PCR & Antigen Cases Spring/Summer 2021
Note that I have switched to reporting Total Tests Reported (via ELR), which means PCR and Antigen testing combined.
|Antigen Cases||Total Cases||Total Tests|
|% Positive |
PCR & Antigen Cases Fall/Winter 2020
|Antigen Cases||Total Cases||Reported |
|% Positive |
* Antigen cases for the first two weeks are the net change in total antigen positive cases from Friday to Friday (10/23 – 10/30 and 10/30 – 11/6), which is the time period used on the the County Indicator Reports before antigen positive cases started being reported daily, beginning 11/3. All other weekly totals use the net change from Saturday to Saturday, which is how my weekly reports are always calculated.
** Recent Data Dumps
The state is working to get providers (like urgent care networks) reporting on a consistent basis instead of in large batches. However, data dumps do occur and can inflate weekly numbers.
- During the week of 3/13, the data dumping from previous weeks slowed down, but there were still over 1,000 old antigen cases reported and over 1,000 old PCR cases reported.
- During the week of 3/6, it appears that the addition of old confirmed (PCR) cases has stopped, but antigen cases are still being dumped in. It slowed by the end of the week, but has not stopped entirely.
- On 2/23, old cases, particularly from Dec 2020 and Jan 2021 started being reported, causing an artificial increase in daily cases by about 20-25% daily for several days in a row. (As of 2/28, this daily case dumping is still occurring.) It is possible these cases are coming from the dump of old tests reported on 2/18.
- On 2/18, there was a got a huge dump of old test data (est. ~25K old tests), but without an equivalent large dump of cases. As a result, I removed this day’s data entirely from the weekly test positivity calculation. Reported testing is artificially high for this week.
- On 11/21, there was a large data dump of antigen positive cases from the past 2 weeks.
- On 11/17, there was a huge dump of PCR tests and confirmed cases from the previous 2 weeks.
- On 11/10, there was a data dump of around 1000 antigen cases from October.
- On 1/6/21 and 1/8/21, there were large dumps of non-ELR cases, so testing this week is underreported by over 50K tests. On 1/6, over 800 of the new cases were prior to October.
- On 2/5-2/7, there was a data dump of old tests in Clayton County, artificially increasing their case rates.
PCR Testing & Cases: June – October
|New Cases||Reported PCR Tests||% Positive PCR Tests |
|June 6||5,028||68,019||4.3% (est.)|
% Positive Methodology
Starting with the week ending June 6, we know the number of PCR and antibody tests without estimating because DPH began providing total viral tests and antibody tests daily. However, we learned on June 2 that testing totals only include tests that are reported electronically to DPH. Some cases are reported to GA DPH outside of the electronic reporting system via fax/email/phone, and they don’t know how many total tests were done to identify those cases. As a result, I now calculate % positive from electronically reported testing numbers only.
NOTE: The % positive for the week of June 6 is an estimate because we don’t have electronic reporting data for the full week.
*** Adjusted % Positive Due to Data Dump
- Data dumps from labs that use the Electronic Lab Reporting system artificially inflate testing, cases, and % positive numbers by reporting a backlog of old test data all at once.
- On 8/24, a very large backlog of old tests (over 20K according to one source) was reported from Navicent Health. These old labs generated over 10,000 tests and almost 800 new cases in Bibb County alone. It likely also affect numbers for nearby counties, including Baldwin, Houston, Crawford, Twiggs, Monroe, Jones, Wilkinson, and Peach. However, the majority of cases were from July or earlier.
- On 9/16-17, 9/19-9/21, and 9/30, there were large data data dumps from a NE GA lab. The backlogged tests seemed focused in four counties: Hall, Habersham, Stephens, and Union.
- Testing and case numbers for these weeks are inflated as a result of the data dump, but it’s unclear by how much. As a result, I did not adjust the case and test numbers.
- For the week ending August 29, I excluded the data from August 24 when calculating % positive for the week.
- For the weeks ending September 19, 26, and October 3, I excluded the test data from these four counties on the days of the data dump when calculating % positive for the week.
Spring Testing Data Prior to Antibody Tests Removed
Estimated Antibody Testing
In late May, it was discovered that GA DPH was including antibody testing in total test numbers. On May 27, they reported that 77,835 antibody tests are included in the total tests. I deducted a rough guess of the number of antibody tests that each week from April 19-May 30 might include, weighted much more heavily to later weeks, assuming that antibody testing expanded as as total test numbers surged in Georgia, in order to estimate percent positive. Even after removing estimated antibody tests, there were still huge increases in testing from May 10 – 23. Numbers in italics are based on my antibody testing estimates.
|New Cases||PCR Tests||Antibody Tests||% Positive PCR Tests|
|16,198 (prior to 5/27)|
6,968 (after 5/27)
Total: 23,166 (27%)