Effective screening strategy against SARS-CoV-2 on self-collected saliva samples in primary school setting: A pilot project

Dear Editor,

Recent articles published in this Journal highlighted the potential of using antigen-detecting rapid diagnostic tests (Ag-RDTs) on saliva samples for massive screening of asymptomatic subjects and epidemiological surveillance.^{1 , 2} Saliva has entered the shortlist of clinical samples to which the current laboratory tests can be applied, due to increasing evidences of comparable sensitivity and specificity with respect to nasopharyngeal swabs (NPS).[3], [4], [5], [6] We previously described the performance of an automated chemiluminescence-based antigen assay applied to saliva samples in identifying individuals with high viral loads, underscoring the need for confirmatory molecular testing for SARS-CoV-2 antigen-positive cases in a setting of low prevalence.⁷ Based on these results, we carried out a pilot project for SARS-CoV 2 screening on saliva in primary schools, aimed at evaluating the feasibility of the proposed algorithm, detecting any critical issues to be overcome before expanding the experience to a territorial scale, virtually to the entire region. The testing algorithm included a two-step approach: a chemiluminescence-based Lumipulse G Sars-CoV-2 Ag assay (Fujirebio) followed by molecular confirmation of positive samples using Simplexa Covid-19 direct assay (Diasorin). Both tests were currently the only ones CE-IVD marked for this specific sample matrix, and had been preliminarily evaluated.^{3 , 7} The pilot study was coordinated by the Regional Health Authority, within the context of surveillance and prevention activities implemented for school re-opening after summer vacations. The study involved five primary schools in Rome, hosting 2522 students overall; 1905 students participated (75.5%) [median age: 9, ranging from 2 to 15; 970 males (50.9%) and 935 females (49.1%)]; sample collection was performed in 9 scheduled days during the period October 6th–November 2th, 2020.

Three-4 days before sampling, the school administrative offices acquired the informed consent from the parents of the children, as well as all data necessary to comply with the test recording procedures on the regional COVID-information platform. A barcode was generated by this procedure, that was applied on both the saliva collecting device and on the corresponding informed consent signed sheet. Saliva samples were collected at the school sites using Salivette® (Sarsted) devices, composed of two concentric tubes and a cotton sponge able to absorb saliva. The sponge should be kept in the mouth until it is well soaked (2–4 min) and then put back in the inner container, that is then sealed. Students had been asked to abstain from food or drinks or cleaning the teeth for at least 30 min preceding saliva collection. Student assistance in sample collection, transportation to the Microbiology Laboratory of the Saint Camillus Hospital, as well as pre-testing clerical work were performed by skilled health care worker teams named USCAR (Special Rehabilitation Care Continuity Units). Samples were registered on the local Laboratory Information System (LIS) and processed as follows:

• 1Salivette®: centrifugation at 1000xg for 2 min.
• 2A 250 μl aliquot of the liquid obtained after the centrifugation, lying in the external tube, was transferred to a new tube to obtain a 1:2 dilution with the lysis buffer of the Lumipulse G SARS-CoV-2 Ag kit and then loaded on the Lumipulse G 1200 instrument. The cut-off, based on the ROC curve, is set by the producer at 0.67 pg/mL; therefore, all results ≤0.67 pg/mL were considered negative. Only samples with antigen results above the cut-off were immediately trasferred to the Laboratory of Virology of the National Institute for Infectious Diseases “L. Spallanzani” (INMI), that is about 200 mt distance from the previous one, for confirmatory molecular testing. Among the 1905 collected samples, 1856 (97.43%) provided sufficient volume and underwent antigen testing. Height samples (0.43%) coming from 4 different schools, resulted positive for the presence of SARS-CoV-2 antigen, 4 of which were confirmed positive with the molecular assay (Table 1 ). As shown in the Table, the 4 confirmed samples showed N antigen concentrations ranging between 1.24 and 44.36 pg/mL, in line with the Ct obtained with the molecular assay (range 23.4–30.4). The low confirmation rate is not surprising, but is consistent with the positive predictive value established during the evaluation of clinical performances if the antigen test in settings of very low prevalence.^{7 , 8}

Table 1.

Description of the 8 saliva samples resulting antigen-positive with Lumipulse® G SARS CoV-2 Ag.

Confirmatory test out come	ID	School	Gender	Age	Lumipulse® GSARS-CoV-2 Ag (pg/mL)	Simplexa™ COVID-19 Direct assay(Ct values)
Positive	6283	n. 1	M	13	44.36	23.4
	4298	n. 2	F	5	33.19	24.8
	4385	n. 2	F	8	1.82	29.1
	4278	n. 3	M	9	1.24	30.4
Negative	4425	n. 3	M	11	1.32	>40
	4404	n. 3	F	9	1.18	>40
	5481	n. 4	F	9	0.69	>40
	5465	n. 4	F	10	0.69	>40

All the results were communicated before the scheduled time (median time: 4:16 PM for negative and 7:15 PM for positive results), allowing the prompt adoption of isolation measures for positives and to quickly carry out contact tracing. The communication through e-mail to the ASL was preceded by a direct telephone call notification only for positive results.

In a low prevalence setting, the availability of an antigen test and of a molecular test that can be performed on the same saliva sample to confirm positive results, without requiring the subject to be recalled for sampling repetition, was a determining factor in the choice of the strategy adopted in this pilot study. The screening test adopted for the program is a laboratory test and, therefore, required transport and processing in the laboratory, thus implying longer times than a point of care test (POCT). However, this did not represent a major obstacle to achieve same day diagnostic definition, due to the timely organization and information flow. On the other hand, the disadvantage of longer process was overcome by the greater accuracy of a laboratory test compared to a POCT^{2 , 9 , 10} and the ability to quickly perform the confirmation test with a system compatible with "urgent" execution.

Overall, the adopted strategy did not evidence critical elements, and the workflow control mechanisms resulted to be appropriate, as they made it possible to timely monitor the process in all phases, from sample collection to the delivery of final results. Our results suggest that the model can be replicated and expanded in a modular way also to different settings that have similar logistics.

CRediT authorship contribution statement

Licia Bordi: . Gabriella Parisi: . Giuseppe Sberna: Formal analysis. Alessandra Amendola: Formal analysis. Bruno Mariani: Formal analysis. Guido Meoni: Formal analysis. Daniela Orazi: . Pierluigi Bartoletti: Data curtion. Lorella Lombardozzi: Funding acquisition. Alessandra Barca: Funding acquisition. Maria Rosaria Capobianchi: . Fabrizio D'Alba: . Francesco Vaia: .

Declaration of Competing Interest

The authors report no declarations of interest.