National and regional modeling of distinct RSV seasonality thresholds for antigen and PCR testing in the United States

Abstract Background PCR tests now outnumber antigen tests for the diagnosis of respiratory syncytial virus (RSV) infection in the US. Recent analyses have shown that the traditional 10% positivity threshold to define an RSV season by rapid antigen testing was inappropriate for real-time PCR testing, for which 3% positivity appeared more appropriate. Objective To respectively model antigen (10%) and PCR (3%) positivity thresholds at national and regional levels using a large dataset of RSV testing results from US hospital-affiliated laboratories. Study Design From 2011–2016, 599 laboratories participated in a national RSV surveillance program (RSVAlert®). For laboratories with ≥10 tests for ≥30 weeks of a season, national and regional test numbers and positivity were summarized by test type overall, by season, and weekly within each season. Test type positivity thresholds were used to calculate season onset and offset. Results A seasonal average of 543,387 RSV tests was reported. PCR testing increased from 26% in 2011–2012 to 72% in 2015–2016. Overall, national positivity was 15.6% for antigen and 8.3% for PCR testing. National RSV season onsets and offsets were comparable using the 10% antigen and 3% PCR thresholds, but PCR-defined seasons generally started and ended later than antigen-defined seasons. Regionally, there were fewer outlier estimates of RSV season length when the predominant regional test type was used to define the season. Conclusion RSV positivity rates differed by test type, likely due to differential clinical use of the tests. These findings support the use of distinct positivity thresholds by test type.