This is an addendum to the April 02 global covid-19 severity graph I published earlier. I had some trouble locating a public daily statistics aggregate for the individual states. In the end I had to lift the data from Wikipedia, so relevant reservations apply.
The graph is explained more extensively in a previous post. Here it should suffice to mention that I am plotting the daily growth rate — i.e., the inverse of the doubling time of deaths in days — of the published cases of covid-19 related deaths against the cumulative number of deaths in each state. By eliminating time from the graph it is possible to see the relative severity of the epidemic in each state, when severity is defined as the expected total number of deaths once the local epidemic has run its course.
We know by now that an uncontrolled epidemic typically has a growth rate somewhere from 0.3 to 0.5 (doubling every three to two days, respectively) and even higher for communities with a large number of daily personal interactions of the type that can transmit the virus. Thus, that is where the growth rate is at the start of an outbreak (i.e., at zero deaths) and more importantly, that is where it will stay unless measures are taken to inhibit the transmission of the infection. It is in the interest of every state to turn their trajectory down as fast as possible but there is a fair amount of inertia: what you are doing now will only have a measurable effect about two weeks down the track, and the sharper the sought turn, the more severe measures are required.
With that in mind, the interpretation of the graph is fairly simple. Going down to zero is the goal; the faster the better. Starting low and going steadily down is good. Starting high but going steadily down is the second best option, at the cost of increased total number of casualties. Starting anywhere and staying level is bad, meaning that the outbreak is not in control and if unopposed will grow exponentially until reaching saturation. Starting anywhere and going up is very bad, meaning that any efforts to contain the local situation are inefficient and the epidemic is in the process of exploding.
Because of statistical noise, values below about 100 cases should be treated as inconclusive.