Evaluating Interlaboratory Variability in Wastewater-Based COVID-19 Surveillance

Abstract

Wastewater-based environmental surveillance enables the monitoring of SARS-CoV-2 dynamics within populations, offering critical epidemiological insights. Numerous workflows for tracking SARS-CoV-2 have been developed globally, underscoring the need for interlaboratory comparisons to ensure data consistency and comparability. An inter-calibration test was conducted among laboratories within the network monitoring SARS-CoV-2 in wastewater samples across the Lombardy region (Italy). The test aimed to evaluate data reliability and identify potential sources of variability using robust statistical approaches. Three wastewater samples were analyzed in parallel by four laboratories using identical pre-analytical (PEG-8000-based centrifugation) and analytical processes (qPCR targeting N1/N3 and Orf-1ab). A two-way ANOVA framework within Generalized Linear Models was applied, and multiple pairwise comparisons among laboratories were performed using the Bonferroni post hoc test. The statistical analysis revealed that the primary source of variability in the results was associated with the analytical phase. This variability was likely influenced by differences in the standard curves used by the laboratories to quantify SARS-CoV-2 concentrations, as well as the size of the wastewater treatment plants. The findings of this study highlight the importance of interlaboratory testing in verifying the consistency of analytical determinations and in identifying the key sources of variation.

Keywords: SARS-CoV2; detection methods; generalized linear models; interlaboratory ring test; wastewater environmental surveillance.

Figure 1

Workflow followed by the laboratories during the interlaboratory ring test (created in BioRender. https://BioRender.com/b09r836, accessed 14 February 2025).

Figure 2

Linearized relationships of the log-transformed concentrations (g.c./μL) across the laboratories responsible for the analytical phase. Full regression statistics for the linear relationship depicted in the chart are provided in Table S1.

Figure 3

Log-transformed detections of N1 gene fragment copies/µL concentration: differences among laboratories concerning the analytical and pre-analytical phases. Full regression statistics for the linear relationship depicted in the chart are provided in Table S2.

Figure 4

Log-transformed detections of N3 gene fragment copies/µL concentration: differences among laboratories concerning the analytical and pre-analytical phases. Full regression statistics for the linear relationship depicted in the chart are provided in Table S3.

Figure 5

Log-transformed detections of ORF1ab gene fragment copies/µL concentration: differences among laboratories concerning the analytical and pre-analytical phases. Full regression statistics for the linear relationship depicted in the chart are provided in Table S4.

Figure 6

Model estimated marginal means of the log-transformed detections of ORF gene fragments: it can be observed that Lab2 shows the most significant variability with respect to the other laboratories and with respect to the WWTP.

Figure 7

Linear relationships of the log-transformed gene copy values and Cq for the three gene fragments (N1, N3, and ORFab) of the two different RT-PCR systems (e.g., AgPath and QuantaBio).

Figure 8

Interlaboratory comparison of standard curves before (upper chart) and after (lower chart) the harmonization process.