Implementation of Practical Surface SARS-CoV-2 Surveillance in School Settings

Abstract

Surface sampling for SARS-CoV-2 RNA detection has shown considerable promise to detect exposure of built environments to infected individuals shedding virus who would not otherwise be detected. Here, we compare two popular sampling media (VTM and SDS) and two popular workflows (Thermo and PerkinElmer) for implementation of a surface sampling program suitable for environmental monitoring in public schools. We find that the SDS/Thermo pipeline shows superior sensitivity and specificity, but that the VTM/PerkinElmer pipeline is still sufficient to support surface surveillance in any indoor setting with stable cohorts of occupants (e.g., schools, prisons, group homes, etc.) and may be used to leverage existing investments in infrastructure. IMPORTANCE The ongoing COVID-19 pandemic has claimed the lives of over 5 million people worldwide. Due to high density occupancy of indoor spaces for prolonged periods of time, schools are often of concern for transmission, leading to widespread school closings to combat pandemic spread when cases rise. Since pediatric clinical testing is expensive and difficult from a consent perspective, we have deployed surface sampling in SASEA (Safer at School Early Alert), which allows for detection of SARS-CoV-2 from surfaces within a classroom. In this previous work, we developed a high-throughput method which requires robotic automation and specific reagents that are often not available for public health laboratories such as the San Diego County Public Health Laboratory (SDPHL). Therefore, we benchmarked our method (Thermo pipeline) against SDPHL's (PerkinElmer) more widely used method for the detection and prediction of SARS-CoV-2 exposure. While our method shows superior sensitivity (false-negative rate of 9% versus 27% for SDPHL), the SDPHL pipeline is sufficient to support surface surveillance in indoor settings. These findings are important since they show that existing investments in infrastructure can be leveraged to slow the spread of SARS-CoV-2 not in just the classroom but also in prisons, nursing homes, and other high-risk, indoor settings.

Keywords: COVID; SARS-CoV-2; environmental sampling; public health; qPCR.

FIG 1

Comparison of SDS/Thermo and VTM/PE pipelines on contrived samples. Average Cq values of contrived samples with the SDS/Thermo pipeline versus the Average Cq of matched samples processed through VTM/PerkinElmer pipeline. A linear regression was overlaid on the measured data (in blue) (Pearson correlation, m = 1.16, b = −1.14, r = 0.81, P = 2.87 × 10⁻⁶). The gray line represents the expected Cq values where x = y, i.e., if the two assays performed identically on the same samples.

FIG 2

Ili mapping of positive detection events across pipeline combinations on a representative 3D render of the rooms swabbed. This 3D rendering represents the relationship between rooms and features that were swabbed in Apt C (Table 1). The color scale represents the number of positive detection events returned across the combinations of extraction and RT-qPCR facilities.

FIG 3

Comparison of SDS/Thermo and VTM/PE pipeline combinations on real samples. (A and B) Scatterplots showing the performance of the Thermo (UCSD) and PerkinElmer (PHL) RT-qPCR workflows on surface samples extracted at both facilities. Empty x’s and o’s next to the sample name indicates that no viral signal was detected in that sample for that combination of extraction and RT-qPCR facility. (C) Swarm plots showing that the sensitivity of the Thermo RT-qPCR workflow is higher than that of the PE pipeline (Kruskal-Wallis, P < 0.01). Post hoc analysis showed that there was no significant difference between samples that underwent RT-qPCR at the same facility (P > 0.1) but there were differences between RT-qPCR facilities (P < 0.05). (D) Venn diagram showing the number of positive samples detected by each of the extraction facility/RT-qPCR pipeline combinations.