Table 2.

Limits of detection

	Number of tested samples	50% limit of detection*(log10 SARS-CoV-2 RNA copies per swab†)	95% limit of detection*(log10 SARS-CoV-2 RNA copies per swab†)	Adjusted and converted 95% limit of detection*‡(SARS-CoV-2 RNA copies per swab§)
Abbott	105	5·61 (5·27–5·95)	7·45 (6·79–8·20)	3·52 × 106
RapiGEN¶	45	9·51 (8·84–12·26)	11·10 (9·71–17·01)	1·57 × 1010
Healgen¶	105	4·48 (3·41–5·32)	7·27 (6·27–8·40)	2·33 × 106
Coris BioConcept	105	7·60 (7·37–7·82)	8·36 (8·00–8·76)	2·86 × 107
R-Biopharm	105	5·40 (4·99–5·77)	7·22 (6·57–7·96)	2·07 × 106
nal von minden	105	7·19 (6·97–7·43)	7·87 (7·52–8·23)	9·27 × 106
Roche-SD Biosensor	115	5·64 (5·28–6·00)	7·68 (6·96–8·50)	5·98 × 106

AgPOCT=antigen point-of-care test.

^*Mean concentration that yields 50% or 95% positive AgPOCT results according to a binary logistic regression analysis; numbers in parentheses denote the 95% highest posterior density interval determined by the Bayesian binary logistic regression model.

^†Concentration per swab presumes that swabs are resuspended in 1 mL fluid during pre-analytical processes in RT-PCR used to measure viral loads.

^‡Due to a systematic pre-analytical dilution factor in our AgPOCT evaluations, the projected mean concentrations at which 95% hit rates were achieved were corrected to be 8-fold lower (a cumulative correction factor representing all correction factors between the actual volume input in our validation studies and the volume input as per manufacturer's instructions); input volumes in all cases are subject to great variability due to the undefined volumes of viscous respiratory tract specimens taken up by swab sampling devices, and our statistical evaluation suggests a level of precision that does not reflect the clinical reality in AgPOCT use.

^§Values have been converted to non-logarithmic number, as normally reported in clinical practice.

^¶Model fit was suboptimal due to a large difference in the number of positive and negative test results (figure 2).