Ending my own data analysis

As of this weekend, the Oregon Health Authority will no longer provide Covid test and body count data on the weekend.

Since the end of March 2020 (initially OHA claimed they could not publicly disclose data), I have made daily updates and at the peak, produced up to over 40 charts analyzing various aspects of the state’s pandemic situation.

In the fall I discovered the state was calculating the percent positive value incorrectly. When cases were low, no one noticed the error. When daily news cases increased by a factor of 10 the error became obvious. While I did not contact OHA, a TV station did contact them and they identified the error. Overnight, OHA went from blaming the public for not following guidance to thanking us for following guidance.

In December I noticed a discrepancy in how the state reported deaths. By mid to late January, it looked like January deaths would exceed December deaths. This occurred because OHA’s body count was based on “day it was reported” but not the “actual day someone died”. Because reports were delayed by a random amount, this time shifted all death counts into the future. As OHA was reporting increasing deaths in January, actual death counts were going down. I did inform them – but did not receive an acknowledgement. However, about 2 weeks later, they began publishing a chart showing corrected “Deaths by actual date” rather than by “day of report”.

(Many states made this same error. Some call it out, publicly, in their reports, while others pretended it did not matter.)

Using the charts I made, and other calculations, I spotted peaks and troughs 4 weeks in advance of official acknowledgement of those peaks or troughs. I correctly projected the large drop in cases in January (while the experts were projecting “surges upon surges”).

I may do some minor updates over the next few days but plan to discontinue my analysis. (1) OHA is no longer doing updates on the weekends, (2) case counts have fallen, hopefully for good, (3) many are now vaccinated, (4) there have always been many revisions to past data that skew this chart by perhaps 3% from real numbers. People who previously died turned out not to have died. Some turned out to have died of non-Covid and sometimes other deaths were re-classified. Daily test counts fluctuate constantly going back months. There is no ways collate the corrections. Finally (5) this is part of scaling back paying any attention to public health as I no longer believe anything they do or say.

The “core chart” showing daily new cases, hospitalizations, ICU and Ventilators in use counts. Previously I tracked deaths by actual date of death – but in January when their data reporting error on deaths was identified they initially stopped reporting actual death dates for a period. There was an outcry from the media and some politicians and they restored that information but I had discontinued tallying the deaths by then.

This next chart helps identify changes in the trajectory of the pandemic. Growth in new cases (and also R-0) peaked in November. While daily cases were still going up, the increase each day over day was becoming smaller. The news media, and apparently the epidemiology experts, only watched the daily number, not the rate of change – and called for a Christmas surge that never happened. In fact, deaths peaked the week of December 6th – not late January as they initially implied. OHA later acknowledged there was no holiday surge in Oregon (and as I have shown elsewhere on this blog, there was no holiday surge anywhere in the country. Myth busted.)

Anyway, this chart was critical to spotting changes in the trajectory.

This chart gives an indication of whether new cases of disease are getting worse or getting better. At the far right we see that there is a slight increase in hospitalizations – but this is because daily new cases have plummeted while some patients remain in the hospital for an extended period of time.

Initially, this ratio was very high – possibly because we should have been doing 5x more testing in the spring of 2020. However, the CDC insisted on a government lab monopoly which failed to provide adequate test availability (and produced erroneous test results throughout February due to now documented incompetence). Eventually, private testing was permitted and gradually became more available.

Hospital beds in use, ICU beds in use and ventilators are the best metrics we have. Daily positive test cases are useless for short term (2+ weeks) due to fluctuations caused by such things as testing all college students in the state during the first week back in school, or decreased testing on weekends. These activities skew daily counts.

As of today, statewide, there are 17 people on a ventilator, and all of those are in ICU, representing over half of the Covid patients in ICUs. Also, as of today, the state has over 790 ventilators available, about the same number as in the spring of 2020. Remember when we were going to run out of ventilators and the government spent billions having General Motors make 200,00 ventilators that were never used?

As of today, just 2% of the state’s ventilators are used by Covid patients.

This is why I tracked the data myself. Much of the public health hysteria was uncalled for, not true, out of date, and some like death reporting, was simply wrong.