Interesting study using a data-driven model to estimate the impact of Covid-19 vaccination in 50+ in NSW by @JamesTrauer and others [1/n] journals.plos.org/plosone/articl… Methodology is appealingly simple: calculate death rate each week for those with 0/1/2/3 doses. Then simulate counterfactuals by changing the death rate applied to each category (e.g. no booster = apply 2 dose rate to 3 dose category, etc)

Since that US veterans reinfection study seems to be popping up again, a thread on some of its major limitations


1. This study doesn't compare 1st infection to reinfection. It compares reinfection to no reinfection.nature.com/articles/s4159… Authors say this:
"Our analyses should not be interpreted as an assessment of severity of a second infection versus that of a first infection,

@MichaelSFuhrer @DrPieterPeach @Pseudorandom75 Yes it is a very nice paper. Interestingly I think the endemic equilibrium situation is actually simpler - at least in the simple well mixed SIRS case – real life is of course a lot more complicated! [1/n] @MichaelSFuhrer @DrPieterPeach @Pseudorandom75 At equilibrium, incidence of new infections per unit time in a closed pop of size N is N/tc*(1-1/R0) where tc = time spent non-susceptible after infection (generation + waning time)

Since there is some confusion about this, a thread with some scientific evidence on RATs and how to interpret a positive result.

TLDR: RATs are actually very good at telling when you’re infectious, which is what matters most for reducing spread First, what about genuine false positives in a person that hasn’t been recently infected? These are extremely rare - less than 1 in 3000 chance according to this estimate assets.publishing.service.gov.uk/government/upl…

It’s clear that the omicron variant can infect more people and spread faster than delta but is it milder or more severe? And what difference will that make? A short thread on transmissibility vs virulence with some hypothetical numbers. Suppose we have two variants: one spreading relatively slowly with R=1.1 and a hospitalisation rate of 4%, the other spreads twice as fast but is half as virulent so R=2.2 and hosp rate of 2%.

This will be well known to many, but here’s a thread on what we can learn about the #COVID19 outbreak from these equations
dS/dt = -βSI
dI/dt = βSI - γI This is known as an SIR model: S stands for people who are susceptible to the disease (everyone who hasn’t had it yet), I is people who are currently infectious, and R is people who are “removed”, meaning they’ve had the disease but are no longer infectious.

Your handy reminder about how herd immunity works. Let's say a single infected person in a fully susceptible population would infect R0 other people on average. Now if a proportion v of the population is immunised then R0*v of those potential infections don't happen (1/n) and so we only get R0*(1-v) new cases. If R0*(1-v) > 1 then the first case infects more than one other person. Each of those secondary cases also infects more than one other person and so on. We have an outbreak on our hands