Antibody tests show what parts of NYC were hit hardest

[1] https://www1.nyc.gov/site/doh/covid/covid-19-data.page

[2] https://opendata.cityofnewyork.us/

[3] https://datasmart.ash.harvard.edu/news/article/new-york-city...

> The accuracy of antibody testing varies widely. Moreover, most antibody tests were done for people seeking them out, which means that those who got tested are a self-selecting group and not a random sample.

> It also appears that New Yorkers in some neighborhoods with lower infection rates were more likely to seek antibody testing.

> In some wealthy and largely white ZIP codes in Manhattan, some 30 percent of people may have gotten an antibody test. But in Corona, for instance, where the positive rate was highest, less than a quarter of people got tested.

The test for me was $165, and my insurance would not cover it.

My recovery was punctuated with almost daily calls from the local hospital i tested at asking me to come back for a follow-up antibody screen for research purposes.

This was $400. It was not covered by insurance so I skipped.

I was then contacted by my insurance company asking if I had completed the antibody test and told it was covered, so I passed the antibody test and went back to work. Three weeks later I wound up fighting $600 from a hospital collections agency for covid ab lab work.

My disease advice is this: If possible, avoid being a working class American.

23k deaths, thus a 1% mortality rate.

I guess this the final nail in the "this is just a flu" .

My read is that test subjects were entirely self-selected, which mean strong bias to those who suspected (or feared) exposure, had access to testing, and (where and when necessary) could pay for same (I don't know whether or when NYC instituted universal free test access).

That's in addition to any issues with test accuracy, procedures, or reporting itself.

The information tells a story. It's not immediately evident just how accurate that story is, an uncertainty which may further muddy conversations and decisionmaking regarding the outbreak.

As any statistician worth their salt will tell you, sample size is vastly less relevant than sample method, and a rigorous, though small (a few hundred, perhaps 1,000 or so randomly selected casess) would be highly valuable. The alternative (at far greater cost) is near-total testing. Itally have conducted comprehensive testing within specific small towns. Some smaller countries and territories (Andorra, Faroe Islands, Monaco, Luxembourg) have attained 100% test coverage. These still suggest 1-6.% CFR.

New York City's test coverage, whilst large, could still remain quite biased.

Really it comes down to 3 dimensions:

1. Where do we direct tests

2. Where do we allocate non-pharmaceutical interventions

3. When ready, where do we direct a vaccine

These are all questions that we can find a signal for within the data. There are obvious constraints, such as reagent and pharmaceutical availability and the viability of certain policies.

I’m flabbergasted and agitated that on the federal level we haven’t come to a consensus on testing and data. If there is anything certain to increase harm relative to where it could be, it’s this “slow down testing” narrative.