Search

Background The Centers for Disease Control and Prevention recommends serial rapid antigen assay collection within congregate facilities. Although modeling and observational studies from communities and long-term care facilities have shown serial collection provides adequate sensitivity and specificity, the accuracy within correctional facilities remains unknown. Methods Using Connecticut Department of Correction data from 21 November 2020 to 15 June 2021, we estimated the accuracy of a rapid assay, BinaxNOW (Abbott), under 3 collection strategies: single test collection and serial collection of 2 and 3 tests separated by 1–4 days. The sensitivity and specificity of the first (including single), second, and third serially collected BinaxNOW tests were estimated relative to RT-PCRs collected ≤1 day of the BinaxNOW test. The accuracy metrics of the testing strategies were then estimated as the sum (sensitivity) and product (specificity) of tests in each strategy. Results Of the 13 112 residents who contributed ≥1 BinaxNOW test during the study period, 3825 contributed ≥1 RT-PCR paired BinaxNOW test. In relation to RT-PCR, the 3-rapid-antigen-test strategy had a sensitivity of 95.9% (95% CI: 93.6–97.5%) and specificity of 98.3% (95% CI: 96.7–99.1%). The sensitivities of the 2- and 1-rapid-antigen-test strategies were 88.8% and 66.8%, and the specificities were 98.5% and 99.4%, respectively. The sensitivity was higher among symptomatic residents and when RT-PCRs were collected before BinaxNOW tests. Conclusions We found serial antigen test collection resulted in high diagnostic accuracy. These findings support serial collection for outbreak investigation, screening, and when rapid detection is required (such as intakes or transfers). [ABSTRACT FROM AUTHOR]

The article reflects the views of author on clinical testing of risk factors for SARS-CoV-2 in a Statewide Correctional System . Topics include Correctional settings are considered to be high-risk environments for transmission of the severe acute respiratory syndrome coronavirus 2; and confidence intervals adjusted for multiplicity and should not be used to draw inferences about definitive associations.