Performance of at-home self-collected saliva and nasal-oropharyngeal swabs in the surveillance of COVID-19

Abstract

Background: SARS-CoV-2 quickly spreads in the worldwide population, imposing social restrictions to control the infection, being the massive testing another essential strategy to break the chain of transmission. Aim: To compare the performance of at-home self-collected samples - saliva and combined nasal-oropharyngeal swabs (NOP) - for SARS-CoV-2 detection in a telemedicine platform for COVID-19 surveillance. Material and methods: We analyzed 201 patients who met the criteria of suspected COVID-19. NOP sampling was combined (nostrils and oropharynx) and saliva collected using a cotton pad device. Detection of SARS-COV-2 was performed by using the Altona RealStar® SARS-CoV-2 RT-PCR Kit 1.0. Results: There was an overall significant agreement (κ coefficient value of 0.58) between saliva and NOP. Considering results in either sample, 70 patients positive for SARS-CoV-2 were identified, with 52/70 being positive in NOP and 55/70 in saliva. This corresponds to sensitivities of 74.2% (95% CI; 63.7% to 83.1%) for NOP and 78.6% (95% CI; 67.6% to 86.6%) for saliva. Conclusion: Our data show the feasibility of using at-home self-collected samples (especially saliva), as an adequate alternative for SARS-CoV-2 detection. This new approach of testing can be useful to develop strategies for COVID-19 surveillance and for guiding public health decisions.

Keywords: PCR; Saliva; coronavirus; infection control; primary health care; telemedicine.

Figure 1.

Comparison of RT-PCR cycle thresholds between naso-oropharyngeal and saliva samples in 37 patients with positive results in both samples. The coefficients of the regression lines are 0.79 (P < 0.001) for gene E and 0.74 (P = 0.002) for gene S.

Figure 2.

(a/b) – Violin plots showing the distribution of cycle thresholds in nasal-oropharyngeal swabs (NOP) and saliva samples for the two genes (E and S) amplified by RT-PCR. Boxplots shows median, interquartile range and range as standard. Analysis of the 37 patients with positive results in both sample types by comparing the distributions of cycle thresholds between NOP and saliva samples. Paired Wilcoxon’s rank-sum test was used, in which P-values were <0.001 for genes E and S. (c/d) – Distribution of SARS-CoV-2 RT-PCR cycle thresholds in the 52 positive nasal-oropharyngeal swabs (NOP) samples stratified by RT-PCR results in saliva (NOP+/saliva- versus NOP+/saliva+). Boxplots shows median, interquartile range and range as standard. Distributions were compared by using paired Wilcoxon’s rank-sum test, in which P-values were 0.21 and 0.35 for genes E and S, respectively

Figure 3.

Relationship between illness course (i.e. time elapsed between symptom onset and sample collection) and cycle threshold values for nasal-oropharyngeal swabs (NOP) (left-hand panels) and saliva samples (right-hand panels). In the NOP samples, the regression coefficients for cycle threshold (delay of log₂-days) for genes E and S were 0.5 (P = 0.72) and 1.1 (P = 0.42), respectively; the regression coefficients for saliva samples were 0.04 (P = 0.98) and −0.26 (P = 0.87) for genes E and S, respectively