Background: Accurately identifying COVID-19 episodes was crucial during the pandemic for evaluating interventions. Results from diagnostic tools like PCR, rapid antigen test (RAT) and serology are affected by factors such as timing of tests and vaccination status. The BRACE trial developed an algorithm integrating these diagnostic tools for illness episode classification.

Methods: In the BRACE trial, 3988 participants reported 5512 febrile/respiratory illness episodes and provided longitudinal blood samples over one year. SARS-CoV-2 diagnosis relied on a three-component algorithm: (1) a serology algorithm assessing anti-SARS-CoV-2 nucleocapsid antibody seroconversion, (2) a PCR/RAT algorithm, and (3) an episode interpretation algorithm combining serology and PCR/RAT results to categorise episodes as COVID-19, Not COVID-19 or Uncertain. The algorithms accounted for vaccination status and timing of testing relative to symptom onset to refine episode classifications.

Results: Of 5512 illness episodes, 890 (16%) were classified as COVID-19, 3852 (70%) as Not COVID-19, and 770 (14%) as Uncertain. Compared to relying solely on PCR/RAT results, integrating serology in the algorithm reduced the proportion of Uncertain classifications by more than half. Among the COVID-19 episodes, 89% were identified by positive PCR/RAT results, and the remaining 11% (with missing or negative PCR/RAT tests) were identified by serology. Discordance between PCR/RAT and serology occurred in 9% of episodes.

Conclusions: An algorithm integrating PCR/RAT and serology results in the context of test timing and vaccine status enabled the accurate identification of COVID-19 episodes and minimised the number of episodes that would otherwise have been classified as Uncertain.

Trial registration: The BRACE trial: BCG vaccination to reduce the impact of COVID-19 in healthcare workers.

Clinicaltrials: gov NCT04327206, Registration date:27 March 2020.