Workflow determinants of recovery efficiency in wastewater-based surveillance of respiratory viruses

Wastewater-based epidemiology (WBE) is becoming a key tool for population-level surveillance of respiratory viruses, yet variability in recovery efficiency (RE) across analytical workflows limits the comparability and accuracy of disease-burden estimates. We systematically assessed the determinants of RE using 835,208 quality-filtered records from the United States CDC National Wastewater Surveillance System (NWSS), covering SARS-CoV-2, influenza, and respiratory syncytial virus (RSV). The reported method frequencies and RE values reflect NWSS practice and are not necessarily representative of the broader independent WBE literature. Across seven workflow stages, 358 unique end-to-end workflows were identified, with the three most common pathways accounting for 31.2% of samples. Because the native viral load entering the workflow cannot be quantified in field samples, measured RE reflected recovery of spiked process control surrogates. Of twelve surrogates used, workflow effects on RE were consistent across surrogates. Spike matrix location and spike concentration introduced further structured variability of comparable magnitude, with bovine coronavirus vaccine spiked at the dewatered solids stage yielding the highest and most reproducible recovery. Random Forest regression with mean decrease impurity, permutation importance, and SHAP analysis consistently ranked nucleic acid extraction method as the dominant driver of RE (62.2-70.9% of feature importance), followed by concentration method (14.7-21.1%) and PCR platform (7.4-13.2%). Viral target contributed minimally once methodological factors were accounted for, supporting standardised multi-pathogen workflows. Optimising nucleic acid extraction therefore represents the most effective step for improving WBE sensitivity, and routine reporting of RE with its process-control parameters is essential to enable cross-laboratory comparability.