning of the term “influenza pandemic” is debatable, as is the question of whether a mild flu pandemic should be called a pandemic at all. When WHO changed some of its definitions and descriptions of flu pandemic phases in ways that de-emphasized severity, it opened the door to suspicion that it had “changed the definition of a pandemic” in order to make sure H1N1 would qualify.
• Failing to acknowledge – until June 2010 – that WHO transparency about conflicts of interest had become inadequate. WHO’s earlier response to conflict-of-interest charges was to explain its policies and offer reassurance that the policies work. It wasn’t until after the two recent attacks that WHO began to concede that in the face of the public’s profound loss of trust, it may need to be both tougher and more transparent about its expert advisors’ conflicts of interest.

We must immediately concede that “failing to acknowledge” in these three bullet points is an overstatement. In their millions of words about the swine flu pandemic, WHO officials have periodically made statements that can be read as acknowledging all the points we’re accusing them of failing to acknowledge. We would be on safer ground claiming that WHO officials have failed to get these acknowledgments across. But that would imply that they were trying to do so. They weren’t. The main thrust of WHO pandemic communications has been, and continues to be, that H1N1 is a pandemic of moderate intensity that requires a continued “Phase 6” response; that H1N1 unambiguously meets the consensus definition of an influenza pandemic which has not been changed; and that WHO deliberations about how to manage H1N1 have been sufficiently transparent and self-evidently free of dangerous conflicts of interest.

That, we believe, is why WHO’s credibility is seriously threatened, and its ability to warn the world about future pandemic threats seriously compromised.

To understand what is behind the serious recent damage to WHO credibility, we are going examine these three failures in detail.

Such a detailed examination is worthwhile, we think, for two reasons:

• First, we hope to make a contribution to the world’s understanding, and to WHO’s understanding, of what happened – of how the World Health Organization managed to do severe damage to its own credibility – not by the way it handled the technical side of the twenty-first century’s first influenza pandemic (it did a creditable job), but by the way it handled its risk communication during that pandemic. WHO has appointed an external review committee chaired by Dr. Harvey Fineberg to assess the “Functioning of the International Health Regulations (2005) in Relation to Pandemic (H1N1) 2009.” We hope this long column will help Dr. Fineberg’s committee see the pivotal role risk communication has played in WHO’s pandemic credibility crisis.
• Second, we believe we are writing not just about what happened, but also about what happens. The H1N1 pandemic is not the only time WHO’s top scientists’ inattention to risk communication has been its Achilles heel, and WHO is not the only organization that has been damaged by its technical leaders’ inattention to risk communication. Many of the more specific phenomena discussed in this article – for example, WHO’s reluctance to make changes that might be seen as caving in to pressure – are also generic. We hope readers who don’t share our longstanding fascination with the World Health Organization and with influenza pandemics will nevertheless find parts of this analysis illuminating.

For readers who don’t want to delve that deep, we hope the foregoing introduction will have been of interest. For the remainder of this lengthy analysis, use the following links:

January 17, 2010

Pandemic Interruptus: It Ain’t Over Till It’s Over

What should pandemic risk communicators be saying in January 2010, when swine flu looks to non-experts like it’s disappearing, when swine flu vaccine is a drug on the market, and when swine flu news has taken on a distinctly skeptical tone?

It has been a wild ride so far, from high drama to bored skepticism in a mere nine months. The novel H1N1 influenza virus was first identified in samples from two U.S. children in mid-April 2009, and then a week later in samples from Mexico. April is a month when there’s normally very little flu in the northern hemisphere. From April through December, the U.S. had more flu than in the average flu season. In most of the country, but not all, we saw a first wave in the spring, then a summer decline (but still much more flu than usual in summer), and then a bigger second wave in the fall. Then the fall wave plummeted.

By mid-January in most years, seasonal flu numbers are increasing in the U.S. But in this pandemic year, the pandemic flu numbers are decreasing, and there is virtually no seasonal flu around yet. The same thing is happening in most of Europe. Pandemic H1N1 is way down, and except for a few isolated cases seasonal flu has yet to appear.

Naturally the media have virtually decreed the pandemic over. And some commentators have said, in hindsight, that it wasn’t a pandemic at all.

In the ecosystem of public opinion, all niches are filled. At one extreme are those who confidently predicted a rerun of 1918 or worse, some of whom even now maintain that it happened and health officials somehow managed to cover it up. At the other extreme are those like Marc Siegel and Michael Fumento who were pandemic scoffers from the outset and are now crowing that they were right all along.

Many in the media have happily oscillated from one extreme to the other. Some of the newspapers and broadcast stations that headlined the direst predictions now headline preposterous charges that it was a “false pandemic” manufactured by a conspiracy of public health officials and pharmaceutical companies.

Okay, so what should pandemic risk communicators be saying?

1. We don’t know what’s coming next.

Although people with normal interests may think of flu as ordinary and boring, public health experts see influenza as endlessly surprising, among the most fascinating of diseases. And influenza is never more surprising than when a new flu virus emerges and launches a pandemic.

Experts see flu as unusually unpredictable – but considerable unpredictability is a hallmark of nearly all crisis situations. That’s why acknowledging uncertainty is a core principle of crisis communication.

The only sensible way to plan pandemic response – and the only sensible way to talk about pandemic planning – is probabilistically. We don’t know what’s next, but we can estimate probabilities and act accordingly.

Probabilistic thinking is the very essence of risk assessment and therefore of risk management. And probabilistic language is an essential part of risk communication.

Thinking and talking probabilistically about risk requires asking six core questions:

• What outcomes are possible?
• What is the estimated probability of each possible outcome? (No one can really know, but experts can make informed guesses.)
• What are the predicted effects of each possible outcome, and how bad is each set of effects? (This can be modeled.)
• What can be done, at what cost, to mitigate each set of effects? (This can be modeled too.)
• How confident are we about the answers to these questions? (Not very.) What are we likeliest to be wrong about? In which direction?
• Based on all of that, what does it make sense to do? (This will be a debatable judgment based partly on values and partly on the guesses and modeling above.)

It doesn’t follow that the wisest course of action is to prepare only for the likeliest set of outcomes. Dire outcomes justify preparedness (if preparedness is possible and cost-effective) even if they’re pretty unlikely. That’s why people buy fire insurance for their homes – not because a big fire is the likeliest outcome, but because it’s a very bad outcome that isn’t vanishingly unlikely.

Inevitably, then, a lot of preparedness will turn out unnecessary or excessive. When your home doesn’t burn down, you don’t cancel your insurance, nor are you angry that you wasted your premium. You’re glad your home didn’t burn down last year, and also glad you’re insured in case it burns down this year.

And fire insurance salespeople don’t claim that their prospective customers’ homes are going to burn down. They ground their sales pitch not in the probability of such a disaster, but in its magnitude. They’re selling a hedge to lessen the impact of a profoundly undesirable scenario – one that is pretty unlikely, but not so unlikely as to constitute a negligible risk. And they’re selling the peace of mind that comes with knowing you have hedged against such a possible disaster.

What’s missing from most people’s pandemic thinking is this probabilistic mindset.

Sports fans think probabilistically. Baseball fans, for example, know that there are times when the smartest thing for a batter to do is try to bunt the runner into scoring position. If the strategy doesn’t pan out, a commentator may remark that the batter probably wishes in hindsight that he’d swung for the fences … but nobody says he should have done so when the odds said bunting was the right bet. There are obvious analogues in virtually every sport: The right play remains the right play even when it didn’t work this time.

Similarly, we all understand that weather forecasters play the odds, and advise us to play the odds. When there’s a 70 percent chance of rain we take an umbrella to work and cancel our picnic plans – and we’re not especially outraged if the sun shines all day. When there’s even a 20 percent chance of a hurricane we buy extra food and check the flashlight batteries – and we feel neither foolish nor victimized if the hurricane weakens or changes course.

But in public health, non-experts often indulge in outcome-biased thinking instead of probabilistic thinking. Thus many people – and many journalists and politicians – seem to believe that officials should have prepared for exactly the pandemic we got (so far), no more, no less … as if they could have known in advance what sort of pandemic would come along.

If the 2009 pandemic had turned out as severe as it looked last April in Mexico, Congress would long since have launched investigations into why the CDC was insufficiently prepared, why we let our public health infrastructure decay, etc. If we end up with a severe third wave, those same questions will be asked. Even the mild pandemic we have had so far was sufficient for “tough questions” about the slow pace of vaccine production. But since the pandemic looks like it’s over (for now) and since it wasn’t very bad (for most people), many are asking instead why public health officials scared us unnecessarily.

I’m not going to burden readers with endless quotations from commentators and critics in the U.S., Europe, India, and elsewhere. They’re all grounded in the same false reasoning:

1. Officials warned us that things might get really bad, and urged us to take precautions.
2. Things didn’t get that bad (at least not yet). Those who took precautions feel foolish. Those who didn’t take precautions feel vindicated.
3. Officials must have known that things wouldn’t get that bad. They misled us on purpose. Here’s why they did it….

Outcome-biased thinking isn’t confined to public health. Consider three other examples.

• After hurricane Katrina, most people came to believe in hindsight that the government should have done more to enable the New Orleans system of levees to withstand heavy floods in a strong hurricane. They didn’t think so beforehand, and they still don’t want to retrofit other cities against other catastrophes. There’s not much demand to prepare New York City for a tsunami or the New Madrid Fault region for an earthquake. Just New Orleans for a hurricane.
• After the financial system imploded in 2008-2009, just about everyone agreed that there should have been tougher regulation of credit default swaps, securitized mortgages, and whatever else needed regulating in order to keep the economy on track. Not that we wanted more government regulation at the time. Nor do we want more government regulation now … except for regulating whatever caused the economy to tank last year.
• After the Christmas Day bomb attempt on Northwest Airlines Flight 253, a near-consensus emerged that the government should have seen it coming and stopped Umar Farouk Abdulmutallab from getting onto the plane. Does that mean we want all people put on the no-fly list if they’re from countries with significant Muslim populations and their fathers think they’re too radical? Nope – just Abdulmutallab and everyone else who’s going to try to blow up a plane.

I’m not asserting that it’s a bad idea to spend more money preparing urban areas for natural disasters, or to regulate financial institutions and especially innovative financial instruments more strictly, or to be more rigorous in the screening of international air travelers. There are pros and cons to each of these measures. And there are counterexamples – horror stories of money spent preparing for natural disasters that never came, of industries that seem unable to compete and innovate effectively because of over-regulation, of innocent people who have struggled unsuccessfully to persuade officials to take them off the no-fly list.

My point is that you can’t prepare only for a hurricane in New Orleans, regulate only the industries that are about to implode, and keep only the actual terrorists from flying. And you can’t get people all worked up about only the pandemics that are going to turn out to have been severe.

You can be as risk-averse or as risk-tolerant as you want in how you play the odds. But you can’t avoid playing the odds.

In all these examples where post-hoc outcome-biased thinking prevailed over real-time probabilistic thinking, the situation was scary enough that many people felt powerless and vulnerable. Sometimes a degree of psychological regression happens when people feel powerless and vulnerable. We get childish and petulant. Like children, we want mommy and daddy to make it right.

So if something bad happens, we complain bitterly that mommy and daddy should have protected us better. And if something bad doesn’t happen, we complain nearly as bitterly that mommy and daddy shouldn’t have disrupted our playtime with unnecessary warnings and precautions.

These examples have something else in common. Discussions of sports and weather refer constantly to playing the odds – but discussions of emergency preparedness, government regulation, counterterrorism, and pandemic response do not.

Early on in the pandemic, public health officials should have been saying – again and again – something like this:

We don’t know how severe this pandemic is going to be. Yet we have to make decisions now – for example, decisions about how much vaccine to order, and decisions about whether to advise closing schools and canceling public events. So some of these decisions may end up wrong. We are trying to err on the alarming side. We’d rather prepare too much than too little. But if things get very bad, critics will say that we should have done more – and in hindsight they will be right. And if the pandemic turns out mild, critics will say that we should have done less – and in hindsight they will be right too.

Officials did say things like this from time to time. At the very start of the CDC’s second pandemic press briefing on April 24, Richard Besser (then acting director) put it superbly:

First I want to recognize that people are concerned about this situation. We hear from the public and from others about their concern, and we are worried, as well. Our concern has grown since yesterday in light of what we’ve learned since then.

I want to acknowledge the importance of uncertainty. At the early stages of an outbreak, there’s much uncertainty, and probably more than everyone would like. Our guidelines and advice our [are] likely to be interim and fluid, subject to change as we learn more….

We do not know whether this swine flu virus or some other influenza virus will lead to the next pandemic; however, scientists around the world continue to monitor the virus and take its threat seriously.

Media coverage of the Besser press briefing was substantial, but very few stories quoted this passage. Reporters went for the hard news: what the CDC thought was happening and what it was going to do about it. “Uncertainty claims” – explicit statements that the situation is uncertain, that officials are playing the odds, and that in hindsight their response may turn out too aggressive or not aggressive enough – are hard to get into the media. Officials almost always end up sounding more certain in news stories than they sounded during the news conference or the interview.

Later on, commentators who missed the qualifiers and uncertainty claims write, “Remember when they said we were all gonna die?”

Even if uncertainty claims make it into the media, they are hard to get into people’s heads – especially the heads of people who feel powerless and vulnerable, who want officials to be confident and definite.

And of course most officials are less committed than Dr. Besser was to communicating their uncertainty. Especially in crisis situations, it’s awfully tempting to project certainty instead (as if that were a stand-in for competence) – to give the anxious public what it seems to want.

But nothing is more important in pandemic risk communication than persuading the public (and the politicians) to think probabilistically. Public health officials need to insist on their uncertainty; they need to make uncertainty the message, not the preamble to the message.

Uncertainty about the future should have been stressed, over and over, early on in the pandemic – far more than it was. But maybe we’re still “early on” in the pandemic. As Yogi Berra (a probabilistic thinker) taught us, it ain’t over till it’s over. So probabilistic thinking should be stressed now as well – both about the decisions that have been made up till now, and about the decisions we face today.

As the 1957 Asian Flu pandemic was looming from a distance, the U.S. Surgeon General at the time, Leroy Burney, said:

I am sure that what any of us do, we will be criticized either for doing too much or for doing too little…. If an epidemic does not occur, we will be glad. If it does, then I hope we can say… that we have done everything and made every preparation possible to do the best job within the limits of available scientific knowledge and administrative procedure.

I doubt that it’s ever advisable to make “every preparation possible.” But Burney’s first sentence is right on target.

Uncertainty is Message #1.

2. It’s a real pandemic.

After many weeks of information-gathering, discussions with Member States, and expert debate, the World Health Organization (WHO) declared the swine flu outbreak a pandemic on June 11, 2009. In January 2010, some commentators are trying to “undeclare” it.

They’re wrong. It’s a real pandemic.

Like a lot of terms in public health, the word “pandemic” isn’t rigorously defined. But there is general agreement that an influenza pandemic has three defining characteristics.

Experts continue to debate the close cases. The Russian Flu of 1977 is a good example. It was caused by the re-emergence (probably from a lab) of a 1950 strain of human H1N1, and it quickly spread globally. But it mostly affected people younger than age 23 – people who hadn’t been around when that strain was circulating previously. And it did not cause excess mortality in any age group. A few experts consider the Russian Flu to have been a pandemic, but most do not.

In 1995, influenza expert Edwin D. Kilbourne said that “defining a pandemic is a little like defining pornography – we all ‘know it when we see it,’ but the boundaries are a little blurred.” Dr. Kilbourne’s wonderfully readable article discusses several examples of such blurriness, especially regarding the concept of a “novel” virus.

If worldwide disease surveillance had been good enough in 1977 to identify the Russian Flu outbreak at its inception, it might have been declared a pandemic – and then “undeclared” when more was known. But it would take a very fringe expert indeed to recommend “undeclaring” the H1N1 pandemic of 2009.

3. It’s a mild pandemic, at least so far.

Sometimes it seems like the world (the world of people interested in flu, at least) is divided into two camps: the people who think swine flu is too mild to call it a pandemic versus the people who think it’s a pandemic and how dare anyone call it mild!

I’m in the third camp, which feels like the smallest camp: the people who keep insisting that it’s a mild pandemic so far.

It’s certainly not – so far – the pandemic that health officials were expecting and dreading. That expectation was shaped by two anchoring frames.

The first standard of comparison is the pandemic of 1918, the most severe pandemic of modern times. The 1918 pandemic is estimated to have killed about 675,000 Americans – virtually all within a single year, though the pandemic actually lasted 27 months. Its case fatality rate in the U.S. was roughly 2% – compared to roughly 0.02% for the 2009 pandemic so far, about a hundred times lower.

The second standard of comparison is the incredibly deadly novel H5N1 (“bird flu”) virus that emerged in 1997. Bird flu has not gone pandemic (so far); it has infected fewer than a thousand people worldwide. But it killed nearly 60 percent of them – a case fatality rate 30 times worse than the 1918 pandemic, and 3,000 times worse than the 2009 pandemic so far. The nightmare that influenza experts have been living with since 1997, and are still living with today, is that bird flu will mutate in a way that makes it as transmissible as ordinary flu (or as swine flu), only thousands of times deadlier.

Flu experts prepared for the 2009 pandemic in the shadows of 1918 and bird flu. They shied away from the most horrific possibilities, but they never even considered the mildest possibilities.

U.S. experts, for example, developed a Pandemic Severity Index (PSI) that had five categories. The PSI assumed that a pandemic would infect about 30 percent of the U.S. population. A Category 1 pandemic would have a case fatality rate of less than 0.1%, adding no more than 90,000 U.S. fatalities – still more than twice as bad as the average flu season. A Category 5 pandemic would have a case fatality rate of greater than 2 percent, meaning more than 1,800,000 U.S. fatalities – basically a rerun of 1918 or worse, with a much larger population. A bird flu pandemic would be off the scale in one direction. The swine flu pandemic we got is off the scale in the other direction … which may be why officials haven’t mentioned their PSI in quite some time.

I have been describing the swine flu pandemic as “mild” since before it was declared a pandemic – almost always adding the crucial qualifier: “so far.” I based my use of the word “mild” on evolving published data about this pandemic, compared to prior pandemics and average flu seasons, not on my own non-existent influenza expertise.

On May 6, I wrote: “Swine flu looks to be an extremely mild pandemic if it goes pandemic at all, despite WHO warnings that it may ‘come back with a vengeance’ in the fall.” On June 4, I wrote that it was “still mild.” On June 17, I wrote: “The big public health risk isn’t the relatively mild flu that’s circulating now…. Swine flu could come roaring back in a much more virulent second wave.” On July 21, I proposed three core pandemic messages. One was: “Pandemic H1N1 looks very mild so far.” Another was: “We must prepare for the possibility that pandemic H1N1 could become more severe.”

I kept arguing that a communicator who failed to acknowledge the current mildness of the pandemic could not credibly warn about its possible future severity.

Finally, on December 2, I entitled my Swine Flu Pandemic Communication Update: “It’s Official (sort of): The Swine Flu Pandemic Is Mild So Far.” The CDC’s own data showed the mildness of the pandemic, I said, and the CDC was unwisely refusing to say so.

This time I was criticized for insisting so aggressively on the pandemic’s mildness. (See my December 15 acknowledgment of this criticism.) I understand some of the reasons why the term “mild” strikes many as offensive:

• It is insensitive to the impact of the pandemic on those who lost a loved one to H1N1, or were severely ill, or even just worked 90-hour week after 90-hour week in a local health department.
• It ignores the reality that the pandemic has already killed several times as many children as the average flu season. (The average flu season kills mostly the elderly.) That’s a fact that has understandably and justifiably worried many parents, and left them feeling that the pandemic was anything but mild.
• It also ignores the reality that many more people under 65 were hospitalized and in intensive care units during the 2009 pandemic than during the average flu season. Often concentrated over a short period of time in specific “hot spots” around the country, these hospitalizations constituted an unusual burden on the U.S. medical system.
• It focuses on one aspect of pandemic severity, the number of deaths. But pervasiveness matters too. My college professor daughter, for example, says more of her students were out sick with the flu last fall than any semester in her memory. As far as she knows, none of them died. But she certainly experienced the fall as a severe flu season.

Despite all of that, the fact remains that in terms of overall mortality the 2009 pandemic has been very mild so far – so mild that some are denying that it’s a pandemic at all. The time has come for officials to acknowledge and insist on the middle ground. Yes, it’s a pandemic – so far a mild one overall, albeit tragic for tens of thousands of people around the world.

In fact, official acknowledgment of the pandemic’s mildness is long overdue. Both the CDC and the World Health Organization have steadfastly avoided the word “mild” in their pandemic communications. Both have passed up many chances to breathe an audible sigh of relief: when the alarming initial news from Mexico was not borne out, when the pandemic case fatality rate came in lower than the rate of prior pandemics and lower than the rate of the average flu season, etc. Both have seldom said publicly what I think they must be saying privately: “So far, so good. We’re not out of the woods yet, and the pandemic could still take a turn for the worse, but to date it has been much less devastating than we dared hope.”

In part because of their failure to acknowledge the pandemic’s mildness, officials are now reaping the whirlwind. Millions of ordinary citizens have seen for themselves that (with tragic exceptions) this pandemic is not such a big deal. If the CDC and WHO think otherwise, if what we have experienced over the past nine months is really the sort of pandemic health officials consider serious, then it makes sense to shrug off their pandemic warnings altogether.

Officials’ failure to acknowledge that the pandemic has been mild so far thus justifies public skepticism about officials’ warnings that this pandemic or some future pandemic could be far more severe. It even gives a semblance of credence to the absurd allegation that officials have been promoting a fake pandemic for ulterior purposes.

It’s a real pandemic, but so far a mild one. Officials need to say so.

4. It’s probably not over – but we don’t know what’s coming next.

I’m not a virologist, and I’m not entitled to an opinion about where the 2009 pandemic is headed in 2010. In fact, many virologists think they’re not entitled to an opinion either. They say it’s anybody’s guess.

The least likely possibility, I’m told, is that swine flu will simply disappear. Influenza is such an unpredictable disease that the experts aren’t ruling anything out, not even that. But the H1N1 virus transmits easily from person to person. And it has lots of people left to infect – people who have neither had the disease nor been vaccinated against it. So most experts expect to see more swine flu.

One question is when swine flu will surge again in the U.S. (It hasn’t disappeared. It is still circulating in low levels here, and at higher levels in some countries around the world.) There are three main possibilities:

A more important question than when swine flu will resurge is how the novel H1N1 virus might change along the way. The scariest possibility is that it could become more virulent, mutating in a way that makes it much less mild. One nightmare scenario: Swine flu and bird flu mix-and-match genetic material, producing a new virus as infectious as swine flu and as deadly as bird flu.

All changes are possible. The swine flu virus could become more or less virulent. It could become more or less infectious. It could attack different age groups or people with different medical conditions. It could become resistant to antiviral drugs. It could (and almost certainly will, over time) drift genetically so the existing vaccine no longer works very well.

Or, of course, it could stay pretty much the way it is – with more and more people becoming immune to it as a result of previous exposure or vaccination. Influenza strains mutate incessantly, so “staying the same” is considered a pretty unlikely scenario. But as the experts all say, the only thing you can be sure of with influenza is that it will surprise you. Staying the same is one possible surprise.

And here’s yet another question: What’s going to happen to the seasonal flu strains? Precedent says one or both of the currently circulating influenza A strains will probably disappear, supplanted by the new flu in town. But that’s not guaranteed either.

If the U.S. is going to have its usual flu season this winter, it’s already a little late … but not yet ridiculously late. So:

• Maybe we’ll see no more pandemic flu this winter, but the seasonal flu will come back as usual.
• Maybe we’ll have to endure both simultaneously – a pandemic third wave plus the usual flu season.
• Maybe the previous seasonal influenza A strains will disappear for good, and swine flu will be our new seasonal influenza A until the next pandemic.

If swine flu does supplant the earlier seasonal A strains (whether it happens this winter or a year or two from now), that could be very good news, especially for seniors. In recent years seasonal influenza has been made up of two influenza A strains: H3N2 (still circulating since it caused the 1968 pandemic) and the 1977 version of H1N1. (There are also two kinds of influenza B in circulation. Influenza B doesn’t seem to compete with influenza A strains, and isn’t known to cause pandemics.) H3N2 is currently the deadlier of the two A strains, especially to the elderly. If H3N2 gets wiped out, and if pandemic H1N1 becomes seasonal and stays as mild as it is so far, we’re unlikely to reach our annual average of 36,000 U.S. flu deaths in the coming years.

In other words, by out-competing and replacing the deadlier seasonal strains (especially H3N2), the 2009 pandemic could end up saving lives!

Like everything else about influenza, this isn’t guaranteed either. Maybe swine flu will end up coexisting with one or both of the two seasonal influenza A strains that have been circulating since 1968 and 1977. Maybe it will supplant those earlier strains but become deadlier itself (just as H3N2 is deadlier today than it was during the pandemic of 1968).

Or maybe swine flu will disappear after all, leaving us back where we started.

Key Messages

So what are the key messages of the moment about the swine flu pandemic? I think there are four.

December 15, 2009

What the CDC Is Saying about Swine Flu Severity

This update draws some inescapable tentative conclusions from the most recent (December 10) tentative estimates of U.S. pandemic flu cases, hospitalizations, and deaths provided by the U.S. Centers for Disease Control and Prevention (CDC).

The update also contrasts the CDC’s estimates with CDC communication about them, and with the Department of Health and Human Services’ new H1N1 vaccination campaign, “Together we can all fight the flu,” launched on December 7.

I don’t have any opinion on whether the estimates the CDC reported on December 10 are reliable, valid, and useful. But I am certain that the arithmetic I have performed on those estimates to show what they mean is solid. And I am certain that what the CDC’s estimates mean diverges significantly from what the U.S. government is telling people about the pandemic.

Bottom line one is what the CDC’s December 10 estimates mean:

• Children (0–17) and adults (18–64) are much more affected, in terms of both deaths and hospitalizations, than they are in an average flu season. Seniors (65+) are much less affected than in an average flu season.
• Nonetheless, the pandemic so far is much less severe than an average flu season in terms of total deaths and hospitalizations per million cases. That is, a much higher percentage of people with swine flu than of people with the seasonal flu recover, and do so without requiring hospital care. This is true mostly because the pandemic is enormously less dangerous than the seasonal flu for people over 65, who account for 90% of all seasonal flu deaths in an average year.
• The pandemic case attack rate is much higher for children than for adults and seniors. So far, kids are likeliest to catch swine flu, by far.
• The pandemic case fatality rate is much higher for adults and seniors than for children. So far, kids are least likely to die if they catch swine flu, again by far.
• The pandemic population mortality rate is also higher for adults and seniors than for children. This is the most useful parameter for comparing the relative risk of different age cohorts. So far, adults and seniors are significantly likelier than kids to first catch swine flu and then die of a flu-related illness.

Bottom line two is what the public is being told:

• The official HHS pandemic vaccination campaign frequently says children under 18 are most at risk.
• The vaccination campaign appropriately targets high-risk subgroups among young adults, but never mentions that, in general, adults 18 and older are at higher risk than children under 18.
• No vaccination campaign messages have prioritized seniors, even high-risk seniors, despite the fact that seniors’ population mortality rate is only slightly lower than that of adults 18–64, and much higher than that of kids under 18.
• The CDC’s news briefing on its December 10 estimates focused on the risk to children and young adults, implying that the risk was much lower to older adults and seniors. (The CDC may have information suggesting that younger adults in the 18–64 age group are at higher risk than older adults in that age group, but the December 10 CDC estimates do not include this information.)
• To the extent that the public continues to believe, mistakenly, that swine flu is deadliest to children and least deadly to the elderly, it seems likely that many older Americans may underestimate their own risk and therefore decide not to get vaccinated, even when vaccine finally becomes available to them.

This is my second time analyzing the CDC’s pandemic risk estimates compared to the CDC’s pandemic risk communication. On December 2, I posted an update that examined the implications of previous CDC estimates from the first six months of the pandemic. That update also criticized the CDC for failing to acknowledge the implications of its estimates, particularly in its November 12 press briefing.

The CDC’s first estimates included data through October 17, about a week or so before the peak of the second U.S. wave. The CDC’s second estimates, released December 10, include data through November 14, a couple of weeks after that peak. But there is still a lot more influenza (more cases, hospitalizations, and deaths) to come – and to be estimated – during the subsequent downslope weeks of the second wave.

Though the numbers all increased substantially from the first to the second set of CDC estimates, the proportions did not change much. In fact, little has changed in the implications of the CDC’s estimates, in what the public is being told, or in the gap between the two.

I made three main risk communication points in the earlier update:

• The CDC has conveyed a misimpression of how deadly the pandemic is and which age cohorts are most at risk of dying from it.
• If and when people become aware that they have been given a misimpression, the CDC risks sacrificing credibility – credibility it will need if this pandemic turns more severe; if another, more severe, pandemic arises; if controversies arise about vaccine safety; etc.
• The goal of motivating people (especially young people) to get vaccinated may make it tempting not to correct these misimpressions. But the price of reduced transparency and reduced credibility is too high. Sustainable public health leadership requires a very high level of candor, even if that candor may diminish achievement of a specific public health objective.

These points remain on target.

Analysis of the Three Age Groups

Children 0–17	The CDC estimates 16 million cases, 71,000 hospitalizations, and 1,090 deaths. U.S. Census data show that 24.3% of the U.S. population of 308 million is under 18; that comes out to 75 million children. 16 million cases out of 75 million children = 21.3% case attack rate       (21 sick children out of every hundred children) 1,090 deaths out of 16 million cases = 0.007% case fatality rate       (70 dead children out of every million sick children) 21.3% (CAR) × 0.007% (CFR) = 0.00149% population mortality rate       (14.9 dead children out of every million children in the population) 71,000 hospitalizations out of 16 million cases = 0.44% case hospitalization rate       (4.4 hospitalizations out of every thousand sick children) 1,090 deaths out of 71,000 hospitalizations = 1.5% inpatient mortality rate       (15 dead children out of every thousand children hospitalized)
Adults 18–64	The CDC estimates 27 million cases, 121,000 hospitalizations, and 7,450 deaths. U.S. Census data show that 62.9% of the U.S. population of 308 million is 18–64; that comes out to 194 million adults. 27 million cases out of 194 million adults = 13.9% case attack rate      (14 sick adults out of every hundred adults) 7,450 deaths out of 27 million cases = 0.028% case fatality rate      (280 dead adults out of every million sick adults) 13.9% (CAR) × 0.028% (CFR) = 0.00389% population mortality rate      (38.9 dead adults out of every million adults in the population) 121,000 hospitalizations out of 27 million cases = 0.45% case hospitalization rate      (4.5 hospitalizations out of every thousand sick adults) 7,450 deaths out of 121,000 hospitalizations = 6.2% inpatient mortality rate       (62 dead adults out of every thousand adults hospitalized)
Seniors 65+	The CDC estimates 4 million cases, 21,000 hospitalizations, and 1,280 deaths. U.S. Census data show that 12.8% of the U.S. population of 308 million is 65 and over; that comes out to 39 million seniors. 4 million cases out of 39 million seniors = 10.3% case attack rate       (10 sick seniors out of every hundred seniors) 1,280 deaths out of 4 million cases = 0.032% case fatality rate       (320 dead seniors out of every million sick seniors) 10.3% (CAR) × 0.032% (CFR) = 0.00330% population mortality rate       (33.0 dead seniors out of every million seniors in the population) 21,000 hospitalizations out of 4 million cases = 0.52% case hospitalization rate       (5.2 hospitalizations out of every thousand sick seniors) 1,280 deaths out of 21,000 hospitalizations = 6.1% inpatient mortality rate       (61 dead seniors out of every thousand seniors hospitalized)

Comparisons by parameter

Case attack rates:

Children    21.3%
Adults       13.9%
Seniors      10.3%

The pandemic is likeliest to sicken children, and least likely to sicken seniors.

Case fatality rates:

Children    0.007%
Adults       0.028%
Seniors      0.032%

The pandemic is likeliest to kill sick seniors, and least likely to kill sick children.

Population mortality rates:

Children    0.00149%
Adults       0.00389%
Seniors      0.00330%

The pandemic is likeliest to first sicken and then kill adults; it’s nearly as dangerous to seniors, but much less dangerous to children.

Case hospitalization rates:

Children    0.44%
Adults       0.45%
Seniors      0.52%

The differences are small, but sick seniors are likelier to be hospitalized than sick adults or children.

Inpatient mortality rates:

Children    1.5%
Adults       6.2%
Seniors      6.1%

Hospitalized adults and seniors are much likelier to die than hospitalized children. (Note that some people die of influenza without having been hospitalized first; these deaths are part of the CDC’s estimated total number of deaths. Thus the number of deaths divided by the number of hospitalizations, reported here, is a rough and somewhat inflated measure of inpatient mortality rates.)

Some conclusions based on the December 10 CDC estimates

For the H1N1 pandemic:

• The age-specific case attack rates tell us that so far children 0–17 are most likely to catch swine flu.
• The age-specific case fatality rates tell us that so far seniors 65+ are most likely to die if they do catch swine flu.
• The population mortality rates tell us that so far adults 18–64 are most likely to do both: to catch swine flu and then die as a result.

The data do not permit assessment of younger versus older children (0–4 versus 5–17) or of younger versus older adults (18–49 versus 50–64) or of younger versus older seniors (65–74 versus 75+), though these finer distinctions may make a difference.

The pandemic is attacking a higher percentage of children than of adults and seniors. But among the three groups, the children are least likely to die.

It is nonetheless true that the pandemic is more deadly than the seasonal flu to children 0–17 and adults 18–64:

• Approximately 90% of the 36,000 seasonal flu deaths in an average year are 65 and over, leaving only 3,600 deaths among children and adults under 65.
• By comparison, 11% of pandemic deaths so far are children, 76% of the deaths are adults, and 13% are seniors. The U.S. population is made up of 24% children, 62% adults, and 13% seniors.

In other words:

• The seasonal flu discriminates overwhelmingly against seniors (65+) – seniors are 13% of the U.S. population and 90% of U.S. seasonal flu deaths.
• By contrast, the pandemic so far is discriminating against adults (18–64), though less overwhelmingly; adults are 62% of the population and 76% of the swine flu deaths.
• For both the seasonal flu and the pandemic so far, children (0–17) make up a much smaller percentage of the death toll than of the population.

It is tragic that the pandemic is killing more children – a lot more children – than usually die from the flu. It is nonetheless deducible from the CDC estimates that the pandemic is less dangerous to children than it is to adults and seniors.

Analysis of the Pandemic versus the Seasonal Flu

While the preceding analysis of the pandemic itself is based entirely on the CDC December 10 estimates plus census data, the comparisons with seasonal flu estimates are not nearly as crisp. The CDC’s methodologies for estimating the seasonal flu burden and its distribution among age cohorts are necessarily quite different from its methodologies for estimating the burden to date of the pandemic. So we are not comparing apples with apples.

But if it is approximately correct that about 90 percent of seasonal influenza-related deaths are in people 65 and older – as the CDC frequently reports – then the comparisons above between how the seasonal flu affects different age groups and how the pandemic so far affects different age groups should be approximately (but only approximately) on target.

The same is true for the comparisons below about overall seasonal flu severity versus pandemic severity so far. If it is approximately correct that about 5–20% of the U.S. population catches the flu in an average year, about 200,000 of them are hospitalized, and about 36,000 of them die, then the severity comparisons below should be approximately (but only approximately) correct as well.

Cases:	There have been an estimated 47 million pandemic cases so far, 15.3% of the U.S. population of 308 million. The seasonal flu averages roughly 31 million cases per year, 10% of the 308 million total U.S. population. The CDC says the seasonal flu case attack rate ranges from 5% to 20%. At 15.3% and climbing, the pandemic is already near the upper end of the seasonal flu range.
Hospitalizations:	There have been an estimated 213,000 pandemic hospitalizations so far – versus roughly 200,000 hospitalizations in the average seasonal flu year.
Deaths:	There have been an estimated 9,820 pandemic deaths so far – versus roughly 36,000 deaths in the average seasonal flu year.

Thus the number of pandemic H1N1 cases (15% of the population) from the start of the pandemic through November 14 was already getting close to the upper end of the CDC’s estimated range of seasonal flu cases (5% to 20% of the population). And that was at a point when the second wave was far from over. Through November 14, those cases had led to 106% as many hospitalizations as the average seasonal flu (213,000 pandemic H1N1 hospitalizations so far versus a seasonal average of 200,000 hospitalizations), which had led to only 27.3% as many deaths as the average seasonal flu (9,820 pandemic H1N1 deaths so far versus a seasonal average of 36,000 deaths). Remember: These comparisons are all methodological “apples and oranges.” But they are all we’ve got so far.

Here they are without any numbers: The pandemic has already sickened a lot more people than the seasonal flu average. Despite a lot more people getting sick, only slightly more people than the seasonal flu average have needed hospitalization. And with slightly more people hospitalized, far fewer people – less than one-third as many – have died.

If “mild” refers to the number of people who get sick, the pandemic is not milder than the seasonal flu average. In those terms it is already more severe than the seasonal flu average, and climbing. If “mild” means the percentage of the population or the percentage of sick people who end up dying, then the pandemic – so far – is significantly milder than the seasonal flu average.

But what’s happening in this pandemic to children under 18, and to adults 18-64, is certainly not mild, compared to the much smaller impact of the seasonal flu on those age groups.

What the Government Is Saying

The following three quotations are all from CDC Director Thomas Frieden’s December 10 press briefing, at which the CDC released the latest estimates. They are Dr. Frieden’s only remarks at the briefing bearing on the significance of the estimates.

The bottom line is that by November 14th, the day up to which those estimates include, many times more children and younger adults, unfortunately, have been hospitalized or killed by H1N1 influenza than occurs during a regular flu season….

What we have seen so far reiterates that people under the age of 65 are most heavily impacted by influenza. By November 14th, many times more children and younger adults, unfortunately, have been hospitalized or killed by H1N1 influenza than happens in a usual flu season. Specifically, there have been, we estimate there have been nearly 50 million cases, mostly in younger adults and children. More than 200,000 hospitalizations which is about the same number that there is in a usual flu season for the entire year. And, sadly, nearly 10,000 deaths, including 1,100 among children and 7,500 among younger adults. That’s much higher than in a usual flu season. So as we’ve seen for months this is a flu that is much harder on younger people and fortunately has largely spared the elderly until now….

In terms of comparison of this year’s flu with H1N1 influenza with seasonal flu, we know that it’s much milder for older people. It’s much less likely to result in death because older people are much less likely to get infected. But it has been a much worse flu season for people under the age of 65, particularly younger adults and children. The estimate we have – the estimate that we’re releasing here is not done in the same way that gives us the 36,000 estimate. That estimate is a different methodology. And will give a slightly larger number than this number would give. But if you were to compare, even though it’s not a directly applicable comparison, under 50 in that estimate, there are less than 1,000 deaths a year in age under 50. We didn’t break out in this – we’re not able to at this time, the 50 to 64 versus [under] 50. But a large portion of those adults are under 50. So it is really many times more severe in terms of severe illness and hospitalizations are several times higher for children and young adults as well in H1N1 than in a usual flu season….

Dr. Frieden’s comments convey the strong impression that pandemic risk is highest for children 0–17 and adults 18–50. As we have seen, the CDC’s December 10 data say that the pandemic is deadliest so far to the 18–64 age group, slightly less deadly to the 65+ age group, and least deadly to the 0–17 age group.

The vaccination public service announcements released by the Department of Health and Human Services on December 7 emphasize three groups: children, pregnant women, and young adults.

Here’s the text of one 10-second television PSA: “Children are especially at risk for the H1N1 flu virus. So get yours vaccinated. Learn more at flu.gov. Together, we can all fight the flu.”

Are children really “especially at risk” for the pandemic virus?

• If you think “especially at risk” means “more at risk than from the average seasonal flu,” then the statement is true.
• If you think “especially at risk” means “more at risk than their parents and grandparents,” then the statement is false according to the CDC’s estimates.

I don’t want to speculate on which message, the true one or the false one, the writers of this PSA meant their audience to hear. Either message would presumably help motivate parents to get their children vaccinated. But which message, the true one or the false one, will adults and seniors hear when they listen to the PSA, and how will that message affect their intention to get vaccinated?

And here is a line from a 30-second PSA distributed by HHS: “Kids run a higher risk of getting the H1N1 flu and developing serious complications.” What message are people in all age groups likely to get from the second half of that sentence – that children are likelier to have serious swine flu complications than older people (false) or that children are likelier to have serious swine flu complications than during an average flu season (true)?

The CDC website currently describes the U.S. government’s pandemic vaccine priority groups as follows:

These target groups include pregnant women, people who live with or care for infants younger than 6 months of age, healthcare and emergency medical services personnel with direct patient contact, infants 6 months through young adults 24 years of age (especially children younger than 5 years of age), and adults 25 through 64 years of age who are at higher risk for 2009 H1N1 complications because of chronic health disorders or compromised immune systems.

I am not making any policy recommendation about vaccine prioritization. It’s not my field. And there are several coherent reasons for prioritizing children (and the youngest adults) even if they are less at risk than their parents and grandparents:

• Young people have more years of life expectancy left.
• Young people are our future.
• Young people may be more likely to transmit infections than the elderly.
• Young people usually have a better antigenic response to influenza vaccines than the elderly.
• Surveys show most Americans would rather protect the young than the elderly.

The question of whether children should be prioritized for vaccination is separable from the question of whether the public should be given misleading information about the relative risk to different age groups.

Good risk communication requires that messaging about vaccine prioritization should always be compatible with the latest available information about the relative risk to the various age groups. Then, if there are reasons for choosing not to align vaccine prioritization with age-related risk, those reasons should be explained. And to the extent that those reasons may be controversial, they should be debated. For people to participate usefully in that debate, they need sound information about both the scientific data (e.g. children are least at risk) and the related values concerns (e.g. children have their whole lives ahead of them).

As the vaccine supply increases to the point where prioritization will no longer be necessary, the question of what to tell people about the relative risk to different age groups will be more important than the question of which age groups to vaccinate first – not just from the perspectives of integrity and credibility, but also from the perspective of public health: motivating precautionary behavior in people not previously prioritized. Many of my friends – mostly in their sixties – have decided they don’t need to be vaccinated against the pandemic virus because they have received the clear impression that their risk from that virus is very low. This misimpression made them comfortable going to the back of the line when the vaccine supply was inadequate. Now it makes them unmotivated to get in line at all.

I want to end this update with a story about U.S. public health resource prioritization in the early history of a different pandemic, one that continues to afflict the world. The punch lines are quotes from a medical ethicist and a former New York City health commissioner arguing for candor.

In the early years of the AIDS pandemic, cases of HIV in the United States were highly concentrated in certain identifiable groups. Efforts to prevent the disease from spreading more broadly – which has since happened – should logically have focused in those early years on the most at-risk groups.

But according to a 1997 Pulitzer Prize-winning Wall Street Journal article by Amanda Bennett and Anita Sharpe, the U.S. government made a political decision to conduct a broad-based anti-HIV campaign instead. It did so despite “unpublished research by the CDC itself [which concluded] that ‘the most effective efforts to reduce HIV infection will target injecting drug users on the Eastern seaboard, young and minority homosexual and bisexual men, and young and minority heterosexual women and men who smoke crack cocaine and have many sexual partners.’”

Bennett and Sharpe asked: “Shouldn’t the public hear the truth, even if there might be adverse consequences?” Their story quotes statements from Boston University medical ethicist George Annas and former New York City Health Commissioner Stephen Joseph that answer their question:

Annas: “When the public starts mistrusting its public health officials, it takes a long time before they believe them again.”

Joseph: “Political correctness has prevented us from looking at the issue squarely in the eye and dealing with it. It is the responsibility of the public-health department to tell the truth.”

Like Joseph, CDC Director Thomas Frieden is also a former New York City Commissioner of Health. During his tenure in New York, he worked courageously to fight HIV, sometimes speaking out in ways that were politically hazardous. In 2005, for example, he quickly announced the discovery of a new, multi-drug-resistant strain of HIV, braving criticism from some scientists for going public too soon, and from some gay activists for promoting sexual stereotypes and trying to frighten gay men into practicing safe sex. Jody Lanard and I wrote a column praising Dr. Frieden for exemplary, candid risk communication.

I hope he will do it again.

My wife and colleague Jody Lanard assisted on this update.

December 2, 2009

It’s Official (sort of):
The Swine Flu Pandemic Is Mild So Far

by Peter M. Sandman

People who get swine flu are one-sixth as likely to die as people who get the ordinary seasonal flu. That’s according to estimates released by the U.S. Centers for Disease Control and Prevention (CDC) on November 12, 2009.

But the CDC didn’t say so. And journalists didn’t do the math. So nothing like the previous paragraph appeared anywhere in the mainstream media on the day of the CDC announcement or the days that followed.

Here I am, three weeks later, a risk communication expert not a flu expert, telling you that the CDC “secretly” (well, quietly) thinks the swine flu pandemic is mild.

So far, anyway. Influenza is famously unpredictable, and influenza pandemics are even more so. Don’t bet the farm (or your life) that swine flu will stay mild, because it could turn more virulent at any time. Still, the big news virtually everybody has missed is that U.S. public health officials think it’s mild so far.

This update is partly the evidence that that’s really what the CDC is saying-but-not-saying, and partly an assessment of why the CDC isn’t saying it more candidly and why reporters aren’t hearing it at all.

The Basic Swine Flu Numbers

The obvious question to ask about any risk is: How serious is it? Risk assessors long ago partitioned seriousness into two components:

• Probability (or frequency) – how often does it happen; how many people does it happen to?
• Magnitude (or consequence) – when it happens, how bad is it?

For an infectious disease, the probability measure is the case attack rate (CAR) – what percentage of the population gets the disease in some unit time? The most important magnitude measure is the case fatality rate (CFR) – what percentage of those who get the disease die from it? – though hospitalization rates and absenteeism rates are also pretty important.

On November 12, for the first time, the CDC provided its best estimate of these two numbers for the first six months of the swine flu pandemic.

Well, not really. It didn’t actually provide a CAR or a CFR. It could have, but chose not to. But it did provide the data needed to calculate these two crucial statistics.

The CDC estimated that about 22 million Americans (with a range of 14–34 million) were sick with swine flu between late April when it first appeared and October 17. The U.S. population is currently 308 million. So a little over seven percent of Americans have had swine flu so far (22 million is 7.1% of 308 million). That’s the swine flu case attack rate for the first six months of the pandemic. It can only go up, as more people in the same population get sick.

By contrast, the average flu season sickens 5–20% of Americans. But of course the next six months could tell a very different story. It’s too soon to tell.

Out of the 22 million swine flu cases so far, the CDC estimated, about 3,900 (with a range of 2,500–6,100) died. That’s a case fatality rate of 0.018% – 180 swine flu deaths for every million swine flu illnesses.

The pandemic case fatality rate could also get higher in the months ahead, especially if the swine flu virus mutates to become more virulent. But unlike a case attack rate, which automatically keeps climbing as the unit time grows longer (as long as the disease remains in circulation), there’s no particular reason to expect a case fatality rate to increase, decrease, or remain the same. So far, at least, the pandemic CFR in the United States is approximately 0.018%.

For an influenza pandemic, a case fatality rate of 0.018% is an astonishingly low number. It’s a lot lower than the CFR of previous pandemics, and a lot lower than the CFR of the average flu season.

These numbers are the biggest pandemic news in some time. They prove that so far the pandemic is mild.

The CDC didn’t interpret the numbers that way, and the media didn’t cover them that way. That’s the main thing I want to talk about.

But I want to talk about four other things first:

• The CDC provided not just overall estimates of the number of U.S. cases and deaths, but also separate estimates for three age groups. I want to get those estimates (and CARs and CFRs based on them) onto the table too, so we can see whether the pandemic is severe for some even though it’s mild overall.
• Risk estimates take on meaning mostly in comparison to other risks that are larger or smaller. I want to put the pandemic’s CAR and CFR figures into context by comparing them in detail to comparable figures for the seasonal flu, the 1918 pandemic, and bird flu. (You can skip this part if you don’t like numbers and you’re willing to take my word for it.)
• The meaning of the word “mild” is central to this update. I want to detail what I mean by the word. In particular, I want to talk about the ways in which this pandemic looks less dangerous than it did at first, the ways in which it could get worse, and the ways in which it could get still less serious … and could even end up saving lives.
• Nearly everything in this update is based on CDC estimates. I want to explain why we need to rely on estimates, why we don’t have solid counts of flu cases and flu deaths.

Then, finally, I can turn to the risk communication issues:

• Why does the CDC virtually never use the word “mild” to describe the pandemic virus?
• Why do local health officials and journalists settle for the meme that pandemic H1N1 is “about like the seasonal flu”?
• Why did the CDC decide to interpret its November 12 estimates as more alarming than reassuring?
• Why did the media go along with the CDC’s misinterpretation?
• And why do I think it’s important for health officials to be more candid about the fact that the pandemic is mild so far, as well as the possibility that it could turn more virulent.

Ages of Swine Flu Victims

The CDC divided the U.S. population into three age groups: 0–17, 18–64, and 65+. For convenience, I’ll call them children, adults, and seniors. Here are the relevant estimates. For each age group, the first line comes directly from the CDC. The other three lines are my arithmetic to calculate a case attack rate (for the first six months of the pandemic) and a case fatality rate. The only additional information used is the U.S. census data.

Children:	CDC estimates 8 million cases and 540 deaths 24.3% of the U.S. population of 308 million = 75 million children 8 million cases out of 75 million children = 10.7% CAR       (11 sick children out of every hundred children) 540 deaths out of 8 million cases = 0.007% CFR       (70 dead children out of every million sick children)
Adults:	CDC estimates 12 million cases and 2,920 deaths 62.9% of the U.S. population of 308 million = 194 million adults 12 million cases out of 194 million adults = 6.2% CAR       (6 sick adults out of every hundred adults) 2,920 deaths out of 12 million cases = 0.024% CFR       (240 dead adults out of every million sick adults)
Seniors:	CDC estimates 2 million cases and 440 deaths 12.8% of the U.S. population of 308 million = 39 million seniors 2 million cases out of 39 million seniors = 5.2% CAR       (5 sick seniors out of every hundred seniors) 440 deaths out of 2 million cases = 0.022% CFR       (220 dead seniors out of every million sick seniors)

Okay, so what stands out?

These conclusions are in conflict with two things you may have heard.

Overall Comparison to Seasonal Flu

The seasonal flu isn’t a single phenomenon; there are severe flu seasons and mild flu seasons. So the seasonal flu numbers below aren’t just estimates (like the pandemic numbers above); they are estimates of averages.

According to the CDC, about 10% of the U.S. population gets the seasonal flu in an average year. The number ranges between 5% and 20%, but we’re going to stick with the 10% average. So the seasonal flu case attack rate is about 10%. That means there are 31 million flu cases a year, on average, in the current U.S. population of 308 million.

The estimated average number of annual U.S. deaths from the seasonal flu, again according to the CDC, is 36,000. If 31 million people get sick and 36,000 die, that’s a case fatality rate of 0.12%.

Of course the actual CFR varies with the severity of each year’s flu strains, the effectiveness of each year’s vaccine, the number of people who get vaccinated, and other factors. It ranges from 0.05% to 0.2%, but is usually shorthanded to an average for a typical year of about 0.1%. We’ll stick with 0.12%, 36,000 deaths divided by 31 million cases. That’s 1,200 deaths for every million cases of flu.

If we multiply the average CAR of 10% by the average CFR of 0.12%, we get an average seasonal flu PMR (population mortality rate) of 0.012%. For every million people in the United States, roughly 120 of them die every year of a flu-related illness.

So here’s the overall comparison for all three parameters, case attack rates, case fatality rates, and population mortality rates:

	2009 Pandemic (estimated)	Seasonal Flu (estimated average)
CAR	7.1% (6 months)	10% (annual)
CFR	0.018% (180 in a million)	0.12% (1,200 in a million)
PMR	13 in a million (6 months)	120 in a million (annual)

The comparison of case attack rates (and therefore population mortality rates) isn’t terribly useful yet. The seasonal flu numbers are annual averages, while the pandemic has lasted only six months so far. There’s no way to predict what will happen over the next six months. If the pandemic virus continues infecting people at the same rate as the first six months, that would yield a first-year CAR for the pandemic of 14.2%, significantly more than the average seasonal flu. But there’s no reason to expect the next six months to be identical to the first six months. The pandemic could accelerate; we’re approaching winter, and influenza tends to transmit better in winter. Or the pandemic could ebb; as I write this, it seems to be ebbing. Or it could accelerate and then ebb. Or it could ebb briefly and then come roaring back in a third wave in late winter or spring.

For what it’s worth, previous flu pandemics have ended up sickening one-quarter to one-half of the population over roughly two years. And the first year has usually been bigger than the second year. So this pandemic’s 7.1% case attack rate for its first six months is a little on the low side for a pandemic … but it’s early days yet.

Still, unless it fades soon, the pandemic is on course to sicken more than 10% of the population by the end of its first year, beating out the average flu season.

Note that the typical U.S. flu season runs from November through April, usually peaking in January and February. Pandemics, by contrast, have no season (though they tend to be worse in cold weather than hot weather). In the U.S., the swine flu pandemic started in late April, had a spring spike, then receded but didn’t disappear over the summer, then spiked again in the fall, and now seems to be receding again (though there’s no way to know if it will keep receding).

Statements pointing out that the amount of pandemic illness is very high “so early in the flu season” are accurate if they mean to suggest that only during pandemics do we see a lot of flu in spring and fall. And it’s reasonable to worry that the pandemic might get worse as the weather gets colder. But it is misleading to compare the pandemic’s first six months with the seasonal flu’s off-season in order to imply that the pandemic is making more people sick than the seasonal flu. To compare the pandemic’s CAR with the seasonal flu’s CAR, the most valid comparison is a year versus a year, and this pandemic’s first year started in late April 2009.

The more enlightening comparison for now is the pandemic’s CFR with the seasonal flu’s CFR. Those numbers don’t include any time unit. The average seasonal flu kills 1,200 Americans for every million it makes sick. So far, the swine flu pandemic has killed only 180 Americans for every million it has made sick. In other words, the average seasonal flu is more than six times as deadly as swine flu if you get it.

There’s a word for a pandemic virus that is less than one-sixth as virulent as the average seasonal flu virus. The word is mild.

Comparison to Seasonal Flu by Age Groups

What about the three age groups? Does the picture change when we look separately at children, adults, and seniors?

We can’t do a rigorous comparison. I can’t find published seasonal flu attack rates separately for children, adults, and seniors. And the CDC only started collecting data on pediatric seasonal flu deaths a few years ago.

Here are the two key published CDC estimates for seasonal flu: It kills an average of 36,000 people a year in the U.S., and 90% of them are 65 and over. Let’s work with just those two numbers to get two population mortality rate estimates for the seasonal flu, one for Americans 65 and over and one for Americans under 65:

1. 90% of 36,000 U.S. flu-related deaths are in people 65 and over. 90% of 36,000 = 32,400. So an estimated 32,400 Americans 65-and-over die of a flu-related illness in the average flu season.
2. 12.8% of the U.S. population of 308 million is 65 and over. 12.8% of 308 million is 39 million. So about 39 million Americans are 65 and over.
3. Dividing #1 (32,400 flu-related deaths in people 65 and over) by #2 (39 million people 65 and over) yields an estimated average seasonal flu-related annual mortality rate for U.S. seniors of 0.083%. (Reminder: This isn’t the case fatality rate. It’s the 65-and-over population mortality rate, the percentage of the population 65 and over who die of a flu-related illness in an average year, not the percentage of that year’s 65-and-over flu cases who die.)
4. 10% of 36,000 U.S. flu-related deaths are in people under 65. 10% of 36,000 = 3,600. So an estimated 3,600 Americans under 65 die of a flu-related illness in the average flu season.
5. 87.2% of the U.S. population of 308 million is under 65. 87.2% of 308 million is 269 million. So about 269 million Americans are under 65.
6. Dividing #4 (3,600 flu-related deaths in people under 65) by #5 (269 million Americans under 65) yields an estimated average seasonal flu-related annual mortality rate for U.S. non-seniors of 0.0013%. (Again, this is an population mortality rate, not a case fatality rate.)

The estimated average seasonal flu population mortality rate for Americans 65 and over is 0.083% – 830 in a million. The estimated average seasonal flu population mortality rate for Americans under 65 is 0.0013% – 13 in a million. If you divide 0.083 by 0.0013, you get 64. An American in the 65-and-over age group is roughly 64 times (0.083 divided by 0.0013) as likely as an American under 65 to die from a flu-related illness.

These population mortality rate calculations are as firm (or as flimsy) as the CDC’s “36,000” and “90%” estimates are. If the CDC is right that about 36,000 Americans die in an average flu season and about 90% of them are 65 and over, then I must be right also that the seasonal flu is 64 times as likely to kill a U.S. senior as a younger U.S. resident.

But I don’t have any age-related case attack rate data for the seasonal flu. To calculate a very approximate CFR, therefore, I’m going to assume a 10% CAR for both age groups. I have already defined a disease’s population mortality rate as its case fatality rate times its case attack rate. It follows that the CFR is the PMR divided by the CAR.

So our estimate of the average seasonal flu CFR for Americans 65 and over is 0.083% divided by 0.1 = 0.83%. And our estimate of the average seasonal flu CFR for Americans under 65 is 0.0013% divided by 0.1 = 0.013%.

Now we can compare the average seasonal flu to the first six months of the swine flu pandemic. Since our seasonal calculations distinguish only two groups, under 65 and 65+, we need to combine the pandemic data for children and adults. Here’s what we get:

	Pandemic	Seasonal
CAR – under 65	7.4% (6 months)	10%* (annual)
CAR – 65+	5.2% (6 months)	10%* (annual)
CAR overall	7.1% (6 months)	10%  (annual)
CFR – under 65	0.017%	0.013%*
CFR – 65+	0.022%	0.83%*
CFR overall	0.018%	0.12%
PMR – under 65	13 in a million (6 months)	13 in a million (annual)
PMR – 65+	11 in a million (6 months)	830 in a million (annual)
PMR overall	13 in a million (6 months)	120 in a million (annual)

(The four asterisks denote estimates that are grounded in the probably false assumption that the seasonal flu attacks under-65s and 65-and-overs with the same frequency.)

The most stunning comparison here is the four CFR estimates. Look at them:

• 0.017% (pandemic CFR for Americans under 65)
• 0.013% (seasonal flu CFR for Americans under 65)
• 0.022% (pandemic CFR for Americans 65+)
• 0.83%   (seasonal flu CFR for Americans 65+)

As they say on Sesame Street: One of these is not like the others.

What jumps out from the chart is that the seasonal flu is a pretty scary disease for people 65 and over. If you catch it, the odds of it killing you are 0.83% – 83 in a thousand, nearly one in a hundred. (And lots of seniors catch it.) Of course 83 in a thousand is the average. The seasonal flu kills more seniors in years dominated by seasonal H3N2 flu than in years dominated by seasonal H1N1 flu.

The other three CFRs are all so much lower that they look almost indistinguishable by contrast: 0.013%, 0.017%, and 0.022% – from just over one in a thousand to just over two in a thousand, compared to 83 in a thousand.

But these three CFRs are worth distinguishing:

• Americans 65+ who get swine flu are a little likelier to die than Americans under 65 who get swine flu (0.022% versus 0.017%).
• Americans under 65 who get swine flu are a little likelier to die than Americans under 65 who get the seasonal flu (0.017% versus 0.013%).
• But they are all enormously less likely to die than Americans 65+ who get the seasonal flu (0.013%-0.022% versus 0.83%).

The PMR (population mortality rate) estimates are the most reliable numbers; they’re not based on the assumption of equal seasonal flu attack rates. In a mere six months, the CDC’s data say, swine flu has already killed as many Americans under 65 as the seasonal flu kills in an average year: 13 per million population. And it has killed almost the same percentage of Americans 65 and over: 11 in a million. So, yes, swine flu really is deadlier to Americans of all ages than the seasonal flu is to Americans under 65.

That’s because the seasonal flu isn’t very deadly to Americans under 65. It’s the fourth PMR that really stands out: In an average year, the seasonal flu kills not 11 seniors in a million, not 13 in a million, but 830 seniors in a million! (Remember: The average seasonal flu kills 36,000 people in the U.S.: 32,400 people 65+, and 3,600 people under 65.)

The communication problem these numbers present is a very standard problem in epidemiology: how to talk about relative risk so people don’t get a misperception of absolute risk.

Suppose there are two cancers. One is pretty rare, killing only a dozen or so people for every million in the population; the other is much more common, killing nearly a thousand people for every million. Now suppose some lifestyle choice (grilling steak over charcoal, say) increases the incidence of the first cancer, the rare one, by about a third. How worried should barbecue lovers be? On the one hand, you’re increasing the cancer rate by about a third! On the other hand, if you increase the incidence of a pretty rare cancer by about a third you still have a pretty rare cancer. A much tinier percentage increase in the incidence of the common cancer would be a much bigger absolute health threat.

In this analogy, seasonal flu for people 65 and over is the common cancer. The difference between pandemic H1N1 and the seasonal flu for people under 65 is the effect of frequent barbecuing on the incidence of the rare cancer. An ethical epidemiologist helps people interpret relative risk data based on the size of the absolute risk: “The number goes up by about a third” is ultimately less informative than “the number goes up by a few extra cases in a million.”

In a nutshell: So far, the swine flu pandemic is a little more serious than the average seasonal flu for Americans under 65. But the average seasonal flu is a life threat almost exclusively to people over 65. No age group is hit anywhere near as hard by the pandemic (so far) as seniors are hit by the seasonal flu.

Or look at it this way. People over 65 are 12.8% of the U.S. population. According to the CDC’s November 12 estimates, they make up 11.3% of U.S. pandemic deaths so far (440 out of 3,900 estimated deaths). The two percentages – 12.8% and 11.3% – are pretty close. But the CDC says people over 65 – 12.8% of the population, remember – make up 90% of U.S. seasonal flu deaths.

The main thing going on isn’t that the pandemic discriminates on the basis of age, attacking mostly the young and sparing the elderly. The difference in case attack rates is fairly small. The main thing going on is that the seasonal flu discriminates on the basis of age, killing mostly the elderly while it mostly gives their children and grandchildren a miserable week in bed.

Some seasonal flu vaccination campaigns are candid about the age differences in vulnerability, but most are not. When officials simply say that we should all get vaccinated every year because flu kills 36,000 Americans a year, they are grossly understating the risk to grandma and grossly overstating the risk to her kids and grandkids. (Children under four are also at increased risk of seasonal flu complications, compared with the rest of the under-65 group.) One reason officials perpetuate this misleading impression is that children are thought to be the main transmitters of seasonal flu. And flu vaccines are more effective in young people than in seniors. One way to protect grandma is to get her kids and grandkids vaccinated.

Individual doctors take their cue from public health professionals. The New England Journal of Medicine recently published an article by internist Danielle Ofri that offers a pristine example of a doctor who regularly oversells the risk of seasonal influenza to her patients under 65.

Dr. Ofri writes that her patients who “routinely refused the seasonal flu vaccine” were initially desperate for pandemic H1N1 vaccine last spring, when the virus first emerged. But six months later her patients were more worried about vaccine safety than about pandemic flu.

“How to explain this dramatic shift in 6 short months?” she wonders. “It certainly isn’t related to logic or facts, since few new medical data became available during this period. It seems to reflect a sort of psychological contagion of myth and suspicion.”

Leave aside that “new medical data” have in fact become available since early in the pandemic, data suggesting that the pandemic is milder than it originally looked in Mexico City. I want to focus on Dr. Ofri’s unsuccessful efforts to persuade her patients to get the seasonal flu vaccine every year.

Presumably, those patients are mostly under 65, since most Americans 65 and over do get the seasonal vaccine. How does Dr. Ofri try to persuade them to get vaccinated? Frustrated by her lack of success, she earnestly tells us her approach: “No matter how often I trotted out the statistics of 30,000 to 40,000 annual deaths from influenza, the patients would not be moved.”

It’s a safe bet that when Dr. Ofri tells her under-65 patients about the 30,000–40,000 annual deaths from influenza, she deemphasizes the fact that 90% of those deaths are in people 65 and over. Dr. Ofri probably doesn’t perceive herself as exaggerating the seasonal flu risk to her under-65 patients, but that is exactly what she is doing. It’s worth wondering whether her apparent willingness to mislead her patients in this way might have something to do with their resistance to her advice about vaccination.

I think the decision of public health authorities and individual doctors to underplay the age differences in seasonal flu vulnerability is a culpable decision. Lots of Americans under 65 get the flu every year, and 99.987% of them recover. Lots of Americans 65 and over get the flu every year, and nearly one percent of them die. Rather like breaking your hip, influenza is a big deal for the elderly and an inconvenience for the young. There are exceptions, of course; 3,600 annual flu-related deaths in the under-65 population aren’t nothing. But they pale to insignificance beside the 32,400 annual flu-related deaths in the much smaller 65-and-over population.

It’s very difficult for officials to talk coherently about the seriousness of the pandemic without first coming clean about the seasonal flu. Seasonal flu is basically two diseases: a severe one for huge numbers of seniors and a mild one for nearly all of their children and grandchildren. The swine flu pandemic so far is – for both age groups – a little more severe than the seasonal flu is for people under 65, and a lot milder than the seasonal flu is for seniors.

For the record: Despite the critique above, I think that 3,600 seasonal deaths a year in people under 65 are well worth preventing. Most of those deaths are likely to be preventable with the seasonal flu vaccine – which I believe is very safe, and which is thought to be 70% to 90% effective in young healthy people when the vaccine match is good. It is the misleading, mistrust-inducing communication that needs changing, not the important effort to increase vaccination rates.

I also need to acknowledge that there are at least two ways in which the swine flu pandemic is (or may be) worse than the average seasonal flu.

• Swine flu is especially likely to be serious for people with specific underlying medical conditions: pregnant women, asthmatics, diabetics, etc. Statistics aren’t available to permit a direct comparison, but for some of these conditions it is possible that swine flu is worse than the seasonal flu.
• The percentage of doctor visits that were for influenza-like illnesses has already reached higher peaks during the first six months of the swine flu pandemic than it usually reaches during non-pandemic flu seasons. In other words, the pandemic (and news about the pandemic, and concern about the pandemic) is sending a lot of people to the doctor with respiratory complaints who would have toughed it out at home during an ordinary flu season.

Neither of these facts is anywhere near as important as the fact that so far swine flu is much less deadly than the seasonal flu. No other fact about the swine flu pandemic is as important as that fact.

Two More Comparisons

The seasonal flu is the most useful yardstick for assessing the seriousness of the swine flu pandemic. But let me add, quickly, two other yardsticks: the pandemic of 1918 (also an H1N1 strain) and the bird flu (H5N1) threat that has hung over the world since 1997.

The 1918 pandemic is estimated to have sickened about 30% of the U.S. population, and killed about 675,000 of them – virtually all within the space of a single year, though the pandemic actually lasted 27 months in all. The U.S. population at the time was 103 million. (1918, by the way, was the only year since 1900 in which the U.S. population actually declined, thanks to the 1918 pandemic.)

So the 1918 pandemic sickened 31 million Americans (30% of 103 million). The 675,000 who died are 2.2% of the 31 million sickened, and 0.66% of the 103 million U.S. population.

Now compare the two pandemics:

	Pandemic of 2009	Pandemic of 1918
CAR	7.1% (first 6 months)	30% (worst 12 months)
CFR	0.018%	2.2%
PMR	13 in a million (first 6 months)	6,600 in a million (worst 12 months)

The pandemic of 1918 was the worst pandemic in modern times, maybe the worst in history. It infected 30% of Americans, an attack rate the current pandemic may or may not eventually equal. But here’s a number the current pandemic is vanishingly unlikely to equal, unless the swine flu virus mutates and becomes far more virulent: The 1918 pandemic had a case fatality rate of 2.2%, 122 times worse than the swine flu CFR to date.

As for bird flu, there are no U.S. data, of course, since there haven’t been any U.S. cases (at least not yet). Since its emergence in 1997, the H5N1 bird flu virus has infected a grand total of 442 people worldwide, according to the World Health Organization. 262 of them died.

With a world population of 6.8 billion, bird flu’s CAR is absurdly low: 442 out of 6.8 billion is 0.0000071%. But bird flu’s CFR is terrifyingly high: 262 out of 442 is 59%. So far, in other words, H5N1 is almost impossibly difficult to catch – but if you catch it you’re likelier than not to die.

Since 1997, the pandemic worst case scenario in the back of every influenza expert’s mind has been this: What if bird flu mutates in a way that enables it to transmit efficiently from human to human, while staying just as deadly as it is already?

Meanwhile, instead of a bird flu pandemic that kills 59% of its victims, we have been dealt a swine flu pandemic that kills 0.018% of its victims – 3,278 times as mild.

So, just to rub it in:

• Worldwide bird flu CFR               59.%
• U.S. 1918 pandemic CFR               2.2%
• Average U.S. seasonal flu CFR      0.12%
• 2009 pandemic CFR in the U.S.     0.018%

Like I said: mild.

What Does “Mild” Mean Here?

Nearly all of the above numbers are based on CDC estimates. (One exception worth noting: I’m using the conventional estimate of 10% for the seasonal flu case attack rate; the CDC usually uses a range of 5% to 20% instead.)

The CDC estimates are debatable. And in fact they are hotly debated – most notably by vaccination opponents, who argue that the CDC has an incentive to overestimate both seasonal flu deaths and swine flu deaths in order to build the case for vaccination.

I’m neither a virologist nor a biostatistician, and I don’t feel qualified to have an opinion about the validity of the CDC’s estimates. But if the estimates are reasonably accurate – or even if they’re equally inflated – then it is clear that so far the swine flu pandemic is basically a practice run. It is more severe than the seasonal flu for the people least endangered by the seasonal flu (under-65s), but it’s a lot less severe than the seasonal flu for the people most endangered by the seasonal flu (65-and-overs). And it’s a lot less severe than the pandemic we were expecting.

It’s also less severe than the pandemic we thought we were getting at first. A May 2009 article in Science, based on early data from Mexico, estimated the pandemic’s case fatality rate at 0.4% – three to four times as bad as the seasonal flu’s average CFR in the U.S. In the weeks that followed, that number began to seem high, though not as high as it seems now. Here’s what I wrote about what I called “the ‘mildness’ meme” in a late June 2009 email responding to a query from Canadian Press reporter Helen Branswell:

Compared to the seasonal flu, which kills on average about one person in a thousand, swine flu isn’t especially mild. Health officials can’t calculate a precise case fatality rate because they’re not trying to count all the cases – so they don’t have any denominator for the fraction. But it’s looking like swine flu is roughly as deadly as the seasonal flu – about one death in a thousand cases. And while seasonal flu deaths are overwhelmingly the elderly, swine flu deaths tend to be younger.

By July 21, only a month later, I was writing that pandemic H1N1 “looks very mild so far” – with a case fatality rate “apparently lower than the rate for the seasonal flu.”

From the perspective of late November, May’s 0.4% and June’s 0.1% look way too high. Based on the November 12 CDC data, the best U.S. pandemic CFR estimate is 0.018%.

That could change. Flu viruses mutate, and swine flu could mutate in a way that makes it more virulent, increasing its CFR. (A mutation could also make it more transmissible, increasing its CAR. Or it could make it no longer susceptible to antiviral drugs, or to the vaccine we have worked so hard to produce.) Of course a mutation that makes swine flu even milder is also possible, but it’s the possible change for the worse that merits attention. One worst case scenario for swine flu: The swine flu and bird flu viruses mix-and-match genetic material, a process called reassortment, to produce a new flu strain with the transmissibility of swine flu and the virulence of bird flu – which would make it by far the worst flu virus in history. Even without such a catastrophic reassortment, the swine flu pandemic could come roaring back in a third wave that would make the first two waves feel like practice runs.

And that is why it’s so important for the CDC to acknowledge that this pandemic – so far – is mild. The first two waves really are practice runs, a chance to prepare for The Big One. This is not being candidly communicated to the public, so the public is not even thinking about The Big One, let alone preparing for it. Nobody knows whether The Big One is coming next month or next century. But we do know that it’s not here yet.

This is a crucial point. There are many reasons to tell people that swine flu is currently mild:

• Because it’s the truth.
• Because people who are very anxious deserve to be told they can calm down.
• Because people who already think it’s mild deserve to be told they’re right.
• Because the credibility of the CDC and of the public health profession is at stake.

But the biggest reason to tell people that swine flu is currently mild is to help people realize that what we’re experiencing so far isn’t what we have been worrying about, that we’re not out of the woods yet, that the next pandemic or the next wave of this pandemic could turn out to be The Big One.

Nonetheless, “mild” doesn’t mean “trivial.” Swine flu has, after all, killed about 3,900 Americans in six months, more than the September 11 attacks of 2001. It will surely kill more in the months ahead.

And though some think otherwise, to me the swine flu virus is surely more dangerous than the swine flu vaccine. Let the policy experts debate whether the swine flu vaccine manufacturing effort was worth the cost. That’s a puzzle that looks different today than it did last spring when decisions had to be made without knowing the virus would turn out so mild (so far) – and it’s a puzzle that may look different yet again next spring when we have twelve months of pandemic experience to look back on. The fact is, we now have a pandemic vaccine (though not yet enough of it). Despite the mildness of the virus so far, getting vaccinated ought to be a no-brainer.

This is a major conundrum for public health professionals. Surveys show that lots of Americans are already deciding against swine flu vaccination because they suspect the virus is mild. Lots more would presumably make the same decision if the CDC flat-out told them the virus is mild. But the virus is mild. So telling the truth about the mildness of the pandemic virus could undermine swine flu vaccination – and thus kill people in the coming months whose deaths could have been prevented by the vaccine.

On the other hand, continuing to exaggerate the seriousness of the pandemic by failing to be explicit about its mildness could undermine long-term public health credibility – and thus kill people in the coming decades whose deaths could have been prevented if they trusted the advice of public health officials.

Whenever health officials warn the public that something may turn out awful and it turns out pretty minor instead, credibility suffers to some extent. A number of risk communication strategies can minimize the damage: emphasizing the magnitude of the risk rather than overstating its probability; acknowledging that it may turn out minor (and voicing hope that it does); explaining that false alarms are part-and-parcel of preparedness; etc. Still, warnings that in hindsight turn out to have been unnecessary or exaggerated inevitably damage the credibility of the source. (Of course over-warning the public damages credibility a lot less than understating the risk and having the situation turn out dire. That’s why crisis communication experts always advise officials to err on the alarming side. )

So yes, it’s embarrassing and a little damaging to tell people that the situation looks better (so far) than you feared. But if the situation actually looks better (so far) than you feared, telling people so does less damage to credibility than not telling them and waiting till they figure it out for themselves.

If it keeps ignoring its own data and pretending that the pandemic is less mild than it actually is (so far), the CDC will be digging itself a deeper and deeper credibility gap. It really shouldn’t let the situation deteriorate to the point where the only thing that can save the agency’s credibility is if the pandemic virus mutates into something much more severe.

For me, there’s no conundrum here. The CDC was right to warn people that the pandemic might get bad. In fact, I criticized the CDC early on for not warning people aggressively enough, and for not urging the public to do what the government was doing: prepare for the worst while hoping for the best. Now the CDC needs to keep warning people that the pandemic still might get bad – while telling people that so far it is remarkably mild.

Uncertainty is a central characteristic of an influenza pandemic. So candor about uncertainty is a central characteristic of pandemic communication.

The pandemic isn’t guaranteed to continue on its current course.

On one extreme, as we have discussed already, the pandemic virus could turn more virulent – even much more virulent.

On the other extreme, the pandemic could actually end up saving lives.

How is that possible? As has been true in most previous pandemics, the H1N1 pandemic virus seems to be crowding out the seasonal flu viruses. That too isn’t guaranteed. (Very little about influenza is guaranteed, except its unpredictability.) But in the northern hemisphere this past spring and summer, the seasonal strains pretty much disappeared (as they usually do in the summer), while the pandemic strain continued to circulate. Meanwhile, the southern hemisphere was having its June-August winter. Instead of a typical flu season, those countries in the southern hemisphere that do systematic influenza surveillance reported that nearly all of their flu cases and flu deaths were from the pandemic strain, not from any of the seasonal strains. And since the pandemic strain is less deadly so far than the worst of the seasonal strains (H3N2), most countries in the southern hemisphere (among the countries that do surveillance) reported fewer flu deaths than usual.

The CDC says there is very little seasonal flu in the U.S. so far; it’s nearly all pandemic H1N1. But there’s not usually much seasonal flu in late November. If this were a normal year, the flu season would be just barely beginning.

This isn’t a normal year. And no one knows yet if we’ll have a normal flu season or not. Here are some of the possibilities, in addition to the possibility that the pandemic virus could become more virulent:

• The pandemic virus recedes and eventually disappears, at least temporarily, while some or all of the seasonal strains end up circulating as usual. We face a pretty typical winter flu season with mostly just the seasonal strains.
• The pandemic virus remains pervasive enough to out-compete the seasonal strains. It doesn’t become more virulent, though maybe it becomes more pervasive as the weather gets colder. We face a milder-than-usual winter flu season with mostly just the pandemic strain.
• The pandemic virus remains pervasive but the seasonal strains pick up steam anyway. We face a bad winter flu season with both seasonal and pandemic flu strains co-circulating.
• The pandemic virus recedes and disappears, at least temporarily, but the seasonal strains have been knocked out, at least temporarily. We face virtually no winter flu season at all.

From what I’ve read, the experts seem to think the first three possibilities are all entirely possible, and they’re not ruling out the fourth. In other words, they haven’t a clue.

In the second and fourth of the four possibilities – the two with little or no seasonal flu this year – the pandemic will have saved lives in the U.S. (assuming the pandemic virus itself doesn’t get more virulent). Saving lives is about as mild as “mild” can get.

The possibility of little or no seasonal flu this year isn’t a secret – but it certainly isn’t something public health officials like to talk about, for fear of undermining seasonal flu vaccination. Public health officials have been avoiding talking about this possibility for months.

Starting in early November, the CDC began to acknowledge this possibility, at least in its weekly flu report: “It is too early in the influenza season to determine if seasonal influenza viruses will circulate widely or how well the seasonal vaccine and circulating strains will match.”

Jody Lanard and I discussed the “what if there’s no seasonal flu this year” question at some length in our September 26 update, “Overselling Seasonal Flu Vaccination in a Pandemic Season.” So I won’t belabor the point here.

But I do want to quote two key paragraphs from that update:

If the seasonal flu strains disappear, or mostly disappear, there will be lots of media stories about that fact in the months to come. People in the northern hemisphere will learn that flu experts knew it was a real possibility as early as April 2009, as soon as the new soon-to-be pandemic virus emerged. They will learn that the amount of seasonal flu in the southern hemisphere dramatically decreased in that hemisphere’s 2009 flu season – decreased in some countries to minuscule levels, as pandemic flu H1N1 dominated in country after country.

Then people (reporters, bloggers, politicians, citizens) are very likely to ask officials:

• Why did you push the seasonal flu vaccine so hard last fall?
• Why didn’t you at least tell us you didn’t know yet whether it would be necessary?
• Why didn’t you give us enough information to make our own informed decision whether to get the seasonal vaccine early or wait until more was known about the need for it?
• If swine flu suddenly disappeared, or looked like it was disappearing, you wouldn’t be urging so many people to get that vaccine, would you? Or would you?

The last bullet point is worth pondering. It would be an exaggeration to claim that swine flu looks like it’s disappearing, but it has been declining in the U.S. for several weeks now. Despite the downward trend, public health officials doubt that the pandemic virus will disappear when barely seven percent of the U.S. population have caught it so far. They expect to see more waxing and waning before a definitive wane. And they are certainly still urging people to get the pandemic vaccine as soon as it becomes available.

They’re almost certainly right to do so; it’s not over till it’s over, and the new flu virus will probably circulate at some level of virulence for years. But I hope officials will be aggressively candid about the evidence that the pandemic’s second wave may be ebbing. I hope they will base their case for pandemic vaccination on the uncertainty of the future, not on exaggerated claims about the seriousness of the present.

Here is the CDC’s weekly graph tracking the percentage of doctor visits that were for an influenza-like illness (ILI):

The graph demonstrates four points: (1) ILI doctor visits peaked higher during the second wave of the pandemic than during the last couple of flu seasons; (2) they’re much, much higher than usual so early in the season; (3) they took ten weeks to climb from trough to peak; and (4) for the last four weeks they have started precipitously downward, but it’s too soon so say whether they’ll continue down or turn around. (As I’ve already noted, the second point isn’t much of a point, since pandemics aren’t seasonal – though it’s true that cold weather tends to increase flu transmission.)

Based on earlier data pointing in this downward direction, CDC Director Thomas Frieden said in Congressional testimony on November 4: “It’s likely that the current wave of infection will peak, crest, and begin to decline before there are ample supplies” of vaccine. He added: “Whether there’ll be another wave of H1N1 between now and May and whether we’ll get a different strain, only time will tell.”

Frieden was testifying on the vaccine shortage. His prediction that the pandemic wave would soon “begin to decline” (before the vaccine supply was adequate) was interpreted as an admission of defeat vis-à-vis pandemic vaccination. This is pretty stunning when you think about it. The head of the CDC tells Congress that the second pandemic wave may be ending, and Congress is so preoccupied with vaccination program deficiencies that it sees this as bad news, not good news.

Maybe it is bad news after all. Maybe the pandemic’s ebb will make room for the more deadly seasonal flu to make a comeback. Nobody knows.

But for those who are focused on the pandemic itself, it’s got to be good news. Not only have the pandemic’s first and second waves both been mild. The second wave may already be ebbing. And there may be time to get vaccinated before another wave.

If the second wave does in fact continue to decline, there may, of course, be a third wave coming after it. The third wave may or may not be short. It may or may not be mild. But any way you look at it, the big pandemic news is this: So far, we have been very, very lucky. That’s the news – but it has not been the official message, and the mainstream media have missed it entirely.

Why Do We Have Only Estimates?

You may wonder why we have to rely on CDC estimates, instead of actual numbers.

The statistics I have been playing with in this update – case attack rates (CARs), case fatality rates (CFRs), and population mortality rates (PMRs) – require knowing two things about the disease in question:

• how many people in a particular population of known size catch it in some period of time; and
• how many of those people die from it.

The first number divided by the population size gives you the CAR (case attack rate); the second number divided by the first number gives you the CFR (case fatality rate); the CAR times the CFR gives you the PMR (population mortality rate). All you need to know is how many people get sick with the disease and how many die from it.

For influenza, pandemic or seasonal, there is simply no way to count the people who get sick from the disease. Here’s why:

• Lots of people have mild symptoms and don’t go to the doctor. They’re missed unless authorities do various types of surveys to pick them up.
• Others go to the doctor and are diagnosed with an “influenza-like illness” but don’t get a flu test. (Flu experts shorten “influenza-like illness” to “ILI”; the fact that they have extensive use for this abbreviation tells you how seldom they actually know whether a case is actually flu or some other ILI.)
• Still others get a screening test for flu, but not a more expensive and sensitive test to identify which kind of flu they have (or had). So once a pandemic has started you can’t be sure which cases are pandemic flu and which are seasonal flu unless you test for that – though nearly all such specific tests in the U.S. since the summer have either been positive for pandemic flu or negative for flu; there’s very little seasonal flu out there so far this year.
• Many of the tests, even some of the more expensive ones, have high rates of false positives and false negatives.

If you count only people who have a laboratory-certified case of influenza, you’re grossly underestimating the number of flu cases. And if you count everybody who recalls having an ILI, you’re grossly overestimating the number of flu cases. There are known to be people who had flu-like symptoms, got treated for flu, and when they were eventually tested turned out to have a bad rhinovirus (a cold). There are known to be people who had flu-like symptoms, figured it was a cold and just stayed home, and when they were eventually tested turned out to have flu. Obviously lots of people who are never tested may be getting misclassified.

So a few months ago, the CDC stopped counting flu cases (swine flu or seasonal flu), and reverted to flu surveillance and monitoring systems that are closer to its methods during non-pandemic years. It does careful laboratory analyses of small samples from around the country and then extrapolates. That in one oversimplified sentence is how it comes up with its estimates of the number of flu cases. On its website, the CDC offers a more thorough explanation of its multiple surveillance mechanisms.

Figuring out the number of flu deaths (again, pandemic or seasonal) raises some of the same problems: people who die without being tested or whose tests may not be accurate. But it also raises a different sort of problem. The number of flu deaths is an informed guess, mostly because flu victims don’t necessarily die of “pure flu.” They often die of a preexisting disease that the flu made worse, or they die of a new disease they got because the flu made them weak, or they die of the flu because other diseases made them unable to fight it off, or they die of several things at once including the flu. Even when it is established that someone got the flu and later died, whether to attribute that death to the flu is a judgment call.

The CDC has been estimating seasonal flu cases and deaths for years. But until the November 12 estimates were released, the CDC had only its tally of laboratory-confirmed swine flu cases and laboratory-confirmed swine flu deaths to go on.

Journalists loved these swine flu tallies, and used them endlessly, even though the CDC kept warning that the former number, especially, was way low because it omitted cases that had never been tested. (The World Health Organization still maintains worldwide tallies of laboratory-confirmed cases and deaths, and the media still dote on them, especially in developing countries.) A very low estimate of the number of swine flu cases so far made the pandemic look less pervasive than it actually was. It also made the pandemic look more deadly than it actually was, since the smaller number of recorded cases made for a higher apparent case fatality rate (not that journalists ever bothered to do the math).

From time to time various CDC officials tried to correct the record with seat-of-the-pants guesstimates of the “real” number of U.S. cases so far:

• “more than 100,000” on May 22;
• “more than a million” on June 26;
• “more than two million” on August 18;
• “millions of people” on September 25;
• “millions and millions” on October 20;
• “many millions of people” on October 29;
• “many, many millions of cases” on October 30;
• and, finally, “22 million people” on November 12.

Although the CDC didn’t stress the point nearly as much, its tally of pandemic deaths was also lower than the real number, mostly because it relied on a strict definition of what constituted a death from pandemic flu – a definition that ruled out many deaths from pneumonia, organ failure, etc., brought on or exacerbated by influenza. By contrast, the CDC’s estimate of 36,000 average seasonal flu deaths includes these “flu-related” deaths. It seemed sensible to use the same criteria for both definitions.

Not that the CDC wanted to facilitate comparisons between pandemic deaths and seasonal flu deaths. It didn’t. It really didn’t. The November 12 press briefing that unveiled the new swine flu statistics never mentioned the 36,000 figure for seasonal flu – a notable omission, considering how often officials flog that figure when trying to persuade people to get the seasonal flu vaccine.

Although the 36,000 figure never came up at the November 12 briefing, the CDC’s Anne Schuchat did mention a much less commonly used number: the 3,600 people under 65 who die in an average flu season. Her point was that swine flu is tougher on the young than the seasonal flu:

If you think about, that only about 10% of the seasonal flu deaths occur in people under 65, that would be about 3600 deaths, we do think we are having a pretty severe amount of deaths so far from the h1n1 virus.

The fact that swine flu is somewhat more deadly to the young than the seasonal flu was featured often in the news briefing. The fact that it is much less deadly overall than the seasonal flu wasn’t mentioned.

But the November 10 New York Times “scoop” by Donald G. McNeil did use the 36,000 figure. Entitled “Recalculating the Tally in Swine Flu Deaths,” the article included this paragraph:

The new estimate will be a more accurate comparison to the 36,000 deaths from seasonal flu each year, [CDC spokesman Glen Nowak] said. That estimate is also based on confirmed cases as well as hospital reports of people who appear to have died after a bout of flu. Over 90 percent of seasonal flu victims are over 65, and many are bedridden or in nursing homes or have serious medical problems like cancer or heart disease that the flu worsens.

McNeil’s story stops short of directly comparing 36,000 annual deaths from the seasonal flu with 3,900 swine flu deaths in six months. The story does include both numbers (though it uses “about 4,000” instead of 3,900). It tells readers that the new CDC swine flu numbers are a lot higher than the old ones. It tells readers that the change isn’t a change in the pandemic, but in how pandemic cases and deaths are counted. It tells readers that the change will facilitate comparison to the seasonal flu. It tells readers the most important seasonal flu number for comparison. And then it doesn’t make the comparison.

Still, McNeil’s story comes closer than most. Like the CDC, most reporters didn’t focus on the new evidence that so far the pandemic was milder than the seasonal flu. They found other, more alarming aspects of the November 12 numbers to emphasize instead.

The Forbidden Word: “Mild”

Never once to my knowledge has anyone at the CDC pointed out publicly that so far this pandemic looks less deadly than the seasonal flu. Never once.

Of course CDC officials couldn’t have pointed that out back in May, when the limited early evidence from Mexico suggested that the pandemic was way more deadly than the seasonal flu. But by late July the tally of confirmed deaths already looked awfully low relative to the guesstimated number of cases – certainly lower than the analogous comparison for the seasonal flu – and still the CDC avoided saying so. It still isn’t saying so.

I understand at least some of officials’ reasons for hesitating to say so:

• The numbers are so squishy. Prior to the November 12 estimates, they were really squishy. Officials didn’t want to do math with seat-of-the-pants guesstimates – a textbook case of “garbage in, garbage out.” The numbers are presumably less squishy now, but still pretty squishy.
• The pandemic is a moving target. Even an accurate measure of the pandemic CFR so far wouldn’t constitute a firm prediction of the pandemic CFR in the months to come – but most reporters would undoubtedly interpret it as one. And many politicians would perceive it, or pretend to perceive it, as a promise – especially later if things get worse. Why give out a number if the number’s going to keep changing, and if people are going to misperceive and reify it?
• As we have seen, the pandemic isn’t actually milder than the seasonal flu for people under 65. The only reason it turns out milder than the seasonal flu overall is because it is enormously milder than the seasonal flu for the elderly, who are the principal seasonal flu victims. Children and young adults, by contrast, rarely die from the seasonal flu and somewhat less rarely die from the pandemic flu. Wouldn’t trumpeting the overall comparison, accurate though it is, mislead the majority under 65 into thinking they, too, were less at risk from the pandemic virus than from the seasonal flu?
• The CDC’s own vulnerability is high. If officials say anything now about how mild the pandemic is so far, they’re likely to be accused of trying to over-reassure the public … or of trying to wriggle out from accountability for having hyped the pandemic ever since last spring … or of trying to reduce people’s outrage at the inadequacy of the vaccine supply. They’re also likely to undermine preparedness efforts, including the vaccination program itself (when more vaccine becomes available).
• The worst-case scenario is still on the table. Although there are few if any signs of it yet, the pandemic H1N1 virus could still mutate or reassort into a more severe strain. If health officials state outright that the pandemic is “mild so far,” how many of us would lose track of the “so far”? And then if things turned horrific in the months ahead, how many of us would come to feel that we had been over-reassured and badly misled? Isn’t it better to let people overreact to the mild pandemic we have so far than to leave them even less prepared for the severe pandemic we might still end up with?

And as I have noted already, it’s almost impossible to explain the pandemic-versus-seasonal differences without first coming clean about the fact that the seasonal flu is a threat mostly to the elderly. That might undermine future seasonal vaccination campaigns. And consider how much damage such an admission could do to the credibility of public health agencies, which would be accountable for all those past seasonal vaccination campaigns that conveniently neglected to mention how seldom the flu kills people under 65.

If the CDC is going to explain that the pandemic is less dangerous so far than the seasonal flu, it will also have to explain that that’s not the case for people under 65. That will raise an obvious paradox: How can the pandemic be less dangerous than the seasonal flu overall when it is more dangerous than the seasonal flu for the bulk of the population, everybody under 65? To unravel the paradox, it will have to come out of the closet about the seasonal flu, admitting that in terms of deaths the seasonal flu is almost entirely a disease of the elderly.

Explaining all that wouldn’t just be embarrassing about past seasonal flu vaccine communications. It would also undermine the annual push for seasonal flu vaccination. The only way to avoid this chain of inferences is to shut up about the fact that the pandemic is turning out milder than the seasonal flu, and hope the media don’t out you.

I still think public health officials should come clean. Whether the media out them or not, if the pandemic stays mild the public will eventually realize it (many realize it already, even without the November 12 data), and will hold officials accountable for not having said so sooner. Not coming clean will damage official credibility far more than coming clean. But I do understand why coming clean will be very, very difficult.

Some officials will undoubtedly claim that they’ve been coming clean all along. After all, the CDC certainly acknowledges that 90% of seasonal flu cases are 65 and over. It’s right there on the CDC website for all to see … all who look. But when officials talk to the public, and especially when officials talk to the public about the importance of vaccination, “36,000 deaths a year” is a much more widely used statistic than “90% of them elderly.”

Even the CDC website reflects the bias. The remaining 10% of those 36,000 seasonal flu deaths, the ones who are under 65, add up to an average of 3,600 deaths a year. That’s the most relevant statistic for the 87.2% of the U.S. population that’s under 65. So search the CDC website for “36,000” and “influenza,” and then do it again for “3,600” and “influenza.” Most of the 36,000s refer to the annual average number of flu-related deaths; most of the 3600s do not refer to the annual average number of flu-related deaths in people under 65. They’re about something else entirely.

Is it possible that the CDC collected the new data about the severity of the pandemic, posted the data on its website, planned to discuss the data at that day’s press briefing … and never thought to use the data to calculate the pandemic’s case attack rate, case fatality rate, and population mortality rate so far, so it could compare those estimates to the analogous numbers for the seasonal flu?

I guess it’s possible. But it seems a lot likelier that the CDC did at least some of the calculations, realized that its data were telling it the pandemic is significantly milder than the seasonal flu so far … and decided not to say so – presumably for the reasons I have listed above, plus some others I haven’t thought of.

The November 12 CDC news briefing “explained” the new data in ways that look designed to emphasize the severity of the pandemic, not its surprising mildness.

For one thing, the CDC’s generally excellent briefer, Dr. Anne Schuchat, refreshingly used the word “pandemic” 22 times. In previous CDC briefings, the word has been used very sparingly; I have always assumed the reason was a fear at some level of government that “pandemic” might be an excessively scary word. On November 12, just when the data were showing that so far swine flu is much milder than the seasonal flu, Dr. Schuchat unleashed the P word.

A reporter responded to the new numbers by saying: “With these numbers seeming to triple literally overnight … it’s fair to say when they hear this, they [the public] will be quite alarmed. What is your message to the public, parents and adults, when they hear these new numbers?” This was a chance for Dr. Schuchat to say, no, actually these are very reassuring numbers. Here’s what she said instead, her whole answer to that question:

Influenza is serious.  Vaccination is the best effort to protect one’s self or family.  These efforts give a bigger picture of what’s going on and reaffirm the priority recommendations we’ve given.  We focus vaccination on younger people because they are disproportionately affected with the virus.  Things haven’t really changed from last week to this week; […] we’re finally able to update the public on how big a toll this virus is having so far.  Of course, this is just the first six months and I am expecting all these numbers unfortunately to continue to [rise].

Another reporter actually asked about the comparison to seasonal flu, specifically regarding children: “I’m wondering at this point are you able to say definitively that this [pandemic] is producing more serious illness in more children than seasonal strains….” I have quoted Dr. Schuchat’s answer already, but here it is again: “If you think about, that only about 10% of the seasonal flu deaths occur in people under 65, that would be about 3600 deaths, we do think we are having a pretty severe amount of deaths so far from the h1n1 virus.”

In response to another question asking for historical perspective – another chance to say the pandemic is milder than the seasonal flu so far – Dr. Schuchat answered as follows:

We’ve been tracking influenza for decades. We have some systems that go back to 1958. Others of our systems are newer. What we are seeing in 2009 is unprecedented. People know we haven’t had a pandemic since 1957. So it’s not surprising that what we’re seeing is unprecedented. To have very high rates of influenza-like illness in September and October is extremely unusual. So I think the report in the MMWR focuses in on that. When we look back year after year, we don’t see a fall [like] this.

It’s certainly true that it would be unprecedented to have as much seasonal flu in the fall as we have pandemic flu this fall. But why would Dr. Schuchat consider this fact more noteworthy (and newsworthy) than the fact that we’re experiencing a pandemic that is killing far fewer people so far than the seasonal flu usually kills?

While Dr. Schuchat chose not to compare the overall swine flu death toll so far to the average seasonal flu death toll, she concluded the November 12 briefing by comparing it to the 1918 pandemic death toll:

What does this look like compared to previous pandemics.  The estimates I’m giving you are the first six months.  This is April through the middle of October.  We have a long flu season ahead of us.  In typical seasonal flu we see disease from December to May, it’s only November.  So exactly what we will see as a full toll of illness from this pandemic is very difficult to say.  I can say, though, that what we’re seeing with this h1n1 virus is nowhere near the severity of the 1918 pandemic.  That caused much larger numbers six months in.  I think projecting out forward is difficult.  And we’re really keen to get vaccine given as rapidly as possible to those in greatest need so we can limit the full toll this pandemic takes.  So thank you, everybody. 

All things considered, it was courageous of the CDC to produce the November 12 numbers at all, and to feature them in its November 12 press briefing. Yes, the agency interpreted the new numbers chiefly as evidence that the pandemic is more pervasive and more deadly than previously reported, not as evidence that it is milder than the seasonal flu. But at least it produced the new numbers.

Perhaps in the weeks to come the CDC will find additional courage to interpret the numbers more candidly. Or perhaps state and local health officials will take up the task, even if the CDC does not. Or perhaps the media will give the numbers a second look. So far, I see no signs of any of these developments.

Notwithstanding all the reasons to avoid the word “mild,” something is stunningly wrong when the evidence says we’re facing a pandemic that is six times milder so far than the ordinary seasonal flu, and the government officials in charge of pandemic preparedness decline to tell us so.

“About Like the Seasonal Flu”

So have the media told us so? Very, very seldom – even after the CDC handed reporters the data on November 12.

I am not accusing the media of hyping the pandemic. The dominant media meme about pandemic severity has been that it’s “about like the seasonal flu.” Reporters keep hearing those words from local health officials, and keep repeating them. They seemed approximately true back in June, when there were no real data to go on. And they’re still approximately true – experientially – for the typical victim. Both kinds of flu usually lead to a miserable week in bed and then you’re better. But “about like the seasonal flu” misses three key facts about this pandemic:

“About like the seasonal flu” doesn’t just miss the first of these three key facts. It misses them all.

Both public health officials and journalists seem terrified of saying anything that suggests that so far the pandemic is a pussy cat, and equally terrified of warning people that it could become a raging tiger at any moment. They’re overstating how bad it is and understating how bad it could get.

When health officials and reporters say that the pandemic is turning out a lot like the seasonal flu, then, what do they think they mean?

• That it’ll probably kill around 36,000 people in a single year? That’s an awfully high guess, given the pandemic’s current behavior. Having killed 3,900 people in the first six months, it would have to kill another 32,100 (more than eight times as many) in the next six months to get to 36,000.
• That it’ll probably kill mostly the elderly? That would be a huge surprise.
• That it’s nearly guaranteed not to mutate suddenly into an enormously more dangerous form? That would be nice, if only it were so.
• That it’ll give most of its victims a really rotten feverish week and then they’ll be fine? That’s solid, at least so far.

More importantly, when officials and reporters say the pandemic is turning out a lot like the seasonal flu, what does their audience think they mean? I don’t know what percentage of the U.S. population actually realizes that the seasonal flu currently kills about 36,000 Americans a year, 90% of them elderly. I suspect most Americans would guess a far lower mortality figure and a far more balanced age distribution.

Maybe all the audience is getting out of the comparison is the part that’s actually so: Most pandemic patients, like most seasonal flu patients, feel awful for a week and then recover.

The more you know about the seasonal flu, the more misled you would be if you imagined that the swine flu pandemic resembles it.

Partly because most seasonal flu deaths are elderly, most seasonal flu deaths are virtually invisible. Of course every death is a tragedy for the loved ones left behind, but seasonal flu is the Rodney Dangerfield of diseases, unable to command the public respect that 36,000 annual deaths ought to command. So when a local official says the swine flu pandemic is similar to the seasonal flu, the official may mean “Wow, expect it to kill around 36,000 people!” But the public may hear: “No biggie. It’s just like a normal year.”

Perhaps that’s why so many comparisons to the seasonal flu say that the pandemic will probably be roughly as severe as a typical flu year, and then go on to explain why it should be considered worse than a typical flu year.

On August 7, 2009, for example, the President’s Council of Advisors on Science and Technology (PCAST) stated: “The case-fatality ratio (i.e., proportion of infected individuals who die as a result of the infection) appears to be similar to seasonal influenza … perhaps 0.05 to 0.2 percent of all symptomatic cases.” PCAST went on to say that the pandemic would probably kill more people than the seasonal flu (because it would probably infect more people), and offered a “plausible scenario” for “planning purposes” of 30,000-90,000 deaths.

It’s the 90,000 number that got most of the media coverage – not PCAST’s mistaken guess that the pandemic would have about the same CFR as the seasonal flu. Neither the reporters who were alarmed by the 90,000-deaths scenario nor the commentators who berated that scenario as alarmist noted that it was far from a worst case scenario.

Though the 90,000-deaths figure is way lower than any real worst case scenario, it is also way higher than the pandemic death toll will be if things stay the way they are. In fact, so is the low-end number in PCAST’s planning scenario. If things stay the way they are, the swine flu pandemic won’t kill 30,000 Americans during its first year either.

As far as I can find, the CDC itself has never come right out and said that the swine flu pandemic has or will have a case attack rate, case fatality rate, or population mortality rate similar to those parameters for the seasonal flu. But CDC officials have seemed comfortable with hundreds of other statements by health agencies, experts, and journalists equating pandemic H1N1 with the seasonal flu.

This comparison grossly overstates the current seriousness of the pandemic overall. It slightly understates the pandemic’s current seriousness for people under 65. And it irresponsibly ignores the possibility of a major change: a major change for the worse if the pandemic virus becomes more virulent, or a major change for the better if the pandemic remains mild and supplants the seasonal strains.

The comparison is thoroughly misleading to people who have a decent sense of what the seasonal flu is like. It’s very difficult to guess what it might mean to people whose impression of the seasonal flu is wrong to start with.

From time to time, even before November 12, some flu expert has expressed the view that that the swine flu pandemic might actually be milder than the seasonal flu – and managed to get a little media coverage for that view. On September 16, 2009, for example, Reuters ran a story about Marc Lipsitch of Harvard University, citing Dr. Lipsitch’s estimate that swine flu’s case fatality rate so far was somewhere between 0.007% and 0.045% – which nicely matches the 0.018% figure based on CDC estimates two months later. Even though the Reuters story explicitly compared Lipsitch’s much lower estimate to the conventional seasonal flu case fatality rate average of 0.1%, ABC News ran the story under the headline “Swine Flu Death Rate Similar to Seasonal Flu: Expert.”

It didn’t make much of an impression, and certainly didn’t alter the media meme that swine flu is about like the seasonal flu.

Off-Base Coverage of the November 12 Data

It is hard to blame the media for mishandling the pandemic-versus-seasonal story prior to November 12. Without the CDC to lead the way (and do the arithmetic), journalists weren’t about to calculate their own tentative CARs, CFRs, and PMRs based on the CDC’s case number guesstimates and its laboratory-confirmed death counts. Instead, they endlessly reported that the pandemic was about as severe as a typical flu season.

Then came November 12. At last the CDC had provided surveillance and monitoring data as a basis for estimating the severity of the pandemic, and for comparing it to the severity of the seasonal flu. But the CDC didn’t actually do the CAR, CFR, and PMR calculations, nor did it make the comparisons. Did the media?

No.

I have looked at virtually all the coverage in major national media of the CDC’s November 12 recalculation of the first six months of swine flu cases and deaths. Here are my overall conclusions:

• Nobody missed that this was just a recalculation, not a change. The CDC emphasized that point in its November 12 press briefing, and reporters got it right. Every story was at pains to explain that the pandemic hadn’t suddenly gotten worse.
• Nobody said the pandemic could in fact get worse. The possibility of an increase in virulence wasn’t mentioned at the November 12 briefing, so it wasn’t mentioned in the news coming out of that briefing. Reporters who cover the pandemic story routinely presumably know about that possibility, but they’re busy covering what has happened, not what could happen.
• Nobody used the CDC’s new numbers to calculate a case attack rate, a case fatality rate, or a population mortality rate. In fact, nobody calculated anything. Reporters cited the numbers (some of the numbers) they were given, period. Nobody did any arithmetic.
• Nobody said the new numbers were reassuring. The new numbers were higher than the previous tallies of laboratory-confirmed cases and deaths – so they had to be seen as alarming. The main thrust of virtually every story was: “This doesn’t mean things are worse than they were. But it does mean things are worse than the previous numbers suggested.” I counted alarming, neutral, and reassuring paragraphs in some stories to make sure my impression was accurate. Most paragraphs were neutral. But there were a lot more alarming than reassuring paragraphs.
• Direct comparisons to the seasonal flu were rare. Some stories did include the 36,000 figure for annual seasonal flu deaths, and some mentioned that seasonal flu kills mostly the elderly while the pandemic is killing mostly younger people. But very few stories tried to assess which was more serious – overall or for specific age groups. No story said the seasonal flu is much more deadly to the elderly than the pandemic is so far to any age group.
• One story (a blog, actually, by Jacob Goldstein on the Wall Street Journal website) explicitly said the new numbers might be seen as suggesting that the pandemic is milder than the seasonal flu. The story went on to detail the reasons why the author thought that conclusion would not be justified.
• Most stories implied that the pandemic is worse than the seasonal flu. The facts and statistics most frequently included in the coverage made this inference inevitable. Story after story, for example, stressed that the number of pandemic cases and deaths is “unprecedented so early in the season” – implying that the pandemic would probably get worse as the “season” progressed. Also much-reported was the estimate of pandemic pediatric deaths, and the fact that this number greatly exceeds both the previous count of laboratory-confirmed pandemic pediatric deaths and the count of laboratory-confirmed seasonal flu pediatric deaths.
• Many stories pointed out that the regular flu season hasn’t gotten started yet. Nobody wondered whether there would be a seasonal flu this year. And nobody noted that if the pandemic virus supplanted the seasonal flu strains and stayed mild, the pandemic would actually save lives – albeit the lives of the elderly. (Nobody raised the question of how many elderly deaths equals one pediatric death, though a few stories came close to implying that the seasonal flu mostly kills people who are on their deathbeds anyway.)
• Most stories dwelled on the new numbers fairly briefly, and then segued to the vaccine shortage and other pandemic issues that had been raised at the CDC’s press briefing.

Why have reporters failed to use the CDC’s November 12 numbers to show that the pandemic – based on the CDC’s estimates – is indisputably milder so far than the average seasonal flu? I don’t know. But three hypotheses seem likely to me:

Contact information page:    Peter M. Sandman