Antibody Testing Vs the World

There have now been two large antibody tests in California to estimate the fraction of the population that has been infected with SARS-CoV-2. This one in Santa Clara County and this one in Los Angeles County. They estimate that 2.5% – 4.2% of Santa Clara had been infected as of April 3 – 4 and that 2.8% – 5.6% of LA County had been infected as of April 10 – 11. Those number include many people who had already recovered as of those dates.

By today, April 20, nearly all of these cases should have resolved, in either recovery or fatality. By taking the number of known deaths and the estimated case numbers (midpoint of the above infection rates multiplied by the population) we can get an estimated infection fatality rate (IFR). The authors themselves do a similar calculation (but they used data as of 10-17 days ago and had to extrapolate to the data we have today) and get similar results.

County Estimated Cases Deaths (4/20) Estimated IFR
LA 331,000 617 0.19%
Santa Clara 64,500 83 0.13%

These IFRs are surprisingly low. This paper found an IFR in China of 0.66%, which a co-author and I age-adjusted to approximately 0.8% in the United States here. Was our estimate way too high? Below, I calculate minimum possible IFRs for the hardest hit regions in the world. I take the number of known deaths as of today divided either by the entire population or 70% of the population (roughly the level of herd immunity). These values give a floor on the actual value of the IFR. Uncounted deaths would raise the values. Deaths that happen after today would raise the values. And if less than 70% of these populations have been infected, that would also raise the values.

Location Population Deaths IFR – 100% infection, no more fatalities IFR – 70% infection, no more fatalities
New York City 8,399,000 14,604 0.17% 0.25%
Lombardy 10,060,000 12,376 0.12% 0.18%
Community of Madrid 6,661,949 7,351 0.11% 0.16%

*For NYC only I include the government’s official counts of suspected-but-unconfirmed fatalities. Comparisons to excess fatalities in NYC suggest this is still an undercount. I know of no similar efforts in Italy or Spain to report on suspected-but-unconfirmed COVID-19 deaths at the individual level (vs statistical).

The fatality rates across the entire populations of New York City, Lombardy, and Madrid make the estimated IFRs in Los Angeles and Santa Clara look implausibly low. By definition, at least one of the following statements must be true:

  1. The antibody studies are overestimating the number of infections.
  2. LA and Santa Clara are under counting deaths by a larger factor than NYC, Lombardy, and Madrid.
  3. NYC, Lombardy, and Madrid are over counting deaths by a larger factor than LA and Santa Clara.
  4. The virus has been deadlier in NYC, Lombardy, and Madrid than in LA and Santa Clara.

Which of these explanations is most likely? I’d speculate that it’s some combination of #1 and #4.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s