Mystery Solved

Dear Members of the Baylor College of Medicine Community:

This is not my weekly message, but a quick follow up on last week’s message. Recall last week, I highlighted the following data mystery:

“The state of Texas tracks the total number of people in the Houston metropolitan area who are tested on a daily basis. Over the past 6 weeks, that number has averaged around 12,000 tests per day, and has ranged from 8,000 to 14,000. Suddenly, about a week ago, we saw a sharp increase in the number of people being tested. Over the past 4 days, this has ranged from 24,000 to 31,000, a doubling or trebling of the baseline. The mystery is when I look at the data from individual testing sites … the demand for testing, and the number of tests performed has actually decreased. What is the source of the very sharp increase in testing?”

My hypothesis at the time was this was driven by testing at colleges and universities. Many of you wrote to me with other good and plausible explanations.

It turns out we were all wrong. The Texas Department of State Health Services serves as the primary data aggregator for the state. They receive data in a variety of formats (including faxes), which creates challenges in cleaning up and error checking the submissions. Any error in the data results in a delay in posting results. Most notably, the system was only built to handle 48,000 reports per day, a number the state has been exceeding regularly. This has created a backlog of reporting. The tests are being performed, results are generally available for clinical decision making, but many were not logged into the public data system.

On Aug. 1, the state upgraded its system to increase its capacity from 48,000 to 100,000. Since that time, the testing numbers shot up as backlogged tests were entered, resulting in the apparent upsurge in testing numbers. It is unclear – and we are trying to clarify – when the backlog will be clear and the data will be accurate.

Also per last week’s message:

“I bring this up not because this mystery is particularly important. Rather, it is because it has been so difficult to get a definitive answer. Public health systems are relatively fragmented and data systems variable and poorly integrated. This is not a criticism of our public health officials, who across the country are doing an admirable job in challenging circumstances. But it does cause me to reflect on our national preparation for this pandemic, and more importantly, our preparation for the next one.”

There may be a silver lining to the data cloud. Our community case rate for Harris and surrounding counties has been stubbornly high (our target is 200 or less, and we are several times that level). It is possible some of this reporting backlog has been falsely elevating our community numbers. Perhaps we are doing slightly better than the data would indicate.

For a summary of the data issue, refer to this article from the Texas Tribune.

James McDeavitt, M.D.
Senior Vice President and Dean of Clinical Affairs