Let me begin this post with a caveat. I have never been less confident about the data we are getting on COVID than I am at this point. There are significant delays in getting back test results and fatality reporting is all over the map. Last week Texas made a major change to the way it reported fatalities that added nearly a 1,000 new deaths. And just to give you some idea how inconsistent the data is, as of Saturday, Harris County is reporting 748 fatalities but the state is reporting 1,288 fatalities in Harris County. You know, a 70% difference is pretty significant!
With that caveat in mind, generally, we are seeing the same pattern that has been in place for the last couple of months. Hotspots break out in particular areas and then gradually moderate, only to be replaced with outbreaks in other areas. For the most part the outbreaks are following a fairly consistent curve with a rapid rise in cases followed by a plateau and then a long, slow decline. There is a lot of speculation about why this pattern is repeating itself, but no one knows for sure.
New infections appear to be holding steady but at the elevated level of Week 29. We have apparently reached some limits on testing capacity, as tests have stalled out at about 800,000 per day for the last three weeks. The positivity rate has ticked down slightly.
Fatalities began to catch up with the rise in new cases. Fatalities have about doubled off their lows of a month ago, but are still only about half of the worst weeks in April. There is a fair amount of noise in the fatality numbers from last week as states continue to refine the methodologies, such as the change Texas made to its reporting.. This means the totals will probably more accurate in real time, but the weekly totals are overstated because they include fatalities from previous weeks. There appears to be about a month lag time between a change in new cases and fatalities.
The good news is that the virus continues to become less lethal. The ratio of fatalities to new cases (based on a 4-week lag time) came down dramatically in the early going and has continued to slowly decline. Last week the ratio was slightly over 2%.* But the bad news is that if that 2% holds true for the future, we will be seeing about 1,000 new fatalities per day for some time to come.
I have come to believe that the number of people currently hospitalized, in ICU units and on ventilators, is the best gauge of the extent of the infection’s spread and the damage it is doing. There is no delay in these metrics as they are reporting real-time interactions. However, there is a lot of variation around the country on how these are reported. Some states do not report any data on them at all. The COVID Tracking Project (CTP) has been tracking these from mid-March, but they concede their data is probably understating the numbers because of the inconsistent reporting.
But with that in mind, the numbers from CTP appear to confirm a plateau in the spread of infection in the last two weeks. Hospitalizations and ICU bed utilization have been fairly flat since mid-July and ventilator usage has only modestly increased.
* This should not be confused with the “infection fatality rate” or the “case infection rate”, both of which are estimated much lower because testing only is catching a small percentage of the actual cases. (The CDC currently has them estimated at .26% and .4%, respectively.)