Background: Respiratory syncytial virus (RSV) produces annual epidemics that vary in the timing of season onset, peak, and duration by season and by geographic region. Recent analyses by the US Centers for Disease Control at the national level have demonstrated that polymerase chain reaction (PCR) testing has largely replaced rapid antigen testing as the predominant test type and that the traditional 10% positivity threshold for defining an RSV season based on antigen testing should not be applied to PCR testing, for which the comparable threshold for real-time surveillance was 3%. The aim of this study was to validate and model implementation of the antigen (10%) and PCR (3%) positivity thresholds at regional, state, and local levels in a large national dataset of RSV testing results from US hospitals.
Methods: RSV test results were obtained from 599 laboratories that participated from 20112016 in RSVAlert®, a national RSV surveillance program. For laboratories that provided consistent weekly data (≥10 tests for ≥30 weeks of a season), regional test numbers and positivity were summarized weekly by test type within each season. Season onset and offset were calculated for the 10 US Department of Health and Human Services (HHS) regions and 10 large states plus Hawaii based on 1) antigen only for all seasons, 2) PCR only for all seasons, and 3) the predominant test type used in a specific geography in a season (either antigen or PCR).
Results: An average of 543,340 RSV tests were reported each season. At the regional and state levels, there were fewer outlier estimates of RSV season length when the predominant regional test type was used to define the season (Figures 1 and 2). Exceptions were few and occurred with antigen testing data.
Conclusion: Overall, PCR positivity of 3% was comparable to antigen positivity of 10% at the regional and state levels. Local RSV season determination was most reliable when based on the predominant test type utilized.
Funded by AstraZeneca
C. S. Ambrose,
M. Brandon, None
K. Frye, None
G. Thomson, None