Happy last day of the year – I’ll do the full rundown on the numbers below, but first here are my thoughts on what we’ll see next year. Let’s start with what we know now that we didn’t last March – I’d say there are two major revelations.
- COVID is far less deadly than most people thought last spring. I did my first mortality calculations in April, where I calculated a blended mortality rate of about 0.5%, but I was nearly alone in this thinking. By summer, nearly everyone agreed with my numbers, and now it’s common knowledge that the overall mortality rate is a bit below 0.5%, and the mortality rate for those under age 60 is flu-like.
- COVID is far more prevalent than anyone thought at the beginning. Because testing throughout the spring was focused on those displaying symptoms, a vast number of mild or asymptomatic cases were missed. Studies from antibody testing, autopsies, and non-related blood tests showed a prevalence to reported ratio of anywhere from 10 to 50. I do believe that the prevalence to reported ratio has steadily fallen, and I estimate it’s in the 3x range now (based on triangulating from mortality data).
This early view, that of a deadly disease with few cases, led to containment strategies – tracing and lockdowns. But we now know we’ve been battling a far more widespread, but far less deadly disease. Containment strategies have not been effective. In fact, my working theory for months now is that the disease reaches a saturation point in any given population, and then declines, irrespective of containment measures. Early on I estimated this saturation point was known active cases equal to 1/3 to ½ of one percent of the population, but now I think it’s more like ½ to 2/3 of one percent, primarily as a result of the falling prevalence to reported ratio. I believe the reason for the population ceiling is that’s when the disease begins to become resource constrained (herd immunity). For more on this, see my discussion below from August 2, 2020.
So what does this mean for 2020? Of course, the biggest news is the vaccine rollout. My math tells me that we will need far less vaccine deployment to contain COVID than most of the experts are claiming. I think somewhere less than 100 million inoculated will do it. For more, see my December 9th discussion below. However, any additional inoculated people slows down COVID transmission. To see real declines, we don’t have to reach herd immunity – we just need the transmission rate (Rt) below 1.0. Every person who gains immunity, either through contracting COVID, or through inoculation, slows down the transmission rate. As a result, I predict that we’ll see COVID in rapid decline in 2 months’ time, by the end of February, and life will more or less return to normal by spring 2021. After springtime, we’ll still see breakouts in certain communities based on vaccine acceptance, but it will be even more geographically localized than it is now, and these clusters will burn out quickly.
We have already seen known active cases decline since mid-December. Here is the active case graph. Since mid-December, we’ve shed about a quarter million known active cases. Of the 22 states that I track, only one is still increasing – New York.
Let’s look at the peak numbers for the various states. All of the states from PA left have shown definitive peaks in November or December. Of the states to the right of PA, 5 states have peaked within the past week, so I don’t trust it’s over yet there. These are NC, SC, TX, GA, and VA. New York is still increasing, and NJ, FL, MA, and WA have peaked earlier. The brown line is the average of the high water marks, equal to 0.58%. My conclusion is that the states on the left have the worst behind them, and the states on the right may yet see more growth. Of course, in another 4 weeks we’ll start to see the effects of the vaccine rollout, but January could still be difficult for the states on the right side of the chart.
Below is the national daily death count. The brown line is my projection or peak deaths. This projection is based on 2.75 deaths per 1,000 known active cases with a 21-day lag. This data point has been very stable, but December has been slightly lower than the previous 4 months. As a result, my projection “might” be a little high, but we won’ know until about January 7th, when we’re confident the holiday backlog of death reporting catches up. In any event, I’m expecting daily deaths peak about January 7th or 8th, then begin declining. The current IHME model (updated December 23rd) is very close, projecting peak deaths on January 11th at about 4,000 per day, vs. my projection of about 4,100.
Here is Arizona, a fairly strong decline since the peak on December 14.
SC has more than doubled since the beginning of December. It’s showing signs of leveling off, but still a bit low to have confidence in the peak.
Here is Florida – never reaching July levels, and now down since the 23rd. This could be just irregular reporting during the holiday – we’ll know by the end of next week. This will be true for several states that show declines in the past week, but not before.
California is also showing declines since December 22nd. We’ll have more confidence in this next week, but the peak is high enough to expect it will hold.
Georgia shows a peak on December 25th which could be solely due to holiday data reporting. The peak is also lower than average.
Texas data is a mess, so I can’t reliably interpret anything here. Texas added over 100,000 cases on a single day (Dec 11), which comprised new “probable” cases going back in history. They provided a new time series going back to November 1st only, so they show a huge spike on November 1st (completely unrealistic). As a result, I have no idea what is going on here.
Both VA and NC have declined lately, but no conclusions until next week when the reporting stabilizes.
Washington data is worse than Texas. They have seen a dramatic drop in 2 weeks, but here is the disclaimer (from the COVID Tracking Project):
On December 18th, Washington revised down their total test numbers from 3,432,892 to 2,765,404. This might relate to the current note on the dashboard, stating: “Today’s total case counts may include up to 1,000 duplicates. Negative test results data from November 21, 2020, through today are incomplete, as are positive test results from December 16, 2020, thus testing and case numbers should be interpreted with caution. The Epidemiologic Curves tab is the most accurate representation of COVID activity and is updated daily as new cases are identified and duplicates are resolved.” We are continuing to report the total tests to figure from Washington’s dashboard and will continue to investigate and adjust this figure if necessary.
On December 16, 2020, Washington added all Probable cases reported since June, 2020 to their data for December 16, 2020.
On December 11, 2020, Washington Deaths decreased from 3016 to 2850 with no explanation.
On December 5, 2020, Washington announced that up to 90 of the deaths reported “yesterday” were incorrectly classified and were not due to COVID-19.
Here are NY and NJ – NY continues growth, although slower than early December, while NJ is now in decline for 3 weeks.
Here is Massachusetts. Rapid post-thanksgiving growth, then gradual decline for 3 weeks.
…And here is Michigan, with a dramatic December decline. The high water mark was just a hair over 0.58%, so I think that was the top.
Here is PA with a nice 2 week decline.
Here is Colorado – with just about 1/3 the COVID they had in late November.
Here is Illinois, similar to Colorado – dramatic recovery over the past 6 weeks.
Here is Wisconsin, same pattern.
Here is Alabama, declining for a week, but we won’t know how to interpret this until next week sometime.
And Tennessee… Rapid decline beginning December 21st, and at a high level, so I think this is real.
Here is Ohio – big declines since mid-December.
Here is Indiana, declining since early December.
And finally, here is South Dakota, with about one-quarter of the COVID they had in November.
So that’s it for today. I’ll report again next week sometime.
–Shane Chalke, FSA