That is generally true - but there are some real problems with the classic evidence pyramid thinking.
I would emphasize that “lower-quality” evidence such as meticulously done N-of-1 crossover trials & Mendelian randomization can be very useful if carefully interpreted with a lot of contextual knowledge, while “high-quality” evidence can be useless or harmful.
Here’s an example:
Ivermectin in a meta-analysis was shown to be effective for COVID-19 treatment. This was later shown to be based on fraudulent data.
As for the case of LDL-P/apoB, my current opinion is LDL-P<1000 and LDL 70-90 (if not discordant) would be a reasonable treatment target for a lipidologist to consider, particularly if there are risk-enhancing factors, medication side effects are very carefully considered, and if it cannot be done through lifestyle alone.
Let’s say we compare LDL 55 vs LDL 70-90, assuming no discordance. What is the purported difference in absolute risk reduction on all-cause mortality over a few decades (let’s say 20-30 years) if your position happens to be correct? That should be more informative.
I suspect the difference would be clinically insignificant in a “healthy” individual with no risk factors (maybe a theoretical 1% or less absolute risk reduction) - while waiting for more information over the next 20 years is probably a better option.
That being said, many people will have varying opinions on this. My general rule is to cut my confidence level in half after a lot of research on any purported benefit (expert opinions alone can and very often are incorrect) - unless there are very, very good reasons and I fully understand the subject at hand. We can size a position based on confidence level of the benefits vs harms accordingly.
