A prospective diagnostic evaluation of accuracy of self-taken and healthcare worker-taken swabs for rapid COVID-19 testing

BACKGROUND: Rapid diagnostic tests (RDTs) developed for point of care detection of SARS-CoV-2 antigen are recommended by WHO to use trained health care workers to collect samples. We hypothesised that self-taken samples are non-inferior for use with RDTs to diagnose COVID-19. We designed a prospective diagnostic evaluation comparing self-taken and healthcare worker (HCW)-taken throat/nasal swabs to perform RDTs for SARS-CoV-2, and how these compare to RT-PCR. METHODS: Eligible participants 18 years or older with symptoms of COVID-19. 250 participants recruited at the NHS Test and Trace drive-through community PCR testing site (Liverpool, UK); one withdrew before analysis. Self-administered throat/nasal swab for the Covios® RDT, a trained HCW taken throat/nasal sample for PCR and HCW comparison throat/nasal swab for RDT were collected. RDT results were compared to RT-PCR, as the reference standard, to calculate sensitivity and specificity. FINDINGS: Seventy-five participants (75/249, 30.1%) were positive by RT-PCR. RDTs with self-taken swabs had a sensitivity of 90.5% (67/74, 95% CI: 83.9–97.2), compared to 78.4% (58/74, 95% CI: 69.0–87.8) for HCW-taken swabs (absolute difference 12.2%, 95% CI: 4.7–19.6, p = 0.003). Specificity for self-taken swabs was 99.4% (173/174, 95% CI: 98.3–100.0), versus 98.9% (172/174, 95% CI: 97.3–100.0) for HCW-taken swabs (absolute difference 0.6%, 95% CI: 0.5–1.7, p = 0.317). The PPV of self-taken RDTs (98.5%, 67/68, 95% CI: 95.7–100.0) and HCW-taken RDTs (96.7%, 58/60, 95% CI 92.1–100.0) were not significantly different (p = 0.262). However, the NPV of self-taken swab RDTs was significantly higher (96.1%, 173/180, 95% CI: 93.2–98.9) than HCW-taken RDTs (91.5%, 172/188, 95% CI 87.5–95.5, p = 0.003). INTERPRETATION: In conclusion, self-taken swabs for COVID-19 testing offer an accurate alternative to healthcare worker taken swabs for use with RDTs. Our results demonstrate that, with no training, self-taken throat/nasal samples can be used by lay individuals as part of rapid testing programmes for symptomatic adults. This is especially important where the lack of trained healthcare workers restricts access to testing.