There are actually three ways the accuracy of diagnostic tests is measured. How likely are they (%) to diagnose someone who has the infection (sensitivity), how likely are they (%) to correctly diagnose someone who doesn’t have the infection (specificity) and how much of the virus needs to be in a sample for a test to reliably pick it up.
For this test, clinical trials included 157 clinically sourced positive and 209 confirmed negative NP / OP samples. The positive coincidence rate (sensitivity) was 96.2% and the negative coincidence rate (specificity) was 99.5%. This gives a Positive Predictive Value (PPV) of 99.3%, and a Negative Predictive Value (NPV) of 97.1%.
The limit of detection was calculated using a reference solution of inactivated virus supplied by the Wuhan Institute of Virology, Chinese Academy of Sciences, with an original concentration of 1×104 TCID50/ml. This was used to calculate the limit of detection as 0.02ng/ml (recombinant N protein), or 0.5TCID50/ml (inactivated virus). Using latex agglutination immunochromatography method the calculation is 0.1ng/ml (recombinant N protein), or 1TCID50/ml (inactivated virus).
The MHRA Target Product Profile asks for LoD in RNA copies per ml and by using a number of coefficients to convert from both TCID50 and ng/ml we estimate it to be under 1,000 copies per ml of sample.
For comparison, lateral flow tests generally need 100-300 TCID50/ml present to reliably show a positive – so this test needs 200x – 600x less viral material in a sample to correctly identify a positive.