The methods for the study have been previously described [29]. Briefly, participants were eligible if they were ≥18 years of age, resided in the United States, had never been diagnosed with a bleeding disorder, were able to read and understand English without assistance, were willing to provide valid contact information so that study testing kits could be mailed to participants, had access to a mobile phone, tablet, or computer with a camera, and were willing to be observed by a clinician while completing the specimen collection processes.

Participants were recruited through two methods. First, we offered enrollment to people who had participated in a previous research study of willingness to self-test for SARS-CoV-2 infection and who agreed to be contacted for participation in future research studies [30]. Second, we shared a link with information about the study within networks of people symptomatic for COVID-19 or at risk for SARS-CoV-2 infection, including through networks of first responders. Participants who accessed the link to the information about the study were offered the opportunity to consent to online screening. Those who consented were screened for eligibility, and those who were eligible were provided with informed consent documents and a contact telephone number and email address to ask questions about the study. Participants were offered US $50 for completion of all study activities (eg, baseline survey, observed participant-collection session, return of specimens by mail, and post-collection survey).

The main outcome of interest was the biological sufficiency of the specimens for testing by RT-PCR and for detection of antibodies by serology testing. The biological sufficiency of the OPS specimens for PCR was assessed by evaluating the total nucleic acid in the specimen using ribonuclease P (RNase P) measurements as previously described [31]. Briefly, saliva and OPS specimens were subjected to nucleic acid extraction using the Thermo Kingfisher platform (Thermo Fisher Scientific). Extracts were tested for human RNase P by RT-PCR with the Thermo SARS CoV-2 testing kit v1. We considered saliva and OPS with cycle threshold (C_t) values <30 to contain sufficient collections of nucleic acid (as a proxy for collection of biological material) [29]. We compared the C_t values of the participant-collected and shipped saliva and OPS specimens to a laboratory reference set of 100 saliva specimens and 100 clinician-collected OPS specimens that were transported directly to the laboratory on ice after collection from a separate clinical population, and we described the differences in the median C_t value between the clinician-collected specimens and the clinician-observed, participant-collected specimens. To assess the biological sufficiency of the DBS cards, we performed a three-point quality check on the cards, assessing the visual appearance of the blood spot, whether the blood had soaked through the paper, and whether the circles were filled, according to our previously reported method for other DBS specimens [29].

Approval for this study was obtained from the Institutional Review Board at Emory University, and the specifics of the protocol have been previously published [29].

We enrolled 159 participants in the iCollect cohort pilot study; 61 (38.4%) were male, 91 (57.2%) were female, 1 (0.6%) was genderqueer, and 1 (0.6%) was multiple gender (Table 1). Most were non-Hispanic white/Caucasian (110/159, 69.2%) and were less than 40 years of age (99/159, 62.3%); 13/159 (8.2%) were 60 years or older. The 159 participants reported residence in the US regions of South/Southeast (57, 35.8%), Northeast (43, 28.3%), Midwest (27, 17.0%), West (14, 15.1%), and Northwest (8, 5.0%). Most reported at least one symptom of COVID-19 at the time of the survey: 51 of the 159 enrolled participants (32.1%) reported no symptoms, 56 (35.2%) had 1-3 relevant symptoms, 29 (17.6%) had 4-5 symptoms, and 9 (5.7%) reported 6-8 of the listed symptoms (Table 1).

A total of 228 respondents accessed the registration link. A total of 159 participants were eligible (Figure 1), gave consent, and provided contact information for the kit mailing. We mailed kits to 159 participants; 153 (96.2%) of these participants scheduled a video appointment, and all 153 (100.0%) completed that appointment. The mean time for all video appointments was 32 minutes (median 29 minutes, range 13-143 minutes). Of the 153 participants who attended a video appointment, 143 (93%) completed collections of all three samples (Figure 1). DBS was the most commonly uncollected sample; however, only 8 participants did not collect a DBS card. Thus, the analytic sample for clinician assessment of suitability and the laboratory assessment of sufficiency was 153 saliva specimens, 151 OPSs, and 145 DBS cards (Figure 1).

Clinical observers assessed that 147/153 (96%) of the saliva samples, 146/151 (96.7%) of the oropharyngeal samples, and 135/145 (93.1%) of the DBS samples were of sufficient quality to be submitted for laboratory testing (Tables 2 and 3). Clinician reasons for lack of suitability are also reported in Tables 2 and 3.

Clinical observers also documented compliance with specific steps in the instructions for each specimen type; these data are presented in Tables 4-6. For DBS collection, the most common errors were touching the specimen collection paper when making the spots (29/148,19.6%) and not completely filling all the circles (52/148, 35.1%). The median number of filled spots was 5; 3 filled spots are required for standard serology assessments in our laboratory, and 137/148 (92.6%) of participants filled at least 3 spots.

Data are presented for the first 101 OPSs, first 123 saliva specimens, and first 137 DBS cards processed by the laboratory. For the saliva specimens, all specimens except three had C_t values for RNase P <30 (the value of one specimen was 30.6; 98% of specimens met our pre-specified threshold [29] for sufficient nucleic acid for detection of target RNA). The median C_t for the saliva specimens was 19.5 (IQR 18.8-20.8). For oropharyngeal swabs, all specimens had C_t values for RNase P <30, meeting our pre-specified threshold [29] for sufficient nucleic acid for detection of target RNA. The median C_t for the OPS specimens was 23.9 (IQR 21.0-25.3). We compared the median C_t for OPS patient-collected specimens under clinician observation to the C_t values of 18 clinician-collected OPS swabs processed in the same central laboratory by the Wilcoxon rank sum test. The results indicated that there was no significant difference in the C_t (and, by inference, no difference in the concentrations of nucleic acid in the specimens) between participant-collected and clinician-collected OPS (median self-collected 23.9; median provider-collected 23.7; P=.70). For the 140 DBS cards evaluated, the median number of usable 6 millimeter punches was 3 (IQR 1-5). In terms of saturation, 70/140 (50.0%) were classified as good, 31/140 (22.1%) were classified as fair, and 38/140 (27.1%) were classified as poor; 1 card (0.7%) was assessed as having no blood. In terms of dryness (1=wet, 10=dry), the median dryness was 10 (IQR 9-10). The minimum dryness was 4.

US and global response to the SARS-CoV-2 pandemic desperately requires at-home sample collection both to detect people who are infected with the SARS-CoV-2 virus and for the measurement and monitoring of antibody response to the infection. Unlike nasopharyngeal swab collection, OPS, saliva, and DBS collections do not require any medical training. The level of testing that has been performed to date in the United States is limited for multiple reasons; important solutions are to diversify the types of specimens that have sufficient biological material to be accurately evaluated for SARS-CoV-2 infection (as assessed by RNase P) and immune response (as assessed by saturation and number of usable blood spots) and to diversify the locations in which these specimens can be collected. This study aimed to provide evidence of whether specimens collected at home for SARS-CoV-2 diagnosis are suitable (as assessed by clinical observers) and are sufficient (as assessed by laboratorians). Our results indicate that the collection of the specimens by the participants at home for the diagnosis of SARS-CoV-2 infection and serologic response was suitable as judged by clinical observers. Additionally, the OPS and saliva specimens were judged by objective measures to be sufficient for analysis of SARS-CoV-2 PCR by laboratorians. Most DBS cards contained sufficient samples for testing; however, the laboratorian-rated quality of saturation was variable. Our assessment did not validate these specimens as appropriate specimen types for use for SARS-CoV-2 testing; however, we did assess that the samples had adequate biological material to support testing. Both OPS and saliva have been determined by the FDA to be suitable specimen types for SARS-CoV-2 detection assays [32,33]. Specimens were tested for SARS-CoV-2 RNA by RT-PCR (saliva, oropharyngeal swab) and for IgG and IgM antibodies (dried blood spot, saliva) and IgA antibody (saliva); however, the results are not reported here because the primary intent of this analysis was to describe whether home-collected specimens were suitable and sufficient for RT-PCR and serology testing.

A major finding of our study was that home collection of specimens returned by mail is highly acceptable as a means of submitting specimens for testing for SARS-CoV-2; 143/153 (93.5%) of the participants who were sent kits completed collection of all the specimens and returned the kits. These data confirm findings from a separate study assessing the willingness of people to collect and return specimens for SARS-CoV-2–related testing [30]. In that study, participants were very willing to submit saliva and oropharyngeal swab specimens but were slightly less likely to report willingness to submit dried blood spot specimens. Our study suggests that the extent to which participants actually collect specimens is consistent with previous reports of their willingness to do so, as reported by different participants in an online survey (eg, the saliva collection was the most complete, and participants in a separate study reported being most willing to provide saliva specimens) [30] These data are also consistent with the acceptability of at-home specimen collection for other health conditions, including a long history of the use of at-home dried blood spot collection for HIV diagnosis [34,35]. In our prior work, we found that video instructions may be helpful in increasing the successful collection of specimens, including DBS specimens. We will consider evaluating video instructions as a complement to printed instructions, and we will continue to evaluate the quality of the collected specimens, especially the saturation and completed number of dried blood spot specimens.

To our knowledge, our study is unique in that we used both telehealth to provide clinician observation of participant-collected specimens and rigorous laboratory assessment to determine the sufficiency of those same specimens. We intend that these data will help create a bridge between current regulatory approvals for self-collection of specimens for SARS-CoV-2 diagnosis in clinical settings (eg, OPS; and participant-collected anterior nares swabs, participant-collected OPS, and participant-collected saliva when those specimens are collected under the supervision of a health care provider) and eventual regulatory review of at-home self-collection specimens for laboratory testing. We believe that this study addresses one important component that would support a transition from clinician-observed collection of these specimens to fully unobserved self-collection of specimens that are returned by mail: that the quality of the specimens for diagnostic purposes must be equivalent to clinician-collected specimens. Our evidence in this regard is strong because we incorporated both the professional opinions of clinical observers and objective assessment of the sufficiency of the samples by laboratorians.

However, we recognize that ultimate implementation of at-home self-collection of specimens for diagnosis of SARS-CoV-2 will also be dependent on other important factors. First, it is important that the materials that are sent out in at-home kits are safe, including consideration of the safety of the components of those kits even if they are not used as directed in the test kit instructions. We believe that this can be addressed by review of the material safety data sheets for the components of the kits and by considering modifications of the kits (eg, providing viral transport media in child-resistant tubes) to further improve the safety of the kit components in diverse household settings. Second, stability tests will be required to indicate whether the diagnostic sufficiency of specimens is compromised by conventional shipping processes, delayed shipping, or shipping in extreme environmental conditions. There are well-characterized protocols for such stability studies [36]; ensuring that the test performs as expected under a variety of environmental conditions and after shipping delays is an important part of assuring the diagnostic integrity of the task and, ultimately, the overall performance of the testing approach.

It is also important to view the consideration of deploying at-home participant-collection specimen kits through a broader lens to examine the potential risks and benefits of implementing such a system. We note that Siegler et al [30] documented that a substantial proportion of US respondents indicated they would they would be willing to submit participant-collected diagnostic specimens but were less willing to go to a drive-through, laboratory, or clinical setting to provide specimens. Therefore, the availability of at-home testing may increase our ability to test large numbers of people, including some who may be unwilling to go into clinical settings where they perceive themselves to be at risk for acquiring SARS-CoV-2 infection. Other countries have similar laboratory capacities, and some already use mailout specimens in public health programs: Public Health England uses mailout specimen collection and specimens returned by mail to screen asymptomatic people for sexually transmitted infections [37]. The ability to test people who have no or mild symptoms or are not willing to be tested in clinical settings can also reduce bias in estimates of SARS-CoV-2 prevalence that are generated from testing cohorts that are largely selected for symptomatic disease or the severity of that disease. Finally, there is a substantial benefit to developing and deploying testing methods for SARS-CoV-2 that are not reliant on supplies of rigid swabs, viral transport media, or PPE, all of which have substantial supply chain limitations. The self-collection of specimens at home thus limits the risk of exposure to health care providers, limits the extent to which PPE is used for diagnostic rather than care purposes, and reduces the congregation of people presenting for SARS-CoV-2 testing in clinics, where they run the risk of being exposed to other infectious patients.

Our study has important limitations. Our participants represent a biased group relative to the US population because they were included in the study based on their willingness to self-collect and return specimens. However, most of the 1435 respondents to an online survey reported willingness to collect and submit these specimens [30]; therefore, the extent of this bias may be minimal. We also acknowledge that the behavior of participants when collecting their specimens may have been influenced by the fact that they were being observed by a clinician (eg, a Hawthorne effect [38]). There are potential concerns about the shipping of boxes handled by participants who may be infected with SARS-CoV-2; however, the Centers for Disease Control and Prevention (CDC), the World Health Organization, and the US Surgeon General have indicated that there is no evidence for the spread of the virus that causes COVID-19 through the mail [32]. Our conclusion is that the specimens collected by the participants contained sufficient biological materials to support testing for RNA and antibodies; however, we do not report the results of our testing for SARS-CoV-2 RNA or serology. The CDC considers OPS to be a suitable specimen type if a nasopharyngeal swab is not available [33], and the FDA has granted an EUA for the use of saliva specimens [32].

There are important next steps to realize the promise of participant-collected specimens as one part of a suite of testing options available to address the current global pandemic of SARS-CoV-2. As noted above, it is important to conduct stability testing and to characterize the safety of the kit components before they are sent out to be used for self-collection without clinician observation. There is also a need for further studies to characterize the performance of serology testing for antibodies to SARS-CoV-2, and there are gaps in knowledge about the interpretation of those results. For example, we do not yet know the extent to which antibody responses confer partial, full, or no protection against reinfection. However, the possibility of new mechanisms to collect large numbers of samples from populations in difficult-to-reach places (eg, rural areas, during stay-at-home guidance) and from patients who are not symptomatic could have a practical public health impact. Potential applications of this technology include enabling the collection of specimens from large probability samples, monitoring the antibody status of communities through community sampling, establishing data on antibody kinetics by collecting serial (eg, daily) DBS collections mailed in by people who have been diagnosed with SARS-CoV-2 infection, and conducting screening of populations where it may be impractical to perform frequent health care visits.

We collected and evaluated specimens that were collected by participants observed by clinical observers that can be used for diagnostic testing related to SARS-CoV-2 infection. Our data indicate that participants were willing to collect specimens and that clinical observers believe that the specimens collected only with reference to the provided instructions were suitable for laboratory testing. We believe that these data are generalizable to any participants who need to be tested for SARS-CoV-2 who have access to mail. Additionally, the laboratory assessment indicated that the DBS specimens were sufficient for testing and that the total nucleic acid content of the saliva samples and pharyngeal swabs were sufficient for testing and were consistent with the amounts of nucleic acid in physician-collected pharyngeal swabs and physician-observed saliva specimens. We believe that the potential benefits of the broad availability of participant-collected and mailed-in specimens for clinical purposes and for epidemiological monitoring of the COVID-19 epidemic in the United States outweigh the concerns about whether clinician-collected or clinician-observed at-home specimen collection will produce superior samples. One important issue from a workforce standpoint is defining the level of health care professional who should be recommended to observe self-specimen collection if telehealth-observed self-collection is implemented as a specimen collection method. Based on our observations of the specimen collection behaviors, and bearing in mind that clinicians did not intervene to correct participants who made mistakes, we believe that a broad range of medical professionals, including medical assistants, would be well prepared to fill this role. A final recommendation is to consider feedback from the test kit users; we collected this feedback but did not summarize it as part of this report. Further studies are needed to establish the safety and stability of the specimens during shipment. If procedures can be created that demonstrate safety and stability, we urge consideration of FDA review and approval of the use of participant-collected mail-in specimens for SARS-CoV-2–related diagnostics.