How I Will Learn About the Next Pandemic

February 16, 2023

I am somewhat worried about the H5N1 news. With no human cases yet, this is a good time to worry about being worried. It’s unlikely to be ≥ covid, but my information diet over the past week has me wondering — when I think backwards on the next pandemic, what will be my first recollection?

I spent the first weekend of 2020 at an academic conference in Puerto Rico. My first memory of the Covid pandemicI also have a strong memory, surely constructed, standing on the steps of my aunt’s house in Morgantown, West Virginia on November 23rd, 2019, speaking to one of her Ph.D. students from Wuhan. This sent me down a rabbit hole looking for the first cases; 1 December, 2019 is still the consensus date for the index case, but that comes with enough controversy that I’m fairly certain of community transmission in November. This is robust to the reports of WIV staffers getting sick in autumn 2019, which I’m sceptical of. is drinking something sweet and coconutty on a terrace over the Laguna del Condado — an academic used to flying between Beijing and New YorkI really want to take this flight someday, in the summer. Look at this gorgeous great circle. Unfortunately, there are no direct flights from Argentina to Australia. was complaining about the somewhat invasive health measures being implemented on his last trip.

At the time, a seemed like a minor concern. My previous exposures to potential pandemics had been outbreaks of swine flu or SARS which fizzled out after a few months and a few thousand deaths. When I thought of public health disasters, the examples which came to mind were infrastructural tragedies — Flint’s pipes, or Chipotle’s E. Coli. And just two days before the conference, Puerto Rico had been hit by a 6.4 magnitude earthquake; as I landed, most of San Juan was without power. There were much more urgent disasters looming than an outbreak of pneumonia in a city I’d never heard of.

I got lucky with Covid, in that I heard about it earlier than the average American. The first US coverage about the outbreak came roughly the same week; these stories relied heavily on statements from the Chinese government. The first WSJ9 January. and Vox20 January. articles on the “Wuhan pneumonia outbreak” are good examples. This coverage was far less informed and more credulous than the early news about H5N1, which I attribute to well-learned lessons.A common refrain — which is false — is that we have ‘learned nothing’ from Covid. A great big communal We have learned much much much less than we should have, but a much smaller commentariat we have forgotten just how bad the first year was.

All my methods of hearing about potential pandemics are indirect, relying on high-level distillation efforts. I don’t spend my time reading pre-prints or skimming APHIS databases,With one exception — last year, I was looking for data on US milk, and ended up talking to Dr. Monica Saavoss, a milk economist at APHIS. Dr. Saavoss has the coolest milk data ever, tho for weird bureaucratic reasons, no data on oat milk. There’s a blog post in here, too. and I probably wouldn’t understand these sources deeply enough to spot problems before they emerge. This seems fine — I don’t want or need to be the first, or the millionth, person to hear about a new pandemic.

I first heard about H5N1 on Sunday, 5 February, from a wildlife biology friend. The first news source I read was Zeynep, predictably, and the nascent Metaculus market soon after. I’m pretty happy with this — I learned about a potential pandemic before the first human caseWhich hasn’t happened yet, and hopefully won’t! and before the first case of human-to-human transmission, which are the barriers to this becoming something worth worrying about. If anything, I feel overexposed to news about potential pandemics. On the time scale of the news cycle, pandemics move slowly; is there any reason for me to know about H5N1 this early?

Because I’m not sure if there’s anything I’m “supposed” to do if I learn about an outbreak which scares me more than H5N1. I’ve been in this situation — asking myself, should I take any action with this information? — three times now. And the answer has been “no” all three times. With one small exception last summer, my response to learning about a new viral sword of Damocles has been to continue with my life.

(Should you do anything when you hear about an emerging pandemic? If you’re a policymaker, yes, definitely, stop reading this blog and go do something. Idk what, ask Zeynep or Zvi. But for everyone else, my advice would be painfully generic. Protect your immune system [diet, sleep] and overall health [lose weight through diet and exercise at a healthy rate]. This is a good idea regardless of whatever new virus is coming down the track, but the prospect of overwhelmed A&Es/EDs and pneumonia can be motivating. We all have strong incentives to freeride on reducing transmission in a population, but reducing the chances of yourself getting very sick very much depends on you.)

I’m not sure what information would cause me to take action. H5N1 seems to be an even-higher variance virus than most potential pandemics — there have been 380 human cases since 2003, and more than half of those were fatal. None of these cases resulted from human-to-human transmission. So as long as we’re dealing with this version of H5N1, we’ll continue to see low casesMostly among people who work with dead birds. and a high fatality rate. But if we see a mutated variant which is capable of community transmission, we’re deep in a long tail of nightmare scenarios — H1N1’s fatality rate is an order of magnitude lower. The likelihood of this change to occur is the scariest and most important parameter in whether we should take action or not.I spent an hour trying to get a sense of the value of this parameter; I didn’t develop strong views. I would sell the Metaculus market down to 15%, with concerns about resolution criteria. One lab model — literal guinea pig model — found that mutations on two amino acids from glutamic acid to lysine was sufficient for low-transmissability avian strains to become high-transmissability mammalian strains. The paper is titled ‘Transmission of Influenza Virus in a Mammalian Host Is Increased by PB2 Amino Acids 627K or 627E/701N,’ which I frankly found offensive. How likely are these two amino acids to switch to lysine? There are no models for that question. Last year at GPI, I explored a fascinating literature on unawareness probabilities, which has some nice applications in wildlife biology and genomics. I’ll write more about this sometime.

My concern is that vigilance will fade over time. Now, there are strong social incentives to speak up about the next pandemic, in the same way there used to be incentives to comment “First” on YouTube videos. These incentives fade over time. There are two mechanisms which worry me.

First, an excess of false positives which looks like crying wolf, so that saying “this might be a pandemic” becomes unfashionable.

Second, the people who would tell me about the next pandemic early get bored or distracted. I’m not worried about Florian Krammer or Caitlin Rivers, they’re ride-or-die virologists; I’m worried I’ll be abandoned by people like Zeynep and Zvi, the people I can understand. Zeynep is more interested in social media than viruses, and if something cooler happens on Facebook, she might be tempted to make that her beat, rather than potential pandemics; AI might be the equivalent for Zvi. These particulars seem unlikely — Zeynep probably can’t escape her virologist sources if she wanted to — but the general phenomenon is plausible.There’s a related worry, about the fading of knowledge about how to cover a pandemic. In ten years, even if the pandemic beat is as well-populated as it is today, those reporters will be a decade removed from their last crisis.

For someone with no pandemic action plan, I’m probably overexposed to potential pandemic news. I had fun reading about H5N1 over the past few days, but deciding to be worried a week earlier or later wouldn’t have had any effect on my life. Unless the next pandemic is much faster moving than the usual — which we’d only really expect for engineered viruses — there’s no reason for a layperson or generalist to be this vigilant. The advantage would come from credibly warning vulnerable friends and family to prepare before large media outlets are confidently ringing the alarm bell. This benefit is marginal, and not worth adjusting your information diet.

Note — I wrote the above on 9-10 February, and any forecasts represent my views on those days. I waited to publish this on 2023-02-16, after reading Zvi’s post and personal communication from virology and epidemiology friends. I have made small updates towards being less worried about H5N1 in the past week.

How I Will Learn About the Next Pandemic - February 16, 2023 - Joseph Levine