Search Peter’s website:

Science versus Spin:
How Ron Fouchier and Other Scientists
Miscommunicated about the
Bioengineered Bird Flu Controversy

This article is excerpted with minor changes (and a little reorganizing) from a Guestbook entry I posted on May 17, 2012. Some readers have suggested it raises important issues and would be easier to find as a stand-alone article.

The controversy that this article addresses arose after a little-known body, the U.S. National Science Advisory Board for Biosecurity (NSABB), recommended to the U.S. government that it should recommend to the editors of two scientific journals (Science and Nature) that they edit out some methodological details before publishing two research papers they had already accepted for publication.

The two papers described the first successful efforts to bioengineer the H5N1 “bird flu” virus to transmit through the air (via aerosol or droplets) between mammals. (The mammal used in both studies was the ferret.) The findings of these two studies are thought to shed light on how a potentially catastrophic H5N1 pandemic might someday be launched among humans – whether by nature, by accident, or by intention.

The NSABB was worried that unless the papers were redacted their value to terrorists, disaffected lab workers, and other bad guys might outweigh their value to influenza scientists and pandemic preparedness professionals. This balancing with regard to “dual-use research of concern” (DURC) was very much the sort of task for which the NSABB had been created.

The controversy is pretty much over for now. The authors of the two papers made some revisions; the NSABB met again and decided (under considerable pressure) that the two revised papers should be published after all; as of early June 2012 one is already out and the other is expected shortly.

But of course the underlying issues aren’t settled:

• What sorts of research, if any, shouldn’t be done at all because it’s too dangerous?
• When very dangerous research is done, what precautions should be taken to reduce the risk of accidental or intentional releases?
• After the results are in, should the papers be freely published so everyone, even bad guys, can absorb the lessons to be learned?
• Above all, who should make these decisions? Just scientists? Is there any appropriate role for governments? For the public?

Outrage on Both Sides

The NSABB was never authorized to censor the two papers, nor was the U.S. government. The journal editors have always been legally free to publish what they want, regardless of what the NSABB and the U.S. government recommended.

And the authors of the two papers have always been legally free to self-publish on the Web (at least insofar as the U.S. government is concerned), regardless of what the journal editors decided. But the career implications of publishing a paper after the U.S. government has successfully persuaded a major journal to withdraw its acceptance of that paper would be, to say the least, daunting. Whether or not that qualifies technically as “censorship,” it can certainly feel like more pressure than is consistent with scientific autonomy.

Was the NSABB recommendation a dangerous precedent or a one-off? Is the entire edifice of scientific publication threatened by a single instance? The problem with slippery slope arguments is that it’s very hard to judge in advance which slopes are actually slippery and which are not. To me, as a non-scientist who has worked with government agencies on influenza communication for more than a decade, H5N1 looks uniquely threatening. So when a government advisory body that has never before recommended against publishing a scientific paper decided that two H5N1 papers should be redacted before they’re published, that looked to me like a one-off. But it’s obviously not a one-off for a scientist who writes many papers about H5N1! And anyone who sees H5N1 as just one among many dangerous pathogens would have good reason to see the NSABB’s recommendations (the first set) as a precedent and a foot-in-the-door, not as a unique response to a unique threat.

In my own language: It’s debatable whether this controversy has posed a big or small hazard to the autonomy of science. Either way, the outrage among many scientists was (and remains) substantial.

The outrage on the other side – among those who wanted the papers redacted and the work discontinued – also was (and remains) substantial. Two hazards have preoccupied the critics of this research and its publication: the possibility that bad guys might be inspired and guided by the two papers to create or steal a bioengineered H5N1 virus and launch their own pandemic; and the possibility that expanded research in H5N1 biotechnology might lead to a laboratory accident that launched a pandemic. I am not qualified to assess the size of these two hazards. Maybe they’re small. But people who are outraged about them obviously think they are substantial.

When scientists who support publication come across as cavalier, patronizing, contemptuous, or disingenuous about these two hazards, that increases the outrage of those (scientists or laypeople) who are worried about them. And the increased outrage leads to increased hazard perception. Sooner or later, the increased outrage might also lead to increased interference with scientific autonomy. That’s why I found myself worrying that scientific arrogance could threaten science more than the NSABB did.

But exactly the same dynamic works in the other direction. Scientists whose concerns about publication censorship were dismissed or ridiculed naturally got more outraged as a result, and their increased outrage increased their conviction that scientific integrity was under siege. Similarly, casual references to “arrogance” and even to “mad scientists” and undocumented assertions about biosafety lapses exacerbated the outrage of publication proponents.

Both sides in the publication debate have communicated in ways that exacerbated the other side’s outrage. And both sides have found it difficult to recognize and change how they were communicating because of their own outrage.

Using Evidence as Ammunition

Two clusters of technical questions underlie this controversy.

The first cluster focuses on the danger posed by H5N1. How likely is it to acquire the ability to transmit easily from human to human? If it did acquire that ability, how deadly would it be likely to be to the humans who caught it? How contagious would it be? Would it probably burn itself out or would it probably go pandemic? How long would it take us to mass-produce a vaccine against it? How effective would the vaccine be in various subgroups? How effective would antivirals be in the meantime? How well would supply chains and social institutions stand up to the stress?

The second cluster focuses on the impact of the two papers. Would they make a devastating H5N1 pandemic less likely (by helping guide surveillance, for example)? Or would they make a devastating H5N1 pandemic more likely (presumably by cluing in bad guys or worsening the odds of a lab accident)?

Of course the second cluster doesn’t matter much unless you lean toward a worrisome answer to the first cluster. The impact of the papers is an important question only if H5N1 is a scary virus. That didn’t keep some proponents of publication from asserting two incompatible propositions: that H5N1 isn’t an especially dangerous virus so we don’t have to worry much about lab accidents or bad guys in connection with the two papers; and that it’s crucial to get the two papers published because they can help avert a potentially horrific H5N1 pandemic.

One of my main critiques of both sides in the controversy, in fact, was their tendency to use any argument they could find to support their position, seemingly regardless of whether that argument was compatible with their other arguments – and at least sometimes regardless of whether they actually believed it.

Consider for example these three narrow technical questions:

• How useful are ferrets as an animal model for predicting human flu transmission?
• How many mild or asymptomatic cases of H5N1 have there been that never got diagnosed?
• How likely is it that the subtype of H1N1 that disappeared in the 1950s and then reappeared in 1977 (and circulated until 2009) was released from a laboratory that had samples from the 1950s?

These three questions have literally nothing to do with each other except this: They were all usable as ammunition for or against publication of the two papers.

If you were pro-publication, it would help your case to assert:

• that ferrets are an unreliable model (so efficient ferret-to-ferret transmission in the lab doesn’t necessarily mean the strain could launch a human pandemic);
• that there have been lots of undetected cases of H5N1 (so the disease is far less lethal than the World Health Organization’s 59% figure implies); and
• that the 1977 strain didn’t result from a lab accident (so there’s no precedent of a lab-related global influenza outbreak).

If you were anti-publication, on the other hand, it would help your case to assert:

• that ferrets are a good model (so we have probably created a pandemic flu virus in the lab);
• that there haven’t been very many undetected human H5N1 cases (so the disease is unprecedentedly lethal); and
• that the 1977 strain probably came from a Russian or Chinese lab (so we’ve seen at least one lab accident before that led to a global flu outbreak).

What bothers me is how seldom I have run across an expert with mixed opinions, an expert whose position on all three technical questions (and plenty of others) wasn’t predictable based on his or her position in the publication debate.

I can’t find many experts who said “Even though I agree that ferrets are a good animal model, here’s why I still support publication…” or “Even though I agree that ferrets are an unreliable animal model, here’s why I still oppose publication….” I can’t find many experts who said “You’re right that there haven’t been many mild cases, but I’m still pro-publication…” or “You’re right that 59% is way too high, but I’m still anti-publication….” I can’t find many experts who said “Our lab safety record is terrific, but I have to admit H1N1 in 1977 was probably a lab accident” or “We have a serious lab safety problem, but the 1977 H1N1 outbreak probably didn’t come from a lab accident….”

On issue after issue, I saw scientists choosing up sides and then marshaling their evidence. That’s how lawyers assess evidence: as ammunition they embrace or disdain depending on which side they’re on. It’s not supposed to be how scientists assess evidence. Scientists who use evidence to prove their hypotheses rather than to test them are being deceptive. If they don’t know it, then they’re being self-deceptive as well.

And when scientists communicate, they’re expected to bend over backwards to be fair. Even if all the facts deployed to advance a case are accurate, scientists aren’t supposed to leave out equally accurate facts that might lead the audience to question their conclusion. Instead of cherry-picking facts, scientists pride themselves on acknowledging the flaws in their case and the sound arguments of their adversaries.

Cherry-picking facts isn’t just bad science. It is also bad risk communication. It exacerbates mistrust and increases the outrage of opponents.

Ron Fouchier as a Risk Communicator

A good way to illustrate this evidence-as-ammunition misuse of science is to examine the public communications of Ron Fouchier, a scientist at Erasmus Medical Center in Rotterdam and the senior author of one of the two papers.

For several months in late 2011 and early 2012, Fouchier appeared to be trying to arouse interest in his study. His messaging was all about how dangerous he considered the H5N1 virus and how terrifying (but incredibly useful) he considered his own soon-to-be-published study. But as the controversy over publication grew, Fouchier became less focused on arousing interest and more focused on allaying concern. And his messaging altered to match his new goal.

I’m inferring the goals, of course. But the messaging is on the record.

The change in Fouchier’s public messaging can be dated precisely. It came on February 29, 2012, when he participated in a panel discussion sponsored by the American Society for Microbiology. Actually, the change probably dates back a little earlier, when Fouchier spoke at a February 16–17 World Health Organization meeting in Geneva. But the WHO meeting was confidential, whereas the ASM panel was (and is) on the Web.

Let’s track Fouchier’s risk communication about his own study before, during, and after his ASM presentation.

September 2011

We’ll start with September 12, 2011, when Fouchier presented his research at the Malta meeting of the European Scientific Working Group on Influenza (ESWI). There is no transcript or video of the presentation, but three science journalists covered it.

Here’s how The Influenza Times, the conference newspaper, reported Fouchier’s key result the next day:

“This virus is airborne and as efficiently transmitted as the seasonal virus,” said Fouchier. His research team found that only 5 mutations, 3 by reverse genetics and 2 by repeated transmission, were enough to produce this result. “This is very bad news, indeed,” said Fouchier.

Katherine Harmon’s story in the September 19 Scientific American paraphrased Fouchier on the key result:

It wasn't until “someone finally convinced me to do something really, really stupid,” Fouchier said, that they observed the deadly H5N1 become a viable aerosol virus. In the derided experiment, they let the virus itself evolve to gain that killer capacity. To do that, they put the mutated virus in the nose of one ferret; after that ferret got sick, they put infected material from the first ferret into the nose of a second. After repeating this 10 times, H5N1 became as easily transmissible as the seasonal flu.

The third account of Fouchier’s Malta presentation was Debora MacKenzie’s article in the September 26 issue of New Scientist. MacKenzie also reported that Fouchier said the new H5N1 strain transmitted easily in ferrets. She quoted him directly: “‘The virus is transmitted as efficiently as seasonal flu,’ says Ron Fouchier.” But MacKenzie also reported something that wasn’t in the other two stories: that Fouchier had said the virus was deadly to ferrets when transmitted through the air. Here’s what she wrote:

Then the researchers gave the virus from the sick ferrets to more ferrets – a standard technique for making pathogens adapt to an animal. They repeated this 10 times, using stringent containment. The tenth round of ferrets shed an H5N1 strain that spread to ferrets in separate cages – and killed them.

These two claims – that the new strain transmitted through the air as easily as seasonal flu and that it killed ferrets when thus transmitted – were undisputed for several months, until Fouchier himself disputed them in February 2012. Back in September 2011, neither Fouchier nor anyone who heard him speak at the Malta conference challenged the accounts of his presentation in The Influenza Times, Scientific American, and New Scientist.

Nobody affirmed the three accounts either. But that’s hardly surprising. Reading an erroneous news story about a presentation you heard might prompt you to post a comment. You’re not likely to post one pointing out, “Yes, that’s what I heard too.”

November 2011

In October, the U.S. government asked the NSABB to consider whether Fouchier’s paper and one other should be suppressed or redacted. Although the NSABB’s recommendation to redact wasn’t announced until December, the issue was heating up well before then.

On November 22, Science sent reporter Martin Enserink to Rotterdam to interview Fouchier for a story in its online website, “ScienceInsider.” Enserink’s November 23 story called Fouchier’s bioengineered strain “a man-made flu virus that could change world history if it were ever set free.” It went on:

In a 17th floor office in the same building, virologist Ron Fouchier of Erasmus Medical Center calmly explains why his team created what he says is “probably one of the most dangerous viruses you can make” – and why he wants to publish a paper describing how they did it. Fouchier is also bracing for a media storm. After he talked to ScienceInsider yesterday, he had an appointment with an institutional press officer to chart a communication strategy.

In Enserink’s story, Fouchier was eloquent about the importance of his study in revealing that a catastrophic H5N1 pandemic is possible. He said nothing that suggested a desire to back off the earlier reports that his virus was easily transmissible and deadly, that it could kill a ferret when a nearby ferret coughed.

Could all four science journalists have misheard and/or misquoted Fouchier? I suppose it’s conceivable. Then what about his own institution? Between November 27 and November 29, four separate documents were posted on the Erasmus Medical Center website: a news release and a Q&A, each of them in both Dutch and English. The timing suggests these documents were probably outputs of Fouchier’s November 22 meeting with his press person.

The news release began (after a boldface summary) with this sentence: “Of the 600 people who have to date been infected with the H5N1 virus worldwide, 60 per cent have died.” Nothing later in the release pointed out that this figure is based on confirmed cases only, and omits from its denominator mild or asymptomatic cases that never got diagnosed.

Fouchier is one of many flu scientists who sometimes criticize others for using the 59% (or 60%) figure without qualifiers and sometimes use it without qualifiers themselves … depending on whether they’re trying to tamp down H5N1 concern or trying to arouse it. The World Health Organization, which has been trying to arouse H5N1 concern at least since 2004, is the source of the 59%/60% figure and repeatedly uses it without qualifiers.

The Erasmus Medical Center release also reiterated Fouchier’s claim that his virus transmits easily in ferrets and implied – without quite saying – that it transmits easily in humans as well:

Scientists worldwide have been concerned with the question whether the [H5N1] virus could change into a virus that can spread among humans. “We have discovered that this is indeed possible, and more easily than previously thought,” says Ron Fouchier, researcher at Erasmus MC. “In the laboratory, it was possible to change H5N1 into an aerosol transmissible virus that can easily be rapidly spread through the air. This process could also take place in a natural setting.”

The accompanying Q&A offered an even clearer version of the news release’s extraordinary implication (untested, thankfully) that Fouchier’s virus can spread among humans. The very first sentence read: “Erasmus MC researchers have discovered that the avian influenza virus spreads more easily among humans than previously thought.”

My wife and colleague Jody Lanard and I originally thought that this might be an error – that Fouchier’s press person might have carelessly elided from the fact that transmission among ferrets raises concern about possible human transmission to the implication that transmission among humans has been proved. So in January Jody emailed the Erasmus website, the Erasmus Medical Center press office, and finally Fouchier himself, suggesting that this sentence should be changed. She didn’t get an answer, and it hasn’t been changed.

We also wondered if “spreads more easily among humans than previously thought” might have been a Dutch-to-English translation error. So Jody looked at the Dutch version of the FAQ, which reads: “Onderzoekers van het Erasmus MC hebben ontdekt dat het vogelgriepvirus zich gemakkelijker onder mensen kan gaan verspreiden dan tot nu toe gedacht.”

“Onder mensen kan gaan verspreiden” means “can be spread among humans.” It was not a translation error.

Neither the November 28 news release nor the accompanying Q&A said anything about whether the ferrets on the receiving end of Fouchier’s aerosol transmission experiment died. Two months earlier, Debora MacKenzie had written in New Scientist that the ferrets died. That had become a widespread impression about Fouchier’s research. Here was a perfect opportunity to say clearly whether or not aerosol transmission of Fouchier’s mutated H5N1 was deadly to ferrets. Fouchier and the Erasmus Medical Center press office did not take that opportunity.

January 2012

Flash forward two months to January 20, 2012, when Jeffrey Kofman’s story on the controversy was posted on the ABC News website. (The story was scheduled to air on “World News with Diane Sawyer” on January 20, but it got preempted by other news and didn’t actually air until February 20.) Datelined Rotterdam and headlined, “Researchers Pause Work on Bird Flu That Could Kill Hundreds of Millions,” the story contained these two paragraphs:

ABC News was given an exclusive inside look at some of the testing facilities the Rotterdam researchers used. With Fouchier as our guide, we donned protective clothing and face masks and passed through three levels of security to see the ferrets he uses for testing.

Fouchier explained how his lab assistants exposed the ferrets to the altered virus and placed unexposed ferrets in cages nearby. All 40 ferrets died.

As in the case of Debora MacKenzie’s September 2011 story, neither Fouchier nor anyone else from Erasmus Medical Center disputed the ABC News report that Fouchier killed ferrets via aerosol transmission. As of May 16, 2012, the story is still on the ABC News website. It has 23 comments, all from January 20-23. Most of them focus on the pros and cons of the research; none of them challenges Kofman’s reporting.

Also on January 20, Science reporter Martin Enserink interviewed Fouchier about a 60-day research moratorium declared that day by Fouchier and 38 other flu researchers. Fouchier told Enserink that he was in touch “on a daily basis” with Yoshihiro Kawaoka of the University of Wisconsin, the senior author of the other article the NSABB had recommended redacting.

Perhaps those daily conversations had something to do with Kawaoka’s decision to speak publicly about his own research for the first time. In a January 25 article in Nature, Kawaoka emphasized that his virus was not deadly to ferrets:

Our results also show that not all transmissible H5 HA-possessing viruses are lethal. In ferrets, our mutant H5 HA/2009 virus was no more pathogenic than the pandemic 2009 virus – it did not kill any of the infected animals. And, importantly, current vaccines and antiviral compounds are effective against it.

In its coverage of Kawaoka’s article, Science reviewed the history of the controversy, contrasting Kawaoka’s research to Fouchier’s as follows:

Fouchier, who has discussed his work at scientific meetings and with the media, concocted a transmissible H5N1 in ferrets by both manipulating viral genes and repeatedly passaging the virus through the animals to help it adapt to that host. This virus was highly lethal.

That paragraph didn’t directly state that Fouchier’s virus was lethal to ferrets via ferret-to-ferret respiratory transmission. But like virtually all coverage of Fouchier’s studies until a month later, the reporter, Jon Cohen, had that impression and repeatedly conveyed that impression during his months of covering the controversy.

February 2012

On February 6, New Scientist ran Debora MacKenzie’s summary of the controversy, in which she repeated the no-longer-new “facts” that Fouchier had created a lethal virus that could “spread through the air like ordinary flu, while staying just as lethal.” Again, neither Fouchier nor anyone else disputed these claims.

Only people who were present know exactly what Fouchier said at the February 16-17 World Health Organization Geneva meeting. But in the days that followed, rumors started circulating among flu cognoscenti that Fouchier had given a presentation that significantly altered participants’ understanding of his work. Two weeks later, one person who attended the WHO meeting, the NIH’s Tony Fauci, said that Fouchier’s and Kawaoka’s presentations had provided the WHO participants with “new data on two manuscripts” and “substantially clarified … original data in one manuscript.”

“Substantial clarification” is one way to interpret what Fouchier did when he participated in the February 29 ASM panel. He said his bioengineered H5N1 strain spread among ferrets but not easily. And he said most of the ferrets that caught the virus via aerosol transmission barely got sick, and none of them died.

In his presentation, Fouchier showed charts indicating that the 2009 swine flu (H1N1) virus spreads much more efficiently in ferrets than his mutant H5N1 strain. He referred to “misperceptions” in the media that the mutated virus “would spread like wildfire,” stating that in fact the efficiency of spread “cannot be deduced from our experiments.”

“To then extrapolate that this virus would spread like wildfire in humans,” he said, “is really, really far-fetched at this stage.”

As for lethality, he said:

The second misconception is that the virus would be highly lethal if this would ever come out [of the lab]. But also here there is some facts to explain…. Now the [lab-mutated] virus that we have used does cause disease when we put it in the nose at very high titers…. But if we now look after aerosol transmission, we actually see no disease, no severe disease at all, in any of the seven animals that received virus by aerosol.

Fouchier summarized both points unambiguously in the Q&A:

These [lab-mutated] viruses do not kill ferrets if they are sneezed upon…. If anything, our data suggest that this virus spreads poorly.

At no point in the ASM panel (and at no point since then) did Fouchier indicate that what he was saying now represented any kind of change from what he had been saying all along – at Malta, in media interviews, or in his original paper.

A few days later, Mike Coston summarized Fouchier’s about-face spectacularly on his “Avian Flu Diary” blog. “If after watching the ASM video,” he wrote, “and reading this report, you aren’t thoroughly confused, you obviously aren’t paying close enough attention.”

Fouchier also went out of his way at the ASM meeting to counter his previous alarmist take on the risk of an H5N1 pandemic. No more off-the-cuff remarks about H5N1 being “probably one of the most dangerous viruses you can make,” as he had told Martin Enserink in November. Here’s what he told his ASM audience:

It’s also important to note … that when ferrets are pre-exposed to seasonal flu, they are fully protected from developing severe disease [after exposure to his H5N1 strain]. So if we compare that to humans and you all have been infected previously with seasonal flu, it would be unlikely that you would have no cross-protection against a virus like H5N1. And so very few individuals would actually develop severe disease but most of them would be protected by cross-protective immunity.

Tony Fauci of NIH was on the ASM panel with Fouchier, and used the occasion to announce that NIH had asked the NSABB to reconvene and reconsider its recommendations. Pushed to explain why, he referred to “old data that’s clarified and new data that’s juxtaposed with the old data” – which sounds to me like code for Fouchier’s new messaging.

Fauci and other NIH officials have denied they asked the NSABB to reconsider, emphasizing that they wanted it to look at revised manuscripts, not to reassess the original papers. Although this comes across to most outsiders as a disingenuous distinction, there is no way to actually judge the matter, since outsiders will never be permitted to compare the original and the revised papers.

March–April 2012

On March 2, “ScienceInsider” ran a story by Jon Cohen and David Malakoff entitled “NSABB Members React to Request for Second Look at H5N1 Flu Studies.” It reported reactions to the ASM panel from seven (out of 22) NSABB members.

Nearly all the commenters stressed that their concern about publishing Fouchier’s paper was unabated by Fouchier’s ASM presentation. Irrespective of new or clarified data about lethality, Fouchier’s study extended the host range (to ferrets) and mode of transmission (via aerosol) of a dangerous pathogen – reason enough to think hard before publishing his methods.

But one NSABB member, Michael Imperiale of the Department of Microbiology and Immunology at the University of Michigan Medical School, went further, commenting: “What Ron [Fouchier] is saying now is not what was in the paper. We were led to believe by the paper that aerosol transmission is also lethal.”

This is the only testimony we have from someone who read the paper Science had initially accepted and got the same impression from that paper that MacKenzie got from Fouchier in Malta and Kofman got from Fouchier in Rotterdam: that aerosol transmission of Fouchier’s virus killed ferrets. We have no testimony from anyone who read the paper and got the opposite impression, the one Fouchier offered at the February 29 ASM panel. Reading the original paper could settle the matter. But outsiders will not be permitted to read the original paper.

From February 29 on, Fouchier took many opportunities to tell the world that his experiment wasn’t especially scary, that wild-type H5N1 might not be so scary either, and that any misunderstanding of his work was attributable to media misreporting.

Here, for example, is Fouchier in a March 26 “This Week in Virology” (TWiV) interview with Vincent Racaniello:

Contact information page:    Peter M. Sandman