Surveillance of SARS-CoV-2 prevalence from repeated pooled testing: application to Swiss routine data

Area of Science:

• Epidemiology
• Biostatistics
• Infectious Disease Surveillance

Background:

• Routine surveillance of SARS-CoV-2 using reported positive RT-PCR tests is subject to significant bias due to non-random testing patterns.
• Population-based sampling methods are crucial for accurate prevalence estimation, with pooled testing offering a promising yet analytically challenging approach.

Purpose of the Study:

• To develop and validate a Bayesian model for estimating SARS-CoV-2 prevalence from repeated pooled testing data.
• The model aims to correct for test sensitivity, account for uncertainty in sensitivity, and incorporate temporal and spatial correlations.

Main Methods:

• A Bayesian statistical model was developed to analyze repeated pooled testing data.
• Model validation was performed using simulated scenarios with varying sample sizes, pool sizes, and prevalence rates.
• The model was applied to a large dataset of 1.49 million pooled tests from Switzerland (2021-2022).

Main Results:

• The model demonstrated reliability for sample sizes ≥500, pool sizes <20, and prevalence <5%.
• Analysis of Swiss data revealed similar prevalence dynamics across schools, care centers, and workplaces, peaking at 4-5% in winter 2022.
• Prevalence estimates in schools showed strong correlation with reported cases, hospitalizations, and deaths (0.84–0.90).

Conclusions:

• Pooled testing, when analyzed with appropriate statistical models, provides a reliable and cost-effective alternative for SARS-CoV-2 surveillance.
• The developed Bayesian model effectively addresses analytical challenges associated with pooled testing data.
• This approach has implications for the surveillance of SARS-CoV-2 and potentially other infectious diseases.