World Wide Trends, How Many People Have Been or Currently Are Infected?
There have now been 25,417,886 confirmed cases of COVID-19 world wide. That's 0.33% of the world population. According to WorldOMeter, 6,840,933 of those cases are "active". However, as we discussed in the last update, WorldOMeter's recovery estimate is too low. I estimate that there are currently 4,395,451 confirmed cases that are still "active'.
However, how many people have actually been infected? Many people who are infected are never tested, and never become a "confirmed" case.
One way to estimate infections that I have used before, is to use the number of deaths, and the range of possible infection fatality rates to get a range of possible true infections (and assuming an average 20 day lag between infection and death).
This estimate will only be as good as the approximations to the Infection Fatality Rate (IFR), and the reliability of the death data used. I am currently using the range of IFR estimates given by the CDC.
If we apply this to the World-Wide data, these are the results:
We can see that the number of true infections inferred in this way is significantly larger than the number of confirmed infections. But it's also worth pointing out that this is possibly an under estimate, because it is highly likely that the true death tole world wide is under-reported, especially in less developed regions.
We can then apply the same formula we used last week for estimating recoveries to turn the cumulative number infected into an estimate of the number of active cases, those currently infected:
By this estimate somewhere between 0.15% and 0.3% of the world's population are currently infected with SARS-CoV-2, the virus that causes COVID-19.
Both of these estimates seem unreasonably low. Certainly it is significantly lower than we see in Europe or the US (although the US should hardly be held up as a model of success). Certainly a large portion of the world's population live in China, and they have done remarkably well at limiting the virus' spread.
Nevertheless, it seems likely to me that the true death tole from COVID-19 is significantly under-reported world wide.
Deaths By Day, Estimating the Reporting Lag from Florida:
It can be very helpful when locations report either cases by date of symptom onset, or deaths by date of death, rather than by the date the case or death was reported/recorded. This can give us a better idea of the true shape of the curve... but it also has a lag to it, that always makes it look like cases/deaths are going down while we wait for these cases/deaths to be reported.
When modeling this lag, it's important to have good data on what the lag has been historically. For Florida, there is an excellent archive of historical information showing how the data has come in over time, available here:
https://github.com/mbevand/florida-covid19-deaths-by-day
This is what that delay looks like for Florida:
The proposed fit is 1-e^(-0.1513x).
No comments:
Post a Comment