432 - High concordance between self-collected nasal swabs and saliva samples for detection of SARS-CoV-2

Background: Shortages of swabs and transport medium for sample collection have made identification of SARS-CoV-2 infections challenging. We examined the agreement in SARS-CoV-2 detection between two types of self-collected samples: nasal swabs (NS) and saliva (SA).Figure 1. Bland-Altman plot of SARS-CoV-2-N1 CT values for SA and NS samples
Figure 2. Bland-Altman plot of SARS-CoV-2-N2 CT values for SA and NS samples


Methods: Paired daily self-collected NS and SA samples were collected in May 2020 in an ongoing case-ascertained study of SARS-CoV-2 household transmission in Nashville, TN. After informed consent was obtained, index cases and household members were remotely trained in the self-collection of daily nasal swabs (from both nares and preserved in transport medium) and saliva samples (participants were asked to spit in a sterile urine cup approximately 6 times during a minute). Samples were refrigerated in closed double bags and delivered to the laboratory within ~72–96 hours of collection. NS and SA samples were tested using qRT-PCR at a research laboratory. We computed the agreement in detections between sample types using the McNemar test and compared median qRT-PCR cycle threshold (CT) values between sample types for 2 targets (SARS-CoV-2-N1 and SARS-CoV-2-N2) using a Wilcoxon signed rank test. Bland-Altman plots were used to visually inspect agreement between NS and SA CT values.

Results: Among 144 self-collected samples (72 pairs) from 13 unique participants (5 index cases and 8 household members), there were 30 pairs with concordant positive detections in both sample types, 35 with concordant negative detections in both samples, 1 pair with positive NS only, and 6 pairs with positive SA only. The overall agreement between NS and SA was 90.3% (p=0.1). The median SARS-CoV-2-N1 CT value for NS samples was 33.3 (interquartile range: 30.2–35.3) and for SA samples was 30.5 (29.8–33.3, p=0.04); the median CT value for SARS-CoV-2-N2 was 34.4 (31.5–37.5) and 31.5 (30.4–34.9), respectively (p=0.01). Bland-Altman plots indicated that most observations were contained within the limits of agreement but suggested NS tended to have higher CT values than SA samples (Figures). 

Conclusion: Self-collection of saliva samples provides a simple, non-invasive, and practical strategy for identification of SARS-CoV-2 infections.