A new dashboard approximating population immunity over time.

Rhode Island, Massachusetts, CT, NJ, Vermont, and South Dakota according to this have the highest population immunity. Vermont did it the easy way through vaccination. South Dakota the hard way through infections.

Generally speaking all those states have some of the lowest case rates right now even with Delta prevalence rising. So the estimates pass the smell test.

I don't like to use the "herd immunity" phrase mainly because we don't really know what the herd immunity threshold is for the coronavirus and it is a bit of a moving target as more transmissible variants emerge.

80-85% seem like reasonable estimates of the herd immunity threshold for Delta. Of course herd immunity is a local phenomenon. NYC would likely need a higher herd immunity threshold than a rural town that is less dense.

Some states like Vermont, Washington, and Maine didn't have very many infections this pandemic. Their population immunity is mainly from vaccination.

Other states like South Dakota, Alabama, New Jersey were hard hit by the coronavirus and have a lot of natural infection immunity.

This dashboard isn't perfect, but it tries to determine total population immunity from both vaccination and natural infection per state.

Of course we don't know the true number of infections in a state as we didn't detect all infections. Many states didn't test as much as they should. And it assumes vaccination and infection are independent. There's survey data to suggest some people aren't getting vaccinated because they were infected. And previous infection isn't a guarantee of protection against reinfection, particularly against Delta.

You can play around with the assumptions. This plot is assuming an ascertainment bias of 4, meaning 1 out of every 4 infections was detected. Infection modeling suggests this is a reasonable assumption for the US as a whole.

Here is the dashboard : https://popimmunity.biosci.gatech.edu/

You can play around with it yourself and see the change in immunity over time for each state and the US at large.

I think seroprevalence surveys and test positivity would help narrow it for states better. Though even those are imperfect.

One could maybe use excess deaths as a correlate for infections, but that might be too affected by demographics and health system quality.

In any case gets close enough for my tastes and helps explain the current case map better than just going off of vaccination rates.