Over the last two weeks (ending March 13), all COVID metrics continued to fall at steady rates and vaccinations continued to accelerate. However, there is some indication that the rate of the decline may be leveling off slightly.
Data Note: Unfortunately, the COVID Tracking Project (“CTP”), which had been accumulating data from the States’ websites and reporting to create a national database, ceased operation on March 7. In the future, we will have to rely on data from other sources, primarily the CDC. Most of the datasets have gradually come into agreement over the last few months, however, there are definitional and timing differences so some of the stats I share may not exactly match what I have previously reported to you. For my Texas reports, I will continue to use the daily data published by the State of Texas.
The University of Minnesota’s COVID-19 Hospital Tracking Project is the best replacement I have been able to find for the CTP data. It shows that over the last two weeks, hospitalizations dropped from 47,421 to 37,771, a 20% decline. The CDC’s website shows a lower count because they do not include suspected cases.1 Also, there is typically a one or two day lag in the CDC’s data so the hospitalizations for a specific timeframe? will frequently change several times before the final number is fixed.
The CDC is also now reporting new COVID hospitalization admissions, which is probably a better indication of the rate of the spread of the virus than total hospitalizations. It shows new hospitalizations were down 8.8% for the last week. However, note the elbow in the curve that occurred just after mid-February, indicating that the rate of the decline in new admissions slowed.
Over the last two weeks, the seven-day moving average for daily reported fatalities fell by nearly 40%, (2,037→1.248).
Of course, we always need to keep in mind that reported fatalities is very much a lagging indicator. So, actual daily fatalities are probably quite a bit below that at this point. IHME model is now projecting there will be fewer than 100 daily reported fatalities by the end of June. We may see a real disconnect between the “new case” numbers and fatalities because of vaccines rapidly removing so many of the vulnerable individuals from the equation.
The national vaccination campaign continues to ramp up, with the administration of doses averaging over 2.5 million per day and a stunning 4.5 million reported on Saturday. That is probably a reporting distortion, but impressive, nonetheless. As of Saturday, 11.3% of the US population and 14.5% of the over-18 population had been fully vaccinated. More importantly, 63.4% of the over-65 population had received at least one dose and 35% are now fully vaccinated. With about 80% of the fatalities occurring in the over-65 age group, this is going to really drive down fatalities.
With the loss of the CTP testing data, we are left with even sketchier date on testing and “new cases.” I have described the significant limitations to this metric below.2 Yet it is still the headline number almost always used by the media. Just to make a complete report to you, I will be tracking the CDC’s “new case” metric. At best, it should be considered as a very rough indication of the spread of the virus. Notwithstanding that it is commonly done, I think it is a serious mistake to use it to calculate derivative metrics, like reproduction and fatality rates. With those caveats in mind, this is the CDC’s chart of new cases. The most significant feature of this chart is that it shows the decline in new cases plateaued in mid-February. That roughly corresponds to the decline in new hospitalizations slowing. There is a little distortion from Missouri changing its reporting methodology and adding over 80,000 older cases on March 8, but the flattening would be distinct even without that anomaly.
In August, I wrote a post that discussed the opposing forces of the relaxation of suppression measures and the growth of population immunity. At the time, those forces were roughly in equilibrium, which I thought might hold. That, of course, turned out to be wrong as the virus got the upper hand in the fall. It was not until just before Christmas that the tide turned and the combination of population immunity and suppression efforts began to reign the virus back in. But the decline over the last ten weeks has been stunning and was not expected by any of the experts.
However, there are some nascent indications that the decline is flattening out. There are probably several factors involved. There is a lot of suppression fatigue. When I go to restaurants, they are normally more crowded than the time before. The variants are likely playing some role as well.
But on the other hand, the effect of the vaccines is just beginning to be felt. The percentage of the population that is fully vaccinated is currently growing at little over 2% per week and that is likely to accelerate over the next few weeks as the J&J vaccine becomes more available.
There are a number of the “experts” who are warning that the US is headed for a new, even worse spike in infections. For example, during a February 15 interview, Michael Osterholm, head of the University of Minnesota’s Center for Infectious Disease and a member of the White House’s COVID advisory board, said, “The next 14 weeks I think will be the worst of the pandemic.”
I would not be surprised to see the level of infections plateauing for a while as population immunity from the vaccines grows. But I do not see how it is mathematically possible for another major surge to occur in the US at this point. There is just not enough dry tender in the forest left to start another major fire, unless there is some mutation in the virus that completely evades the existing immunity. I suppose that is possible, but I cannot find any historical example of where a mutation in a virus substantially extended an epidemic.
Early in the pandemic, I heard an immunologist say, “At the end of the day, bet on the human immune system.” Our immune system has been battling pathogens like COVID for millennia – and winning. And that was before we were able to give it a helping hand with vaccines.
That great American philosopher, Yogi Berra, once said, “Predictions are hard – especially about the future.” Even so, it sure seems like the COVID epidemic, at least in the U.S., is rapidly winding down.
Note 1 – One of the problems with CTP hospitalization data was that some states report lab confirmed hospitalizations only, while others report confirmed and suspected, which CTP lumped together. It also reported ICU bed usage and ventilated patients on a national basis even though many states do not report that metric on their website. The CDC’s website data presentation only shows confirmed cases and does not separately show ICU bed usage and ventilated patients. You can get the detailed reports on the suspected cases and ICUs, but that requires daily downloading files. So, going forward, I will just be reporting the CDC’s confirmed only data, which, at least, is not mixing data types.
Note 2 – It has become very difficult to determine daily test results since CTP shut down. The CDC reports “new cases” as opposed to actual testing results. Also, there was nearly a two million test difference between the CDC and CTP when CTP stopped reporting. There are a lot of problem with the new case metric. First, we know that testing only discovers a portion of the actual infections and that portion has changed over time. The PCR tests return many false positive and negative results and there continue to be delays in testing reports coming in. As a result, every day there are “new cases” being reported where the tests occurred days or even weeks before, as we saw with Missouri reporting 80,000 old cases on March 8. And like hospitalizations, some jurisdictions only report confirmed cases as opposed to suspected cases. As a result, the new case metric should only be used in the most general sense and little inference should be drawn from daily fluctuations.