SHEAR Saliva Collection Device Augments Sample Properties for Improved Analytical Performance

SHEAR SCD

The advent of the COVID-19 pandemic resulted in a sudden surge in demand for widespread SARS-CoV-2 testing and supply chain bottlenecks amplified the shortage of PPEs, medical equipment, and materials for diagnostic tests (e.g. nasopharyngeal swabs and reagent required for PCR). SHEAR SCD was designed in response to rigid funnel shortages for collecting saliva, and to address potential procedural bandwidth limitations with regards to sample processing, among other factors during the substantial ramp-up timeframe required to mobilize large-scale testing. SHEAR SCD is made of 3 main components: a soft foldable funnel, a shearing filter, and a collection tube (Fig 1A). Collection of saliva samples with SHEAR SCD involves 6 steps (Fig 1B) and 3 main activities: 1) deposition of saliva, 2) processing of saliva and, 3) funnel disposal. The usage of SHEAR SCD starts with pulling of the pull tabs to open the funnel and the deposition of saliva. Once enough saliva is collected in the funnel, the funnel is sealed and folded to press down and squeeze saliva through the shearing filter into the collection tube. Once enough saliva is collected in the collection tube, the funnel can be disconnected and disposed while the collection tube is closed with a screwcap. SHEAR SCD underwent product design cycles for an enhanced user experience and improved manufacturability. The high-fidelity prototypes, used for all testing in this study were manufactured through injection molding techniques under ISO13485 conditions. The early lab prototype (before the redesign process) is shown in Fig 1C. It was constructed within a week with readily available materials and technology. The filters were 3D printed using an inert and rigid photopolymer with a liquid resin printer. Piping bags commonly used for baking purposes were modified to construct the soft funnel and a hole was bored into a screwcap of a 15 ml centrifuge tube which houses the top filter module, and the centrifuge tube acted as the collection tube.

• Download figure
• Open in new tab

FIG. 1. SHEAR SCD product design and instruction for use.

(A). SHEAR SCD with labelled components. (B). Graphic of the instruction for use of SHEAR SCD. (C). Photo of the early lab prototype of SHEAR SCD.

SHEAR SCD characteristics for saliva collection

The sample recovery test was conducted to investigate the sample retention in SHEAR SCD. When 2g saliva samples were pipetted directly to the base of the funnel, the average weight loss of saliva in SHEAR SCD and commercial, rigid SCD were 0.1710 ± 0.0227g and 0.0986 ± 0.0034g, respectively (Fig 2A). When saliva was pipetted to the middle of the SHEAR funnel the weight loss was 0.202 ± 0.0246g.

• Download figure
• Open in new tab

FIG. 2. Sample recovery, backflow and food particle test results

(A). Weight loss of saliva from sample recovery testing of SHEAR SCD (n=3), SHEAR SCD with sample loading in the middle of the funnel (n=3), Commercial SCD (n=3). (B). Sample recovery of saliva samples from backflow tests; inverted test and shaking test, two SHEAR SCD devices (SHEAR SCD 1 and SHEAR SCD 2, N=3). (C). Size of food particles found in SHEAR (SHEAR SCD-processed sample), Native (Native sample) and Filtered water in the food particle test. (D). Count of food particles found in [count/mm²] of SHEAR (SHEAR SCD-processed sample), Native (Native sample) and Filtered water in the food particle test. N = number of replicates, n = number of devices. Lines in (A-B) represent the mean value. For (C-D) whiskers represent maximum and minimum values and the box represents the median value, 25th and 75th percentile. ****P<0.0001,***P<0.001, **P<0.01 *P<0.05

POC tests commonly include a buffer in the collection tube for enhanced analyte interfacing. Improper handling of a kit during the saliva donation process may lead to accidental ingestion of buffer medium from the collection tube. To test if SHEAR SCD filter stops backflow of saliva (or saliva-buffer mixture) from the collection tube, we performed two sample recovery tests with saliva present in the collection tube: after shaking the tube with SHEAR SCD attached in a horizontal position, and after inverting the tube with SHEAR SCD attached upside-down. The average weight loss of saliva in the shaking test and the inverted test were 0.0268 ± 0.0151g and 0.0932 ± 0.0307g respectively, which corresponded to a sample recovery rate of 98.8% and 95.4% respectively (Fig 2B).

Presence of large food particles in saliva samples may possibly introduce interference into the downstream analytical tests. The food particle analysis was conducted to test the efficacy of the shearing filter in SHEAR SCD to trap particle and reduce contamination. The median size and count of food particles found in SHEAR SCD-processed sample were 44.38 μm² (95% Confidence Interval (CI), 36.38 - 93.86) and 60.86 count/mm² (95% CI, 46.07-64.84) respectively - a 36.1% reduction in size and a 215.8% significant reduction in count (P<0.05, Kruskal Wallis with Dunn’s test at α = 0.05) as compared to the native sample before processing (Fig 2C-D).

SHEAR SCD enables saliva homogenization and analyte release

To assess the sample processing performance of SHEAR SCD, the viscosity of SHEAR SCD processed saliva samples was measured across a shear range of 50-3000s⁻¹ and compared to saliva samples treated with other conventional saliva processing methods and an ART buffer solution. The saliva samples and ART buffer solution exhibited a non-Newtonian, shear thinning behavior; decreasing viscosity with increasing shear stress (Fig 3A). While no statistically significant differences in viscosity of saliva were detected between SHEAR SCD processed saliva and all measured samples (Kruskal Wallis with Dunn’s test at α = 0.05), at low shear rate (50s⁻¹) the viscosity of SHEAR SCD processed saliva was lower (7.20 cP (95% CI, 5.02 - 9.28)) than Native, Freeze-Thawing and DTT processed saliva, but higher to that of Supernatant and ART buffer (Fig 3B). Except for DTT processed saliva, a similar trend of viscosity differences between SHEAR SCD processed and all measured samples was observed at high shear rate (3000s⁻¹) (Fig 3C). Additionally, we calculated the coefficient of variation (CV) of viscosity to assess the uniformity of the measured samples. Enhanced uniformity of viscosity was observed in SHEAR SCD processed saliva; CV of SHEAR SCD processed saliva was lower than all but the Supernatant samples (Fig 3D).

• Download figure
• Open in new tab

FIG. 3 Viscosity and total protein concentration of saliva samples

(A). Viscosity profile of fluid samples measured with rheometer: SHEAR (SHEAR SCD processed saliva, N=6), Native (Native saliva, N=7), F-T (Freeze-Thawing processed saliva, N=5), DTT (DTT processed saliva, N=5), Supernatant (Centrifuged processed saliva, N=5) and ART Buffer (N=5) over a shear range from 50s⁻¹ to 3000 s⁻¹. (B). Viscosity of the fluid samples measured with rheometer at low shear rate: 50s⁻¹. (C). Viscosity of the fluid samples measured with rheometer at high shear rate: 3000s⁻¹. (D). Average sample uniformity of the samples measured with rheometer over the shear range from 50s⁻¹ to 3000 s⁻¹. (E). Estimation of total protein concentration of SHEAR (SHEAR SCD processed saliva, N=5), Native (Native saliva, N=5), F-T (Freeze-Thawing processed saliva, N=4), DTT (DTT processed saliva, N=5), Supernatant (Centrifuged processed saliva, N=5). Whiskers represent maximum and minimum values and the box represents the median value, 25th and 75th percentile. ***P<0.001, **P<0.01 *P<0.05

A bicinchoninic acid (BCA) assay was conducted to assess and compare the total protein concentration of 5 saliva samples treated with different saliva processing techniques. The total protein concentration in SHEAR SCD processed saliva with 1051 µg/ml (95% CI, 837.7 - 1072.3) was higher when compared to Native, Freeze-Thawing and Supernatant but lower than DTT-processed saliva (Fig 3E). Except for DTT processed saliva (p<0.001), the increase was not statistically significant (Kruskal Wallis with Dunn’s test at α = 0.05).

SHEAR SCD pre-processing affects the diagnostic performance of an ART

To investigate the effects of improved saliva’s biophysical properties on ART, paired saliva samples processed with SHEAR SCD and commercial SCD were spiked with SARS-CoV-2 nucleocapsid protein at the limit-of-detection (LOD) concentration level and tested with a commercial SARS-CoV-2 ART kit. Presence of a filtered out yellow residue was detected in the sample well of the ART kit for several individuals with less of that residue detected in the paired, SHEAR-preprocessed samples (Supplementary Fig S1). We measured the change in the intensity of the background after passing of the sample, and the relative intensities of the test and control lines after 20 and 30 mins from dropping the samples into the sample well of an ART cassette. While there was no statistically significant difference in the intensities of the background and the lines between the two SCD types (Wilcoxon signed-rank test at α = 0.05) at either of the timepoints, the samples processed with the commercial SCD led to formation of a visible test line for 6/10 individuals and SHEAR SCD improved that result by facilitating the test line formation in the samples from 9/10 of the same individuals (Fig 4A). No correlation was detected between the change in the background intensity and the relative intensity of the test line.

• Download figure
• Open in new tab

FIG. 4. Test line intensity, loading time and control line formation time

(A). Test line intensity for SHEAR SCD-processed saliva (N=10), Commercial SCD-processed saliva (N=10) and PBS solution (N=3) at 20-minute and 30-minute timepoints. (B). Loading time and control line formation time of SHEAR SCD processed saliva (N=10), Commercial SCD processed saliva (N=10) and PBS solution (N=3). Whiskers represent maximum and minimum values and the box represents the median value, 25th and 75th percentile. No statistical difference detected with Wilcoxon signed-rank test at α = 0.05.

We additionally investigated the effects of SHEAR SCD on temporal dynamics of sample loading and line formation. There was no difference in the median loading time between the two groups (59.36 s vs 55.18 s, for SHEAR SCD and commercial SCD sample group, respectively; p = 0.96 Wilcoxon signed-rank test), however, SHEAR SCD sample preprocessing decreased the variation in the loading time by two times - with a CV of 0.75 vs 1.82 in the SHEAR SCD and commercial SCD test groups, respectively. For comparison, the median loading time of the control samples (PBS; N=3) was 17.5 s with CV of 0.08. We further assessed the time required for the sample after loading to pass the membrane assay across the result window and lead to the formation of the control line. The time required to form the control line was not significantly different between the two groups (3.25 min and CV of 0.42 vs. 3 min and CV of 0.31 in the SHEAR SCD and commercial SCD sample group, respectively) and was half of that in the control group (1.5 min with CV of 0.06) (Fig 4B). Lastly, no correlation was found between liquid migration speed and line intensity for test and control line (R² = 0.035 and 0.095 respectively) (Supplementary Fig S2).

SHEAR SCD is user friendly

The collection of saliva using SHEAR SCD includes an additional processing step to be done by the saliva donor - folding down the funnel to push the saliva sample through the filter into the collection tube. To investigate the amount of pressure required for users to push their saliva through the filter, the pressure generated within funnel during the processing step was measured. The maximum pressure generated within the funnel during the processing step was 7.50 ± 1.55 PSI.

15 healthy individuals participated in the user study of the SHEAR saliva collection device from February 2022-June 2022. There were eight female and seven male participants with a median age of 27 (range 23-60 years). Participants were provided with an instructional video (n1labs.org/shear-scd). Of note they did not require additional help throughout the user study and were able to complete the processing of saliva with SHEAR SCD independently. The participants were asked to donate saliva into a standard, commercial rigid funnel shortly after, within one session. All participants participated in a semi-structured interview which was conducted after the use of SHEAR SCD. 3 main themes emerged from the responses from the interviews and data saturation was achieved with 15 participants – no new themes emerged after participant 11.

SHEAR SCD scalably fulfils basic requirements for saliva collection

A saliva collection device is typically classified as a medical device. Hence, safety and effectiveness are two essential components that need to be demonstrated. While uncommon, several accidental exposures to buffer solutions from diagnostic test kits resulting in minor health outcome have been reported (21). In our backflow tests, 99% and 95% of the saliva sample was retained in the collection tube after the inverted and horizontal backflow tests respectively, demonstrating SHEAR filter’s ability to reduce backflow of the saliva samples and mitigate the risk of spillage leading to lowered incidence of such potential accidental exposure of saliva sample and buffer solution to the user. Furthermore, collection of saliva with SHEAR SCD provides an additional filtering function to reduce food debris preventing potential interference in diagnostic assays (22). In our food particle test, the 36.1% and 215.8% reduction in size and number of food particles observed in SHEAR-processed samples, potentially decrease the chance of particle interference with detection assays. Aside from particle contamination, saliva retention within the funnel itself can lead to a decrease in yield of the analyte of interest, potentially impairing accuracy of the diagnostic test. While, SHEAR SCD demonstrated a higher weight loss of saliva when compared to commercial SCD in the sample recovery test, the sample recovery rate was still above 90%, which is sufficient to perform follow up diagnostic assay.

SHEAR SCD improves the physical properties of saliva for diagnostics analysis

Saliva is mainly produced by three pairs of salivary glands, the submandibular, parotid, and sublingual glands (8) and numerous minor glands in the mouth and throat. The composition and biophysical properties such as viscosity of saliva produced from each gland varies (23), hence the collection methods and types of saliva collected plays a key role in accurate detection of specific biomarkers. Saliva with high viscosity is challenging for both laboratory-based testing and POC diagnostic applications involving lateral flow assay and paper-based analytical devices. The observed reduction in saliva viscosity, due to the mechanical shearing effects of the SHEAR SCD filter, can ease the pipetting procedures in laboratory-based saliva testing reducing pipetting errors, risk of environmental cross contamination and processing time required (10, 11, 13). Furthermore, while the viscosity of supernatant after centrifugation was comparable to the viscosity of SHEAR SCD processed saliva, implementation of SHEAR SCD for POC application is more feasible since it does not require additional equipment. Freeze-Thawing is a commonly used pre-processing method for the preparation of saliva samples to reduce viscosity through the breaking down of mucopolysaccharide (24). However, no decrease in viscosity was observed in the Freeze-Thawing saliva samples in our viscosity testing. This could possibly be due to the absence of the centrifugation step that was present in studies that demonstrated reduction in viscosity which remove precipitates from thawed samples (12, 25).

Color formation with the BCA assay is mainly due to the number of peptide bonds and presence of specific amino acids (26). In our total protein concentration study involving the use of BCA assay, an increase in total protein concentration was observed in all processed saliva samples when compared to the native saliva samples. Breaking down mucopolysaccharide can potentially reduce the viscosity of the saliva sample and release analytes bound to the mucin networks. A previous study has attributed the increase in total protein concentration in mechanically processed (magnetically-beat) saliva samples to the disruption of mucin networks, which released protein for increased detection by BCA assay (14). Specific to COVID-19, a disruption of mucin networks can possibly release SARS-CoV-2 spike protein, which was recently found to bind to the sialic acid glycan end of mucins (27), and enhance the virus’ detection. Aside from DTT-processed saliva samples, the increase in protein concentration was comparable among samples processed with all other methods. DTT-processed saliva samples displayed the highest increase in protein concentration suggesting the greatest disruption of mucin networks. DTT is a commonly used reducing reagent for chemical homogenization of saliva and sputum that homogenizes samples through splitting of disulfide bond of proteins. While effective, sample homogenization with DTT or other harsh chemical reagents is not suitable for every application due to imposing alterations in biochemical properties of the analytes that may lead to their impaired detection with the sensitive detection methods, and even reduce concentration of certain biomarkers. For example, sample homogenization with DTT has been shown to reduce the concentration of e.g. sputum myeloperoxidase (19, 28), a potential biomarker for chronic obstructive pulmonary disease (COPD) management and monocyte chemoattractant protein-1 and macrophage inflammatory protein (18, 20), mediators associated with COPD.

SHEAR SCD effects on ART testing

Saliva samples processed with SHEAR SCD demonstrated a mild improvement in the test line intensity. Interestingly, previous studies demonstrated an increase in signal intensity of test lines in samples with higher viscosity suggesting the amplification of the signal intensity could possibly be due lower sample flow rate which leads to increased incubation time of the analyte at the test lines (29) and elevated reaction time at the conjugate pad (30). Several other studies have corroborated the effects of viscosity on the flow rate of the liquid sample across the nitrocellulose strips of a lateral flow assay (31, 32). In contrast, in our study, no correlation was found between liquid migration speed and line intensity for the test and control lines. Instead, we hypothesize, that as the nucleocapsid protein was spiked into the liquid fraction of the saliva samples after passing through SHEAR SCD and the commercial SCD, it is plausible that the higher uniformity and lower viscosity of the SHEAR SCD-processed samples allowed better dispersion of the analyte and the buffer during the brief, 1 second-long vertexing, enhancing the inclusion of the analyte within the sample dropped onto the ART’s sample well. This may be indicative of the benefit of homogenizing the sample before its interfacing with the collection buffer and an improved analyte distribution for diagnostic of the analytes with varied presence in different fractions of the collected sample.

Of note, the test line intensity was generally low in this study due to the spiking of the nucleocapsid protein at the LOD concentration of the antigen to challenge the test to the extreme cases and reflect individuals with low viral load. The improved visibility of the test lines after sample pre-processing with SHEAR SCD could influence the judgment of the test result and potentially lead to a reduction in false negatives and a limitation of the intervention delays.

The sample well of an ART and the membrane assay allow only the liquid fraction to pass through to interact with the antibody conjugated strips to form the test and the control lines. Accordingly, SHEAR SCD had no effect on temporal dynamics of the sample migration after loading. However, SHEAR SCD lowered the variation in the sample loading time, which facilitate identification of a universal readout period for the most reliable results by lowering the interindividual, natural variation of saliva samples.

SHEAR SCD offers benefits to user experience and safety

Developing an engineering solution for a medical application, benefits from including all stakeholders in the co-development process (33). Through interviews with the users, we sought to identify the attitudes towards SHEAR SCD, particularly as an additional step (folding the funnel) is expected from the saliva donor. In general, participants of the user study commented that SHEAR SCD was straightforward and easy to use, with simple and easy to understand instructions for use. The difficulties reported for SHEAR SCD were mainly attributed to the pushback of saliva during the squeezing and folding process, unclear saliva level indicator and insufficient air trapping within the funnel. Targeted refinements of SHEAR SCD based on the highlighted difficulties will benefit the usability and functionality for enhanced user-experience. In our pressure testing, the measured pressure required to squeeze saliva through the filter was 7.50 ± 1.55 PSI, a pressure that an average 70-year-old female can generate through pulp pinching (34), which was consistent to findings from user study where no difficulties in squeezing of saliva due to pressure required were reported.

Transmission of respiratory viruses can occur through direct contact with aerosol and droplets containing virions (35). It has been found that sneezing and coughing forcefully expels droplets into the surrounding environment (36) hence spitting and hocking actions during saliva collection could potentially lead to release of such aerosols and droplets increasing the risk of transmission of such diseases. Furthermore, in a recent study, an incidence rate of 27.8% of saliva contamination on exterior surface of collection tube was reported (37). Saliva collection with SHEAR SCD covers the user’s mouth and nose during the donation process, containing the dispersion of droplets within the funnel and prevents saliva leakage to the exterior surfaces of the device. In addition, the foldable funnel enables the closing of the opening which leads to a reduction in the exposure time between donated saliva and the environment. While nasal swabbing methods do not lead to release of such droplets, three participants from our user study reported sneezing after the conduct of nasal swab, which is a drawback of nasal swabbing for detection of airborne respiratory viral infections in a mass testing setting. One participant in our study reported history of epistaxis after nasopharyngeal swabbing perfomed by a healthcare worker. While nasal swabbing is a less invasive procedure, there is still a risk of epistaxis. In addition, a participant expressed concerns of nasal swabbing of children and elderly due to underdevelopment of skull and lower tissue integrity, respectively. Compared to nasal swabbing, ease of saliva collection in terms of comfort and easier visual quantification of sample volume facilitates the extraction of the required amount of sample for a follow up diagnostic test. In fact, a study revealed that 16% of their participant were fearful to conduct self-nasal swabbing leading to the occurrence of “fairly shallow swab” in some cases (38), which potentially amplifies the probability of false-negative in samples from those participants. The simplicity of saliva collection, coupled with the use of SHEAR SCD that brings increased comfort level and ease of saliva donation as commented by participants in our user study, points to SHEAR SCD’s potential for home-based and remote testing.

Limitation

While offering multiple benefits, SHEAR SCD is not without limitations. The effect of mechanical shearing is limited, and even though no statistical significance was detected, the viscosity of SHEAR SCD-processed saliva was higher than that of buffer solution used in the ART. Additionally, in our study, while the participants were instructed to donate back throat saliva, drool saliva secreted from parotid gland were inevitably donated in the process, which resulted in the increased variability of viscosity within the individual’s saliva sample. Furthermore, in the total protein and viscosity experiments saliva samples collected from multiple participants were pooled, as both experiments required more amount of saliva than what an individual can donate in a sitting. Both of those factors may have contributed to increased variation in the saliva viscosity. To lower its effects, saliva samples in all groups were gently vortexed for 1 second right before each test.

The ART kit testing was used off label - the saliva sample were not the validated specimen listed in the instruction for use (IFU). Also, spiking of SARS-CoV-2 nucleocapsid proteins into processed healthy saliva samples limits the evaluation of the effects of mechanical shearing on mucin networks, which can potentially release SARS-CoV-2 virions trapped in that network. The benefits of SHEAR SCD for POC kits will require further testing in clinical trials with this study setting a foundation for such endeavour. Lastly, in the user study the mean age of the participants was 34 (ranging from 23-60 years) and only healthy individuals were recruited, hence findings from the study maybe have limited generalizability to elderly, children and people with disabilities. User study with participants representing wider age group and health status may capture a more comprehensive sentiment for SHEAR SCD among the real-world users, however, the results of our study demonstrate general acceptability of the device.

SHEAR SCD’s relevance for saliva-based clinical testing beyond COVID-19

Saliva has an advantage over other diagnostic samples due to the easy, safe, and non-invasive nature of attaining the sample. Similar to the responses recorded from participants in our user study, patients also cited these among reasons for which they prefer oral fluid sampling when given the choice (39). The non-invasive nature of saliva collection may be greatly beneficial for broad deployment under epidemic and pandemic circumstances, but also in a routine care, especially for patients that requires inpatient and/or outpatient longitudinal tests. These factors may potentially increase compliance with the testing regime, lowering the stress exerted on patients, and potentially enabling remote monitoring.

As demonstrated in COVID-19 diagnostics, POC testing, including saliva-based, has a potential for an early detection of the infection, which in turn allows earlier management options to be commenced and a greater likelihood for successful therapy. The benefits extend to other infectious diseases – e.g. saliva-based POC tests can be used for the diagnosis of HIV (40) and Hepatitis C, where the presence of immunoglobulins can inform about the stage of the infection (41). Other infectious diseases, e.g. Hepatitis B, mumps and rabies, also has a potential to be detected in saliva (42). Beyond infectious diseases, salivary diagnostics are clinically applicable to monitoring drug abuse, hormone levels and a range of disease markers. For example, saliva-based monitoring of C-reactive protein (CRP), a non-specific inflammatory response factor, has a potential to assist in detection and tracking of inflammation (43, 44), enabling earlier detection of infection and supporting rapid decision making. Similarly, being able to monitor drug levels with saliva-based POC testing will help reduce the number of invasive blood tests (45).

While salivary diagnostics hold potential, they are not often used as first line tests or stand-alone tests. Broadly deployed applications of saliva-based POC testing are also not common. The challenges stem from minute quantities of the markers and sample variability (46). Improving the biophysical properties of saliva sample for diagnostics at the collection stage with SHEAR SCD, offers a range of benefits and may potentially contribute to the broader realization of saliva’s potential as a clinical sample beyond COVID-19, especially in POC applications.

In summary, we presented a novel saliva collection device with sample homogenization capabilities that enhances interface for downstream analytical processes. SHEAR SCD-processed saliva samples exhibited lowered viscosity with enhanced uniformity, increased total protein concentration and an augmented visibility of test line when analyte was spiked at LOD concentration and detected with an ART. Importantly, participants from the user study highlighted the ease of use of SHEAR SCD and preference of saliva collection over collection of other diagnostic samples. In sum, the strategic implementation of SHEAR SCD may play an important role in enabling rapid and self-contained of saliva processing for diagnostic applications, especially in the context of POC testing during outbreak, epidemic and pandemic conditions.

Manufacturing method of SHEAR SCD

SHEAR SCD is made out of three main components: a soft, foldable funnel, shearing filter with small pores housed in an adapter and a collection tube. The components were fabricated and assembled by an ISO 13485:2016 certified manufacturer (Forefront Medical Technology, Singapore). The soft funnels were made from die-cutting and heat sealing of PolyVinyl Chloride (PVC) sheets with the aid of the 2D computer aided diagram drawn with Solidworks (Dassault Systèmes, MA, USA). The filter was manufactured with micro-injection molding of Polypropylene (PP, K1P38AE) using MicroPower15t machine (Wittmann Battenfeld, Austria) and the adapters were constructed with injection molding of Polypropylene (PP, K1P38AE). Medical grade adhesive (Loctite 3921, USA) was used to bond the filter and the soft funnel to the adapter. Standard collection tubes were used.

Saliva collection method and IRB

Saliva samples were collected from 24 healthy volunteers aged between 21 and 65 years in accordance with protocol approved by the Institutional Review Board of National University of Singapore (NUS-IRB-2021-15); written informed consent were obtained from each volunteer. Body temperature of each volunteer was taken prior to the study and the volunteers that exhibited acute respiratory infection symptoms were excluded. Volunteers were advised to avoid brushing of teeth and consumption of food and beverages for at least an hour before the collection of the saliva. The volunteers were advised to donate deep throat saliva. Each volunteer donated approximately 3ml of the sample. All samples were collected between 0900-1100 and 1300-1400.

Pressure measurement

The maximum pressure generated within the funnel of SHEAR SCD during the folding and squeezing process was measured with a pressure transducer (Analog Pressure Sensor, Gravity) inserted at the bottom of the funnel. 1ml of saliva from three participants were added in each replicate test.

Filter performance experiments

Rapid Antigen Testing procedures

Panbio™ COVID-19 Ag Test Kit (Abbott, USA) was used as the ART kit. Saliva samples of each volunteer were aliquoted into two parts. 500µl of saliva was extracted from each aliquot and processed with SHEAR SCD and commercial SCD (MicroCollect™ Saliva Collection Device, CD Genomics, USA). 1µg/ml SARS-CoV-2 nucleocapsid protein solution were prepared by diluting the 9.5 mg/mL stock solution (47). 4.5µl of SARS-CoV-2 nucleocapsid protein (1µg/µl) and 95.5ul of Panbio buffer solution were added into each processed saliva sample and a control solution - phosphate buffer saline solution (PBS). The mixture was vortexed, and 5 drops of the mixture were loaded onto the sample well of the ART cassette. The liquid mixture was allowed to migrate along the result window for 30 minutes. 30 minutes time-lapse videos (time-lapse interval = 10 seconds) of the development of each cassette was recorded with GoPro Hero 4 (GoPro, USA) under a custom-made lightbox to provide a fixed imaging condition (Fig 5A). The time-lapse videos were aligned and converted into 8bit image stacks with ImageJ (National Institute of Health, USA) for quantification of signal intensity (Fig 5B). The average pixel intensity of the regions of interest at the test line (I_{test line}) and background (I_background) were measured and test line intensity ratio, I_{test line} / I_background, was calculated to reduce background noise (48) at 20 (t₂₀) and 30 (t₃₀) minutes after the sample loading. Test line formation was analysed through the measurement of the average pixel intensity of the region of interest at the test line and 20 pixels before and after test line (I_{test line}, I_{before test line}, I_{after test line} respectively). Test line is determined to be formed if I_{before test line} / I_{test line} and I_{after test line} / I_{test line} and average pixel intensity of I_{before test line} and I_{after test line} is not more than 120 arbitrary units (Fig 5C). Equations used for the ART procedures are shown in Table 1.

• Download figure
• Open in new tab

FIG. 5. Analysis method for antigen rapid test

(A) Photo of the set-up for the capturing of time-lapse video for the antigen test experiment. (B) Antigen test cassette and general image processing step. (C) Description of analysis conducted in the antigen test experiment.

User Study

The user study sought to evaluate the end user’s experience of the SHEAR SCD and a commercial SCD (MicroCollect™ Saliva Collection Device, CD Genomics, USA). Participants were asked to donate 1 ml of saliva to each of the two SCDs, and they were subsequently interviewed based on a semi-structured interview guide (Table 2). The average length of the interview was 21 minutes (range: 18-25 minutes). Participants’ responses were audio-recorded, transcribed verbatim and analyzed using thematic analysis, where the responses were descriptively labelled for primary coding. The labels were then categorized into different groups which were subsequently used to create broader themes.

Statistical Analyses

Statistical analyses were carried out using GraphPad PRISM 9 (version 9.3.1, GraphPad) Statistical significance was determined using the Kruskal Wallis test with Dunn’s test at alpha = 0.05 for the food particle, rheology and protein test, and Wilcoxon signed-rank test at alpha = 0.05 for the examination with ART.