Advertisement

Eyewitness News looks into discrepancies between state, national COVID-19 data

Published: Oct. 23, 2020 at 1:11 PM CDT
Email this link
Share on Pinterest
Share on LinkedIn

TOPEKA, Kan. (KWCH) - The Kansas Department of Health and Environment reported Friday 1,774 new COVID-19 cases since Wednesday, bringing the statewide total to 76,230. The department reported an additional 23 deaths since Wednesday, with the statewide total now at 975, and a percent positive monthly average of 8.1%.

But that last number is a drastic difference that is being reported by The COVID Tracking Project, a national volunteer organization launched from The Atlantic that collects and publishes data “required to understand the COVID-19 outbreak in the United States.”

The group reports Kansas with one of the highest increases in coronavirus cases across the country with a seven-day rolling average spike above 20%. That means one in five people in Kansas test positive for the virus. KDHE data shows the daily tracking rate of positive cases lower than 20 percent and since July, it has never shown a daily percentage of more than 11%.

So which the difference? The answer is complicated.

Coronavirus cases are tracked differently by different agencies. Eyewitness News has used the KDHE dashboard since the beginning of the pandemic.

We asked Governor Laura Kelly about the discrepancy?

“There’s a lot of research groups, organization groups that are doing tracking," she said. "What we have to do in Kansas is pick one and use that because that’s the data that will tell us the differences that are occurring, rather than try to use all these other samples that come in. It’s better for us to use the data we trust.”

Here at Eyewitness News, we take this data seriously because it influences decisions on businesses, schools, and ultimately your family. We reached out to John Hopkins University that works closely with the National COVID Tracking Project.

According to its Frequently Asked Questions (FAQs), The COVID Tracking Project says it obtains almost all of its data “directly from the websites of local or state/territory public health authorities. Where data is missing from these websites, we supplement available numbers with information from official press conferences with governors or public health authorities.”

The group says its numbers may not match state-provided data due to: date lag, hidden data, different data definitions, Different ways of reporting “new” cases, tests, or deaths and backfilled/backdated data.

Johns Hopkins University issued the following information:

About the Data

  • New cases are presented as daily counts as reported by the state; for smoothed data presented as a 7-day rolling average, click here. Due to fluctuations in daily reporting, testing rates are presented as 7-day rolling averages.
  • As guidance evolves on Covid-19 case reporting, some states are modifying their reporting to include both confirmed cases, based on laboratory testing, and probable cases, based on specific criteria for symptoms and exposure reflect. This may cause new case data to “spike.”
  • It is important to note that the quality of testing data varies by state. Click here for more.

Data Sources: Testing data from The COVID Tracking Project and cases data from JHU CSSE.

Our data provider, The Covid Tracking Project, is in the process of changing how it maps states' data to the categories we use for our positivity calculations. These changes mean the category of data we use in our denominator (Total tests) may now include tests previously not included in our calculations, which in turn may result in a test positivity calculation that is lower than what we would have calculated for the state prior to the change.

With this significant change, we will once again review our data inputs and calculations to ensure that our numbers reflect the most responsible public health calculation of test positivity.

HOW WE CALCULATE POSITIVITY:

Positivity Rates: Our calculation, which is applied consistently across the site and predates most states' test positivity tracking efforts, looks at number of cases divided by number of negative tests plus number of cases. We feel that the ideal way to calculate positivity would be number of people who test positive divided by number of people who are tested. We feel this is currently the best way to track positivity because some states include in their testing totals duplicative tests obtained in succession on the same individual, as well as unrelated antibody tests. However, many states are unable to track number of people tested, so they only track number of tests. Because states do not all publish number of positive and number of negative tests per day, we have no choice but to calculate positivity via our approach. We describe our methodology as well as our data source (COVID Tracking Project) clearly on the site.

7-Day Averages: The CRC calculates the rolling 7-day average separately for daily cases and daily tests, and then for each day calculate the percentage over the rolling averages. Some states may be calculating the positivity percentage for each day, and then doing the rolling 7-day average. The reason why we use our approach is because testing capacity issues and uneven reporting cadences create a lot of misleading peaks and valleys in the data. Since we want to give a 7-day average, it is more fair to average the raw data and then calculate the ratios. Otherwise, days when a large number of negative tests are released all at once—and positivity is going to be very low—will have the same weight as days when data was steadily released, and the overall result is going to be lower. Our approach is applied to all our testing data to correct for these uneven data release patterns.

Positivity rates can tell us whether a state’s testing capacity is sufficient. Ideally, a state should be meeting or exceeding the recommended positivity rate, which the WHO has set at 5%. A positivity rate over 5% indicates a state may only be testing the sickest patients who seek out medical care, and are not casting a wide enough net to identify milder cases and track outbreaks.

Percent positivity can also help us determine if an increase in cases is simply the result of expanded testing or if it signals increased transmission of the virus. If we see the percentage of positive tests begin to rise, it indicates insufficient testing to find infections that may be occurring. Not finding these infections may mean that the virus is transmitting without intervention, which can lead to future case growth.

Specifically:

  • If a rise in cases is the result of increased testing, the percent positive line could look flat or like it is decreasing over the time period when cases increased.
  • If a rise in cases is the result of increased transmission, the line could appear to be increasing over that same time period.

Per the Kansas Department of Health and Environment:

To calculate the positivity rate for Kansas, KDHE uses the number of positive tests by laboratory test date divided by the number of negative tests for that laboratory test date to come up with the positivity rate for Kansas. We have 8.1 percent positivity in October thus far. The average we report is not significantly different than the daily positivity rates.

Other sources, such as the COVID Tracking Project, may use different formulas (methodologies) to come up with their own numbers for positivity rate. I believe their rolling average is what you are inquiring about. I cannot speak for why theirs is different --  I encourage you to reach out to them to understand how they are reporting and why they are so different from us. Here is a link in their data explainer that may help you: https://covidtracking.com/about-data/faq#why-doesnt-your-data-match-what-i-see-on-the-official-covid-19-page-for-my-state. The Backfill section, in particular, may help with the explanation.

When KDHE reports our numbers on Mondays, Wednesdays and Fridays, there is a cumulative increase. For example, let’s say– 7,000 negative tests/1,300 positive tests reported since Wednesday (making up an example). However, those 8,300 tests were not all administered between Wednesday and Friday. They were reported to us during this time frame, but could have been much earlier.  It’s a more accurate picture of the disease progression to look at things by the laboratory test date, which is why we do that.

I encourage you to look at our Testing tab on our dashboard. We have the positivity rate by month listed and there is a chart where if you hover over the columns, you can view the positivity rate day by day as well as the number of positive and negative tests.

Copyright 2020 KWCH. All rights reserved.