Self-collection: an appropriate alternative during the SARS-CoV-2 pandemic

21 22

23 Swabs for SARS-CoV-2 are routinely collected by health care workers, putting them at risk 24 of infection and requiring use of personal protective equipment (PPE). Self-collected swabs 25 offer many advantages provided detection rate of SARS-CoV-2 and other respiratory viruses 26 is not compromised. 27

28 In a prospective study, patients attending dedicated COVID-19 collection clinics were offered 29 the option to first self-collect (SC) nasal and throat swabs prior to health worker collection 30 (HC). Two different laboratory services participated, with HC at Site 1 collecting nasal and 31 throat swabs and at Site 2 nasopharyngeal (NP) and throat swabs. Samples were analysed for 32 SARS-CoV-2 as well as common respiratory viruses. Concordance of results between 33

was assessed using Cohen’s kappa (κ ). 34

35 Of 236 patients sampled by HC and SC, 25 had COVID-19 (24 by HC and 25 by SC) and 63 36 had other respiratory viruses (56 by HC and 58 by SC). SC was highly concordant with HC 37 (κ = 0.890) for all viruses including SARS-CoV-2 and more concordant than HC to positive 38

by any method (κ = 0.959 vs 0.933). 39

40 Self-collection of throat and nasal swabs offers a reliable alternative to health worker 41 collection for the diagnosis of SARS-CoV-2 and other common respiratory viruses. High 42 viral load of SARS-CoV-2 throughout the respiratory tract and sensitive molecular methods 43 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted April 11, 2020. ; https://doi.org/10.1101/2020.04.09.20057901doi: medRxiv preprint 3 may explain these findings. Self-collection also provides patients with easier access to 44 testing, reduces the exposure of the community and health workers to those undergoing 45 testing and reduces the requirement for PPE. 46 47

48 On the 11th March 2020, the World Health Organisation (WHO) announced COVID-19 as a 49 pandemic.1 The WHO Director-General issued a call for urgent action and encouraged all 50 countries to ‘innovate and learn’ in their response to this crisis. 51 Demands on health services have increased and a commensurate decrease in availability of 52 personal protective equipment (PPE) has occurred whilst the protection of health staff and the 53 community remain paramount. Self-collected swabs in the community for SARS-CoV-2, the 54 agent of COVID-19, and for other respiratory viruses offers potential significant benefit in 55 the current pandemic by reducing requirement for PPE, and limiting exposure of patients and 56 staff to infection. 57 Self-collection for respiratory viruses is not a new concept. Benefits include increased 58 convenience and access for patients and timeliness of a sample receipt.2,3 Patients report self-59 collected nasal swabs are easy to perform2,4,5 and highly acceptable.2,4 A meta-analysis of 9 60 studies comparing self-collect (SC) and health care worker collect (HC) for influenza testing 61 reported a pooled sensitivity of 87% and specificity of 99% for SC compared to HC6 however 62 sensitivity for other respiratory viruses was not studied. Irving et al7 studied paired samples 63 from 240 adults and found sensitivity using nasal or nasopharyngeal (NP) collection for 64 influenza did not vary significantly when using a highly sensitive molecular test.7 A study in 65 230 children reported equivalent sensitivity for all respiratory viruses except respiratory 66 syncytial virus (RSV) when comparing nasal swab and NP aspirate.8 Larios et al9 67 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted April 11, 2020. ; https://doi.org/10.1101/2020.04.09.20057901doi: medRxiv preprint 4 demonstrated that using flocked swabs and sensitive molecular methods, equivalent 68 sensitivity and specificity was obtained for 76 matched self-collected mid-turbinate nasal 69 swabs and NP swabs in 38 individuals for a range of respiratory viruses including human 70 coronaviruses (hCoV 229E/NL63 and hCoV OC43/HKU1). 71 Recent reports on SARS-CoV-2 in respiratory specimens indicate early high viral loads in 72 symptomatic and asymptomatic patients in a variety of clinical specimens including nasal and 73 throat swabs, sputum and saliva samples.10-14 Wang et al reported that in 205 patients with 74 COVID-19 the highest positive rates were found from bronchoalveolar lavage fluid, sputum 75 and nasal swabs respectively.15 Wolfel14 and colleagues reported that in hospitalized cases of 76 COVID-19 there was no discernible difference between NP and throat swabs with high viral 77 load present in both specimens early in the illness and suggested that simple throat swabs 78 may provide sufficient sensitivity when patients are first tested with mild symptoms of 79 COVID-19. 80 The aim of this study was to compare prospectively the performance of HC with separate SC 81 nasal (SCN) and throat swabs (SCT) and the combination of the two (SCNT) for respiratory 82 viruses including SARS-CoV-2. 83 84

85 This study was conducted across two laboratory sites (Site 1 and Site 2) and had ethics 86 approval from the Western Australian branch of the Australian Medical Association, with all 87 participants providing informed consent. For a period of one week in March 2020, patients 88 presenting for SARS-CoV-2 testing at dedicated COVID-19 collection rooms were offered 89 participation in the study. Demographic data was recorded including the address postcode to 90 assess the Index of Education and Occupation (IEO) which assesses education level based on 91 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted April 11, 2020. ; https://doi.org/10.1101/2020.04.09.20057901doi: medRxiv preprint 5 a scale of 1 to 5 with 5 being the highest level of education.16 A questionnaire assessing 92 acceptability of SC based on that of Akmatov4 was provided to patients. Printed instructions 93 including diagrams were provided on how to collect throat and nasal swab (See 94 Supplementary Information). Self-collection kits included two swab packets each containing 95 a single swab and screw-top container with 2mL liquid Amies medium, a tongue depressor 96 and a zip lock sample bag. SC samples were taken immediately prior to trained HC samples 97 to reduce ‘training bias.’ For SC and HC at Site 1 and SC at Site 2, open-cell polyurethane 98 foam swabs (Σ Transwab® ref MW940, Medical Wire & Equipment (MWE), Wiltshire, 99 England) were used. Throat swabs were collected from the posterior throat and tonsil areas 100 while nasal swabs were inserted as far as comfortably possible and at least 2-3 cm inside one 101 nostril, rotating the swab 5 times and leaving in place for 5-10 seconds. For HC at Site 2, a 102 flocked NP swab and a foam throat swab (Σ Transwab® ref MW819 and MW940) were used. 103 In addition, because the expected SARS-CoV-2 positivity rate at the time was estimated to be 104 less than 1%, a subset of 24 patients recently diagnosed with COVID-19 performed SC in 105 their homes. 106 At site 1, testing for SARS-CoV-2 was on the AllplexTM 2019-nCoV Assay (Seegene, Seoul, 107 South Korea) and followed sample extraction using MagNA Pure 96 (Roche, Basel, 108 Switzerland) with amplification utilising CFX96 Touch RT-PCR Detection Systems (BioRad, 109 Hercules, California USA). Samples were confirmed as SARS-CoV-2 positive if all three 110 gene targets (E/RdRp and N genes) were detected within 40 cycles. At site 2, the same 111 extraction method was used. Testing for SARS-CoV-2 was performed using an in-house 112 developed Taqman assay targeting the E gene.17 All positive samples then underwent 3 113 supplementary RT-PCRs targeting the N gene.18 Both laboratories utilised the Seegene RV 114 Essential assay to detect other respiratory viruses (influenza A, influenza B, parainfluenza, 115 RSV, human metapneumovirus (HMPV), adenovirus and rhinovirus). 116 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted April 11, 2020. ; https://doi.org/10.1101/2020.04.09.20057901doi: medRxiv preprint 6 Statistical methods 117 A positive result on either HC or SC was defined as the benchmark result All Positives (AP). 118 Concordance between HC and SC swabs and AP was calculated using Cohen's Kappa (κ ), 119 which measures agreement between the categorical assignments given by two methods. The 120 statistic takes values typically between zero and one. A κ >0.80 indicates very good 121 agreement, while κ =1 indicates perfect concordance. Cycle threshold (Ct) values were 122 recorded for all positive test results as a surrogate measure for viral load. Mean Ct was 123 compared between HC and SCNT (combined category using the lowest Ct of either SCN or 124 SCT), using linear mixed effects models, with a random effect for patient identification. HC 125 and SC SARS-CoV-2 positivity rates were compared with Pearson’s χ 2 test. 126 From power calculations assuming a significance level of 5% and a null hypothesis of low 127 concordance between the HC and SC methods (H0: κ =0.3), there was at least 80% power to 128 detect a concordance of 0.6 or more with a sample size of 66. Significance level α was set at 129 0.05, however for concordance and regression analyses, a Bonferroni multiple testing 130 correction was applied such that minimum α '=0.05/8=0.0063. Statistical analyses were 131 completed in the R statistical computing environment,19 including the package irr. 132 133

134 A total of 236 participants across the two sites took part in this study. Median age of 135 participants was 40 (range 9-81) years and 60% were female. Twenty-five patients were 136 positive for SARS-CoV-2 and 63 patients positive for other common respiratory viruses. For 137 SARS-CoV-2 cases, 24/25 were detected by HC and 25/25 by SC. For common respiratory 138 viruses 56/63 (89%) were detected by HC and 58/63 (92%) by SC (Table 1). A positive result 139 on either HC or SCNT was included in the group AP. 140 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted April 11, 2020. ; https://doi.org/10.1101/2020.04.09.20057901doi: medRxiv preprint 7 Table 2 summarises the respiratory viruses detected by the different methods of collection. At 141 Site 1, co-detection of rhinovirus (Ct 29) + influenza A (Ct 41) was found in one patient by 142 SC only and RSV (Ct 24) + rhinovirus (Ct 35) in one patient by HC only. Two parainfluenza 143 cases and one rhinovirus case were detected only by SC. Overall the detection rate was 6% 144 higher in SC compared with HC swabs for non-SARS-CoV-2 respiratory viruses which 145 equated to 3/20 (15%) additional positive results. At Site 2, no co-detections occurred. 146 Collection of samples for the 13 SARS-CoV-2 positive patients ranged from 2 to 9 days 147 following onset of symptoms with a mean of 4.8 days. One positive patient retested 6 days 148 after symptom onset using the screening E-gene assay, was detected only on SCN but not the 149 HC. A second positive patient was detected using HC and SCT but not SCN. Of the patients 150 with detectable respiratory viruses other than SARS-CoV-2, at site 1, 8/23 (35%) had virus 151 only detectable on one of SCN or SCT while the proportion was 14/35 (40%) at site 2. 152 153 When all detections by HC and SCNT were compared with AP, the sensitivity of SCNT and 154 HC to detect COVID-19 was 1.0 (95%CI: 0.86-1) and 0.96 (95%CI: 0.8-1) respectively; for 155 other respiratory viruses it was 0.94 (95%CI: 0.87-0.98) and 0.91 (95%CI: 0.83-0.96) 156 respectively. 157 Table 3 summarises concordance between AP and each collection method. Both SCNT and 158 HC showed very high concordance with AP at each site and overall, with SCNT slightly 159 higher (κ =1, 0.934, 0.959 at Site1, Site2, Combined Sites) than HC (κ =0.929, 0.934, 0.933). 160 Additionally, SCNT was highly concordant with HC (κ =0.929, 0.863, 0.890 at Site 1, Site 2, 161 Combined Sites). When Ct values for COVID-19 cases were compared by collection method 162 (Figure 1), mean E-gene Ct did not differ between HC and SCNT or SCN (p=0.236, 0.083, 163 against α '=0.0083) but was significantly higher in SCT compared with HC (β =7.31, p<0.001). 164 Mean N-gene Ct was not significantly higher in SCNT compared with HC (p=0.041; 165 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted April 11, 2020. ; https://doi.org/10.1101/2020.04.09.20057901doi: medRxiv preprint 8 α '=0.0083) but was higher in SCN and SCT (β =4.00, p=0.006; β =7.63, p<0.001). In 166 rhinovirus cases (Figure 2), mean Ct was not significantly higher in SCNT compared with 167 HC (p=0.036; α '=0.017) but was higher in SCN and SCT (β =2.50, p=0.002; β =6.68, 168 p<0.001). In parainfluenza cases, mean Ct differed between HC and SCN (β =4.67, p=0.014) 169 but not the other methods (SCNT v HC, p=0.231; SCT v HC, p=0.119; α '=0.017). 170 171 At Site 1 an analysis of acceptability was performed using a questionnaire and was completed 172 by 42/70 (60%) participants with 31/42 (74%) preferring self-collection over trained 173 collectors, with all considering it acceptable. Analysis of the IEO found that the Median (LQ, 174 UQ) IEO was 3 (2, 4) with participants identified across all educational levels but the 175 majority (30/42, 71%) were in the 3 lowest education levels and a smaller proportion (12/42, 176 29%) in the highest 2 levels. 177 Following this study, Site 1 has since processed a small percentage of SC swabs (7% of all 178 collections). There was no significant difference in the SARS-CoV-2 detections between HC 179 with 242/13851 (1.8%) and SC with 20/1035 (1.9%) (p=0.753 from χ 2 test). 180 181

182 In our group of 236 ambulatory, literate, mostly adult patients, the performance of self-183 collected nasal and throat swabs was at least equivalent to that of health care worker collected 184 swabs for the detection of SARS-CoV-2 and other respiratory viruses. 185 This study included two different sites using two different methods of HC (combined N + T 186 and combined NP + T) and also employed two different molecular strategies for detection of 187 SARS-CoV-2. As such these findings are more widely applicable. 188 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted April 11, 2020. ; https://doi.org/10.1101/2020.04.09.20057901doi: medRxiv preprint 9 At Site 1 where SCNT was compared with HC using the same swab and collection methods, 189 for the 12 patients testing positive to SARS-CoV-2 there was complete concordance between 190 HC and SC samples even though on average 2.5 days had lapsed. In the remaining SARS-191 CoV-2 negative patients, SC detected 3 additional respiratory viruses, with the overall 192 positivity rate increasing from 34% to 40%. However, the additional 3 SC detections were 193 weak positives based on high Ct values (33-40). 194 At site 2 where comparative HC involving a NP and T swab occurred at the same time as the 195 SCN and SCT for the SARS-CoV-2 positive patients, SCNT detected all 13 positive patients 196 while one patient was negative by HC. Detection of other respiratory viruses by SCNT was 197 highly concordant with HC detecting only 1 less respiratory virus and may relate to the fact 198 that SCNT sampling was compared with NP +T sampling. 199 When data from each site was combined, concordance between SCNT or HC with the All 200 Positive rate was very high, slightly favouring SCNT. The similar SARS-CoV-2 percent-201 positivity rate in ongoing comparison data between those having only HC or SC provides 202 further reassurance that SCNT is equivalent to HC. 203 The advantages of self-collection are evident and even more important at a time of global 204 health crisis. Self-collection greatly reduces the number of patients requiring trained health 205 worker collection and PPE, thus preserving the limited supplies of PPE. Access to testing is 206 increased, as swab kits can be provided quickly by clinicians or available at dedicated 207 COVID-19 collection centres aiding timeliness of testing2,3 which is critical in the current 208 pandemic. There is increased safety for both patients and staff using a SC model as exposure 209 to others is limited. 210 Further, data from patients at site 1 suggests that SC is accessible and achievable over a range 211 of education levels with all finding SC acceptable and the majority having a preference for 212 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted April 11, 2020. ; https://doi.org/10.1101/2020.04.09.20057901doi: medRxiv preprint 10 this method over HC as has previously been reported.2,4,5 This may relate to the ability of 213 patients to control the comfort level of throat and nasal collection better than a trained 214 collector can. 215 We chose to trial SCN and SCT swabs rather than NP collections because the latter is 216 technically more difficult and uncomfortable for patients. Literature suggests that collection 217 of mid-turbinate nasal swabs is comparable in performance to collection of NP swabs for 218 respiratory viruses including other coronaviruses.9 We chose to perform nasal swabs given 219 that mid-turbinate swabs with a safety stopping point are generally not as widely used and 220 more uncomfortable than nasal swabs. 221 Recent studies suggest there is a high viral load in patients with early COVID-19 across the 222 upper and lower respiratory tracts, including nasal and throat sites10-12,14 as well as in saliva,13 223 even in asymptomatic, mild or prodromal states. Wolfel et al14 noted no discernible 224 difference between nasopharyngeal and oropharyngeal viral loads and detection rates in 225 hospitalized cases of COVID-19 and noted that simple throat swabs provide sufficient 226 sensitivity in early infections. Given these high viral loads throughout the respiratory tract it 227 may be that requiring NP sampling is not as significant for SARS-CoV-2 as for some other 228 respiratory viruses. It may also be that sensitive and specific PCR methods for viral detection 229 are improving the sensitivity of a range of sample and collection methods as shown for a 230 range of respiratory viruses but also Group A Streptococcal detection.9,10 We hypothesize that 231 the high viral load of SARS-CoV-2 and sensitive molecular techniques may explain the 232 equivalent sensitivity of SC to HC samples in COVID-19 patients. Additionally viral load at 233 different sites may differ with disease evolution and the SARS-CoV-2 positive patients in this 234 study were tested over a range of 2 to 9 days from symptom onset. 235 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted April 11, 2020. ; https://doi.org/10.1101/2020.04.09.20057901doi: medRxiv preprint 11 Our data support the decision by the Communicable Disease Network of Australia (CDNA)21 236 to recommend sampling of both nasal and throat sites for the diagnosis of respiratory viruses 237 including for SARS-CoV-2, due to the concern of a possible missed diagnosis if only one 238 site is sampled. This was the case for two COVID-19 positive patients on SC who were only 239 diagnosed by SCN and another only by SCT. If only one swab site was obtainable, our data 240 suggests the nasal may be the better swab site for the diagnosis of COVID-19 as it had 241 greater concordance with the AP group and showed consistently lower Ct values in the order 242 of 100-1000 fold higher viral load (data not shown). 243

of this study include the limited number of positive SARS-CoV-2 patients and 244 modest number of other positive respiratory virus cases with the exception of rhinovirus. 245 Further data on self-collection would be helpful to confirm these findings. In the setting of 246 limited resources, both in terms of PPE and health care workers, these findings may be 247 important for other health services. Furthermore, we have instituted use of a single swab to 248 sample both throat then nasal sites. This has the potential to preserve limited supplies of 249 swabs and also provide additional efficiencies in the laboratory as only preparation of a single 250 sample per patient is required. 251 252

253 The world is facing unprecedented demands on health care services and health resources 254 during the COVID-19 pandemic. Innovative ways to address this crisis are required and we 255 believe that this study provides early evidence that self-collection of throat and nasal swabs 256 for SARS-CoV-2 offers an acceptable and reliable alternative to health care worker collected 257 samples. This is achieved whilst preserving critically needed PPE supplies, optimizing the 258 time to testing and reducing exposure of health care workers to potentially infected patients. 259 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted April 11, 2020. ; https://doi.org/10.1101/2020.04.09.20057901doi: medRxiv preprint 12 260

261 We thank the Training and Patient Services departments, the Collection and Clinical Area 262 Managers, Clinical Supervisors, Collection staff, and Molecular Laboratory staff without 263 whom this study would not have been possible. 264 265 266 267 268 269 270 271 272 273 274 275 276 277

278 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted April 11, 2020. ; https://doi.org/10.1101/2020.04.09.20057901doi: medRxiv preprint 13 279 1. WHO Director-General's opening remarks at the media brie fing on COVID-19 - 11 March 280 2020. 2020, March 11. at https://www.who.int/dg/speeches/detail/who-direc tor-general-s-opening-281 remarks-at-the-media-briefing -on-covid-19---11-march-2020 .) 282 2. Goyal S, Prasert K, Prapha siri P, et al. The acceptability and validity of self-collec te d nasal 283 swabs for detection of influenza virus infection among older adults in Thailand. Influenza Other 284 Respir Viruse s 2017;11:412-7. 285 3. Jackson ML, Nguyen M, Kirlin B, Madziwa L. Self-Collected Nasal Swabs for Respiratory Virus 286 Surveillance. Open F orum Infect Dis 2015;2:ofv152. 287 4. Akmatov MK, Gatzemeier A, Schughart K, Pessler F. Equivalence of self- and staff-collected 288 nasal swabs for the detection of viral respiratory pathogens. PLoS One 2012;7:e48508. 289 5. Dhiman N, Miller RM, Finley JL, et al. Effectiveness of patient-collected swabs for influenza 290 testing. Mayo Clin Proc 2012;87:548-54. 291 6. Seaman CP, Tran LTT, Cowling BJ, Sullivan SG. Self-collected compared with professional-292 collected swabbing in the diagnosis of influenza in symptomatic individuals: A meta-analysis and 293 asse ssment of validity. J Clin Virol 2019;118:28-35. 294 7. Irving SA, Vandermause MF, Shay DK, Belongia EA. Compari son of nasal and nasopharyngeal 295 swabs for influenza detection in adults. Clin Med Res 2012;10:215-8. 296 8. Heikkinen T, Marttila J, Salmi AA, Ruuskanen O. Na sal swab versus nasopharyngeal aspirate 297 for isolation of respiratory viruses. J Clin Microbiol 2002;40:4337-9. 298 9. Larios OE, Coleman BL, Drews SJ, e t al. Self-collected mid-turbinate swabs for the detection 299 of respiratory viruses in adults with acute respiratory illnesses. PLoS One 2011;6:e21335. 300 10. Zou L, Ruan F, Huang M, et al. SARS-CoV-2 Viral Load in Upper Respiratory Specimens of 301 Infected Patients. N Engl J Med 2020;382:1177-9. 302 11. Pan Y, Zhang D, Yang P, Poon LLM, Wang Q. Viral load of SARS-CoV-2 in clinical samples. 303 Lancet Infect Dis 2020. 304 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted April 11, 2020. ; https://doi.org/10.1101/2020.04.09.20057901doi: medRxiv preprint 14 12. Kim JY, Ko JH, Kim Y, et al. Viral Load Kinetics of SARS-CoV-2 Infection in First Two Pati ents in 305 Korea. J Korean Med Sci 2020;35:e86. 306 13. To KK, Tsang OT, Chik-Yan Yip C, et al. Consistent detection of 2019 novel coronav irus in 307 saliva. Clin Infect Dis 2020. 308 14. Wölfel R, Corman VM, Guggemos W, et al. Virological assessment of hospitalized patients 309 with COVID-2019. Na ture 2020. 310 15. Wang W , Xu Y, Gao R, et al. D etection of SA RS-CoV-2 in Different Types of Clinical 311 Specimens. JAMA 2020. 312 16. Census of Population and Housing: Socio-Economic Indexes for Areas (SEIFA) Australia 2016. 313 2018, March 27. at 314 https://www.abs.gov.au/ausstats/[email protected]/Lookup/by%20Subject/2033.0.55.001~2016~Main%20F315 eatures~IEO~22 .) 316 17. Corman VM, Landt O, Kai ser M, e t al. Detection of 2019 novel coronavirus (2019-nCoV) by 317 real-time RT-PCR. Euro Surveill 2020;25. 318 18. 2019-Novel Coronavirus (2019-nCoV) Real-time rRT-P CR Panel Primers and Probes. 2020, 319 Janua ry 24. at https://www.who.int/doc s/defaul t-source/coronaviruse/uscdc rt-pcr-panel-primer-320 probes.pdf?sfvrsn=fa29cb4b_2 .) 321 19. R: A language and environment for statistical computing. R Foundati on for Statistical 322 Computing, Vienna, Austria. 2019. at https://www.R-project.org/ .) 323 20. Murray MA, Schulz LA, Furst JW, et al. Equal performance of sel f-collec ted and health care 324 worker-collected pharyngeal swabs for g roup a streptoc occus testing by PCR. J Clin Microbiol 325 2015;53:573-8. 326 21. Public Heal th Laboratory Network (PHLN) guidance on laboratory testing for SARS-CoV-2 327 (the virus that causes COVID-19). 2020, April 1. at 328 https://www.health.gov.au/sites/default/files/documents/2020/04 /phln-guidanc e-on-laboratory-329 testing-for-sars-cov-2-the-virus-that-causes-covid-19_0.pdf 330 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted April 11, 2020. ; https://doi.org/10.1101/2020.04.09.20057901doi: medRxiv preprint 15 331 Table 1 : Summary of COVID19 case s, other respiratory case s and nega tive test re sults from both 332 sites, with corresponding detections under the HC and S CNT methods. 333 N = 23 6 T est R esu lt Sit e 1 Sit e 2 Al l P a t ients HC Ne ga ti ve 38 1 17 155 (65 .7 % ) Ot he r R e sp i ra to r y 20 36 56 (23 .7 % ) C O V ID19 12 12 24 (10 .2 % ) SCN T Ne ga ti ve 35 1 18 153 (64 .8 % ) Ot he r R e sp i ra to r y 23 35 58 (24 .6 % ) C O V ID19 12 13 25 (10 .6 % ) AP Ot he r R e sp i ra to r y 23 40 63 ( 2 6 . 7% ) C O V ID19 12 13 25 ( 1 0 . 6% ) HC : Hea l th w o rke r C ol l e ct; S CN T : Se lf C ollec t Na sal a n d T hroat ; AP : A ll P o sit ives (po s i tive resul ts from 334 either HC or S C NT ). 335 336 337 338 339 340 341 342 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted April 11, 2020. ; https://doi.org/10.1101/2020.04.09.20057901doi: medRxiv preprint 16 Table 2 : Summary of COVID-19 and other respiratory illnesses detected under the HC, S CN, SCT, 343 SCNT methods and positives from all methods (AP), at the two collection sites. 344 Site 1 H C S C N SC T SC N T A P R h i no virus 1 5 1 5 1 4 16 1 6 ( 2 2. 9 % ) In f lu en z a B 2 1 2 2 2 (2 .9 % ) RSV 1 1 1 1 1 ( 1. 4 % ) A den ovirus 1 0 1 1 1 ( 1. 4 % ) Pa rai n fl ue n z a 0 2 1 2 2 ( 2. 9 % ) HMPV 1 1 0 1 1 ( 1. 4 % ) Total Oth er R espir at or y 2 0 ( 2 8.6%) 20 19 23 ( 3 2.9% ) 2 3 (3 2. 9 % ) SA R S - C oV- 2 (E,N,RdRp gene) 1 2 ( 1 7.1%) 5/ 5* 5 / 5* 1 2 ( 1 7.1% ) 1 2 (1 7. 1 % ) Tota l u n de r goi ng HC a n d SC 70 ( 1 00%) 7 0 (1 00% ) 70 (10 0%) 345 Site 2 HC S C N SCT S C NT AP R h i no virus 2 3 19 1 7 2 2 2 5 ( 1 5. 1 % ) In fl ue n z a B 1 1 0 1 1 ( 0. 6 % ) RSV 1 1 1 1 1 ( 0. 6 % ) A den ovirus 2 2 1 3 4 ( 2. 4 % ) Pa rai n fl ue n z a 7 4 6 6 7 ( 4. 2 % ) HMPV 2 2 2 2 2 ( 1. 2 % ) Total Oth er R espir at or y 3 6 ( 28. 6% ) 2 9 27 3 5 ( 21. 1%) 4 0 (2 4. 1 % ) SA R S - C oV- 2 (E gene**) 12 (7 . 2 % ) 12 1 1 1 3 (7. 8 % ) 1 3 (7. 8% ) T o t al und e rg o in g H C and S C 166 (100 %) 166 166 166 (100 %) 1 66 (10 0% ) All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted April 11, 2020. ; https://doi.org/10.1101/2020.04.09.20057901doi: medRxiv preprint 17 HC : Hea l th w o rke r C ol l e ct; S C N: Se l f C o l lect Nas al ; S C T : Se lf C o l lec t T h roa t; SCNT: S elf Colle ct N a sal an d T h roat ; A P: All 346 Po sit ives (po s i tive resul ts from eit her HC or S C NT ); RS V: Res p iratory S yn c i tia l V i rus; H MPV: huma n me tapn e um o v iru s. 347 *onl y a subset of 5 p a ti e n ts a t S i te 1 h a d n asa l a n d thr o a t swabs test e d i ndivi dual ly.** A ll pa ti e n ts h ad s u p ple mentary N 348 ge n e testing: H C 13; S C N 13 ; S C T 11; SCNT 1 3 d e te ct ed. 349 350 351 352 Table 3 : Concordance (Cohen's κ) between (i) AP and HC, S CN, SCT and SCNT; and (ii) HC and SCNT. A 353 value of 1 indicates the method detected all COVID-19 and other respiratory cases, while a value 354 above 0.9 indicates a very high level of detection of all respiratory case s (AP). 355 C on c ordanc e with AP H C S CN S CT S CN T Sit e 1 0. 929 0 . 9 05 * 0. 8 7 2 * 1 Sit e 2 0. 934 0. 835 0. 789 0. 9 3 4 C omb i ned S i te s 0 .9 33 0. 858 0. 817 0 .95 9 Co n c o r d a n ce be t w ee n H C a nd S CN T S i te 1 S i te 2 Comb i ned S i te s 0 .9 29 0 .8 63 0 .89 0 356 HC : Hea l th w o rk e r Co l lect; SCN: S elf C o l lec t Na sal; SCT: Self C o ll e ct T hro a t ; S C NT : Se l f Collect Nasa l a nd T hro a t ; AP : A l l 357 Po sit ives (po s i tive resul ts from eit her HC or S C NT ). 358 P- v alue <0.0 01 f o r ea ch c oncord a n c e test. * SCN a nd S C T c o nc orda nce on redu c ed set o f i ndivi duals fo r Si te 1 (o nly 5 of 1 2 359 SARS-C o V- 2 pat ients h a d S CN a nd S CT tes ti ng i ndiv i dua l l y p e rf orm ed. 360 361 362 363 364 365 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted April 11, 2020. ; https://doi.org/10.1101/2020.04.09.20057901doi: medRxiv preprint 18 Figure 1 : E-gene and N-gene Ct values obtained by the diffe rent collection methods for SARS -CoV-2 366 positive patients at both site s. 367 368 369 Figure 2 : Ct values obtained by the different collection methods for rhinovirus and parainfluenza 370 positive patients at both site s. 371 372 373 374 20 30 40 20 25 30 35 Mean Ct Ct Value

HC SCN SCT SCNT SARS-CoV-2 positive Patients E Gene 15 20 25 30 35 40 25 30 35 Mean Ct Ct Value

HC SCN SCT SCNT SARS-CoV-2 positive Patients N Gene 20 30 40 20 25 30 35 40 Mean Ct Ct Value

HC SCN SCT SCNT Rhinovirus Positive Patients 20 25 30 35 40 24 28 32 36 Mean Ct Ct Value

HC SCN SCT SCNT Parainfluenza Positive Patients All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted April 11, 2020. ; https://doi.org/10.1101/2020.04.09.20057901doi: medRxiv preprint