With winter approaching, many governments in the northern hemisphere are stocking up on Tamiflu and fine tuning their civil defense plans to cope with the disruption a bird flu outbreak might cause.
But how likely is an outbreak? While various groups have written about how a pandemic might happen, Rinaldo Schinazi at the University of Colorado says nobody seems to have bothered to work out the probability of such an event.
That sounds unforgivable, given the investment that’s going in to tackling the problem. But in some ways it’s understandable. Schinazi says:
“The occurrence of a pandemic seems to be hopelessly complex in the sense that it depends on a multitude of factors, some known others unknown. Factors mentioned in the literature go from the stability of the current influenza strains to the number of pigs in China! Hence, a model taking into account all known factors would probably be as complex as the phenomenon itself, would not be that accurate and would therefore be useless.“
But Schinazi has taken an entirely different approach:
“We do not attempt to incorporate into the model any of the factors that are believed to provoke a pandemic. Instead we treat the occurence of a pandemic as a purely random phenomenon.“
And the result is a fascinating read. It turns out there have been only 10 pandemics in the last 300 years. This is the data Schinazi uses to prime his models.
Assuming a Poisson distribution, the mean time between pandemics is 30 years but the probability that no pandemic occurs in 60 years is 14% .
Assume a random walk model and the probability that the time between two pandemics is at least 50 years becomes 11% but the probability that this interval is at least 100 years is 8%, which is hardly much less.
Schinazi’s conclusion is both surprising and re-assuring: the next pandemic might not be as imminent we’ve been led to believe.
Ref: arxiv.org/abs/0807.3524: Will the Announced Influenza Pandemic Really Happen?