Tuesday, September 8, 2020

Early College Covid Indications - Lots of Positive Tests, Nothing Bad

Author Daniel Horowitz, writing for the Blaze, summarizes the news on the COVID-19 front and it all seems like good news.  As of September 4th, over 11,000 college students have tested positive for the virus, although that was using the Polymerase Chain Reaction (PCR) test noted for a high false positive rate.  There have been zero hospitalizations and zero fatalities.  Sounds like good news all the way around, as we're in something like Day 200 of the 15 day shutdown to Flatten the curve. 

Remember the goal of flattening the curve? Ensuring that hospitals weren't overrun? Well, what do you call a scenario where thousands of cases result in zero hospitalizations? I'd call it the ultimate flat curve – or downright flat line. Yet rather than recognizing the detection of mild cases among college students as portents of good news, universities continue to sow panic for no good reason.

Dr. Andrew Bostom, a cardiovascular and epidemiology researcher, posted a spreadsheet on twitter of all the positive tests in 17 state university systems as of September 4.  Clearly a small sample of all the colleges in the nation, but the data looks like this:

 Since it's a PCR test, prior experience says the majority of those are false positives.  

Many of them could be false positives, insignificant viral loads, or the dead RNA of a virus that passed weeks ago still being carried around in the student's nasal passages. There is no metric for any of this being monitored in the testing. The irony of the University of Arizona using positive testing of benign cases as the baseline for such draconian measures is that so many of those tests turn out to be false positives. Out of the 13 positive results among members of the university's athletics department last week, 11 of them turned out to be false upon retesting.

That false positive rate (11/13) is  85%, so if we did a worst case analysis and assumed 90% of these college students were really negative, only 10% (1,100) were really positive.   Since it's such a small sample of students and schools, I have hope there really are 11,000 positives out there, although the more people who have had the virus and developed antibodies to it the more likely we achieve "community immunity." 

As always with this situation, the real thing to be afraid of is the State.  

Last Monday, Ohio Health Director Lance D. Himes signed an order requiring even students who test negative to be isolated in a quarantine house on campus. It includes asymptomatic individuals or even those merely "exposed" to a COVID-positive individual. They'd be barred from exiting the quarantine house without written permission from a health official, and individual universities would decide whether parents are even allowed to visit them. This is a mandate for de facto prison – all for an "epidemic" built on false or notional positives with no health risks beyond the ordinary bugs that spread on campuses every year.

By sending your children to Ohio's public colleges, you are essentially sending them off to jail, because it's nearly impossible for them not to be quarantined. Ohio State University is conducting mandatory random testing of 8,000 students each week via their "surveillance testing program." Based on everything we know about false positive or old dead viral RNA, it's a near-certainty that the testing will net dozens of people every week. Now, this order will force numerous friends and dorm-mates to be confined as well.

Because of the amount of latitude variance that the US covers, we bridged two zones in R. Edgar Hope-Simpson's curves of influenza epidemic extent vs. time of year (which has empirically worked for related viruses well - in the sense that it "rhymes," not duplicates). In the first quarter, the worst places to be were high latitude - the sun was low in the sky.  As the sun moved to return to those northern states (April through September), the problem moved south, but the overall numbers of deaths were lower than in the winter; again, similar to Hope-Simpson's work on Influzenza. 

The summer isn't over, so the story hasn't played out, but the following plot shows data as of Sept. 7th. 

(Source: Ivor Cummins in Twitter feed of The Ethical Skeptic.)

This plot is busy, but straightforward.  The top, gray, bar graph is the number of positive tests (they use the word "cases," which I disagree with using since the tests have such high false positives).  Zero is the red line in the middle and reflected below that red line, so that bigger numbers go farther down, is the number of deaths.  They both use date for the horizontal axis.  You can see, for example that in the spring the number of confirmed cases isn't terribly different from the number of deaths.  Because of the CDC screwup on tests, we didn't test people unless we were pretty sure they had the disease. You can see that change completely starting in June when the number of tests tripled while the number of deaths increased but not as much.  Since the peak around the start of August, both positive tests and deaths have been decreasing. 

The obvious good news is that CDC-reported Covid deaths are as low as they were in late March, and headed lower.  Graph at William M. Briggs, "Statistician to the Stars".  Since Gov. Cuomo and other northern governors already killed off the most vulnerable in their states, that high death rate is unlikely to repeat.


  1. It seems like a good idea to account for detectable active disease + positive test, as well as the other results. But that would take away the punch bowl.

  2. This comment has been removed by a blog administrator.

  3. Its clear there must be an “After Action” study into the constitutional abridgments by government and industry as a result of their response to the virus. The courts will be dealing with citizen lawsuits against their cities and states over the individual loss of rights and wrongful deaths for years, and I am certain their will be a Nadler style commission to hang Trump over the medical response by the Trump administration no matter who wins in November.
    But there has to be an accounting over the curtailment of individual rights imposed by government and industry to hold those accountable and to insure they are not repeated in future events

  4. The roughly one-week periodicity on the graphs is clearly an artifact, probably based on reporting requirements. I cannot believe deaths would oscillate like that unless hospitals are adding up deaths over the weekend and reporting them all with Monday's totals.

    1. The data has had that forever and the reason has to be something like that - making up for data not reported. Plus, data is always noisy to some degree. All the sources have that periodicity, but I've never tried to verify the peaks always occur on the same dates.

      Then there's things like the spike on August 31st you'll find at the Florida Department of Health state statistics. Virtually every county - but not all - has a massive spike of positive tests on August 31st. The state went from 1692 on the 30th to 7449 and back down to 2514 on Sept. 1. My county (Brevard) hasn't had a day with over 100 positives in August until the 31st. It went from 32 to 144 to 45. WTF? Has to be an artifact of something.

    2. The shape of the noise in a signal is a strong indicator of problems with the measurement system. I know I'm preaching to the choirmaster, here, but most noise is Gaussian or Rayleigh, depending on the natural processes that create the signal. These case numbers show a recurring peak with a following dropoff that looks a great deal like the decay of a capacitor or inductor when hit with a step function. To me, it reeks of a skewed sampling process.

      Just like the numbers that were reported early on in the scamdemic indicated it wasn't a natural disease progression. The increase rate was linear, which never happens with an exponential spread.

      I called this thing a hoax in early March, and I stand by it.

  5. I meant to add: yes, these artifacts are noise, but it's noise created by bad science, not nature.