**UPDATED 03/26/20**

I have created several charts based on official (State-by-State) statistics on COVID-19 cases in the U.S. that are reported here. The numbers reported there exclude cases and deaths occurring among repatriated persons (i.e., Americans returned from other countries or cruise ships).

The source tables include the U.S. territories of Guam, Puerto Rico, and the Virgin Islands, but I have excluded them from my analysis. I would also exclude Alaska and Hawaii, given their distance from the coterminous U.S., but it would be cumbersome to do so. Further, both States have low numbers of cases and (as yet) no deaths, so leaving them in has almost no effect on the results displayed below.

All of the following charts are current through March 25, 2020.

* * *

The number of cases is about 2/100th of 1 percent of the population of the U.S.; the number of deaths, about 3/1,000th of 1 percent. That is, only about 1.5 percent of cases have thus far resulted in deaths (but see Figure 4). Note the logarithmic scale on the vertical axis. Every major division (e.g., 0.01%) is 10 times the preceding major division (e.g., 0.001%).

There are some comparisons of the U.S. with other countries, but the ones that I have seen use raw numbers of cases and deaths rather than cases and deaths per unit of population. The comparisons therefore make the situation in the U.S. look far worse than it really is.

Nor do the publishers of such comparisons address cross-country differences the completeness of data-collection or standards for identifying cases and deaths as resulting from COVID-19.

In any event, here’s how the coronavirus outbreak compares with earlier pandemics when the numbers for those pandemics are adjusted upward to account for population growth since their occurrence. (Again, note that the vertical axis is logarithmic.) The number of COVID-19 cases is thus far about 1/1,000th the number of swine-flu cases; the number of COVID-19 deaths is thus far about 1/14th the number of swine-flue deaths.

The daily percentage change in new cases is declining. But the daily percentage change in new deaths is holding steady (temporarily, at least). Even if the rate at which new cases continues to decline, it will take a while for the rate of new deaths to drop, given the lag between infection and recovery or death (again, see Figure 4).

The following graph represents an early attempt to find a relationship between new deaths and new cases. There are only 13 observations, so the estimate of 40 deaths per 100 additional cases (implied by the regression equation) isn’t a reliable one. I chose the lag of 12 days between the onset of a case and a death because that lag yields the highest correlation between cases and deaths, given the limited data set. That may well change as the number of observations increases.