Article Text

Oral saliva swab reverse transcription PCR for Covid-19 in the paediatric population
1. Cinta Moraleda1,2,
2. Sara Domínguez-Rodríguez2,
3. Juan Miguel Mesa3,
4. Paula García-Sánchez4,
5. María de la Serna3,
7. Amanda Bermejo6,
8. Gema Sabrido7,
9. Leticia Martínez-Campos8,
11. Marta Illán-Ramos10,
12. Elena Cobos-Carrascosa2,
13. Álvaro Ballesteros2,
14. Juan Carlos Galán11,
15. Francisco Llorente12,
16. David Aguilera-Alonso13,
17. Ana Belén Jiménez14,
18. Pilar Catalán15,
19. Elias Dahdouh16,
20. Ignacio Navarro3,
21. Bárbara Fernández-Garoz17,
22. Pablo Mendoza18,
23. Concepción Pérez-Jorge19,
24. Teresa Cabezas-Fernández20,
25. Daniel Blázquez-Gamero1,2,21,
26. Gonzalo Rivas22,
27. Patricia Gonzalez-Donapetry16,
28. Elena Sáez23,
29. Jovita Fernández-Pinero12,
30. Jesús Lucas-Fernández11,
31. Elisa Pérez-Ramírez12,
32. Paloma Merino24,
33. Sandra Miragaya17,
34. Jorge Lorente25,
35. Irene Iglesias12,
36. Alfredo Tagarro2,3
37. the EPICO working group
1. 1Pediatric Infectious Diseases Unit. Department of Pediatrics, Hospital Universitario 12 de Octubre. Pediatric Research and Clinical Trials Unit (UPIC). RITIP (Translational Research Network in Pediatric Infectious Diseases), Madrid, Spain
2. 2Instituto de Investigación 12 de Octubre (imas12), Fundación para la Investigación Biomédica del Hospital 12 de Octubre, Madrid, Spain
3. 3Paediatrics Department. Pediatrics Research Group, Hospital Universitario Infanta Sofia, San Sebastian de los Reyes, Madrid, Spain
4. 4Emergency Pediatric Department. Instituto Investigación Hospital La Paz (IDIPaz), Hospital Universitario La Paz, Madrid, Spain
5. 5Emergency Pediatric Department. Instituto Investigación La Princesa, Hospital Infantil Universitario Niño Jesús, Madrid, Spain
6. 6Pediatric Department, Hospital Universitario de Móstoles, Móstoles, Madrid, Spain
7. 7Pediatric Department, Hospital Universitario Rey Juan Carlos, Móstoles, Madrid, Spain
8. 8Pediatric Infectious Diseases, Materno Infantil. Hospital Universitario Torrecárdenas, Almeria, Spain
9. 9Emergency Pediatric Department, Hospital Universitario 12 de Octubre, Madrid, Spain
10. 10Pediatrics Department, Hospital Universitario Clínico San Carlos, Madrid, Spain
11. 11Microbiology Department, Hospital Universitario Ramón y Cajal and Instituto Ramón y Cajal de Investigación Sanitaria (IRYCIS). Centro de Investigación Biomédica en Red en Epidemiología y Salud Pública (CIBERESP), Madrid, Spain
13. 13Pediatric Infectious Diseases Unit. Department of Pediatrics, Hospital Universitario Gregorio Marañón. Unidad de Investigación Maternoinfantil Fundación Familia Alonso (UDIMIFFA), Instituto de Investigación Sanitaria Gregorio Marañón (IiSGM). CIBER en Enfermedades Infecciosas (CIBERINFEC), Madrid, Spain
14. 14Pediatrics Department, Hospital Universitario Fundación Jiménez Díaz, Madrid, Spain
15. 15Servicio de Microbiologia Clinica y Enfermedades Infecciosas, Hospital General Universitario Gregorio Marañón, Madrid, Spain
16. 16Clinical Microbiology and Parasitology Department, Hospital Universitario La Paz. IdiPAZ, Madrid, Spain
17. 17Clinic Analysis Department, Hospital Infantil Universitario Niño Jesús, Madrid, Spain
18. 18Microbiology Department, Hospital Universitario de Móstoles, Mostoles, Madrid, Spain
19. 19Microbiology Department, Hospital Universitario Rey Juan Carlos, Mostoles, Madrid, Spain
20. 20Microbiology Department, Hospital Universitario Torrecárdenas, Almería, Spain
22. 22Microbiology Department, Hospital Universitario 12 de Octubre, Madrid, Spain
23. 23Microbiology Department, UR Salud. Hospital Universitario Infanta Sofía. Paediatrics Research Group, San Sebastian de los Reyes, Madrid, Spain
24. 24Microbiology Department, Hospital Universitario Clínico San Carlos, Madrid, Spain
25. 25Emergency Pediatric Department, Hospital General Universitario Gregorio Marañón, Madrid, Spain
1. Correspondence to Dr Cinta Moraleda, Department of Pediatrics, Hospital Universitario 12 de Octubre, Madrid 28041, Spain; cintamoraledaredecilla{at}gmail.com

## Abstract

Objectives To evaluate the performance of oral saliva swab (OSS) reverse transcription PCR (RT-PCR) compared with RT-PCR and antigen rapid diagnostic test (Ag-RDT) on nasopharyngeal swabs (NPS) for SARS-CoV-2 in children.

Design Cross-sectional multicentre diagnostic study.

Setting Study nested in a prospective, observational cohort (EPICO-AEP) performed between February and March 2021 including 10 hospitals in Spain.

Patients Children from 0 to 18 years with symptoms compatible with Covid-19 of ≤5 days of duration were included. Two NPS samples (Ag-RDT and RT-PCR) and one OSS sample for RT-PCR were collected.

Main outcome Performance of Ag-RDT and RT-PCR on NPS and RT-PCR on OSS sample for SARS-CoV-2.

Results 1174 children were included, aged 3.8 years (IQR 1.7–9.0); 73/1174 (6.2%) patients tested positive by at least one of the techniques. Sensitivity and specificity of OSS RT-PCR were 72.1% (95% CI 59.7 to 81.9) and 99.6% (95% CI 99 to 99.9), respectively, versus 61.8% (95% CI 49.1 to 73) and 99.9% (95% CI 99.4 to 100) for the Ag-RDT. Kappa index was 0.79 (95% CI 0.72 to 0.88) for OSS RT-PCR and 0.74 (95% CI 0.65 to 0.84) for Ag-RDT versus NPS RT-PCR.

Conclusions RT-PCR on the OSS sample is an accurate option for SARS-CoV-2 testing in children. A less intrusive technique for younger patients, who usually are tested frequently, might increase the number of patients tested.

• Covid-19
• Microbiology
• Molecular Biology
• Virology
• Infectious Disease Medicine

## Data availability statement

Data are available on reasonable request. Not applicable.

This article is made freely available for personal use in accordance with BMJ’s website terms and conditions for the duration of the covid-19 pandemic or until otherwise determined by BMJ. You may use, download and print the article for any lawful, non-commercial purpose (including text and data mining) provided that all copyright notices and trade marks are retained.

https://bmj.com/coronavirus/usage

## Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

### What is already known on this topic

• SARS-CoV-2 saliva tests have not been widely implemented in children.

• The accuracy of reverse transcription PCR (RT-PCR) on saliva and nasopharyngeal swab is similar.

• Studies in this population are scarce, particularly in preschoolers.

• RT-PCR on oral saliva swab is an accurate option for SARS-CoV-2 testing in children.

• RT-PCR on oral saliva swab shows higher sensitivity than the antigen rapid diagnostic test on nasopharyngeal swab.

### How this study might affect research, practice or policy

• Saliva swab may be more acceptable for younger patients than nasopharyngeal swab, possibly increasing the capacity of testing in this age group.

• Saliva swab could be added to the SARS-CoV-2 diagnostic protocols as an alternative specimen to nasopharyngeal swab in children.

## Introduction

Nucleic acid amplification testing in nasopharyngeal swab (NPS) samples is considered as the gold standard technique for the diagnosis of SARS-CoV-2 infection1; however, this technique requires trained staff, is associated with an increased risk of complications (epistaxis, retained swabs, cerebrospinal fluid leak),2 is unpleasant and generates anxiety mainly in children. Indeed, patients with mild symptoms are often put off accessing the test.2 3

Saliva swab reverse transcription PCR (RT-PCR) is an alternative and minimally invasive test for the diagnosis of SARS-CoV-2 infection. Data from a recent systematic review showed that given a proper RT-PCR kit is chosen, the accuracy of RT-PCR on saliva and NPS is similar, particularly in ambulatory patients,4 yet saliva tests have not been widely implemented in children. Indeed, studies in this population are scarce, particularly in preschoolers.5–10

We performed this study with the hypothesis that oral saliva swab (OSS) RT-PCR is as sensitive as RT-PCR on NPS, which is considered the gold standard. Our objective was to analyse the performance of OSS RT-PCR in comparison to (1) the gold standard (RT-PCR on NPS) and (2) the Panbio SARS-CoV-2 antigen rapid test (Ag-RDT) on NPS, a technique widely implemented, in a symptomatic paediatric population evaluated at the emergency departments of 10 Spanish hospitals.

## Methods

### Study design

Cross-sectional multicentre diagnostic study nested in a prospective, observational cohort, the Epidemiological Study of Covid-19 in Children of the Spanish Society of Pediatrics (EPICO-AEP). Participants were children from 0 to 18 years old with symptoms compatible with SARS-CoV-2 infection of ≤5 days of duration, seen in the emergency departments of 10 secondary and tertiary hospitals, in Madrid and Almería, Spain.

The study was performed between February and March 2021. At that time, the peak incidence of Covid-19 in Spain was 559/100 000 individuals.11 Standards for Reporting Diagnostic Accuracy (STARD) 2015 guidelines for reporting diagnostic accuracy studies were followed throughout.12

### SARS-CoV-2 testing

Trained nurses collected three samples consecutively from each patient following a standardised procedure: two NPS samples (for Ag-RDT and RT-PCR) and one OSS sample for RT-PCR. The Panbio SARS-CoV-2 Ag-RDT was purchased from Abbott Rapid Diagnostics (Jena, Germany). OSS samples were collected using the identical brand of the swab for the NPS samples. The procedure included smearing the swab under the tongue, between the gums and lips and in the inner part of both cheeks, but not the pharynx. Older patients were invited to cough with their mouths closed before sampling. A minimum amount of saliva was not necessary for the oral smear. Clinicians recommended patients not eat or drink at least 30 min before sampling. OSS and NPS to be used for RT-PCR were placed in standard viral transport media (Delta Labs, Barcelona, Spain).

The nurses in charge of collecting samples performed Ag-RDT on-site and results were interpreted following the manufacturer’s instructions, whereas OSS and NPS for RT-PCR were immediately transported to each local microbiology laboratory of each participating centre after specimen collection.

RT-PCR testing in OSS and NPS was performed immediately after specimen collection, targeting at least two viral genomic fragments following the manufacturer’s recommendations in each laboratory. RNA was extracted from samples by using the automated systems in place in each of the participating centres. OSS were processed using the same protocol as NPS in each centre. It was considered indeterminate those samples in which only one of the genes was amplified with a high cycle threshold (Ct). Turnaround time was less than 12 hours at all the centres. Nurses who interpreted the Ag-RDT did not have access to the RT-PCR result. Lab technicians had access to the Ag-RDT result. Clinicians attending each patient had the information of the three tests once available. Clinical information was available for all of them.

Patients with discordant diagnostic test results from any of the three different tests were invited to a second visit and a blood sample was drawn for SARS-CoV-2 IgG detection. Samples were performed in less than 24 hours after the researchers were aware of the results and less than 48 hours from the initial clinical visit. IgG antibodies directed against the SARS-CoV-2 surface S1 domain of the spike protein or the internal nucleocapsid protein were measured in the serum samples using commercial enzyme-linked immunosorbent assay (ELISA) or chemiluminescent immunoassay kits, depending on availability in each centre following manufacturer’s instructions. Besides, in these patients, a study of viral viability was performed whenever possible. All the information related to this additional analysis is included in the online supplemental material.

### Data collection

Clinical and laboratory data were collected using RedCap electronic data capture tools13 14 hosted at the 12 de Octubre Hospital.

### Statistical methods

The intended sample size was 1500 patients to estimate a 70% sensitivity for OSS and 90% of sensitivity for Ag-RDT to reach a statistical power of 80% with acceptable precision.

The study population was described using counts and percentages for categorical variables and medians and IQRs for continuous variables. Categorical variables were compared with χ² or Fisher’s test, and continuous variables with Mann-Whitney U test. The performance of the test was evaluated estimating the sensitivity, specificity, negative and positive predictive values and the kappa index. Additionally, we provided the sensitivity and specificity of the different diagnostic techniques stratified by age (≤3 and >3 years old). The performance of the test was evaluated by estimating the sensitivity, specificity, negative and positive predictive values and the kappa index taking RT-PCR on NPS as the gold standard. A confusion matrix was assessed to calculate the diagnostic accuracy. CIs for sensitivity and specificity are exact Clopper-Pearson CIs. Besides, the performance of the three tests was also estimated by Bayesian Latent Class Models (BLCA) using extensions of the three tests in one population model implemented in a simplified interface application.15 The BLCA16 was used to approximate the prevalence and the sensitivities and specificities of all tests. This model does not assume that any test is perfect but considers that each test could be imperfect in diagnosing the true disease status. The true disease status of the patient population was then defined based on overall prevalence (the probability that a patient with suspected SARS-CoV-2 infection is truly infected with SARS-CoV-2). BLCA estimates the prevalence and accuracy of each test based on the observed frequency of the possible combinations of test results.

All the analyses were performed using R software V.4.1.2,17 including caret R package V.6.0-90,18 MKmisc R package V.1.819 and compareGroups Package v.4.0.17

## Results

Overall, 1186 patients were included in the study. Eight children did not provide the three samples, and four children presented indeterminate RT-PCR results (two for OSS and two for NPS) and were excluded. Thus, 1174 children were included in the final analysis (figure 1). Median age was 3.8 years (IQR 1.7–9.0), and 516/1174 (44.0%) were ≤3 years old. A total of 647/1174 (55.1%) children were males. The median duration of symptoms before emergency department admission was 1.0 days (IQR 1.0–2.0) (table 1).

Figure 1

Flow chart of the study and positive results in each of the three evaluated techniques. One NPS antigen rapid test result missing and seven oral saliva swabs collected but not processed due to laboratory workflow problems. NPS, nasopharyngeal swab; RT-PCR, reverse transcription PCR.

Table 1

Clinical characteristics of the children included in the study

In total, 73/1174 (6.2%) patients tested positive by at least one of the diagnostic techniques. Out of those 73 patients, 68 tested positive by NPS RT-PCR (93.2%; 5.8% of the total), 53 by OSS RT-PCR (72.6%; 4.5% of the total) and 43 were positive by NPS Ag-RDT (58.9%; 3.7% of the total) (figure 1).

The comparison of OSS RT-PCR and the Ag-RDT in NPS samples versus NPS RT-PCR (gold standard) is shown in table 2. The overall sensitivity was 72.1% (95% CI 59.7% to 81.9%) for OSS RT-PCR and 61.8% (95% CI 49.1% to 73.0%) for the Ag-RDT. The specificity was 99.6% (95% CI 99.0% to 99.9%) for OSS RT-PCR and 99.9% (95% CI 99.4% to 100%) for the Ag-RDT. The kappa index for the OSS RT-PCR and NPS RT-PCR was 0.79 (95% CI 0.72 to 0.88) versus 0.74 (95% CI 0.65 to 0.84) for the Ag-RDT. OSS RT-PCR showed similar performance in children stratified by age. By contrast, the Ag-RDT showed lower sensitivity in children younger than 3 years (table 3).

Table 2

Diagnostic accuracy and agreement of oral saliva swab RT-PCR and rapid antigen test compared with nasopharyngeal swab RT-PCR

Table 3

Performance of the studied tests compared with the gold standard test stratified by age

The estimated medians with 95% credible intervals (Crs) in the whole cohort and in the BLCA model are shown in table 2. Sensitivity for OSS RT-PCR was 84.8% (95% Cr 71.5%–93.6%) and 72.5% (95% Cr 58.8%–83.6%) for the Ag-RDT. Specificity for OSS RT-PCR was 99.7% (95% Cr 99.2%–99.9%) and 99.9% (95% Cr 99.6%–100%) for the Ag-RDT.

The median Ct values for those patients with positive RT-PCR results were 21.1 (IQR 17.8–30.0) on NPS samples and 32.9 (IQR 28.8–35.6) on OSS samples. The Cts were higher in OSS than in NPS samples (Ct (NPS)=0.5×(Ct saliva)+4.5), p=0.027).

In total, 36/1174 patients (3.1%) tested positive with the three diagnostic tests, 13 cases (1.1%) tested positive only by NPS RT-PCR, 4 cases (0.3%) only by OSS RT-PCR and 1 case (0.1%) only by Ag-RDT (table 4). Ct values for patients with discordant results are shown in online supplemental table S1.

Table 4

Detection of SARS-CoV-2 by RT-PCR on NPS or OSS and Ag-RDT in NPS in all 1174 tested children

A total of 37 patients had a discordant result in at least one of the three tests (table 3 and online supplemental material). All were invited to provide a blood sample for SARS-CoV-2 serology, and it was available for 22/37 (59.5%) patients.

A total of 19/68 (27.9%) patients were positive for NPS RT-PCR and negative for OSS RT-PCR. Of these 19 patients, 4 were serology positive and could be considered not contagious.20 Four patients had OSS RT-PCR as the unique positive test, but three of them with serology available and positive, suggesting old infections.

## Discussion

This study evaluates the performance of three SARS-CoV-2 diagnostic tests in a large paediatric population with Covid-19 symptoms showing that the RT-PCR on OSS is a valid diagnostic option in the paediatric population, even better than the widely used Ag-RDT on NPS. If the turnaround time is sufficiently rapid, OSS RT-PCR would seem to be the better choice for children. The performance of the test improved when NPS RT-PCR was considered as an imperfect gold standard.21 22

NPS is the most widely used sample for SARS-CoV-2 testing, although it is not always well accepted by patients.1–3 23 Several studies have shown that different sample types of saliva (dribble pots, suck swabs, oral and oropharynx swabs) are a valid alternative for SARS-CoV-2 testing, but most of the studies included only adults.4 To our knowledge, this study is the largest paediatric study comparing OSS with the established gold standard. The results are in concordance with the few studies performed on children.5 7 9 10 24 A recent study performed in Dubai with 476 children demonstrated a sensitivity even higher (87.7%) than found here; however, the authors used a saliva sample rather than saliva swabs, and the children were older (mean 10.8 years old).10 Obtaining direct saliva samples in children younger than 3 years of age and children with special needs, who could be at a higher risk of severe Covid-19 (ie, neurological diseases) is challenging. However, unlike direct saliva, oral swabs can be obtained with practically no collaboration from patients. For this reason, we used oral swabs, which is a sample easily feasible and painless for all ages including those children with special needs. We had a high proportion of children younger than 3 years old, which we believe fills the gap of knowledge on the performance of saliva samples at this age. Moreover, analysis by age showed that the accuracy is similar for both groups (≤3 years vs >3 years), unlike the Ag-RDT, which performed poorly in children under 3 years. Children with special needs were not specifically included in this study so the results cannot be directly extrapolated to them.

We found that OSS RT-PCR could detect some cases that were missed by the NPS RT-PCR; four cases were positive for OSS RT-PCR but negative for NPS RT-PCR, as has been described previously.7–9 Also, three of those four cases had positive serology but were negative for the Ag-RDT. These data suggest that the positivity on saliva RT-PCR could remain even longer than in NPS RT-PCR, but a SARS-CoV-2 reinfection cannot be fully discarded.

A similar finding was observed in the majority of the 19 cases with a negative result on the OSS but positive on the NPS. Of these 19, 4 had positive serology. An observational study showed that the concordance between paired samples (NPS and saliva) decreased with time, with saliva false negatives increasing in older infections.25 We observed a lower sensitivity of the Ag-RDT in children younger than 3 years old. Moreover, among the four cases with negative Ag-RDT and positive OSS, three were younger than 3 years. Thus, younger children might benefit from the OSS RT-PCR in testing guidelines possibly before Ag-RDT.

At the time of performing the study, the cost of SARS-CoV-2 Ag-RDT was US$2–US$5, which was around 10 times lower than the average price of NPS RT-PCR including the swab and reagents. OSS RT-PCR uses the same diagnostic approach as NPS RT-PCR, but in a different sample, so the cost in our study was the same. Additionally, RT-PCR tests require the prior acquisition of expensive devices to perform the analysis and specialised staff. Considering only the cost, Ag-RDT is a better choice than RT-PCR tests. Furthermore, the turnaround is lower (15 min vs 2–12 hours). However, compared with OSS RT-PCR, it has some disadvantages, such as a lower sensitivity and NPS being a more uncomfortable sample than OSS. So, OSS RT-PCR could be a good alternative to NPS RT-PCR in case of aiming a high sensitivity test and comfortable for patients. Besides, a new approach for SARS-CoV-2 RT-PCR (SalivaDirect) that simplifies RNA extraction has been published.26 In this study, the standard RNA extraction and the SalivaDirect procedure were performed followed by the same RT-PCR with similar results. The potential implementation of the simplified procedure could greatly reduce the cost and the time of the OSS RT-PCR testing, making this method even more attractive.

No systematic or standardised information of the acceptance of the patients, families and operators were collected in this study, but several publications support that OSS is the preferred technique by patients.27 28 The acceptability of participating in the study (>99%, data not shown) shows how families acknowledge that finding a more accepted technique to test children is needed. Besides, the collection and processing of samples showed scarce difficulties as only eight children did not provide the three samples, and four children presented indeterminate RT-PCR results.

This study has several limitations. RT-PCR tests and serology were not centralised, which may have introduced some heterogeneity in the results and difficulties in reproducibility. It accounted for only one wave, including the alpha variant as dominant (>70% of isolations), and it may not be fully extrapolated to Omicron or future variants. The possibility of reinfections was not addressed, and not all discordant patients had serology available. Finally, this study included only symptomatic patients (≤5 days of duration), so the results cannot be extrapolated to asymptomatic patients and new studies will be necessary to evaluate this test in SARS-CoV-2 contacts and as a screening test in the general paediatric population. However, as a strength, we evaluate a novel, comfortable and easy-to-collect sample in children, which has not been previously thoroughly evaluated in the paediatric population, mainly in the younger children.

In conclusion, OSS RT-PCR seems a more suitable and friendly technique for younger patients who must be tested very frequently and might help to maximise the number of patients tested, playing an important role in the control of the disease.

## Data availability statement

Data are available on reasonable request. Not applicable.

## Ethics statements

### Ethics approval

This study involves human participants and was approved by Ethics Committee of the “Fundación de Investigación Biomédica del Hospital 12 de Octubre” (code 20/101) and by the Ethics Committees of the other participating hospitals. Participants gave informed consent to participate in the study before taking part.

## Acknowledgments

We thank all the patients and families for their participation in the study, and the laboratory staff and clinical staff members who cared for them and the EPICO working group. We also thank Kenneth McCreath for the meaningful comments and suggestions on English writing. K.M. is a medical English consultant funded by Universidad Europea de Madrid.

• ## Supplementary Data

This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.