[bill_from_tampa]

The encouragement is that the rate of "death from colorectal carcinoma" seems to be reduced in studies of screening. This is a 'disease specific' mortality statistic. Most of us don't care why we die, or what is ultimately written upon our death certificates, we would simply rather still be alive! So reduction in "total mortality" would be a more convincing endpoint. If a study of some screening test for a dread disease does NOT show a reduction in the 'total mortality' of the group screened at some clinically reasonable point of time in the future, one could argue that the screening was pointless. Yes, perhaps less people died from the condition being screened for, but if the same number of persons died in the screened -vs- unscreened group, what has been accomplished?

It has been hard, well impossible, to show that screening for colorectal cancer reduces the total or overall death rate. For example, a recent study published in the NEJM in 2022 did find a reduction in persons who died from colorectal carcinoma after screening with colonoscopy. But they did NOT find that the total or overall death rate had decreased!

"The risk of death from any cause was 11.03% in the invited group and 11.04% in the usual-care group (risk ratio, 0.99; 95% CI, 0.96 to 1.04)." N Engl J Med 2022;387:1547-1556 DOI: 10.1056/NEJMoa2208375V

When reading 'screening' studies, one usually needs to look very carefully at the article and published data to find this statistic. Sometimes it is not even reported at all, it is simply ignored. It's almost like the authors don't want the fact that the screening program does not reduce one's risk of dying to any measurable degree is ... embarrassing?

This problem is not unique to colorectal cancer screening.

[epistasis]

When studies are designed a key parameter is the statistical power that the size of the stud provides: how big of a difference in the key stat can it find?

I would expect that the design of a study will be able to find differences in the key readout: did screening reduce death in the types of cancer it could find.

Colon cancer deaths are only a small fraction of total mortality, and statistically, finding a small change in a ratio close to zero is far easier to see than a that same change in a ratio that's further away from zero.

So I would not expect a study to be powered to be able to find changes in total mortality from all causes. That would take a far larger number of people and generally be considered a waste of resources that would be better spent investigating other things.

So I wouldn't take the lack of statistical significance in total mortality as evidence of anything, unless total mortality was a primary outcome that was being tested.

Pretty much all the good interventions for reducing mortality will not have much impact on total mortality because there are so many things causing total mortality. But add up all the small ones and it starts to make differences in total mortality.