Inferring hospital admissions from SARS-CoV-2 virus loads in wastewater in The Netherlands, August 2020 - February 2022.

Wastewater-based surveillance enables tracking of SARS-CoV-2 circulation at a local scale in near-real time. Here we investigate the relation between virus loads and the number of hospital admissions in the Netherlands. Inferred virus loads from August 2020 until February 2022 in each of the 344 Dutch municipalities are analysed in a Bayesian multilevel Poisson regression to relate virus loads to daily age-stratified (in groups of 20 years) hospital admissions. Covariates include municipal vaccination coverages stratified by age and dose (first, second, and booster) and prevalence of the circulating coronavirus variants (wildtype, Alpha, Delta, and Omicron (BA.1 and BA.2)). Our model captures the relation between hospital admissions and virus loads well. Estimated hospitalisation rates per 1,000,000 persons per day at a virus load of 1013 particles range from 0.18 (95 % Prediction Interval (PI): 0.046-0.48) in children (0-19 years) to 20.1 (95 % PI: 9.46-36.8) in the oldest age group (80 years and older) in an unvaccinated population with only wildtype SARS-CoV-2 circulation. The analyses indicate a nearly twofold (1.92 (95 % PI: 1.78-2.05)) decrease in the expected number of hospitalisations at a given virus load between the Alpha and the Omicron variant. Our analyses show that virus load estimates in wastewater are closely related to the expected number of hospitalisations and provide an attractive tool to detect increased SARS-CoV-2 circulation at a local scale, even when there are few hospital admissions. Our analyses enable integration of data at the municipality level into meaningful conversion rates to translate virus loads at a local level into expected numbers of hospital admissions, which would allow for a better interpretation of virus loads detected in wastewater.

PCR standard curve quantification in an extensive wastewater surveillance program: results from the Dutch SARS-CoV-2 wastewater surveillance.

Since the start of the COVID-19 pandemic in 2020, wastewater surveillance programs were established, or upscaled, in many countries around the world and have proven to be a cost-effective way of monitoring infectious disease pathogens. Many of these programs use RT-qPCR, and quantify the viral concentrations in samples based on standard curves, by including preparations of a reference material with known nucleic acid or virus concentrations in the RT-qPCR analyses. In high-throughput monitoring programs it is possible to combine data from multiple previous runs, circumventing the need for duplication and resulting in decreased costs and prolonged periods during which the reference material is obtained from the same batch. However, over time, systematic shifts in standard curves are likely to occur. This would affect the reliability and usefulness of wastewater surveillance as a whole. We aim to find an optimal combination of standard curve data to compensate for run-to-run measurement variance while ensuring enough flexibility to capture systematic longitudinal shifts. Based on more than 4000 observations obtained with the CDC N1 and N2 assays, taken as a part of the National Sewage Surveillance program at the Dutch National Institute for Public Health and the Environment, we show that seasonal and long-term shifts in RT-qPCR efficiency and sensitivity occur. We find that in our setting, using five days of standard-curve data to quantify, results in the least error prone curve or best approximation. This results in differences up to 100% in quantified viral loads when averaged out over a nationwide program of >300 treatment plants. Results show that combining standard curves from a limited set of runs can be a valid approach to quantification without obscuring the trends in the viral load of interest.