Hacker News Reader

Hacker News

Comment by scotty79 | Hacker News Reader

Full thread

scotty79

It's really hard to report on correlation without implying causality.

So whenever you see research popping up in your news feed just flip the order of statements in the title.

"People at greater risk of dying walk less"

Then read the paper carefully for any signs of something that doesn't fit this flipped version.

"Correlation consists even when controlled for smoking and comorbidities"

It still can be true because smoking is just one factor out of thousands factors contributing to early deaths, and comorbidities probably bunched up so many unrelated stuff together that you could control for some random stuff and get really close results. And even if they accounted for all the comorbidities precisely still people with each might be in different condition and those farther from death might want to walk more.

View on HN

Agree. But even this meta-study of 175370 person-years only encountered 2310 deaths.

Imagine what it would cost to make an interventional study and track how accurately people report they are walking.

But even the credibility of this is somewhat higher than a pure "correlation" or cross-section study, because the dying is in the future, compared to when the walking is counted ("prospective" cohort studies).

And there is no possible blind study, because the participants know they are walking, and you can't have a "placebo walk".

> no possible blind study,

Not sure I agree with this.

I am definitely not saying this should be done, but as a thought experiment:

Imagine you worked for Fitbit, and could choose to undercount steps slightly.

You have a cohort of users who will walk each day until they reach 10k steps. You undercount steps for a random subset so that they end up walking 10.5k.

You repeat this experiment so you make the folks with a 5k goal actually walk 5.5k, etc

This is a pretty good blinded study. The difference is slight enough that the users don't notice the extra increase in distance consciously.

(Obviously you have to take into account the risk of churning people from the service somehow.)

If you were to check all cause mortality somehow (look up public death records? Some other way?) That would seem like a pretty good study design.

Again, perhaps unethical to do this without consent; but not impossible; you could ask a cohort of users to opt in to this sort of study.

If the claim is that it's distant dependent, this would likely take you a lot closer to causality.

This. Most research I see on factors related to activities vs death rate seem to have a pretty big duh! factor--are they healthy enough to be able to do the activity and consider it worth the effort it will take to do. There's no way they can possibly control for comorbidities because not everything is in the medical records, nor do medical records have all the details of severity or how it limits someone.

Even for things where ability doesn't really enter into it there's also the factor of the level of effort one spends on self-care. If I'm near death am I going to bother with the dentist unless something hurts?

Reply on news.ycombinator.com