A new report has some people arguing that we’re “overdue” for a pandemic — for an influenza virus strain of higher than “normal” lethality that moves quickly and overwhelms global healthcare systems. I’m going to argue that it isn’t true, and that, if anything, we’re under-due1Yes, yes, I know. Not a word. Well, overdue is the wrong word in random processes too..
A good place to start is by looking at the historical rate of pandemics. From various sources, here are the dates of the major pandemics over the last few hundred years2We shouldn’t be completely confident in this data, but it’s not terrible either. While we had no CDC or WHO one hundred years ago collecting data, there are archival and other sources sources that tell us about disease outbreaks..
- 1729. Started in Russia, and then spread in waves across Europe.
- 1732. Some argue that was just a continuation of the 1729 outbreak, but let’s be charitable and say it was a new pandemic,
- 1781. This one started in China, spread across Russia and into China. Highly lethal among young adults.
- 1830. Started in China, and then spread into Russia, Europe, and, eventually North America. Generally thought to have had similar lethality to 1918 pandemic.
- 1833. Like in 1732, many argue that this was another and less lethal wave of the 1830 outbreak, but, again, let’s be charitable and say it was its own thing.
- 1889. Started in Russia, and spread rapidly outward, touching entire northern hemisphere within four month.
- 1918. The one, of course, the Spanish flu, which killed at least 50m people worldwide, a staggering figure.
- 1957. The Asian flu, originating in east Asia. Killed something like 1m people worldwide.
- 1968. The Hong Kong flu, named after where it started. It also killed around 1m people worldwide.
And that’s about it. You could, I suppose, argue that SARS (severe acute respiratory syndrome) in late 2003 was an incipient pandemic, but, despite its high lethality, it still killed less than 1,000 people worldwide (and to be somewhat virally pedantic, SARS was caused by a coronavirus anyway, not an influenza subtype).
So, what does this pandemic pattern tell us? It tells us that pandemic flu is regular, bad, and that we haven’t had a serious outbreak in a little over fifty years. Is that a long time? And what does a “long time” mean for a random event like influenza anyway?
Well, one way of answering is to say that in the last few hundred years we have never gone more than forty years without an influenza outbreak. And that’s cautionary, given that we’re now well past that If you say that “normalcy” ended in 1968, then, with nine pandemics since 1729, there is an implied flu frequency of roughly 3 every hundred years or so. Given that frequency, and given the current fifty year gap, should we be worried?
We have methods of analyzing that kind of question in non-arm-wavey ways. Among the better ones is to treat the underlying frequency as a Poisson distribution — similar to how you might think about people randomly arriving at a gas station, or at a grocery store queue — and then ask how likely it is we should see a particular pattern, given the underlying frequency.
The following graphic shows the result for the case of two or less pandemics in a 100-year period, given an observed historical frequency of three per century3There are many Poisson calculators online, but I used this one here: .
So, the likelihood of seeing two or fewer pandemics if the observed historical frequency is three is a fairly whopping 42%. In other words, we shouldn’t be very surprised to see fewer than the usual number of pandemics in any given century4Of course, the reverse is also true: We shouldn’t be surprised, statically speaking, to see more than the “usual” number of pandemics either, but I’m going to circle back to that and argue, while it is statistically true, we actually should be surprised at to see more than the usual number.. (By the way, the Poisson-derived likelihood of seeing zero pandemics in any century is low, around 5%, but definitely not zero.)
What this tells us is that the current 50+ year gap in the pandemic record is unsurprising, nor would an even longer gap. It just isn’t that long, given how time pas hassed, and given the historical incidence rate.
But there are at least two non-statistical reasons to think this pandemic gap should be even less surprising. First, we are getting healthier, and we are better at keep people healthy when things go awry. While we have no cure for pandemic influenza, it’s also less likely to be misdiagnosed, and infectious patients (see SARS) in a lethal outbreak are more likely to be kept separate. Overall, pandemic influenza isn’t magic: it should decline with systemic improvements global healthcare.
The response here is usually …. but, but, global travel! Airplanes! Rapid transmission! Sure. The history of modern pandemics is that, in general, they become more frequent and spread faster and further as transportation networks improved. It’s hard to have a global pandemic when most people don’t travel more than a few dozen miles from where they are born. Russian horses in the 18th century helped a lot, as did 19th century marine trade with China, and so on. Tribespeople on walkabout as patients zero? Not so much.
There is reason to think, however, that increased travel, rather than making pandemics more likely, may be making them somewhat less so. A paper5Increased frequency of travel in the presence of cross-immunity may act to decrease the chance of a global pandemic
R.N. Thompson, C.P. Thompson, O. Pelerman, S. Gupta, U. Obolski
bioRxiv 404871; doi: 10.1101/404871 published last year got considerable attention for making a credible case that by exposing people more regulary to viruses, assuming there is at least some cross-immunity (being exposed to one influenza strain reduces the reaction to another strain), then more frequent air travel is one of the main factors in reducing the incidence of pandemic influenza. While there are many reasons to think air travel makes the world worse, smaller, homogenous, and so on, it seems as if it could also be making the world more resistent to pandemic influenza, which is remarkable.
To return to the original question, should we be surprised that it’s been more than a half century since the last pandemic? Should we be expect one soon? Simple statistics, epidemiological trends, and network analysis suggests we shouldn’t be surprised at all. Yes, a pandemic might still strike at any moment — the probability is and will always be non-zero — but there seems growing evidence that the historical base rates are becoming bad predictors of the flu future.