New analysis of sequence data suggests the current US COVID-19 outbreak sparked in mid-Feb. As well as providing insights how the epidemic started, I think this has some important implications for how we interpret what happens next... statnews.com/2020/05/26/new… 1/
When reports talk about transmission chains and reproduction numbers, it can conjure an image of a clockwork outbreak, e.g. 3 cases infecting 3 more, who infect 3 more... The resulting image is one of a steadily growing outbreak, which may be undetected at first. 2/
But of course we know that COVID-19 doesn't spread like clockwork. Some cases generate a lot of transmission, while most generate little or none. In other words, outbreaks seem to be driven by superspreading. sciencemag.org/news/2020/05/w… 3/
Superspreading actually makes it less likely that a single infection will spark a new outbreak (because most cases don't lead to transmission) - we estimated several independent introductions might be needed before a COVID outbreak eventually takes off thelancet.com/journals/lanin… 4/
To show the difference, here are some simulations of new COVID outbreaks with little variation in transmission (adapted from @C_Althaus' code here: thelancet.com/journals/lanin…) - each one follows similar path over time. 5/
In contrast, if transmission involves superspreading event, many outbreaks fail to spark initially, but when they do, they can grow rapidly. So what does this mean for the next phase of the epidemics in Europe, US and elsewhere? 6/
When we talk about the effects of relaxing measures, we shouldn't think of it like a clockwork process, e.g. relax measures a bit, R immediately goes up a bit; ease them a lot, R immediately goes up a lot. 7/
Instead, we should probably think of it like those early introductions, which took time to spark. Countries may relax measures & initially see little change in cases. But it doesn't necessarily mean risk isn't there - it may just be that, by chance, the spark is yet to come. 8/8
