Probability in disease testing

[PierreDumont]

Recently I found out about a coronavirus test from the company Euroimmun AG, which claims to have sensitivity 0.9 and specificity 1. Considering a prevalence of 5% and applying a simple Bayes' Theorem, we would get a positive predictive value of 1 and a negative predictive value of 0.9947. How is this possible?

 

[tkhunny]

My first alarm went off when EAG claimed its test to have a Specificity of 1. It NEVER produces a false positive? NEVER!? This may not explain your results, but perhaps you have not started with reasonable values.