2.1 Setting and design

For this prospective observational study participants were randomly recruited from September 2021 to July 2022. Recruitment took place at two centers: at the emergency room and the pediatric ward of the Spaarne Gasthuis (hereafter sometimes referred to as “SG”) hospital located in Haarlem (The Netherlands) and at the Public Health Service corona testing facility at Schiphol (hereafter sometimes referred to as “GGD”).

2.2 Participants

Children aged between 0 and 5 years, presenting at one of the centers with symptoms/signs of an URTI, characterized by at least nasal discharge, and an indication for a routine nasal swab, were eligible for participation. Infants were excluded if they had received antiviral therapy in the last 2 weeks before recruitment, were too sick to participate as objectified by a medical health professional, had craniofacial abnormalities and if their parents or caregivers were unable to give informed consent due to insufficient understanding of the Dutch or English language. Among others, gender, age, results of both sampling methods and duration of admission were recorded for each participant. The code of conduct relating to expressions of objection by minors participating in medical research was adhered to.

2.3 Materials and data collection

Two samples were collected for each individual. The eSwab^TM 490CE (COPAN), a swab and a tube with 1 ml liquid Amies preservation medium, was used for collection of the rhinorrhea sample. Using the eSwab nasal discharge was collected at the beginning of the nasal entrance or on the upper lip. The eSwab was taken directly from the nasal discharge or, if this was preferred by the parent(s) or caregiver(s) or the child, from a tissue paper after wiping their nose. When nasal discharge was not present, the child was asked to blow their nose and, when secretion was visible, the sample was taken from the tissue paper. WEPA Satino® prestige 2-layer tissues were used as tissue paper for rhinorrhea collection.

A routine swab was then performed according to standard procedures, generally a (combined) MT nasal (and OP) swab, hereafter termed nasal swab. Nasal swab was chosen since this type of swab was certainly performed in all children.^{16, 18-20} Combined NP and OP swabs were not collected during this study, due to the change in sampling preference during the COVID-19 pandemic.²¹ The iClean® Specimen Collection Flocked swab was used for routine sample collection and the Remel MicroTest^TM M4RT® Transport 3.0 ml tube was used as transport medium. Disposable gloves, face masks and aprons were worn during both sampling procedures.

Ideally the RS and NS were to be taken in a fixed order, immediately after one another, with the RS sample being taken first. However, children did not always have nasal discharge at the exact moment that the routine swab was performed, nonetheless all samples were collected with a maximum of 24 h between two samples.

In addition, a questionnaire was used to evaluate the presenting symptoms of the child's respiratory infection such as the duration of symptoms, the type of symptoms, the color of rhinorrhea and the amount of rhinorrhea. To measure tolerability across both sampling methods, for both children and their parent(s) or caregiver(s), four VAS scores were noted.

2.4 Laboratory analysis

Both samples were stored and transported to the Regional Public Health Laboratory Kennemerland laboratory according to standard procedures at room temperature. Both samples were analyzed within 48 h using multiplex ligation-dependent probe amplification for the presence of 4 bacterial and 18 viral pathogens: bordetella pertussis; chlamydia pneumoniae; legionella pneumophila; mycoplasma pneumoniae; adenovirus; human bocavirus; coronavirus OC43, 229E, and NL63/HKU1; influenza virus A, B, and A(H1N1)pdm09; human metapneumovirus; parainfluenza virus 1, 2, 3, and 4; rhinovirus/enterovirus; RSV A and B (RespiFinder® 2Smart; PathoFinder). In addition, the samples were analyzed for the presence of SARS-CoV-2 using reverse transcriptase PCR. Amplification products were detected with the Roche LightCycler® 480.

2.5 Main outcome measures and statistical analysis

The performance of an RS was compared to that of a (combined) nasal MT (and OP) swab for detecting viral pathogens using multiplex PCR. Data analyses were performed using the statistical software package SPSS version 28. Means with standard deviation (SD) were calculated for normally distributed continuous variables and medians with interquartile range (IQR) for skewed distributed continuous variables. Proportions were calculated for dichotomous and categorical variables. A nonparametric independent sample t-test (Mann−Whitney U test) was used to evaluate the differences in age and duration of symptoms between the two groups. In addition, the chi-square test or the fisher's exact test was used to calculate differences in other characteristics between the two groups.

The primary outcome measures were the sensitivity and specificity of the multiplex PCR and SARS-CoV-2 reverse transcriptase PCR results for both the RS and NS, when using the NS as the gold standard. This was calculated using 2 × 2 contingency tables. The McNemar test was used to test for differences in the proportions. In addition, concordance rates were calculated and confidence intervals (CI) were calculated using epitools.²² Wilson CIs were calculated if the proportion ranged from [5% to 95%] and otherwise Clopper−Pearson exact intervals were calculated. The secondary outcome measures were the tolerability, calculated as the differences in VAS score across both sampling methods, analyzed using the nonparametric Wilcoxon Signed Rank test, assuming a non-normal distribution. The VAS score measurement level was considered as a ratio since absolute differences were evaluated.²³ A p value below 0.05 was considered statistically significant.

3.1 Participant characteristics

A total of 91 children were recruited during a period of 11 months. Three children were not included in the study due to a lack of a reference/NS sample, leading to 88 children being included. The majority of these children (n = 68) were recruited at the emergency department and the pediatric ward of the Spaarne Gasthuis in Haarlem. Twenty children were recruited at the Public Health Service corona testing facility at Schiphol. All children were included in the analysis, including one child with craniofacial abnormality, as there was no contraindication for routine sampling and therefore also not for the RS.

The median age of participants was 1.00 years [IQR 0.00−3.00], ranging from 0 to 5 years, with the majority of children being girls (53%). Characteristics of the participants are presented in Table 1. No significant differences were found between the groups for gender (p = 0.871) or duration of symptoms (p = 0.075), however, there was a significant difference in age (p = 0.003). Median duration of symptoms at the time of presentation was 5 days [IQR 3.00−6.00] with the most commonly reported complaints being coughing (94%), fever (66%), shortness of breath (65%), fatigue (65%) and sneezing (65%). The majority of the reported colors and amounts of rhinorrhea were respectively bright (61%) and mediocre (41%).

When comparing the two centers, some differences in reported complaints were identified. The children recruited at the GGD reported more headaches (35%) and fever (75%) and less earaches (0%) compared to children at the SG center, whose symptoms were more frequently shortness of breath (79%), fatigue (69%), and hoarseness (34%).

Paired swab samples were performed in all included children. MTSs (18%) or combined MTSs and OPSs (49%) were performed in most cases for sample collection, however in 33% of the cases the exact method for sampling of NS is unknown. RS were mainly taken directly from nasal discharge (85%). If nasal discharge was not present at the time of sample collection, but the child was able to blow their nose, the sample was taken from tissue paper (15%). Samples were taken from visible snot or visible moist on the tissue paper and only dry tissues were discarded. Eventually, RSs were taken more frequently after the routine sample collection (70%). An overview of the order and methods of sample collection are described in Table 2. Median time between both sampling methods was 60 min [IQR 1.00−935.00], ranging from 1 min to a maximum of 24 h. Both samples were collected within 1 h in 47 (53%) of the participants and within 6 h in 56 (64%) of the participants.

3.2 Primary outcome measures

Eighty-one (92%) of the 88 participating children were positive for one or more viruses. The RS and NS were both positive for at least one pathogen in 79 (90%) of these children (McNemar's test p = 1.000). Codetections were found in 32 (36%) children, with two and three viruses being simultaneously detected in respectively 19 (22%) and 7 (8%) of the rhinorrhea samples and 19 (22%) and 5 (6%) of the nasal samples. One paired swab sample showed a codetection with four viruses within one participant, namely for rhinovirus, adenovirus, human metapneumovirus, and human bocavirus. Results of the samples are presented in Table 3.

In total, 122 viral pathogens were detected by either sampling method, revealing 15 different viruses. No bacteria were detected with either sampling method. Rhinovirus was the most frequently detected virus, as it was detected by either sampling method in 41 (47%) children. Overall, the RS detected 112 (92%) viral pathogens, compared to 108 (89%) viral pathogens detected by the NS. Adenovirus, human bocavirus and respiratory syncytial virus B were more frequently detected by the RS and human metapneumovirus, rhinovirus and respiratory syncytial virus A were more frequently detected by the NS. The number of pathogens detected by the different sampling methods is presented in Figure 1. Most viruses detected were found between October 2021 and February 2022 and involved viruses such as adenovirus, human bocavirus, human metapneumovirus, rhinovirus and RSV-A & RSV-B. Some viruses were also detected in April, June and July, with the majority being adenovirus, influenza-A, influenza-A(H1N1), parainfluenza-3, rhinovirus, and some RSV-A.

RSs were positive for at least one viral pathogen in 77 of the 79 children with a positive NS, which resulted in a sensitivity for the RS of 97% (95% CI 91%−100%). RSs were negative in 7 of the 9 children with a negative NS, which resulted in a specificity of 78% (95% CI 45%−94%). The results of the RS compared to the reference standard, the NS or (combined) MT nasal (and OP) swab, are presented in Table 4.

Concordance between both samples for the detection of at least one pathogen was 95% (95% CI 89%−99%). Concordance between both samples for a complete match was 78% (95% CI 69%−86%). The concordance rate is presented in Table 5 (further details can be found in the Supporting Information File: Table 5) and an overview of the reported outcome measures is presented in Table 6. In total the RS missed 10 viruses that were detected by the NS. The RS missed rhinovirus five times, human metapneumovirus three times, parainfluenza-4 once, and RSV-A once. Yet when only using the RS as diagnostic sample, 14 additional viruses were found, namely adenovirus in four samples, human bocavirus in three samples, RSV-B in two samples, rhinovirus in two samples, parainfluenza-4 in two samples and influenza-A in one sample. Concordance for a complete match between the NS and rhinorrhea samples taken from tissues was 85% (11/13 samples) and for the NS and rhinorrhea samples it was 77% (58/75 samples).

3.3 Secondary outcome measures

The experience of children, as reported by their parents, during the collection of an RS from nasal discharge or blowing their nose into a handkerchief, was significantly better for the RS (median VAS [IQR]: 2.00 [0.00−3.00]) than their experience with the NS (median VAS [IQR]: 6.00 [4.00−8.00]) (Wilcoxon test p = <0.001). Parents also endured the sampling of RS significantly better (median VAS [IQR]: 0.00 [0.00−2.00]) when compared with that of the NS sample (median VAS [IQR]: 5.00 [2.00−7.75]) (Wilcoxon test p = <0.001). The VAS scores are presented in Figure 2. Significant differences in total median VAS scores were found between the two groups (centers) for both the children (Mann−Whitney U test, p < 0.001) and their parents/caregivers (Mann−Whitney U test, p = 0.001). Sub analysis of VAS scores to correct for sampling order was also performed and can be found in the Supporting Information File: Table 1.

4.1 Strengths and limitations

The present study's crucial strength, as mentioned earlier, is that the use of rhinorrhea samples taken directly from nasal discharge has not been previously studied. Therefore, the evaluation of rhinorrhea samples as a specimen for the detection of respiratory pathogens is a valuable addition to the existing literature and advances the options in this emerging field of noninvasive sampling. Furthermore, a method of sample collection was examined which is less invasive and easier to perform than a routine sampling procedure, which can aid in its capacity for implementation. Moreover, it was hypothesized that the collection of rhinorrhea samples would be better tolerated by young children and their parents, which was also objectified by the VAS score analyses. Finally, the timing of this study, which included sampling during all seasons, perhaps provided a high overall detection rate (92%) of respiratory viruses, including clinically relevant respiratory viruses, such as RSV and influenza. Another virus that became important during the past couple of years has been SARS-CoV-2 and was found in less than 7% (6/88) of the population, all diagnosed at the GGD location. However, the concordance for both sampling methods in detecting SARS-CoV-2 was 100%, meaning RSs could aid in a future pandemic when less invasive methods are required for diagnostics in children.

Despite some encouraging results, it is important to note that the sample size of 88 participants did not allow for pathogen specific conclusions. While cautious conclusions can be drawn concerning the most frequently detected viruses, such as rhinovirus and respiratory syncytial virus, no solid conclusions should be drawn yet regarding other viruses due to the small number of cases found in this population.

Additionally, different sampling methods were used during routine care and in about 1/3 of the cases it was unknown how the nasal swab was performed. The RS was therefore not compared to the gold standard, the combined NP- and OP swab, which might have led to a higher sensitivity for the rhinorrhea samples. Nevertheless, the comparison of rhinorrhea samples with these different sampling techniques is in line with clinical practice and different recommendations.^{16, 18-20} Also, the results should perhaps not yet be generalized to a larger population. This was a two center study and only children admitted to the emergency department, pediatric ward or tested at the Public Health Service testing facility were included in this study. Moreover, while children aged from 0 to 5 years were eligible for participation, median age of the participants was 1 year and therefore these results particularly concern a young population. Practical challenges for implementation in routine care also remain. Children presenting to the emergency department with covid-19 relating complaints, who were admitted to the pediatric ward, were tested for the presence of SARS-CoV-2 using rapid point-of-care testing. These samples were in return then used for additional diagnostics for other respiratory viruses using multiplex PCR. However, the use of rhinorrhea samples for rapid point-of-care testing has not been explored and therefore should be studied as its ability to detect viruses is absolutely present. Nonetheless, RSs might not replace nasal swabs altogether, since nasal discharge is not always present at the moment sample collection and pathogen detection is warranted. A possible solution for this, as mentioned earlier, would be to use another noninvasive sampling method such as saliva which is also considered a strong reservoir for viruses. A possible future study should therefore evaluate the performance of rhinorrhea samples in comparison to saliva samples. In addition to this, multicenter studies in the Netherlands, as well as international studies could be performed to expand the sample size and assist in detection of viral pathogens in third world countries. Moreover, the use of self-testing of nasal discharge (antigen testing) might be a possibility worth exploring, as it might lead to a higher willingness to test in both children and their parents. Lastly, the detection of viral pathogens in nasal discharge could be studied among older children or even adults to see if this could aid in diagnostics in those patient groups.