Comparison of self-collected mouth gargle with deep-throat saliva samples for the diagnosis of COVID-19

Dear Editor,

We read with interest the study by Zhu et al.¹ that proved saliva as an acceptable alternative to nasopharyngeal or oropharyngeal swabs for diagnosis and monitoring of SARS-Coronavirus-2 (SARS-CoV-2) patients. While saliva is a well-accepted self-collected sample, mouth gargle is a potential alternative with characteristics more favorable for laboratory handling. Mouth gargle are non-viscous in nature, which minimizes cross-contamination during re-suspension and transfer that poses risk of generating false-positive results. Furthermore, the non-viscous nature also reduces the chance of clogging of liquid handling systems that would result in testing failure. To date, only a few comparative studies analyzed the analytical performance of mouth gargle in the diagnosis of Coronavirus Disease 2019 (COVID-19)^{2 , 3}. We performed a prospective head-to-head comparison on the analytical performance of self-collected mouth gargle and Deep-Throat Saliva (DTS) samples for the detection of SARS-CoV-2 using four different Nucleic Acid Amplification Test (NAAT) platforms performed in three independent laboratories.

We recruited patients admitted to the Prince of Wales Hospital, Hong Kong SAR, China, with active SARS-CoV-2 infection confirmed by the presence of SARS-CoV-2 RNA in their respiratory tract (nasopharyngeal swab, sputum or DTS) using reverse transcription polymerase chain reaction (RT-PCR). Recruitment started from November 23 to December 1, 2020 for patients aged above 10 years who can follow instructions of gargle collection. Disease severity was classified into asymptomatic, mild, moderate, severe, and critical as previously described^{4 , 5}. Informed consents were obtained from all participants. The study was approved by The Joint Chinese University of Hong Kong – New Territories East Cluster Clinical Research Ethics Committee.

Mouth gargle and deep-throat saliva (DTS) samples were collected at least 30 min apart. Mouth gargle samples were collected with sterile bottles pre-filled with 4 mL of 0.9% sterile normal saline according to the illustrations printed on the instruction sheet (Supplementary figure 2). DTS samples were collected according to the instructions produced by the centre for Health Protection, HKSAR (https://www.chp.gov.hk/files/pdf/information_sheet_on_dts_en.pdf). DTS samples were diluted with 3 mL of Phosphate Buffered Saline (PBS) that ended up in about 4 mL similar to the final volume of gargle samples that were undiluted. Samples were mixed and aliquoted into equal portions for head-to-head comparison using four different NAAT platforms by three independent laboratories (Supplementary figure 1): (i) In-house assay quantitative real-time reverse-transcriptase polymerase chain reaction (qRT-PCR), Roche Cobas 6800, and Cepheid GeneXpert by local Public Health Reference Laboratory. (ii) In-house RT-PCR assay by University Laboratory A; and (iii) Roche Cobas 6800 and Cepheid GeneXpert assays by University Laboratory B (Supplementary data).

We recruited 49 patients (30 females) aged 12–81 (median: 61) years, with seven asymptomatic, 20 mild (no pneumonitis), 15 moderate (pneumonitis), and four had severe disease that required oxygen support, and three were critical and required ventilator support. All participants recovered and discharged. Samples were collected between 1 and 19 days (mean 7 ± 4) from symptom onset.

A total of 109 mouth gargle-DTS sample-pairs (mean of 2 ± 1 sample-pairs per patient) with at least one positive result for SARS-Co-V-2 RNA were analyzed. The overall positive rate ranged from 89.9% to 96.3% (Fig. 1 ) with no significant difference between mouth gargle and DTS samples by all four assays. Six sample-pairs had discrepant results of either having detectable RNA in mouth gargle only or in DTS only, all with Ct values of >30. Analysis of diagnostic yield reflected by 1/Ct showed strong positive correlation between the paired mouth gargle and DTS samples regardless of the assay used, and with Spearman's correlation index ranged from 0.662 to 0.727 (Fig. 2 B). Of note, DTS had a significantly higher diagnostic yield than mouth gargle (Fig. 2A).

Fig. 1.

Positive rates of SARS-CoV-2 RNA detection between mouth gargle and saliva samples using four different assays in three different laboratories

Reference Laboratory (Cobas 6800) positive rate for mouth gargle: 96.3% (95% CI 90.0–98.6), and DTS: 95.4% (95% CI 89.7–98.0). Reference Laboratory (GeneXpert) positive rate for mouth gargle: 94.5% (95% CI: 88.5–97.5), and DTS: 96.3% (95% CI: 90.9–98.6). Reference Laboratory (In-house method) positive rate for mouth gargle: 95.4% (95% CI 89.7–98.0), and DTS: 96.3% (95% CI 90.9–98.6). University Laboratory A (In-house method) positive rate of for mouth gargle: 89.9% (95% CI 82.8–94.2), and DTS 93.6 (95% CI 87.3–96.9). DTS – Deep-throat saliva samples; Gargle – mouth gargle samples. Fisher test was used to assess the difference in positive rates between detection assays and/or specimens. The 95% confidence interval (CI) was calculated using epi.conf (ctype=“prop.single") in epiR.

Fig. 2.

Comparison of diagnostic yield between sample types, assays, and laboratories

A. Comparison of diagnostic yield of the 109 mouth gargle and DTS sample pairs by test method. Midline: median; Box: interquartile range; DTS, Deep-throat saliva; Gargle, mouth gargle. **** p ≤ 0.0001. Comparisons of viral concentration between detection assays and/or specimens were performed using non-parametric Mann-Whitney Wilcoxon rank-sum test (unpaired) or Wilcoxon signed rank test (paired).

B. Correlation analysis between 109 mouth gargle and DTS sample pairs by test method. Correlation performed by Spearman's correlation index; R2adj, Adjusted R-squared. Spearman's rho and linear regression were used to evaluate their associations. A two-sided p value of < 0.05 was considered statistically significant.

C - Inter-laboratory comparison of Cobas 6800 and GeneXpert of 26 paired samples by Reference Laboratory and University Laboratory B. Correlation performed by Spearman's correlation index; R2adj, Adjusted R-squared. DTS – deep-throat saliva; Gargle – mouth gargle. Spearman's rho and linear regression were used to evaluate their associations. A two-sided p value of < 0.05 was considered statistically significant.

D - Inter-assay comparison of 26 paired samples by analyzing E genes detection by Cobas 6800 and Genexpert. Midline: median; Box: interquartile range; DTS – deep-throat saliva; Gargle – mouth gargle. Comparisons of viral concentration between detection assays and/or specimens were performed using non-parametric Mann-Whitney Wilcoxon rank-sum test (unpaired) or Wilcoxon signed rank test (paired). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

A total of 26 sample-pairs that had adequate volume were additionally tested for inter-laboratory and inter-assay consistencies. We found a strong inter-laboratory correlation when using Cobas 6800 and GenXpert with both sample types with correlation coefficients of > 0.8 (Fig. 2C). Inter-assay performance comparison by E gene detection using Cobas 6800 and GeneXpert performed in Public Health Reference Laboratory and University Laboratory B again showed no significant difference in diagnostic yield using mouth-gargle and DTS (Fig. 2D). When analyzing the diagnostic yield among various clinical situations, we found that the diagnostic yield of DTS was significantly higher in patients with respiratory symptoms, but there was no significant difference in asymptomatic patients and those who were symptomatic but without respiratory symptoms (Supplementary figure 3)

Our study found that the positive rate for SARS-CoV-2 RNA detection from mouth gargle samples was similar to paired deep-throat saliva (DTS) samples collected from patients with active COVID-19 infection irrespective of the laboratory or assay used.

Mouth gargle had been shown to be a suitable sample for diagnosis of COVID-19 and respiratory pathogens^{6 , 7}. Previous studies have shown that gargle samples are comparable, but with slightly higher cycle threshold (Ct) values than those of nasopharyngeal and oropharyngeal swabs⁸. Mouth gargle samples are easy to collect, acceptable to patients², and have the advantage over saliva that it is non-viscous and can dilute inhibitors present in the samples.

Our study has the strength of being prospective, inclusion of a wide age range (12 to 81 years), wide spectrum of clinical severity (from asymptomatic to critically ill), and covering all stage of illness (from 1 to 19 days of onset). There are limitations in our study. First, our cohort were all confirmed COVID-19 patients, thus we cannot analyze specificity. Secondly, we did not specify the sequence of mouth gargle and DTS collection; however, we gave explicit instructions to separate the collection of the specimens to be at least 30 min apart, which should negate the collection sequence bias. In conclusion, our findings suggested mouth gargle showed excellent correlation with DTS and can be a choice of self-collected specimen for mass screening of asymptomatic individuals.

Declaration of Competing Interest

None declared