Abstract
Accurate detection of respiratory viruses is essential for infection control, patient management and public health response. Nasopharyngeal swabs (NPS), collected by healthcare workers (HWC-NPS), remain the gold standard for Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) detection but require trained personnel and can be uncomfortable for patients. Self-collected swabs, such as oral–nasal swabs (SC-ONS) and mid-turbinate swabs, offer scalable alternatives suitable for mass testing. Previous mid-turbinate swab designs included a plastic stopper and were not compatible with automated laboratory testing. This study compared the performance of (i) SC-ONS, (ii) an automation-friendly, redesigned version of mid-turbinate swab (SC-MTS) and (iii) HWC-NPS in detecting SARS-CoV-2. Between April and June 2022, paired NPS, ONS and MTS samples were collected from 100 participants at a Coronavirus disease 2019 (COVID-19) assessment centre in Hamilton, Ontario. Samples were tested for SARS-CoV-2 by reverse-transcriptase PCR. Compared to HWC-NPS, SC-ONS demonstrated 82.1% sensitivity and 100% specificity, while SC-MTS showed 75.0% sensitivity and 100% specificity. Agreement with HWC-NPS was strong for both SC-ONS (κ=0.863) and SC-MTS (κ=0.804). Agreement between SC-ONS and SC-MTS was nearly perfect (κ=0.944). Cellular material yields as well as SARS-CoV-2 viral loads were lower for self-collected swabs than HWC-NPS, while viral load comparisons revealed no significant difference between SC-ONS and SC-MTS. Our study aligns with previous work demonstrating the use of self-collected mid-turbinate swabs for the detection of respiratory viruses, while also demonstrating the compatibility of the newly designed swabs with automated laboratory instruments. Our results, while conducted in a relatively small sample, suggest that self-collected ONS and MTS are reliable alternatives to HWC-NPS, offering practical, less invasive, automation-friendly options for large-scale respiratory virus surveillance and pandemic preparedness.
Keywords: SARS-COV-2, PCR
Data Summary
In accordance with ethics review board approval, individual-level patient data are not available to maintain patient anonymity.
Background
Since December 2019, there have been over 750 million confirmed Coronavirus disease 2019(COVID-19) cases and over 7 million COVID-19 deaths worldwide [1]. Accurate detection of respiratory viruses, including SARS-CoV-2, is crucial in limiting the spread of infection, implementing appropriate public health measures and managing patients who are severely or critically ill with respiratory viral infections. Currently, standard respiratory virus testing relies on the collection of nasopharyngeal swab (NPS) [2] specimens and amplification of viral genomic targets by real-time, reverse-transcriptase PCR.
While the sensitivity of NPS testing is high, collection must be performed by licensed healthcare professionals, and some patients may experience discomfort during the procedure. Non-NPS, self-collected methods of swabbing are ideal for scenarios where individuals do not require physician assessment. Over the course of the COVID-19 pandemic, self-swabbing has facilitated large-scale COVID-19 testing programmes at schools and workplaces and self-collected oral-nasal swabs (SC-ONS) are an adequate alternative to healthcare worker-collected NPS (HWC-NPS), with 90% sensitivity for detecting SARS-CoV-2 [3].
Additionally, mid-turbinate nasal (MT) swabbing has previously been shown to have 80–90% agreement with NPS swabbing for SARS-CoV-2 detection; however, most studies were performed using universal flocked swabs, and collection was performed by a healthcare provider [4,5]. Previously, Copan Italia developed a contoured mid-turbinate swab (MTS), designed specifically for self-swabbing with a stopper 5.5 cm along the shaft. The MTS consists of a tapered cone-shaped design, intended to sample a larger surface area of respiratory mucosa [6]. In previous studies, it was found to be comparable to HWC-NPS in terms of epithelial cell counts, and self-collection using this swab was able to detect several seasonal respiratory viruses with >70% sensitivity [6,7]. Due to the implementation of high-throughput robotic liquid handling systems to increase sample testing capacity during the SARS-CoV-2 pandemic, the swab was redesigned without the stopper to be compatible with laboratory automation. While the use of the MTS has been well-established in the literature, the redesigned MTS has not yet been evaluated on automated liquid handling instruments for the detection of SARS-CoV-2. In this study, we compare HWC-NPS, SC-ONS and self-collected MTS (SC-MTS) using the redesigned MTS for the detection of SARS-CoV-2.
Methods
MTS, oral–nasal swab (ONS) and NPS samples were collected in parallel from patients presenting to a Hamilton COVID-19 assessment centre between April and June 2022, for a prospective comparison of the three sample types. Any individual presenting to the assessment centre for COVID-19 testing was eligible for the study. Ethical approval was received from the Hamilton Integrated Research Ethics Board, and verbal consent was obtained from all study participants. NPS samples were collected by a healthcare professional first in all cases and were retrieved from the regional laboratory following clinical testing for SARS-CoV-2. Participants were provided instructions for self-collection of ONS and MTS samples by a member of the research team. Supervised ONS samples were collected by first inserting the swab into the buccal area and rotating on each side of the mouth, followed by inserting the swab into the anterior nares and rotating. Self-collection of the MTS or ONS swab was alternated in a randomized fashion to account for possible sampling variability due to swab collection order [8]. All swabs were collected into McMaster Molecular Medium and tested for SARS-CoV-2 using a validated, laboratory-developed RT-PCR assay targeting the 5′-UTR and E- genes of SARS-CoV-2 and RNaseP, a human gene, used as a sample adequacy control. Quantitation was performed using an N1 standard (Exact Diagnostics) and an RNaseP standard (gBlock standard, Integrated DNA Technologies). Swabs were considered positive if two or more of the SARS-CoV-2 targets were positive with a CT value <40. The 5′ UTR and E-gene regions of SARS-CoV-2 with similar Ct cut-off values for positivity were well utilized for SARS-CoV-2 testing during the COVID-19 pandemic in Canadian laboratories [9]. Sensitivity and specificity of the ONS and MTS were calculated in comparison to the NPS as the gold standard collection method, and agreement was assessed using Cohen’s kappa coefficient. Formulas for the calculation of sensitivity, specificity and accuracy are available in Supplementary Material 1 (available in the online Supplementary Material). NPS samples that could not be retrieved for the research study were excluded from sensitivity, specificity and agreement calculations. Sensitivity and specificity of the MTS were also calculated in comparison to the ONS as the gold standard due to its widespread use. One-way ANOVA and paired t-tests were used to compare the mean number of cells collected (Log10RNaseP copies/reaction) and mean viral load (Log10N1 copies/reaction) for each swab type. Assumptions of normality were assessed using the Shapiro–Wilk test and visual inspection of Q-Q plots, while homogeneity of variance was evaluated using Levene’s test. All samples were processed on the Hamilton Microlab Star (Hamilton Company, Reno, NV) liquid handling instruments without the removal of the swabs before processing.
Results
One hundred participants were enrolled in this clinical evaluation. Age and sex characteristics of the enrolled participants are available in Table 1. Overall, 28% (28/100) of participants tested positive by at least one swab type. Eighty-nine participants had all three swab types available for analysis; 11 NPS specimens were not available after standard-of-care testing. ONS and MTS swab pairs were available for all 100 patients (Table 1). For detection of SARS-CoV-2, compared to HWC-NPS, the sensitivity and specificity of SC-ONS were calculated to be 82.1% (95% CI=63.1, 93.9) and 100% (95% CI=94.1, 100), respectively, with an overall agreement of 94.4% (κ=0.86; 95% CI=87.5, 97.6, P<0.001) (Table 2). The sensitivity and specificity of the SC-MTS were calculated to be 75.0% (95% CI=55.1, 89.3) and 100% (95% CI=94.1, 100), respectively, compared to HWC-NPS, and overall agreement was 92.1% (κ=0.80, 95% CI=0.67, 0.94) (Table 3). Overall agreement between the SC-ONS and SC-MTS was 98.0% (95% CI=93.0, 99.5), κ=0.94 (95% CI=0.87, 1.00). Kappa values indicate strong agreement between SC-ONS and SC-MTS with HWC-NPS and near-perfect agreement between SC-ONS and SC-MTS for the detection of SARS-CoV-2.
Table 1. Characteristics of enrolled participants and overall swab testing results.
| Characteristic | No. (%) | |
|---|---|---|
| Number of participants | 100 | |
| Age (years) | Min-max | 20–76 |
| Mean (sd) | 39.4 (14.3) | |
| Median | 36.5 | |
| Sex | Male (%) | 23 (23) |
| Female (%) | 77 (77) | |
| ONS SARS-CoV-2 result | Positive (%) | 24 (24) |
| Negative (%) | 76 (76) | |
| MTS SARS-CoV-2 result | Positive (%) | 22 (22) |
| Negative (%) | 78 (78) | |
| Number of participants with HCW-NPS available for testing | 89 | |
| NPS SARS-CoV-2 result | Positive (%) | 28 (31.5) |
| Negative (%) | 61 (68.5) | |
Table 2. Comparison of SARS-COV-2 results for different swab types.
| Statistic | NPS vs. ONS (n=89) |
NPS vs. MTS (n=89) |
ONS vs. MTS (n=100) |
|||
|---|---|---|---|---|---|---|
| Value | 95% CI | Value | 95% CI | Value | 95% CI | |
| True positive | 23 | – | 21 | – | 22 | – |
| False positive | 0 | – | 0 | – | 0 | – |
| True negative | 61 | – | 61 | – | 76 | – |
| False negative | 5 | – | 7 | – | 2 | – |
| Sensitivity | 82.1% | 63.1% to 93.9% | 75.0% | 55.1% to 98.3% | 91.67% | 73.0% to 99.0% |
| Specificity | 100.0% | 94.1% to100.0% | 100.0% | 94.1% to100.0% | 100.0% | 95.26% to100.0% |
| Accuracy | 94.4% | 87.5% to 97.6% | 92.1% | 84.6% to 96.1% | 98.0% | 93.0% to 99.5% |
| PPV | 100.0% | – | 100.0% | – | 100.0% | – |
| NPV | 92.4 | 84.5% to 96.8% | 89.7 | 84.5% to 96.8% | 97.4% | 90.98% to 99.31% |
| kappa | 0.863 | 0.747 to 0.979, P<0.001 | 0.804 | 0.669 to 0.941, P<0.001 | 0.944 | 0.868 to 1.020, P<0.001 |
Table 3. Mean, minimum and maximum Ct values by gene targets for the three swab types.
| Swab | Measure (unit) | N | Mean (sd) | Minimum | Maximum |
|---|---|---|---|---|---|
| SC-ONS (n=100) | E-gene (CT) | 24 | 24.78 (5.06) | 16.07 | 35.50 |
| UTR (CT) | 24 | 25.82 (5.00) | 17.05 | 36.35 | |
| N1 (CT) | 24 | 23.17 (2.04) | 15.55 | 32.95 | |
| Log10N1 (copies/reaction) | 24 | 4.72 (1.51) | 1.83 | 6.99 | |
| RNaseP (CT) | 100 | 24.41 (2.04) | 20.61 | 32.98 | |
| Log10RNaseP (copies/reaction) | 100 | 5.53 (0.78) | 2.35 | 6.92 | |
| SC-MTS (n=100) | E-gene (CT) | 22 | 23.45 (4.71) | 16.74 | 32.74 |
| UTR (CT) | 22 | 24.86 (4.64) | 18.16 | 34.13 | |
| N1 (CT) | 22 | 22.24 (4.77) | 15.57 | 33.45 | |
| Log10N1 (copies/reaction) | 22 | 5.00 (1.42) | 1.69 | 6.89 | |
| RNaseP (CT) | 100 | 24.39 (1.59) | 20.09 | 29.30 | |
| Log10RNaseP (copies/reaction) | 100 | 5.54 (0.58) | 3.67 | 7.36 | |
| HWC-NPS (n=89) | E-gene (CT) | 27 | 23.53 (8.04) | 11.64 | 38.85 |
| UTR (CT) | 26 | 24.14 (7.43) | 13.16 | 39.16 | |
| N1 (CT) | 28 | 21.13 (8.31) | 11.06 | 39.83 | |
| Log10N1 (copies/reaction) | 28 | 5.32 (2.45) | −0.19 | 8.31 | |
| RNaseP (CT) | 89 | 22.72 (1.82) | 18.80 | 32.38 | |
| Log10RNaseP (copies/reaction) | 89 | 6.15 (0.66) | 2.56 | 7.53 |
To assess whether the three swab types collected equivalent mean amounts of cellular material, a one-way ANOVA was performed using Log10RNaseP copies/reaction as the dependent variable and swab type as the independent variable. Analysis revealed that there was a statistically significant difference in mean Log10RNaseP values between at least two groups [F(2, 286)=27.0, P<0.001]. Post hoc Tukey’s Honestly Signficant Difference (HSD) analysis found that the mean Log10RNaseP value was significantly different between both SC-ONS and HWC-NPS (−0.63, 95% CI=−0.86, –0.39, P<0.001) and SC-MTS and HWC-NPS (−0.62, 95% CI=–0.85, –0.39, P<0.001). No statistically significant difference was observed between the mean Log10RNaseP value of the SC-ONS and SC-MTS (−0.01, 95% CI=−0.23, 0.22, P=0.998) (Table 3, Fig. 1a). Subsequent paired t-tests revealed similar findings with statistically significant differences in mean Log10RNaseP copies/reaction between both SC-ONS and SC-MTS compared to HWC-NPS, but not between SC-ONS and SC-MTS.
Fig. 1. (a) RNaseP quantitation for different swab types in Log10RNaseP copies/reaction. Mean Log10RNaseP for SC-ONS, SC-MTS and HWC-NPS are 5.53±0.78, 5.54±0.58 and 6.16±0.66 copies/reaction, respectively. Mean difference in Log10RNaseP copies/reaction per Tukey’s Least Significant Difference (LSD) post hoc: SC-ONS vs. HWC-NPS=−0.63 (CI= −0.86,–0.39, P<0.001); SC-MTS vs. HWC-NPS=−0.62 (CI=−0.85,–0.39, P<0.001); SC-ONS vs. SC-MTS=−0.005 (CI=−0.23, 0.22, P=0.998). Circles (○) represent mild outliers calculated as greater than 1.5 times the interquartile range above Q3 or below Q1. Asterisks (*) represent extreme outliers calculated as greater than three times the interquartile range above Q3 or below Q1. (b) SARS-CoV-2 viral load (based on Ct values for N1) by swab types in Log10N1 copies/reaction. Mean Log10N1 for SC-ONS, SC-MTS and HWC-NPS are 4.72±1.51, 5.00±1.42 and 5.32±2.45 copies/reaction, respectively. Mean difference in Log10N1 copies/reaction by paired t-test: SC-ONS vs. HWC-NPS=−1.49 (95% CI=−2.11, –0.88, P<0.001); SC-MTS vs. HWC-NPS=−1.48 (95% CI=−2.11, –0.78, P<0.001); SC-ONS vs. SC-MTS=−0.02 (95% CI=−0.28, 0.4, P=0.870). Circles (○) represent mild outliers calculated as greater than 1.5 times the interquartile range above Q3 or below Q1.
One-way ANOVA comparing mean viral load of the three swab types revealed no statistically significant difference between mean Log10N1 values in any of the pairings [F(2,72)=0.725, P=0.488]. However, a subsequent paired t-test analysis revealed a statistically significant difference in mean Log10N1 copies/reaction between SC-ONS and HWC-NPS (−1.49, 95% CI=−2.11, –0.88, P<0.001) as well as SC-MTS and HWC-NPS (−1.48, 95% CI=−2.11, –0.78, P<0.001). No statistically significant difference was observed for mean Log10N1 values between SC-ONS and SC-MTS (−0.02, 95% CI=−0.28, 0.24, P=0.870) (Table 3, Fig. 1b).
Discussion
While NPS sampling remains the gold standard for detection of SARS-CoV-2 and other respiratory viruses, collection is limited to healthcare workers and can be uncomfortable for the patient. SC-ONSs have previously been shown to be an adequate alternative for at-home or resource-limited swabbing and have been implemented into practice. Mid-turbinate swabbing has also been shown to be a good alternative to nasopharyngeal swabs for the detection of respiratory viruses, but previous studies have included swabs collected by a healthcare worker [4,5]. Here, we support findings from previous studies, which show SC-MTS and SC-ONS to be acceptable alternatives to HWC-NPS and demonstrate the utility of the automation-friendly MTS [10,11].
In this study, we demonstrate that SC-ONS and SC-MTS have good agreement with HWC-NPS for the detection of SARS-CoV-2. Further, SC-ONS and SC-MTS had near-perfect agreement for the detection of SARS-CoV-2 when compared to one another. Comparison of means analysis showed no statistically significant difference in the amount of cellular material collected by SC-ONS and SC-MTS. However, HWC-NPS collected a significantly higher (P<0.001) amount of cellular material when compared to both the SC-ONS and SC-MTS, which may be due to the deeper nature of the nasopharyngeal swab, although this did not contribute to any differences in positivity for the RNaseP target. When comparing viral loads, statistically significant differences were seen when comparing HWC-NPS and SC-ONS and HWC-NPS and SC-MTS, but only in paired specimens. Lower viral loads observed in SC-ONS and SC-MTS may contribute to the slightly lower sensitivity observed in these swab types.
While SARS-CoV-2 has now shifted from a pandemic to an endemic virus, ongoing surveillance remains important, especially when coupled with testing for additional pathogens during the respiratory season. Given the shared nature of specimen collection and testing for many respiratory bacteria and viruses, the findings of this study could be expanded to the detection of other targets, including emerging pathogens. Additionally, the COVID-19 pandemic has taught us the importance of preparedness and redundancy. At the height of the pandemic, swabs were in short supply and labs across the world were compelled to use whatever swabs were available. Many of these swabs were not compatible with high-capacity automated instruments, which meant that an upfront manual step, such as sample aliquoting or swab removal, was required. The newly designed mid-turbinate swab (SC-MTS) is compatible with the Hamilton Microlab STAR liquid handling instruments, allowing for high-throughput testing. This is an important feature as automation can reduce turnaround times and increase the scalability of testing. Further, as previous studies evaluating mid-turbinate swabs consisted of swabs collected by a healthcare professional [4,5], we demonstrate the potential use of SC-MTS for a future pandemic scenario.
One limitation observed in this study was the relatively low number of SARS-CoV-2-positive specimens and wide confidence intervals in sensitivity calculations for the SC-ONS and SC-MTS. Further, the study was conducted in a single centre, and generalizability may be affected due to the small sample size and demographic skew towards female participants. Additionally, for several participants, the HWC-NPS was not available for further testing after standard-of-care. This reduced the sample size for pairwise comparisons.
Overall, self-collection using the redesigned MTS may be an adequate alternative to an HWC-NPS since it can be self-collected with less discomfort. Self-collection using the redesigned MTS is shown to be a comparable alternative to the SC-ONS, which is currently in widespread use. While testing requirements for COVID-19 have decreased, the insights from this study may still apply to novel variants or other respiratory viruses in a future pandemic scenario.
Supplementary material
Abbreviations
- HWC-NPS
healthcare worker-collected NPS
- MT
mid-turbinate nasal
- MTS
mid-turbinate nasal swab
- NPS
nasopharyngeal swab
- ONS
oral–nasal swab
- SC-MTS
self-collected mid-turbinate swab
- SC-ONS
self-collected oral–nasal swab
Footnotes
Funding: This work received no specific grant from any funding agency.
Ethical statement: This study was reviewed by the Hamilton Integrated Research Ethics Board.
Author contributions: J.G.: conceptualization, formal analysis, investigation, methodology, project administration, writing – original draft and writing – review and editing. K.W.: investigation and writing – review and editing. S.M.: investigation and writing – review and editing. D.W.: investigation and writing – review and editing. K.V.D.: investigation and writing – review and editing. K.B.: investigation and writing – review and editing. A.M.: investigation and writing – review and editing. D.B.: conceptualization, project administration, methodology, supervision and writing – review and editing. M.R.H.: writing – review and editing. M.S.: conceptualization, supervision, project administration and writing – review and editing.
Contributor Information
Jodi Gilchrist, Email: gilchrij@mcmaster.ca.
Kelly Waters, Email: waterk1@mcmaster.ca.
Sarah Marttala, Email: smarttal@stjoes.ca.
Doris Williams, Email: dwilliams@stjoes.ca.
Kaitlyn Van Deventer, Email: kjv.883@gmail.com.
Keltie Baldwin, Email: keltiebaldwin@gmail.com.
Alexandria Martindale, Email: a_martindale@outlook.com.
David Bulir, Email: bulirdc@mcmaster.ca.
Mohammad Rubayet Hasan, Email: hasanm51@mcmaster.ca.
Marek Smieja, Email: smiejam@mcmaster.ca.
References
- 1.World Health Organization (May 2024) World Health Organization; 2024. WHO COVID-19 Dashboard.https://data.who.int/dashboards/covid19/deaths?n=o [Google Scholar]
- 2.Miller JM, Binnicker MJ, Campbell S, Carroll KC, Chapin KC, et al. A guide to utilization of the microbiology laboratory for diagnosis of infectious diseases: 2018 update by the infectious diseases Society of America and the American Society for Microbiology. Clin Infect Dis. 2018;67:e1–e94. doi: 10.1093/cid/ciy381. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Kandel CE, Young M, Serbanescu MA, Powis JE, Bulir D, et al. Detection of severe acute respiratory coronavirus virus 2 (SARS-CoV-2) in outpatients: a multicenter comparison of self-collected saline gargle, oral swab, and combined oral-anterior nasal swab to a provider collected nasopharyngeal swab. Infect Control Hosp Epidemiol. 2021;42:1340–1344. doi: 10.1017/ice.2021.2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Berenger BM, Fonseca K, Schneider AR, Hu J, Zelyas N. Clinical evaluation of nasopharyngeal, midturbinate nasal and oropharyngeal swabs for the detection of SARS-CoV-2. Diagn Microbiol Infect Dis. 2022;102:115618. doi: 10.1016/j.diagmicrobio.2021.115618. Epub 2021 Dec 16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Vos MB, Gonzalez MD, Stone C, Cleeton R, Figueroa J, et al. Comparison of mid-turbinate nasal swabs, saliva, and nasopharyngeal swabs for SARS-CoV-2 reverse transcription-polymerase chain reaction testing in pediatric outpatients. Arch Pathol Lab Med. 2022;146:1056–1061. doi: 10.5858/arpa.2021-0625-SA. [DOI] [PubMed] [Google Scholar]
- 6.Smieja M, Castriciano S, Carruthers S, So G, Chong S, et al. Development and evaluation of a flocked nasal midturbinate swab for self-collection in respiratory virus infection diagnostic testing. J Clin Microbiol. 2010;48:3340–3342. doi: 10.1128/JCM.02235-09. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Larios OE, Coleman BL, Drews SJ, Mazzulli T, Borgundvaag B, et al. Self-collected mid-turbinate swabs for the detection of respiratory viruses in adults with acute respiratory illnesses. PLoS One. 2011;6:e21335. doi: 10.1371/journal.pone.0021335. Epub 2011 Jun 23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Van Wesenbeeck L, Meeuws H, D’Haese D, Ispas G, Houspie L, et al. Sampling variability between two mid-turbinate swabs of the same patient has implications for influenza viral load monitoring. Virol J. 2014;11:233. doi: 10.1186/s12985-014-0233-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.LeBlanc JJ, Gubbay JB, Li Y, Needle R, Arneson SR, et al.COVID-19 pandemic diagnostics investigation team of the canadian public health laboratory network (CPHLN) respiratory virus working group. real-time PCR-based SARS-cov-2 detection in canadian laboratories J Clin Virol 2020128104433.Epub 2020 May 13 10.1016/j.jcv.2020.104433 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.D’Andrea EL, Cossu AM, Scrima M, Messina V, Iuliano P, et al. Efficacy of unsupervised self-collected mid-turbinate FLOQSwabs for the diagnosis of coronavirus disease 2019 (COVID-19) Viruses. 2021;13:1663. doi: 10.3390/v13081663. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Lim HJ, Baek YH, Park MY, Yang JH, Kim MJ, et al. Performance analysis of self-collected nasal and oral swabs for detection of SARS-CoV-2. Diagnostics . 2022;12:2279. doi: 10.3390/diagnostics12102279. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.