key: cord-1015876-xagm8y0p
authors: Khiabani, Kazem; Amirzade-Iranaq, Mohammad Hosein
title: Are saliva and deep throat sputum as reliable as common respiratory specimens for SARS-CoV-2 detection? A systematic review and meta-analysis
date: 2021-03-24
journal: Am J Infect Control
DOI: 10.1016/j.ajic.2021.03.008
sha: d0ee414fdbb68adf51ec904556d966e176c17a33
doc_id: 1015876
cord_uid: xagm8y0p

OBJECTIVE: The COVID-19 pandemic raises an urgent need for large-scale control through easier, cheaper, and safer diagnostic specimens, including saliva and sputum. We aimed to conduct a systemic review and meta-analysis on the reliability and sensitivity of SARS-CoV-2 detection in saliva and deep throat sputum (DTS) compared to nasopharyngeal, combined naso/oropharyngeal, and oropharyngeal swabs. METHODS: This systematic review and meta-analysis was performed according to the PRISMA statement. The inclusion criteria were studies that specifically assessed a sample of saliva or deep throat sputum (DTS) with at least one other respiratory specimen in patients with COVID-19 infection, based on RT-PCR tests. The DerSimonian-Laird bivariate random-effects model analysis performed using STATA software with the "metaprop" package. RESULTS: From 1598 studies, we retrieved 33 records, of which 26 studies were included for quantitative analysis. We found an overall sensitivity of 97%(95%CI,86-100) for bronchoalveolar lavage fluid, 92%(95%CI,80-99) for double naso/oropharyngeal swabs, 87%(95%CI,77-95) for nasopharyngeal swabs, 83% (95% CI 77-89) for saliva, 82% (95%CI,76-88) for DTS, and 44% (95%CI,35-52) for oropharyngeal swabs among symptomatic patients, respectively. Regardless of the type of specimens, the viral load and sensitivity in the severe patients were higher than mild and in the symptomatic patients higher than asymptomatic cases. CONCLUSION: The present review provides evidence for the diagnostic value of different respiratory specimens and supports saliva and DTS as promising diagnostic tools for first-line screening of SARS-CoV-2 infection. However, sampling, storing, and laboratory assay needs to be optimized and validated before introducing a definite diagnosis tool. Saliva, DTS, nasopharyngeal, and even double naso/oropharyngeal swabs showed approximately similar results, and sensitivity was directly related to the disease severity. This review revealed a relationship between viral load, disease severity, and test sensitivity. None of the specimens showed appropriate diagnostic sensitivity for asymptomatic patients.

COVID-19 disease, caused by the SARS-CoV-2, is a severe infection causing morbidity and mortality worldwide. More than 200 countries and territories are affected, more than 64,000,000 are infected, and approximately 1,500,000 deaths are reported by December 2020 1 .

It has now been acknowledged that early detection, isolation, and management of infected individuals will play a critical role in stopping the pandemic's further escalation.

Nasopharyngeal swab (NPS) followed by RT-PCR laboratory confirmation is the most recommended diagnostic method for COVID-19 detection. 2 The collection of common respiratory samples such as nasopharyngeal and oropharyngeal specimens require trained medical personnel 3 , which exposes staff to a high risk of infection. 4 . While these tests are not always successful at first, shortages of swabs, sample transport media, and personal protective equipment are frequently reported. 3, 5 Mass testing requires an increased number of trained personnel at specimen acquisition sites. Also, nasopharyngeal sampling causes discomfort to patients 4 , and there are several contraindications, such as coagulopathy or anticoagulant therapy and significant nasal septum deviation 3 . Considering the high rate of disease transmission and the drawbacks of common respiratory sampling techniques, the use of more flexible, less invasive, and facile specimens for RT-PCR diagnosis tests is crucial.

Despite the heterogenic origin of saliva, it is informative to identify various oral and systemic conditions and viral infections such as severe acute respiratory syndrome (SARS) and the Middle East respiratory syndrome (MERS) 4, 6 .

Therefore, since early January 2020, there has been a growing interest in using salivary secretions and deep throat sputum (DTS) as alternatives for common respiratory samples to diagnose COVID-19 infection. The literature has been indicated the possible use of saliva or DTS as a diagnostic specimen for detecting SARS-CoV-2 based on RT-PCR tests. [7] [8] [9] [10] Several systematic reviews and meta-analyses aimed to conclude this diagnostic samples' efficacy and compare it to other specimens such as NPS and Oropharyngeal swab (OPS). [7] [8] [9] [10] The present comprehensive systematic review and meta-analysis aimed to overcome the limitations of the small sample sized studies and heterogenic outcomes associated with the different clinical course of the disease to estimate the diagnostic sensitivity of different oral and pharyngeal-based specimens and compare the ability and reliability of different respiratory specimens for detection of the SARS-CoV-2.

We systematically searched four major databases for this systematic review and metaanalysis (PubMed, Scopus, WoS, and PMC). We also manually searched Google Scholar and pre-print archives, references of included studies, cited and citing papers of the relevant studies for relevant results, and sought suggestions from experts to supplement the database searches.

We considered original articles (published or pre-print) and conference proceedings without time restriction. (last updated August 2020)

We used the following search terms and their variations: "COVID-19", "SARS-COV-2", "novel coronavirus", "2019 novel coronavirus", "new coronavirus", "diagnosis", "diagnostic", "diagnostic test", "diagnostic assay", "saliva", "sputum", "oral fluid", "oral secretion", "Deep throat saliva", "Deep throat sputum", "oropharyngeal saliva", "Deep throat secretion", and other terms combined with Boolean operators "AND" and "OR".

We included studies in this systematic review and meta-analysis if they met all the following eligibility criteria: 1. records published or under-publishing in scientific journals (including pre-print studies); 2. patients diagnosed with or screened for COVID-19; 3. the diagnosis based on RT-PCR method; 4. studies designed to specifically use samples of saliva or oropharyngeal sputum or oral secretion or oral fluids or pharyngeal secretion for quantitative or non-quantitative comparison of diagnostic methods and viral loading in SARS-CoV-2 infected patients; 5. studies that assessed at least two respiratory specimens; 6. studies conducted on the previously confirmed COVID-19 patients or compared simultaneously with matched (paired) specimen. Exclusion criteria: (1) publications with no primary outcomes such as reviews, guidelines, and recommendations; (2) publications dated before January and after 11 August 2020.

Two authors (KK and MHA) independently screened titles and abstracts of all publications identified through the literature search, reviewed potentially eligible full-text papers using the predefined criteria, extracted data from included studies, and assessed methodological quality. Discrepancies were resolved through consensus between two researchers. Unsolved cases were referred to a third reviewer, and duplicate studies were excluded. The following data were extracted and calculated from the text and tables and transferred to the preconstructed data extraction form: author's name, place of study, method of diagnosis, sampling technique, matched or reference specimen, population size, viral load, and the following outcome parameters: numbers of total, positive and negative saliva or DTS tests regarding disease status and numbers of total, positive and negative NPS, OPS, double NPS/OPS regarding disease status.

Two reviewers (KK and MHA) evaluated independently the risk of bias in each study using the "Diagnostic Precision Study Quality Assessment Tool" (QUADAS-2) recommended by the Cochrane Collaboration. Our assessment consisted of evaluating the risk of bias in four domains: (1) patient selection, (2) 

Regardless of the number of tests performed for a patient, each test result is crucial to assess different specimens' sensitivity. Since the difference in the number of tests between sample-based and patient-based reports was small, we included both types in the analysis. All case reports presented a single participant's data were excluded. One of the following methods have been used to confirm the COVID-19 infection in the studies included for quantitative or non-quantitative assessment of oral or retropharyngeal specimens:

1. Study was performed on patients who had been confirmed for COVID-19 infection by RT-PCR.

2. Diagnosis was based on a reference test, NPS or OPS or double NPS/OPS, collected in parallel with saliva or sputum (matched/paired sampling).

3. Infection was confirmed based on pooled event rates (positive and negative results) of saliva/ DTS and other respiratory specimens.

In the present study, to reduce the heterogeneity in the diagnosis methods, the second type was changed to the third type based on the pool of positive and negative saliva/DTS results and the reference/matched sample. Sensitivity, defined as the probability that a test result will be positive when the disease exists (true positive rate), was calculated as TP/ (TP + FN).

The DerSimonian-Laird bivariate random-effects model analysis was performed with STATA software version 16 (StataCorp, Texas, USA) with the "metaprop" package written by Victoria N. Nyaga. For analyses involving studies with a small sample size and sensitivity value too high (towards 1) or low (towards 0), we incorporated the Freeman-Tukey Double Arcsine Transformation method to stabilize the variances by-study confidence intervals.

A common confusion in the studies that use oral & retropharyngeal fluids specimens is the unclear definition of "saliva". To achieve greater consistency in this analysis, we divided studies into two main categories: saliva-based and DTS-based studies. DTS contains upper and lower respiratory tract secretions and collected in the same way as recommended for lower respiratory tract sputum. Saliva (oral fluid) sampling includes two common techniques: sampling through frequent spitting out (Drooling technique) and direct sampling from the oral fluid pool.

In general, the drooling technique should not be considered pure saliva sampling because bronchoalveolar secretion, nasopharyngeal discharge, and other intraoral substances are added to the saliva. Direct saliva sampling includes using saliva sampling kits, instruments, and swabs to collect saliva from the salivary pool under the tongue tip near the orifice of major salivary glands, exactly posterior to the lower anterior teeth. In this study, saliva and oral fluids and oral secretion were used interchangeably. All extracted studies that mentioned sampling of oral fluid or sampling without referring to oropharyngeal secretion were included in the salivary group.

Deep throat (upper respiratory) sputum sampling is performed by throat clearing and coughing up and out the secretion and sputum of the retropharynx. All extracted studies sampled deep throat specimens with or without coughing were included in the DTS group. However, a few studies that referred to retropharyngeal sampling without cough and sputum were analyzed separately (DT-secretion).

From 1598 articles retrieved from the initial search, 57 saliva/DTS-related studies were identified in the title and abstract screening after duplicate removal. According to inclusion criteria, 33 studies were meet the review's aims. The "Saliva" group included 19 articles 14-32, and the "DTS" group included 14 articles. [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] (Supplementary Material Tables S1 & S2) For the quantitative analyses (meta-analysis), 16 and 10 trials in the Saliva and DTS groups were included, respectively. [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] The remaining studies, which did not report quantitative information, were only systematically reviewed. [30] [31] [32] [43] [44] [45] [46] (Supplementary Material Figure S1) 

Seventeen studies provided appropriate quantitative information to calculate the saliva sample sensitivity. 14-30 However, the Chen et al. study, which was the only study using a different collection method to obtain almost pure saliva, was not included in the statistical analysis. 30 The summary of diagnostic results from saliva and other respiratory specimens is listed in Table 1 . The remaining articles were included in systematic reviews. 31, 32 Ten studies assessed the diagnostic test results in the previously laboratory-confirmed COVID-19 patients, 14-21, 30, 31 , while eight studies used the pooled event rates of salivary test results with another specimen to obtain the sensitivity. [22] [23] [24] [25] [26] [27] [28] [29] (Supplementary Material Table S1) Fourteen studies performed sampling concurrently from saliva and other areas (matched samples). 14-17, 20, 22-30 Four studies reported non-matched sample results. 18, 19, 21, 31 Nasopharyngeal swab (NPS) 14-19, 23, 25, 27, 28, 31 , Oropharyngeal swab (OPS) 21, 22, 30 , combined Naso/oropharyngeal swabs 20, 22, 24, 26, 29 were used as a reference or as a matched specimen with saliva sample in eleven, three and five studies, respectively.

In the "Saliva" group, most studies have used the drooling technique to collect saliva. 15-17, 22-25, 27-29 Two studies collected saliva after pooling in the mouth, 14, 18 and Two studies have used swab sampling 16, 31 , one study utilized a saliva-collecting kit 25 , and one study has used one of the following techniques: drooling, pipette or swab for saliva sampling. 16 however, five studies did not describe the collection method at all. 19-21, 26, 32 Two, five, and eight studies evaluated patients in severe 15, 16 , severe in combination with mild to moderate, 17, 19, 22, 29, 30 , and mild to moderate 14, 18, 20, 23-25, 27, 28 conditions, respectively. Also, twelve, two, and one studies included symptomatic [14] [15] [16] [17] [18] [19] [22] [23] [24] [25] [26] [27] , symptomatic combined with asymptomatic 20, 31 , and asymptomatic patients 21 .

Sixteen eligible studies, including 1052 patients with 1056 salivary tests based on the previously confirmed patients (859 positive vs. 197 negative SARS-CoV-2 samples), provided quantitative analysis information. Most of the studies reported patient-based results. 14, [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] However, Wyllie et al. conducted a sample-based study. 15 The saliva test's sensitivity possessed a wide range from 25% 31 to 100% 16 . (Table 1) The saliva test's overall sensitivity in symptomatic patients showed 83% (95%CI= 77-89; I 2 = 79.04%) ( Figure 1 ). However, it decreased to 81% (95% CI=74-87; I 2 = 81.26%) when asymptomatic patients were included. 

The highest sensitivity (100%) for saliva specimen test among all studies was reported by Azzi et al. from 25 severe patients. 16 (Table 1) . For severe patients, pooled sensitivity assessed 90% (95% CI= 76-99; I 2 = 72.43%). (Figure 1 ) Studies in which the cases were all mild to moderate patients or a small number were severe, categorized as "mild to moderate" disease subgroup. 14, 17, 18, 20, [22] [23] [24] [25] [26] [27] [28] In this category, most of the studies reported higher viral load for NPS than saliva specimen 14, 17, 23, 25, 27, 28, [30] [31] [32] . (Supplementary Material Table S1 ). In mild to moderate stage patients, pooled sensitivity estimated 81% (95% CI= 73-88; I 2 = 80.77%). ( Figure   1 ) Two studies tested asymptomatic patients. 20, 21 (Table 1) 

Ten studies provided quantitative information to calculate the sputum specimen sensitivity. [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] Other remaining studies were systematically reviewed. (Supplementary Material   Table S2 ) [43] [44] [45] [46] . The summary of diagnostic results from DTS and other respiratory specimens is listed in Table 2 . Nine studies evaluated the test sensitivity in patients with previously confirmed infection [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] and one study with pooled event rates 42 . Five and six studies performed paired 33, 36, 39, 40, 42 , and non-paired sampling [34] [35] [36] [37] [38] 41 , respectively. However, one study used both types of samplings. 36 (Table 2) Six studies collected DTS through throat clearing and coughing up/out into a sterile container. 33-35, 38, 40, 42 One study used a throat washing technique to collect deep throat secretion. However, quantitative information was not available for including in meta-analysis. 43 One study used throat gargling with saliva to obtain deep throat secretion. 38 . The clearing of the throat without coughing and trough, making the noise of "Kruuua" used to collect posterior oropharyngeal secretion in another study. 39 In the remaining other studies, the technique of sputum sampling was unavailable. 36, 37, 41 Patients in severe 34 , severe in combination with mild to moderate 33, [35] [36] [37] [38] [39] [40] [41] , and mild to moderate 42 conditions were evaluated in included studies. Nine symptomatic [33] [34] [35] [36] [37] [38] [39] [40] [41] and one symptomatic combined with asymptomatic patients 42 studies were evaluated, respectively. respectively. Among all the studies, the sputum specimen's sensitivity was ranged from 53% 36 to 91.6% 34 . (Table 2 ) The DTS overall sensitivity in symptomatic patients showed 82% (95% CI= 76-88; I 2 = 67.57%). (Figure 2 ) However, it decreased to 79% (95% CI=72-85; I 2 = 80.46%) when studies with cough-free secretion were included. The sensitivity of cough-free samples from deep throat secretion was 64.37%.

The highest sensitivity was in studies with severe patients when all cases 34, 37 , or about half of the patients were severe. 35 (Table 2 ) Wang et al. reported the highest sensitivity rate (91,6%) for the DTS specimen for 12 severe patients. 34 Wang to et al., in a quantitative viral load study 35 , reported higher viral load in severe cases than mild cases. Yu Xia et al. found a close relation between viral load and the severity of the disease so that the viral load was higher in severe cases and in patients who became severe during hospitalization. 45 reported higher viral load in NPS than DTS. 33 32 Focusing on mild to moderate patients, pooled sensitivity was estimated as 80% (95% CI= 73-87; I 2 = 76.65%). (Figure 2) 

Altogether, some studies provided quantitative information of NPS [15] [16] [17] [18] [19] 23 (Table 1&2) Several studies provided quantitative information for NPS sensitivity assessment from critically ill 15, 16, 19, 37 , mild to moderate 17, 18, 23, 25, 27, 28, 33, 36, 37, 39, 42 , and asymptomatic 42 patients, respectively. Overall NPS sensitivity is estimated 87% (95% CI=77-95; I 2 = 93.33%).

Pooled sensitivity was estimated 83% (95%CI=73-91,I 2 =55.10%) regarding the patients in severe stages. Also, for patients in mild to moderate stages, pooled sensitivity was estimated at 88% (95% CI= 76-97; I 2 = 94.97%). Overall double NPS/OPS sensitivity was estimated 89% (95% CI= 78-97; I 2 = 90.96%). (Figure 3 ) Also, overall OPS sensitivity, estimated 44% (95% CI=35-52; I 2 = 77.91%). The BALF overall sensitivity was assessed 97% (95% CI=86-100).

(Supplementary Material Figure S4) 

We assessed the risk of bias in 16 

Upper respiratory specimens such as NPS/OPS and lower respiratory specimens such as sputum are recommended specimens for COVID-19 laboratory diagnosis. 2, 47 The sensitivity of the tests depends on the type of specimen, sampling procedures, different viral loads in different anatomic sites, the clinical course of the disease, and the variation in viral RNA sequences. 7, 48 The NPS is the most recommended and widely used diagnostic specimens, 2, 49 but suffers from a lack of an optimal basis for reliable RT-PCR assay. 5, 10, 48 Up to 29% false-negative results have been reported from upper respiratory samples. 50 In a pre-print meta-analysis, the sensitivity of NPS and double NPS/OPS was reported 40-70%, and 70-80%, respectively. 8 It appears that a positive test is highly suggestive of true SARS-COV-2 infection, but a negative test is insufficient to rule out COVID-19. 51 In this regard, failure to diagnose is more consequential in asymptomatic individuals and contributes significantly to further contamination. In addition to the shortage of supply chain, 3, 34 the sampling sequences of common respiratory diagnostic specimens such as NPS and OPS are invasive and may induce bleeding, nausea, vomiting, coughing, and sneezing, which these side effects generate aerosols and create a risk of contamination. 4, 34, 48, 52 Clearly, more flexible and less invasive reliable sampling techniques for screening purposes are crucial to informing clinical and public health systems.

Oral fluid (saliva) and DTS are candidates as non-invasive and easy collectible specimens with advantages, including low cost, ease-to-obtain, self-collectability, safety, and no need for highly trained staff. 4 The saliva secreted 90% from the major and rest from the minor salivary glands and mainly consists of 99% water and remaining of electrolytes, mucus, and digestive and protective proteins. 9, 53 The fluid collected from the oral cavity known as saliva is a mixture of glandular secretions, gingival crevicular fluid, expectorated surface liquid from the upper & lower airway, oral mucosa and upper airways' epithelial and immune cells, and oral microbes and viruses. 9, 54 Interestingly, the saliva sample composition is informative for diagnostic purposes to identify various oral and systemic diseases. 6, 9 An ideal role for saliva has been reported by isolating proteins, peptides, and even sheds of numerous viruses such as Ebola, Zika, influenza A and B, and the recently emerged coronaviruses responsible for SARS and the MERS. 6, [55] [56] [57] Following the outbreak of SARS-COV-2 contamination, several studies have indicated saliva's diagnostic role. [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] However, the source of SARS-CoV-2 in the saliva is not clearly defined. The most convenient and probable case is the entry of virus-infected secretions from the posterior oropharynx, which is the site of mixing and exchanging secretions and debris in the nasopharynx and lower respiratory tract. 34, 35 The second one is salivary gland involvement and direct viral shedding into the saliva. However, precise information is unavailable in this matter. 9 Involvement of epithelial cells lining of salivary gland ducts, as an early target for SARS-COV, has been seen in the infected rhesus macaque model. 58 Additionally, the secretion of SARS-CoV-1 specific secretory immunoglobulin-A into animal models' saliva has been reported. 59 The third one is indirect viral shedding from blood plasma into the oral cavity inside the gingival crevicular secretions. 60 However, it is known that blood plasma with a detection rate of 7.3% is a minor source of SARS-CoV-2. 7 The last one is the direct infection of oral mucosal endothelial cells via ACE2, a specific receptor of SARS-CoV-1 & 2.

Overexpression of ACE2 receptors on the oral cavity mucosa has been stated recently. 61 Many studies have called specimens collected from the oral cavity or oropharyngeal region saliva despite their different origins and samplings. We categorized the collected samples into pure saliva, saliva, deep throat secretion, and DT-sputum (mentioned as DTS in the text) based on the collected fluids' source and location to maintain evidence consistency. Pure saliva is sampled from the salivary gland orifice. Saliva is the fluid available in the oral cavity. The coughing up/out accompanied by a throat-clearing maneuver can introduce more lower respiratory secretion and sputum into the DT-secretion, resulting in a DT-sputum sample.

Therefore, DT-sputum is mixed with the upper and lower respiratory samples and may reveal the sputum specimen's diagnostic potential. 62 Remarkably, reporting various specimens' different sensitivity has led to significant uncertainty and confusion for diagnosis, estimation of tests' accuracy, and monitoring of SARS-CoV-2 infection. In the present study, we performed a comprehensive systematic review and meta-analysis to compare different upper respiratory specimens' feasibility to detect SARS-CoV-2 RNA. Compared to the previous analysis, 7-10 this study has several advantages, including the classification of specimens by considering saliva-based or sputum-based samples, the use of a large number of studies in the meta-analysis and systematic review, classification and analysis of the findings based on disease severity, analysis of asymptomatic patients, analysis of respiratory diagnostic samples simultaneously with saliva or DT-sputum, evaluation of the relationship between viral load, various samples, and disease severity.

Considering the RT-PCR test's specificity and sensitivity, in vitro analyses demonstrated its high specificity and sensitivity for SARS-CoV-2, 63 despite the questionable SARS-CoV-2 detection rate resulting from upper airway specimens in clinical settings. 51 In this regard, very high specificity and moderate sensitivity (40-78%) have been estimated for NPS. 8, 51 Numerous studies have reported negative results for the NPS, while the saliva or DTS tests positive. 14-16, 18, 22, 23, 25, 27, 33, 42 So that, according to current knowledge, none of the approved SARS-COV-2 diagnostic samples is accurate enough to be considered as a gold standard. The WHO and CDC have recommended the lower respiratory tract's sputum specimen as a diagnostic test for patients developing a productive cough. 2, 47 This sampling is the same as the technique of collecting DTS. 33-35, 40, 42 Also, the emergency use authorization (EUA) of a saliva-based diagnostic method was approved by the Food and Drug Administration (FDA) in April 2020 for screening. 64 In this review, to maintain consistency in the analysis of evidence, not losing true positive results while the reference test is negative, and also to match the results of different studies with each other, we considered pooled event rates (negative and positive results) of all samples as reference for sensitivity calculation. Furthermore, considering the lack of accurate reference tests and limited available studies to determine false-positive results, specificity calculation was not feasible in the present analysis.

In this study, the highest overall sensitivity for SARS-CoV-2 in symptomatic patients was 97%(95%CI=86-100) for bronchoalveolar fluid lavage, 92%(95%CI=80-99) for double naso/oropharyngeal swabs, 87%(95%CI=77-95) for NPS, 83%(95%CI=77-89) for saliva, 82%(95%CI=76-88) for DTS, and 44%(95%CI=35-52) for OPS, based the pooled event rates. 14 studies, which all are included in our analysis 10 . We found similar sensitivity for OPS and higher sensitivity for NPS and DTS comparing to the estimates reported by Mohammadi et al. 8 They reported 43%(95%CI=34-52%) for OPS, 54%(95%CI=41-67%) for NPS, and 71%(95%CI=61-80%) for sputum in a meta-analysis of 11 studies. We found lower sensitivity for sputum and higher sensitivity for saliva than the study from Boger et al., which reported a sensitivity of 0.972 (0.903-0.997) for Sputum and 0.623 (0.545-0.696) for Saliva samples through analyzing limited records. 7 Furthermore, the present study showed lower sensitivity for saliva compared to the meta-analysis of Czumbel et al., which included four studies with severe patients (95%CI=80-99%). 9 In the present study, considering the method of diagnosis, no difference was found between previously confirmed patients and pooled event rates group in saliva sample (Supplementary Material Figure S3 ), which was consistent with the study by Ricco et al. 10 The sensitivity rate was directly related to the severity of the disease, which means that a positive test's probability increased as the disease's severity increased. The highest sensitivity The NPS appeared to be more sensitive in mild patients. According to the subgroupanalysis, the reason was the lack of studies based on the pooled event rates in severe patients compared to mild patients. Hence, to achieve greater consistency in the assessment, we excluded these studies from the group of mild patients, and as a result, the results changed in favor of severe patients.

Overall, the lowest rate of sensitivity was found in asymptomatic patients. However, the number of studies that reported test results on asymptomatic individuals was limited. Saliva showed 83%(95%CI=77-89) sensitivity in symptomatic patients, approximately two times more than asymptomatic patients (46% with 95%CI=27-66). The influence of asymptomatic patients on the estimation of overall sensitivity was also evident so that if the number of samples from asymptomatic patients increases, the diagnostic tests' sensitivity decrease. (Supplementary Material Figure S2) In general, sensitivity was directly related to the viral load so that as the viral load increased, the probability of a positive test increased. Regardless of sample type, several studies found higher viral load and consequently sensitivity in severe patients than mild patients. 16, 35, 37, 41, 44, 45 So that the mean viral load of severe cases was found around 60 times higher than mild cases. 65 Isolation of the virus has been reported from a considerable fraction of respiratory specimens during the first week from mild patients, whereas no isolates were obtained with a reduction in viral load after the first week. 46 Noteworthy, patients with higher baseline viral load more likely to become severe. 45 In this regard, Magleby et al. found that higher viral load was associated with an over 6-fold higher risk of death and a nearly 3-fold higher risk of intubation. 66 Viral load was also highest during the first week after symptom onset and subsequently declined with time 26, 34, 35, 38, 41, 46 , and it is significantly higher during early and progressive stages than the recovery stage. 34, 35, 38, 41, 46 Up to 50% reduction in SARS-CoV-2 detection rates from saliva samples has also been reported during the convalescent period. 18, 25 Mohammadi et al. found that early sampling following the onset of symptoms was associated with improved detection rates. 8 In this regard, asymptomatic patients show significantly lower viral load and faster viral clearance than symptomatic patients 15, 20 , thus less likely to test positive. 20 Considering viral shedding in the recovery period, NPS showed around three times more sensitivity than saliva. 18 Iwasaki et al. reported lower viral load and earlier viral clearance in saliva than NPS 23 (Supplementary Material Table S1 ). Simultaneously, sputum was slightly more sensitive than NPS in the recovery phase in the study of Yang et al. 37 . The sputum has been suggested for patient monitoring and discharge management due to the prolonged viral shedding. 34 Based on the current knowledge, the source of saliva contamination is unknown 67 , so sampling from different anatomical sites may lead to different results and sensitivity.

Interestingly, sampling from pure saliva revealed only 12.9% positive results, so that 60% of samples from severe patients and 3.8% of mild patients resulted positive. 30 Also, direct sampling from bilateral buccal mucosa showed significantly lower sensitivity and lower viral load in saliva samples than NP swabs in infected children. 31 These findings may indicate a low probability of involvement of the major and minor salivary glands and consequently secretory saliva, and in case of involvement, it occurs in the more severe stages of the disease. 30, 31 Despite considerable heterogeneity in sampling techniques and composition, the present meta-analysis showed similar saliva and DTS sensitivity. It appears that the proximity of the mouth and throat (The latter is where the secretions of the oral cavity, nasopharynx, and lower respiratory tract meet) 34, 35 , and the constant mixing of the contents of the oral cavity and deep throat through swallowing, clearing the throat, and even during the sample collection procedure, may lead to a similar composition of saliva and DT-sputum.

Several limitations arising from the structure of analyzed included studies could have been addressed, while none can be sufficiently solved due to the lack of information. We are aware of the potential bias resulting from the insufficient methodological quality of included studies. The high heterogeneity in the analysis of the results was probably caused by differences in samples and origins, sampling techniques, sample number, sampling on different days following symptoms onset, disease severity, RNA shedding and viral load, purification, and diagnostic PCR kits. Other issues that make "an in-depth evaluation" complex and even impossible are the lack of methodological homogeneity, inadequate reporting of methods and outcome parameters, inconsistent quantitative test results, unavailable viral load reports, and inconsistent results in some studies. Also, a significant share of included studied was retrieved from pre-print platforms (e.g., medrxiv.org). Furthermore, in the lack of an accurate reference test to determine false-positive results, the specificity calculation was not feasible in the present study.

The present meta-analysis provides evidence regarding different respiratory specimens.

Within the limitation of the present study, saliva and DTS are valuable diagnostic specimens for COVID-19 diagnosis. Self-collected Saliva and DTS as a non-invasive sampling method allow a more facile, cheaper, safer, and broader population screening than current respiratory specimens.

They also improve patient acceptance and decrease the risk to healthcare workers. The results support the use of saliva and DTS as a suitable alternative first-line screening method of SARS-CoV-2 infection based on RT-PCR assay; however, the methods of sampling, storing, and laboratory assay need to be optimized and validated before introducing as an appropriate standardized procedure for definite diagnosis and viral load monitoring in clinical applications.

Based on meta-analysis and systematic review of the present study, it seems that the following conclusions can be reached:

 BALF, double NPS/OPS, NPS, saliva, and DTS, showed the highest sensitivity, respectively.

 Saliva, DTS, and NPS showed similar results.

 Double NPS/OPS show higher viral detection than NPS; however, given the 5% difference in diagnosis, a rational and scientific decision is needed to continue to use combined NPS/OPS based on cost and benefit.

 OPS is the most unreliable respiratory sample.

 Viral load and disease severity and SARS-CoV-2 detection rate are directly related.

 Viral load and sensitivity are higher in severe patients than mild patients.

 Viral load and SARS-CoV-2 detection rate are significantly lower, and viral clearance is significantly faster in asymptomatic patients than in symptomatic individuals  It appears DTS-based assay reveals more reliable results compared to the saliva-based assay for patient monitoring and during the recovery stage.

 None of the diagnostic specimens showed appropriate diagnostic sensitivity in asymptomatic patients; however, DTS and double NPS/OPS appear more promising. 

WHO Coronavirus Disease (COVID-19) Dashboard

Interim Guidelines for Collecting, Handling, and Testing Clinical Specimens for COVID-19 2020

RNA and antibodies in diverse samples: protocol to validate the sufficiency of provider-observed, home-collected blood, saliva, and oropharyngeal samples

A review of salivary diagnostics and its potential implication in detection of Covid-19

False Negative Tests for SARS-CoV-2 Infection-Challenges and Implications

Human saliva: Non-invasive fluid for detecting novel coronavirus (2019-nCoV)

Systematic review with meta-analysis of the accuracy of diagnostic tests for COVID-19

SARS-CoV-2 Detection in Different Respiratory Sites: A Systematic Review and Meta-Analysis. medRxiv

Saliva as a Candidate for COVID-19 Diagnostic Testing: A Meta-Analysis. medRxiv

Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement

Saliva as a non-invasive specimen for detection of SARS-CoV-2

Saliva is more sensitive for SARS-CoV-2 detection in COVID-19 patients than nasopharyngeal swabs

Saliva is a reliable tool to detect SARS-CoV-2

Sensitivity of nasopharyngeal swabs and saliva for the detection of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)

EasyCOV: LAMP based rapid detection of SARS-CoV-2 in saliva

Comparisons of viral shedding time of SARS

The natural history and transmission potential of asymptomatic SARS-CoV-2 infection

Rapid implementation and validation of a cold-chain free SARS-CoV-2 diagnostic testing workflow to support surge capacity

Saliva sampling and its direct lysis, an excellent option to increase the number of SARS-CoV-2 diagnostic tests in settings with supply shortages

Saliva as an alternate specimen source for detection of SARS-CoV-2 in symptomatic patients using Cepheid Xpert Xpress SARS-CoV-2

Evaluation of symptomatic patient saliva as a sample type for the Abbott ID NOW COVID-19 assay

Saliva offers a sensitive, specific and non-invasive alternative to upper respiratory swabs for SARS-CoV-2 diagnosis. medRxiv

Detection of 2019-nCoV in saliva and characterization of oral symptoms in COVID-19 patients. Available at SSRN 3556665

Clinical Utility of Buccal Swabs for Sars-Cov-2 Detection in Covid-19-Infected Children

Clinical Significance of a High SARS-CoV-2 Viral Load in the Saliva

Evaluating the use of posterior oropharyngeal saliva in a point-of-care assay for the detection of SARS-CoV-2

Temporal profiles of viral load in posterior oropharyngeal saliva samples and serum antibody responses during infection by SARS-CoV-2: an observational cohort study

Detection of SARS-CoV-2 in different types

Posterior Oropharyngeal Saliva for the Detection of Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2)

Comparison of throat swabs and sputum specimens for viral nucleic acid detection in 52 cases of novel coronavirus (SARS-Cov-2)-infected pneumonia (COVID-19)

Quantitative detection and viral load analysis of SARS-CoV-2 in infected patients

Self-Collected Oral Fluid and Nasal Swabs Demonstrate Comparable Sensitivity to Clinician Collected Nasopharyngeal Swabs for Covid-19 Detection. medRxiv

effect of throat washings on detection of 2019 novel coronavirus

Viral load of SARS-CoV-2 in clinical samples. The Lancet Infectious Diseases

SARS-CoV-2 viral load in sputum correlates with risk of COVID-19 progression

Virological assessment of hospitalized patients with COVID-2019

Laboratory testing for coronavirus disease 2019 (COVID-19) in suspected human cases: interim guidance

Diagnostic methods and potential portable biosensors for coronavirus disease 2019

COVID-19 care-lab testing

Accessed 9

False-negative results of initial RT-PCR assays for COVID-19: a systematic review

The appropriate use of testing for COVID-19

Nasopharyngeal swabs are more sensitive than oropharyngeal swabs for COVID-19 diagnosis and monitoring the SARS-CoV-2 load. Frontiers in medicine

Oral saliva and COVID-19

Current developments in salivary diagnostics

Human saliva can be a diagnostic tool for Zika virus detection

Saliva as a diagnostic specimen for testing respiratory virus by a point-of-care molecular assay: a diagnostic validity study

Find the right sample: A study on the versatility of saliva and urine samples for the diagnosis of emerging viruses

Epithelial cells lining salivary gland ducts are early target cells of severe acute respiratory syndrome coronavirus infection in the upper respiratory tracts of rhesus macaques

effect of mucosal and systemic immunization with virus-like particles of severe acute respiratory syndrome coronavirus in mice

Quantitative proteomic analysis of gingival crevicular fluid in different periodontal conditions

High expression of ACE2 receptor of 2019-nCoV on the epithelial cells of oral mucosa. International journal of oral science

Infectious Diseases Society of America guidelines on the diagnosis of COVID-19

64. (FDA) FaDA. Coronavirus (COVID-19) Update: FDA Authorizes First Diagnostic Test Using At-Home Collection of Saliva Specimens

Viral dynamics in mild and severe cases of COVID-19. The Lancet Infectious Diseases