Lessons from applied large-scale pooling of 133,816 SARS-CoV-2 RT-PCR tests

Pooling multiple swab samples before RNA extraction and real-time reverse transcription polymerase chain reaction (RT-PCR) analysis has been proposed as a strategy to reduce costs and increase throughput of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) tests. However, reports on practical large-scale group testing for SARS-CoV-2 have been scant. Key open questions concern reduced sensitivity due to sample dilution, the rate of false positives, the actual efficiency (number of tests saved by pooling), and the impact of infection rate in the population on assay performance. Here, we report an analysis of 133,816 samples collected between April and September 2020 and tested by Dorfman pooling for the presence of SARS-CoV-2. We spared 76% of RNA extraction and RT-PCR tests, despite the frequently changing prevalence (0.5 to 6%). We observed pooling efficiency and sensitivity that exceeded theoretical predictions, which resulted from the nonrandom distribution of positive samples in pools. Overall, our findings support the use of pooling for efficient large-scale SARS-CoV-2 testing.