Summary: People think there are a huge number of undetected COVID19 cases out there due to lack of testing and symptoms. This analysis suggest there are not. US cases may be 40-350% higher.
There is a general perception that no country is testing enough of its population to know its true infection fatality rate (IFR). People believe there are many mild asymptomatic cases. This has two implications. It makes people feel less at risk and that their population is closer to reaching the level of herd immunity.
The USA is believed to be under testing and currently has a case fatality rate (CFR) of 2.7% (4Apr) based on data from worldometers.info. South Korea has been described as doing extensive testing as part of its effort to eradicate the disease, yet its CFR is not much lower than the US at 1.7%. The hypothesis would be that with extensive testing the denominator of the CFR would increase and move the CFR closer to the true Infection Fatality Rate (IFR).
Countries throughout the world vary in the degree to which they test their populations for COVID-19. These countries are also at different stages in their epidemics. If there is a pattern in CFR with the degree of testing, it should show up in the relationship between CFR and testing hit rate, i.e., the deeper the testing in population the more negative tests and the more complete picture of the CFR denominator.
Figure 1 shows a “surviving” rate (1-CFR). It is not the actual survival rate as this can’t be assessed until all patients recover or pass away.
Here is the same data upside down plotted versus CFR:
Early on the US has had highly restrictive criteria on symptoms and travel history that limited testing to patients who were highly likely to have severe symptoms and to test positive and thus have a higher fatality rate than if all mild cases were identified. Different countries have different degrees of testing availability. The less selective the country is, the more likely it is to get an assessment of the true underlying infection rate in the country. The plot above implies that once a country has sufficient testing available such that more than 90% of tests come back negative, the CFR plateaus to somewhere between 06% and 1.9% rather than continuing to drop. Not seeing the CFR trend downward further supports t also supports the premise that COVID-19 is much more deadly than fly by perhaps 10-20x.
The United States has gotten a positive result in 20 percent of the tests conducted and is seeing a CFR of 2.7%. If are testing was more extensive one could estimate that we would find between 2.7/1.9-1=0.4 and 2.7/0.6-1=3.5 more cases per current case.
While having true cases be 3.5 more than measured is a fairly big underestimation, because cases are doubling every 5.1 days, it just means our true situation and case count are just 11 days behind reality.
One complicating factor that I suspected early is the question of how many US deaths are being undercounted. One way that the US can have a relatively low CFR despite limited testing availability is missing a nearly proportional amount of deaths as cases (see NY Times and other articles below)
NY Times : https://www.nytimes.com/2020/04/05/us/coronavirus-deaths-undercount.html
NY Serological testing lines up with CFR/0.66 correction factor for total infections: