A new plot. Deaths overlaid with new cases for the nation. And, deaths, overlaid with infection density for the nation. Both cases deaths is time shifted 27 days. A thread. 1/

If you follow me, you know actual infections are not reflected by confirmed cases. You need to adjust them by test positivity. You also know there are 5-10 times as many cases as are confirmed. Now, people are wondering why deaths are not tracking confirmed cases. 2/

There are two reasons that definitely exist. First, confirmed cases do not reflect actual cases. Infection density is closer to the truth. The second is that the delay between confirmed cases reporting and death reporting has changed. 3/

Death reporting lagged confirmed cases 7-8 days in mid March. Today it lags by 27 days. The differences are likely 1) time from inoculation to testing 2) testing delays 3) test reporting delays. 4/

So when people line up the exponential increase in cases between cases and deaths in March, it is misleading in June. 5/

The other difference is that infection density is adjusted for test positivity. This difference adjusts for the number of cases you miss because you don't test enough. When we align both of these factors, in the bottom plot top of this thread, deaths tracks cases pretty well. 6/

As an aside, I've been using the exponent 0.5 to estimate the cases, I think the true exponent is closer to 0.4 resulting in some lack of fit, but I didn't have time to analyze this properly this weekend. THE END 7/7