Interesting attempt to measure true number of infections using cases and positivity rate:
https://covid19-projections.com/esti...ue-infections/
Quote:
Knowing the true number of people who are infected with COVID-19 in the US is an essential step towards understanding the disease. But estimating this number is not a simple task. The true number of infections is many times greater than the reported number of cases in the US because the majority of infected individuals do not get tested due to several reasons: 1) they are asymptomatic, 2) they are only mildly symptomatic, 3) they do not have easy access to testing, or 4) they simply do not want to.
On this page, we introduce a simple square root function to estimate the true prevalence of COVID-19 in a region based on only the confirmed cases and test positivity rate: true-new-daily-infections = daily-confirmed-cases * (16 * (positivity-rate)^(0.5) + 2.5). We will also introduce the implied infection fatality rate (IIFR), which is a metric derived by taking a region’s reported deaths and dividing it by the true infections estimate (after accounting for lag).
Using this method, we estimate that the true number of new infections peaked at close to 500,000 new infections per day in July, compared to 300,000 new infections per day in March. This means that the peak of infections after reopening is 60% higher than the initial peak in March. In total, by the end of July 2020, we estimate over 35 million (1 in 10) Americans have been infected at some point by the SARS-CoV-2 virus.
|
Quote:
|
Once we have a reasonable estimate of the true number of newly infected individuals per day, we can use the reported deaths to compute the implied infection fatality rate (IFFR). The IIFR for the US was above 1% in March, stabilized at around 0.6% in April-May before decreasing to ~0.25% in July. Note that our IIFR estimate does not take into account excess/unreported COVID-19 deaths, so it is likely a lower bound for the true IFR. This is further explained below.
|
This meshes with everything else I've been reading. Of course the main driver of IFR decreasing is younger/healthier people getting it while older/at-risk people isolate much better than they were back in Feb/March.
Quote:
|
Note that our use of the term infection fatality rate (IFR) refers to true deaths divided by true infections. It is not age-adjusted. As a result, if there is an increasing prevalence of the disease in a younger population, then the IFR will decrease, despite the deadliness of the virus remaining unchanged among a particular age group. It is likely that the fatality rate for a given age group have not changed significantly.
|
Better treatment is helping too.