Thread by @DaveKeto

1/ I'm looking forward to covering Bradford Hill in an upcoming presentation. I find there's often a misconception that he introduced a low standard to claim causality with epidemiological data. Actually, he was far stricter than he's been portrayed... edwardtufte.com/tufte/hill

2/ Many are quick to claim they've helped to demonstrate causality using his criteria (sometimes just a few criteria, not all). But he's already remarked on many occasions how important each of these are taken together -- and likewise given examples to demonstrate.

3/ In other words, I think he himself would likely disagree with the vast majority of the modern claims made about meeting his criteria.

The first two, Strength and Consistency, are arguably the most intuitive and yet hardest to meet.

4/ But they can appear easy to meet if selecting the evidence that supports it and excluding the evidence that doesn't (which is in direct violation of Consistency itself).

He actually addresses this quite directly under that criterion...

5/ "The lesson here is that broadly the same answer has been reached in quite a wide variety of situations and techniques. In other words, we can justifiably infer that the association is not due to some constant error or fallacy that permeates every inquiry."...

6/ "And we have indeed to be on our guards against that."

Clearly if there are other major influences at play in the outcome of interest and certain "situations and techniques" can help to illuminate this -- they can't be ignored.

7/ Of course, this comes up regularly with my focus on looking at high LDL-C/-P/ApoB in normal populations without Atherogenic Dyslipidemia (low HDL-C and high Triglycerides).

For example, would diet-induced high LDL+high HDL-C+low TG result in high cardiovascular disease?

8/ We're presented with an extraordinary opportunity to study this directly with the Lean Mass Hyper-responder study through the CitizenScienceFoundation.org