Thread by @igdstrachan

I drew some epidemic curves to show the effect of the much talked about "False positives". It shows there is no such thing as a "False Positive Pseudo-epidemic", much though @ClareCraigPath would want you to believe. /1

I derived the blue curve using the commonly used S-I-R model for the progress of an epidemic. We'll take this as the actual cases (which we can't measure directly). en.wikipedia.org/wiki/Compartme… /2

The green curve gives the corresponding number of "cases" (seen as PCR positives). This is with the widely cited figures of 80% sensitivity and false positive rate (FPR) of 0.8%. At low levels, this curve exceeds the actual curve, but not by much. /3

At high levels of prevalence, the number of cases detected is lower than the actual number (because it misses 20% of cases). But the trend still shows an increase in prevalence, and these are NOT false positives. /4

Now you could increase Ct so the sensitivity gets to 99%, but there is a trade off - the FPR also increases. The orange curve assumes FPR of 10%. This means that at the peak the PCR positives slightly exceed the true number. But not by much /5

At low levels of prevalence, the false positives are dominant, and most positives are false. But when the epidemic peaks, the effect of false positives is negligible, even with a massive 10% false positive rate. /6

These curves can be simply calculated by the application of Bayes's theorem.

Don't believe people who tell you it's a pseudo-epidemic or a "casedemic". That's a load of nonsense.

Bottom line is that at the peak of the epidemic, false positives are not significant. /7