From April until the first week of December, I had been plotting the positivity rate for COVID-19 tests on my blog. These numbers were calculated from the regular updates provided by the Pennsylvania state and Allegheny County numbers on the number of tests and the number of positive test results. Through the Spring and Summer the numbers published were consistent with what mainstream media publications were reporting. However, in the fall while some media reports were consistent with the positivity values I was calculating, other reports were discrepant.
The Allegheny County Health Department has been publishing daily updates on their Facebook page that include wording such as this:
This is the COVID-19 Daily Update from the Allegheny County Health Department (ACHD) for December 7, 2020.
In the last 48 hours, 1,470 new cases were reported to the Health Department. Of these, 1,461 are confirmed cases from 4,661 new PCR tests.
The positivity rate for my blog was calculated by dividing the number of confirmed new cases by the number of new PCR tests.
$\frac{1461}{4661}=31\%$
Many news outlets were also using this method in some of their articles, including the Post-Gazette, the Pittsburgh Tribune-Review, and WPXI.
However, the Pennsylvania state health department has been reporting positivity rate numbers that are consistently lower than the result from this method at its Early Warning Monitoring System (EWMS) Dashboard. For example, between November 27 and December 3, it reported a positivity rate of 12.6%.
In the weekly COVID-19 update from the Allegheny County Health Department on December 2, 2020, Health Department Directory Dr. Debra Bogen provided an explanation for the discrepancy. Dr. Bogen stated that the EWMS positivity numbers are the ones that should be reported and used.
When the county refers to as ‘new PCR tests’, it is really referring to the number of people tested for the first time. This is because the data is deduplicated so if a person is tested multiple times they are only counted once. The county should use different nomenclature in their daily updates to communicate more clearly.
The number of negative PCR test results reported for Pennsylvania by the state health department is deduplicated in the same way. Therefore the state positivity numbers on my blog were also higher than what was reported by the EWMS.
As of December 8, 2020, 6,114,009 PCR tests have been administered in Pennsylvania and 3,363,022 people in Pennsylvania have received at least one PCR test. Therefore 26% of Pennsylvania’s population has received at least one PCR test, and a substantial number of Pennsylvanians have been tested multiple times.
As a result, a substantial number of new test results are for people who have previously been tested, but these tests are not included in the number of new tests reported by the county.
The EWMS only reports positivity numbers every Monday for the seven day average ending the preceding Thursday, and the average positivity rate for the seven days prior to that. Earlier positivity rate data is not retained on the website. Therefore I am unable to reconstruct a plot of the positivity rate over time. I emailed the state health department which responded that this data is not publicly available at this time.
Throughout the months of August, September and October the state health department published a weekly press release that listed the positivity rate for the state, as well as for every county that had a positivity rate above 5%. Allegheny County’s positivity rate was below 5% for the entirety of those two months.
I searched through the websites of local papers to find articles such as this one from the Post Gazette that record the EWMS positivity rate for Allegheny and/or Westmoreland County for previous weeks. I was able to develop the following plot that partially reconstructs the past positivity rates.
Pennsylvania has also published the number of total test samples collected each day here. The data is for the date the test samples are collected. The archived data for new cases per day is the number of new cases reported, which occurs after the samples have been processed. I divided the number of new positive test results reported per day by the number of test samples collected per day to calculate an estimated positivity rate. The following figure compares this result with the official positivity rate reported by EWMS.
Data is available to calculate the estimated positivity rate at earlier dates than I was able to find EWMS data.
Going forward with my daily COVID-19 number updates I am just going to publish the EWMS data. I am going to record the positivity rate every Monday as new data is published. My blog will be the only website I am aware of that has a historical record of past positivity rates reported from this source.
