This post addresses the trends in the 7-day lag case fatality rates (CFR). I am modeling the positive cases and then extrapolating that to future deaths using the CFR (7-day lag). I estimate the CFR for each region that I am modeling (e.g. if I am looking at New England, it’s the New England CFR. If I am looking at IA, it’s the IA CFR.) If states are ramping up testing and getting many more mild cases, then the CFR would be dropping (and the reverse if they limit testing to only hospital cases).

There is no evidence of that happening, at least not in any major way. States seem to have settling into a testing procedure (meaning who to test) and the 7-day lag CFR is steady after March across (most) states. The plot below shows the 3-day average new deaths divided by the 3-day average new cases 7 days prior. Normally you compute CFR with all deaths to date. Because I want to see if CFR is changing, this is not doing that. It is a point estimate of CFR; 3 days of new deaths divided by 3 days of new cases, 7 days prior.

I show four representative states that illustrate the pattern that 7-day lag CFR stabilized to 4-7% after April 1st, 2020 across multiple states. This is the pattern across almost all states.

There are a few strong exceptions. First the bad trends. In MN and MI, the CFR has gone up to 10%

Across the states that have been having outbreaks in meat processing plants, the CFR has been declining, suggesting more extensive testing associated with these (and other agricultural) plants. SD is the real standout and they had an early meat processing plant outbreak. The CFR’s are about half what we see in other states (1-3%), suggesting more extensive testing program. The meat processing sector has had a lot of outbreaks, and I have been seeing reports of ramped up testing in this sector. If asymptomatic or mild symptomatic people are tested, then the CFR goes down.

In a number of states, there were really high CFRs (over 15%) early and then these fell dramatically. In this case, the outbreak was detected in a more vulnerable population (elderly) and testing ramped up to cover the wider population later.

All the states

If you want to look up your state.