Background. To limit viral transmission, COVID-19 testing strategies must evolve as new SARS-CoV-2 variants (and new respiratory viruses) emerge to ensure that the specimen types and test analytical sensitivities being used will reliably detect individuals during the pre-infectious and infectious periods. Our accompanying work demonstrated that there are extreme differences in viral loads among paired saliva (SA), anterior-nares swab (ANS) and oropharyngeal swab (OPS) specimens collected from the same person and timepoint. We hypothesized that these extreme differences may prevent low-analytical-sensitivity assays (such as antigen rapid diagnostic tests, Ag-RDTs) performed on a single specimen type from reliably detecting pre-infectious and infectious individuals.
Methods. We conducted a longitudinal COVID-19 household-transmission study in which 228 participants collected SA, ANS, and OPS specimens for viral-load quantification by RT-qPCR, and performed an ANS Ag-RDT (Quidel QuickVue At-Home OTC COVID-19 Test) daily. We evaluated the performance of the Ag-RDT (n=2215 tests) to detect infected individuals (positive results in any specimen type by RT-qPCR) and individuals with presumed infectious viral loads (at or above thresholds of 10^4, 10^5, 10^6, or 10^7 copies/mL).
Results. Overall, the daily Ag-RDT detected 44% (358/811) timepoints from infected individuals. From 17 participants who enrolled early in the course of infection, we found that daily Ag-RDT performance was higher at timepoints when symptoms were reported, but symptoms only weakly correlated with SARS-CoV-2 viral loads, so ANS Ag-RDT clinical sensitivity remained below 50%. The three specimen types exhibited asynchronous presumably-infectious periods (regardless of the infectious viral-load threshold chosen) and the rise in ANS viral loads was delayed relative to SA or OPS for nearly all individuals, which resulted in the daily ANS Ag-RDT detecting only 3% in the pre-infectious period and 63% in the infectious period. We evaluated a computationally-contrived combined AN-OP swab based on viral loads from ANS and OPS specimens collected at the same timepoint; when tested with similar analytical sensitivity as the Ag-RDT, this combined swab was predicted to have significantly better performance, detecting up to 82% of infectious individuals.
Conclusion. Daily ANS rapid antigen testing missed virtually all pre-infectious individuals, and more than one third of presumed infectious individuals due to low-analytical-sensitivity of the assay, a delayed rise in ANS viral loads, and asynchronous infectious viral loads in SA or OPS. When high-analytical-sensitivity assays are not available and low-analytical-sensitivity tests such as Ag-RDTs must be used for SARS-CoV-2 detection, an AN-OP combination swab is predicted to be most effective for detection of pre-infectious and infectious individuals. More generally, low-analytical-sensitivity tests are likely to perform more robustly using oral-nasal combination specimen types to detect new SASR-CoV-2 variants and emergent upper respiratory viruses.
