Chapter 4: Examples of COVID-19 Real-World Evidence Studies

Diagnostic Test Utilization Study

Study citation: Mack CD, Osterholm M, Wasserman EB, Petruski-Ivleva N, Anderson DJ, Myers E, Singh N, Walton P, Solomon G, Hostler C, Mancell J, Sills A. Optimizing SARS-CoV-2 Surveillance in the United States: Insights From the National Football League Occupational Health Program. Ann Intern Med. 2021 Aug;174(8):1081-1089. doi: 10.7326/M21-0319. Epub 2021 Jun 15. PMID: 34125571; PMCID: PMC8252091.7

Study objective: Evidence to understand effective strategies for surveillance and early detection of SARS-CoV-2 was limited throughout the first year of the pandemic. Limited data have been available describing frequent testing for SARS-CoV-2 in a closed population. The purpose of this study was to describe a large COVID-19 testing and surveillance program.

PICO: The National Football League (NFL)/NFL Players Association (NFLPA) COVID-19 Testing and Surveillance Program instituted routine testing in a closed population of NFL players and team staff, most of whom were tested daily. Several diagnostic tests, including point-of-care (POC) antigen testing and reverse transcription polymerase chain reaction (RT-PCR), were used. Test results were compared against final clinically adjudicated case status, as well as across diagnostic machines (e.g., Roche Cobas compared to Hologic Panther and antigen POC compared to RT-PCR). Outcomes were the agreement of test results and descriptive statistics of test results, including Cycle threshold (Ct) values.

Data source: This large observational study included 11,668 NFL players and staff

who underwent testing as part of the program between August 1, 2020 and November 14, 2020. Individual characteristics, symptom data, longitudinal testing data, and case adjudication associated with 632,370 tests were included.

Study period: NFL Season, August through November 2020

Key sources of error and how they were handled: This study drew from a real-world program performing daily RT-PCR tests in a large, closed population with almost complete clinical follow-up for all individuals testing positive, serial testing after an initial positive result, and case adjudication. This, combined with frequent testing and same-day testing on multiple platforms for those with positive tests, provided transparency into the utility of newly authorized diagnostics, notably showcasing the ability of RT-PCR testing to detect early and asymptomatic or presymptomatic infection. Ct values are not often used in real-world application, and high values are often disregarded as not true infection; this study showed that quantitative values proved useful for understanding test results and that high Ct values can signal early infection. Importantly, because antigen POC tests were often administered on the same day as RT-PCR tests and positive results were clinically adjudicated with serial testing, the frequency of false results for POC testing could be evaluated. This study is limited in its generalizability; notably it was performed within a healthy working population and is not diverse in gender (largely male). Race and ethnicity data were not collected. Additionally, application to non-daily testing environments should be considered; these real-world results within a surveillance system administering daily testing in an ideal setting may not be applicable to for-cause testing but are useful for understanding the utility of the diagnostic tools.