No, you’re not crazy and you’re not doing the math wrong. The ADH dashboard went a bit wonky on Saturday. ADH is doing some database cleaning this weekend that will result in big changes to our overall numbers and to certain counties in particular. Dr. Jose Romero is scheduled to address this in a press conference this week, but I’ll give a quick rundown of the reasons for the changes I gleaned from my conversation with Dr. Austin Porter from ADH.
Removing records related to people who live out of state
Dr. Porter said the state will remove “a little over 1,000” positives from people who have been determined to live out of state. Most of these were people who live in border counties of neighboring states that were tested in Arkansas, he said. A good portion have been counted under the “Missing County” category on the state dashboard. (You can see that on Saturday, the state removed 1,093 positives from this category.)
At the heart of much of this is a system wherein the Arkansas Department of Health may receive very little information from private labs. For example, a lab may only send the state the person’s name and birth date – no phone number, no address, no other contact information. The state first tries to match up the person using its own resources for contact tracing, but if unsuccessful, sends the information to a third-party vendor for address information.
These cases go back to May, Dr. Porter said. I asked Dr. Porter why this is happening all at once instead of over the last few months. He said this is something the state has always known is a possibility and they send the data to the third party in batches with thousands of names, then have to wait sometimes weeks to get the information back.
The state notifies the neighboring states of positives for their residents when it learns the correct address, and vice versa. The notification system between states “has gotten a lot better,” Dr. Porter said.
In addition to positive cases, the state is also removing negative results and recoveries associated with out-of-state cases.
Removing duplicate records
The state endeavors to ensure that each positive or negative test reported equates to one person, regardless of how many times that person is tested (or how long the time lapse is between the tests). The same third-party vendor referenced above was able to help the state confirm duplicate records. For example, if the state gets a positive result from Jon Smith one day, and a second positive result for Jonathan Smith the next week, the vendor confirmed whether or not Jon was Jonathan for de-duplication purposes.
All together, ADH plans to remove around 44,000 negative results that were out of state or duplicates.
Adding negative results from private lab
The above removals are offset by the addition of around 50,000 negative test results from a private lab that wasn’t syncing with the Arkansas Department of Health system. The lab sent the test report to the patient/physician, but the information wasn’t processing through to the Department of Health. Therefore, while the patient knew they were negative, their results weren’t being counted in state or county totals or figured into percent positive rates.
NWA lab issue
Finally, there was a separate issue with a lab in Northwest Arkansas from which ADH also wasn’t receiving information. Dr. Porter said he thinks fixing this issue may help even out the persistent testing discrepancy between Benton and Washington counties.
Clean data-set forthcoming
Dr. Porter said he plans to release a clean, historical data-set this week. I’m hopeful this will go down to the county level, so we can have an accurate overview of individual trajectories and changes over time. Until then, the charts on my website and others are going to look really, really wonky.
The fact that these data errors existed is problematic. We knew there was something off about the Northwest Arkansas data, and for months we were told that Washington County residents just weren’t seeking testing as frequently as Benton County residents. In fact, it seems likely that the results weren’t being either communicated to or by the Arkansas Department of Health. It’s not ideal.
But here’s the thing: the leadership at the Arkansas Department of Health has changed since then. We have a much stronger team of contact tracers in place with a higher likelihood of catching these issues. Data communication from ADH has improved in both big and small ways since Dr. Jose Romero has taken the helm. I don’t know if this clean-up is his effort or someone else’s, but it’s a correction of errors that were birthed under different leadership and are being fixed under his.
I don’t begrudge anyone’s right to look back and point out past mistakes. But I’m inclined to acknowledge the state’s attempt to correct its mistakes and not to continue or to cover them up.
That said, I’ll be watching the data over the weekend to ensure the changes make sense when implemented and provide more detailed explanations about what exactly those changes look like and mean to the state and individual counties.