It’s been a couple of weeks since I posted stats on the Covid-19 epidemic. I’ve been waiting for the numbers to show something definitive. They are tantalizingly close, but at present we can’t say whether we’ve hit the peak or not. I’ll show you what I’ve got, and I’ll venture some opinions. A caveat: I have no medical or epidemiological training, and only have access to public data. So take my opinions with a fair amount of salt.
There are lots of unknowns, but the questions I’d really like answers to are:
- What is the case fatality rate (CFR, the number of people who die from the disease over the number of confirmed cases).
- What is the infection fatality rate (IFR, the number of people who die from the disease over the number of people infected).
- What is the N0 (the number of people infected by the average person) for various amounts of social distancing.
- Have the mutations in the virus affected any of the above?
- Do recovered patients have immunity? To what degree, and for how long?
I have some answers, and some educated guesses:
- The CFR has dropped to about 8% nationwide, and 5% in California. The reason for the drop is probably more testing, not better treatment — yet. The New York City fatality rate is much higher than the California rate. This could be do to more testing in California (which increases the denominator), or overloaded medical facilities in New York (which increases the numerator).
- The IFR is probably somewhat less than 0.5%, maybe a lot less. We won’t know until we do a lot more testing.
- Testing is rolling out slowly, but we are testing more people as time goes by. I have numbers for California.
- With the testing improving over time, it’s hard to track the spread of the disease, but even with the increased amount of testing, there has still been considerable curve-flattening over the last month of six weeks.
- In spite of the social distancing that has been taking place, we are not seeing much in the way of decreases in either daily number of cases of daily number of deaths. We are at somewhat of a plateau. It is unclear what direction we’ll go from here. With relaxed social distancing, the daily cases could start climbing again.
- Another way of saying the above is that the N0 is not much less than one for the social distancing that we’ve seen over the last few weeks.
- Although the virus is widespread in the US, the per-capita case rate is very uneven across the country. The reason for that is unclear; it can’t all be explained by social density.
- By looking at overall deaths and comparing them to historical records, it is clear that the number of Covid-19 deaths is been underreported. We are currently at about 65,000 deaths nationwide, but I think the real number is closer to 100,000.
Let’s look at the US numbers first, then California, then Monterey County. Here are the nationwide cumulative cases and deaths.
This is a semilog plot, so exponential growth would plot as a straight line. So we are flattening the curve. If we look at the new daily case and death numbers, instead of the cumulative ones, we get this:
I’ve plotted 7-day moving averages as well as the daily numbers, and you can see that these curves look nothing like the modeled plots we saw in March, which rose rapidly to peaks, and fell almost as rapidly on the far side of those peaks. Part of this might be that the disease is spreading across the US and drops in some regions are balanced by increases in others, but you’d think the modelers would have taken that into account. It looks like we’ve passed the peak of deaths, but that’s because in the middle of April, New York decided to put all suspected Covid-19 deaths into the database with confirmed Covid-19 deaths. That caused a huge spike in deaths on one day, which created the peak in both the daily plot and the 7-day moving average. For completenss, here is the weekly new case graph.
From 100 cases to almost 100,000 cases, we had exponential growth, then the curve plateaued.
Looking at the number of deaths as a proportion of confirmed cases 10 days earlier:
In California, using the California Department of Public Health numbers, things look somewhat different, but the same trends apply.
The California plateau is not really flat. The number of cases is still rising, even with strong social distancing rules (adherence to those rules is spotty, however).
Deaths look about the same:
The California CFR is lower than the national one:
Let’s take a look at how much testing has been done in California.
There are two steps in the completed tests (orange) curve that need explaining. The first one, at the end of March, was the result of an effort to clear the test backlog. The second one, in the third week of April, resulted from the CDPH’s decision to count each and every test, where before they had only counted the tests on each person. If one person had been tested three times before the third week in April, the test would have been counted once. After that, the tests would be counted three times. The grey curve is the daily moving average. In the first part of April, it was stuck at about 10,000 tests a day, and now it is about 30,000.
The percent positive test graphs is interesting:
This shows the ratio of positive tests to test performed dropping dramatically. The testing policies have changed a great deal over the time span of the graph. Earlier, they were much more restrictive. Now, they allow for more testing of health workers as well as of the general public.
In Monterey County, with respect to daily new cases, we had a flat curve for more than two weeks, then an increase, followed by a gradual roll-off.
Thankfully, the Monterey County number of deaths has been too small for graphical analysis.
Maybe in another couple of weeks, we’ll have a better idea whether we’re on a plateau that stretches out there as far as the eye can see, at the top of a falling new cases slope, or at the bottom of another wave of new cases.