Observer bias

"On reading," by Simon Wain-Hobson, is a weekly discussion of scientific papers and news articles around gain of function research in virology.

Since January 2024, Dr. Wain-Hobson has written weekly essays for Biosafety Now discussing risky research in virology. You can read his entire series here.

Sea urchins, winter, Côte d’Azur

On reading the Fouchier and Kawaoka GOF papers on bird flu H5N1 viruses.

---

It’s been mentioned in numerous essays that these two papers, along with similar experiments on bird flu viruses, couldn’t deliver on the promises made at that time, which was the prediction of the next pandemic strain with the corollary that preventive drugs and vaccines could be made.

One of the hidden features of this work is worth explaining for it is a major reason, albeit not the only one, why these experiments have no predictive power, so it’s worth spelling out. To do so, a few words about the experimental set up are needed.

A ferret - call it the donor animal - is injected with a H5N1 flu virus and put inside a cage. In an adjacent cage is an uninfected or receiver ferret. Between them are two perforated screens so that the animals can’t touch or spit at each other. Air flows from the infected ferret’s cage to that of the receiver. As ferrets are one of the few animals that sneeze when they have the flu droplets laden with virus will move to the cage of the uninfected animal. Click here to see a photo of this experimental setup.

For the moment H5N1 is not transmissible between humans. However, an engineered flu virus mutant might be capable of airborne transmission and so will infect the uninfected animal. Virus is taken from the receiver ferret and used to start another cycle. This can go on for 5 cycles, sometimes more, with each cycle being performed in triplicate and occasionally more.

Flu virus mutants can produce varying symptoms from mild weight loss to severe respiratory distress. If the researcher systematically takes virus from the receiver ferret with the severest symptoms, they will end up with a highly virulent transmissible virus.

Conversely, if the researcher systematically takes virus from the receiver ferret with the mildest symptoms they will end up with an asymptomatic transmissible virus.

ALSO, the astute lay science observer, has already spotted the plethora of intermediate scenarios.

Two questions. One, which scenario is the most likely to occur in nature? Two, which of the two results outlined above is the most likely to ensure publication in a top scientific journal? Let’s take the second question first. The genesis of an engineered transmissible flu virus that is highly virulent in the ferret model has a much greater chance of getting published. Why? Because microbiologists are focused on, and understandably fascinated by, highly pathogenic organisms. They are not interested in a wimpy virus because it is not attractive. How can you ‘sell’ the findings? Plus, funding agencies will not see their role fighting disease fulfilled. This was touched upon in Too Much.

Back to question one which is more important. No idea especially as there are numerous intermediate outcomes. You could conclude by saying yes, it is possible to generate a transmissible H5N1 flu virus between ferrets but in no way can the ferret experiment predict the next pandemic virus nor say anything about its severity.

The experimental set up reeks of observer bias, with a premium for publication going to highly pathogenic transmissible viruses.

To understand the situation fully consider a parallel example. Humans have been breeding dogs starting with wolves for some 10,000 years. The variety of shapes and forms never ceases to amaze. Would Mother Nature have come up with the dachshund? While a rhetorical question there is a credible answer. So long as there are marauding wolf packs and big cats on the planet, the dachshund doesn’t have a chance. In biology, there is a plethoric literature showing that you get what your experimental set up was designed for. Whether this will happen naturally is different question. The flu community doesn’t seem to be aware of this distinction.

These flu experiments never had predictive power. At best they showed that bird flu H5N1 viruses could mutate to become airborne transmissible, which wasn’t entirely unexpected, even though it has not happened in the 28 years since H5N1 infected a human being. Yet both studies were published in top scientific journals, created a stir while huge amounts of ink and angst were expended. Still are. Odd, no, for a qualitative result tainted by observer bias?

Yet we’re not done. What can we do with this information? Not much by virtue of being qualitative. Yet stocks of these viruses of unknown potential remain in freezers. That represents is a small yet finite risk. It is very different to having stocks of nasty natural viruses in a freezer. These are needed as part of society’s long-term goal of beating the hell out of them. Stocks of these unnatural viruses of unknown potential could, indeed should, be destroyed (Rinderpest).

Finally, as this information is on the cloud, what about the other side, the dual use part of DURC that no virologist talks about apart from the biosecurity people? They don’t seem to be heeded. It’s not that microbiologists don’t get biosecurity; they don’t want to know.

Now, this is unwise. On reading came across this recently: "Russia is endangering our European security not only with its illegal war of aggression against Ukraine, but also with severed cables, displaced border buoys, disinformation campaigns, GPS jammers and, as we have seen, dilapidated oil tankers," the German foreign minister said in a statement.

This isn’t going to stop. There are reports that they seem to be revamping their bioweapons capabilities. The DURC issue must be addressed by academia. Let’s hope some sense sinks in soon.

Conclusion: something as simple as observer bias is a key element in this macabre comedy supported by the NIH.

This must be combined with unfalsifiable results, unfulfilled promises, infectious disease scientists not even ‘seeing’ dual use side of their work, effectively zero self-governance, chilly draughts from the top of the NIH and an increasingly dangerous world - and no insider moves?

It’s not surprising that the Risky Research Review Bill has made its way to the US Senate floor. Not caring for way too long, 20 years at least but almost certainly more, has effectively handed biosafety and biosecurity concerns over to the lawmaker.

---

Aside 1

Apropos the rhetorical question, in less than 50 years the Soviet geneticist Dmitri Belyaev (1917-1985) selected Siberian silver foxes for reduced fear of humans.

---

Aside 2

While infectious disease researchers are understandably fixated on pathogenic organisms, not all microbes cause disease. TTV was first reported in 1997. In some countries almost everyone is infected. It can be found in blood donors yet causes no trouble as infection is asymptomatic. The blood banks don’t even screen for it.

---

Simon Wain-Hobson is an emeritus professor at the Institute Pasteur, Paris, from which he retired in 2021. He and his colleagues were the first to sequence the genome of HIV, and Wain-Hobson has published more than 230 papers on virology and cancer.