Seven questions people ask about our daily COVID-19 data dashboard

The recently expanded “Daily COVID-19 outbreak summary” shows the latest data for test results, hospitalizations and deaths in King County and in each city, town and zip code. Dashboard users can use “tabs” to filter by race, age and gender for each location. They can also compare how each city or zip code area compares to the overall rates for King County. (Most unincorporated areas are also included, and more will be added.)  

The dashboard includes demographics for race, age, and gender.

(Don’t miss the other COVID-19 data dashboards, showing the status of the pandemic and impacts on the community.)

Keep in mind that the dashboard is a snapshot at one moment in time – it’s not a final report. New information is constantly arriving, and when we update any of the data dashboards, it represents a snapshot of what we knew at that moment.

These are some common questions people ask about the daily outbreak summary, with answers from our lead epidemiologists.

Why does data that was posted a day (or a few days ago) sometimes change?

Investigations into COVID-19 cases and deaths are ongoing. And Public Health’s epidemiologists are continually improving the accuracy of the data we receive from multiple sources. Every time we find a test result that had inaccurate data about an individual, we correct the data, and that leads to adjustments on the data dashboard.

In addition, some information takes longer for labs and hospitals to collect and send to Public Health.

For example, hospitalizations and deaths may not be reported immediately. If someone dies today, the COVID-19 test may happen after the death, and then another day or two would pass before the test result is available from the lab.  In contrast, someone else may die today, but that person was tested a week before dying. Because they already tested positive, that death would get recorded immediately as a COVID-19 death. Both individuals died the same day, but they are added to the dashboard on different days, when the information becomes available to Public Health.

Hospitalizations related to COVID-19 may require four to five days to confirm. The process involves a case investigator interviewing a person who tested positive for COVID-19. The investigator learns that they had been hospitalized, and then the case is matched to hospital data. Sometimes, a person who tests positive was in the hospital data – but it turns out the reason for hospitalization was unrelated to COVID-19 (such as a car crash). This process of matching and correcting might take four or five days after the positive test was first reported on the dashboard. 

Why would test results include inaccurate information? 

Test results begin with samples taken at many different testing locations (health care providers), and then sent to many different laboratories. Each location collects the data differently, and it’s common for a test result to have incomplete information. If you’ve ever skipped a question or left a field blank, you understand how this can happen.

More than a dozen different labs send results to the Washington state Department of Health, which merges them together. When Public Health receives the data for King County residents, the epidemiologists then cross-reference the results with data from hospitals, geographic systems, and other databases that can help fill in missing data. During this process, over the course of a few days, individual errors are corrected and gaps are filled.

A wrong address is one of the most common errors:

  • If a person’s address field was left blank, the laboratory will insert the location where the test happened.
  • If an address is along the border of two cities or counties, it might initially get mis-categorized.
  • If a person is living in a long-term temporary location, such as a nursing home, the form might still have their previous permanent residence. But it’s important to record illnesses acquired in long-term care facilities.

As the case investigation leads to more accurate information, we update the original data. When this happens, a positive case or a death might be correctly moved from one city to another, and the date might change. 

Why does the race and ethnicity data have higher levels of uncertainty than other data?

Typically, about 60% of all lab results are missing race/ethnicity. This is one of most common fields left blank on the original form collected at testing sites.

For people who test positive, a health department case investigator will reach out to the individual to interview them. During the interview, they will ask about race/ethnicity. Some people decline to answer. After the interviews, about 20% of positive cases are still missing race/ethnicity. When there are high rates of missing data, then calculations based on the remaining information are less precise.

Other uncertainties come from people who are of mixed race. They may choose to answer the question in many different ways.

Why don’t the total numbers for all the cities add up to the total for King County? 

These are three common reasons:

  • No street address: Some King County residents don’t have a street address. For example, they might be homeless, or they might use a Post Office Box as their address. They will be included in King County totals, but not in any city. To make things more complicated, people with a P.O. Box are included in zip code data, but they can’t reliably be added to a specific city without having a precise street address.
  • Unincorporated neighborhoods: Many people live in unincorporated areas, and the county total includes all of the cities plus unincorporated areas.
  • Subtractions: Sometimes, the column called “new since yesterday” includes not only newly reported individuals, but also data corrections – which include subtractions! If a specific case gets removed from a category (such as deaths or hospitalizations), or assigned to a corrected address, “new since yesterday” could include a subtraction in one city and an extra addition in a neighboring city. Some cities even cross over into neighboring counties, so a small correction in an address can move someone in Bothell from King to Snohomish County. 

Are there any other reasons why it might appear your numbers are off?

Our daily dashboard is updated with data as soon as it becomes available, and there could be errors on any given day, as Public Health receives data that is incomplete or not yet fully investigated. Please be aware improvements may look like mistakes – such as address corrections, or updated info about hospitalizations or deaths. Sometimes, there may be differences in an underlying data source that is not publicly available that may cause differences in calculations on any given date. 

When comparing one table or map to another, beware that the rates may be presented with different denominators in different tabs. And it’s always worth clicking “refresh” on your web browser to make sure you are seeing the most current information.

Why aren’t the results from antibody tests included in the data?

These tests don’t show whether someone is currently infected – so they don’t help us know how much the virus is spreading at this moment in time. And not all antibody tests that are available are reliable.

An antibody test is a blood test that shows the presence of antibodies after someone has been infected and recovered, but the antibodies are only detectable weeks after the person was infected. Antibody tests don’t tell you if you are infected now – only if you were at some time in the past. The infection date cannot be determined from an antibody test. In addition, there are many antibody tests on the market, and not all of them have been validated to be accurate. Adding false positives would also distort our picture of the outbreak. 

Reliable antibody tests are helpful to inform Public Health about what portion of the overall population has had exposure to novel coronavirus (SARS-CoV-2). In the future, when antibody tests become more utilized and more reliable, we intend to include that information separately.

How often are the other data dashboards updated? 

Each dashboard is on a different update schedule – which is now noted near the top of each screen. They are all compiled and checked manually, to catch many of the errors that would be included if they were automated.

Originally posted 6-23-20