Thread by @DaveKeto

Welcome, to my first Tweetorial! This subject is indeed near and dear to my heart for many reasons you'll soon come to know. So without further ado, I give you:

If a drug reduces your risk of dying by cardiovascular disease, will it necessarily reduce your risk of dying overall?

Three risk hypotheticals:
1/3: Would rock climbing without safety equipment reduce your chance of dying by cancer?

Three risk hypotheticals:
2/3: Would playing in the traffic reduce your chance of dying by Alzheimer's Disease?

Three risk hypotheticals:
3/3: Would skydiving without a parachute reduce your chances of dying by infection?

How can it be that a drug that *reduces* your risk of CVD could possibly *increase* your risk of dying by something else?
(A) The drug can inhibit something beneficial in the body
(B) The drug can increase risk of death by other means

How can we be sure something reduces risk of the targeted disease and does not increase risk of death by other means?
(A) You measure death by targeted disease
(B) You measure death by everything other than the targeted disease

This “overall death” is commonly referred to as All-Cause Mortality, often abbreviated to ACM. As in, death by *any cause*.

This is measured in many different ways from tracking for death certificates to mortality databases in the country of origin.

Is there any possible way to determine the mortality benefit of a drug, activity, or diet from tracking one form of death alone?

Many point out collection of ACM isn’t always reliable, that it can be costly to the trial, and it can force the trial to take longer to gather statistical significance. Given these reasons, should some trials that claim a mortality benefit be excluded from having to collect ACM?

Last, BONUS question...
For those who follow me, which of these do I advocate:
(A) Having ACM in every study that claims a mortality benefit
(B) Including ACM in the opening paragraph of an abstract and/or results
(C) A day without explaining why ACM is important