Biotech Values

[iwfal]

EXAS -

On the other hand, a non-invasive, no-prep test such as EXAS’ ColoGuard can be repeated often, allowing the cumulative sensitivity of a series of tests to be impressively high for both cancer and pre-cancer even when the sensitivity of each individual test is only middling.

That's only true if the individual tests are largely independent - and I'd bet they aren't. E.g. Some tumors never shed much. Some tumors don't have a genetic profile easily detected by the test. And most tumors of any given type are only detected when they reach a size threshold. And note that the company has released statistics consistent with your thesis - but I'd bet a LOT of money that it is a theoretical calculation based upon complete independence (since it EXACTLY matches the theoretical calc - maybe that is where their name comes from? -g-). So they are either ignorant or disingenuous (not good either way).

Also note - at first glance it would seem to be extremely difficult to equate Cologuard statistics with colonoscopy statistics since the former is probably about detecting any polyp/tumor, but the former is about detecting any individual polyp/tumor. In order to equate them you need to know more data than the company has (that I can find) released (and perhaps more than they know? - e.g. for the false positives, are they sure they are false positives, or are they something beyond the reach of the conventional detection?)

None of this is to say that Cologuard might not be a valuable addition or even replacement for colonoscopy - but the simple stats they publish don't seem to tell the story. (A side benefit is that more comparative stats would help them to better sell the test to docs.)