This.

And even your willing to exercise changes during live. It was easier and fun to exercise when was a teenager and so on. Become lot less when gained almost 40 pounds in a month due to medication. Become even less interesting when would get hurt for exercising while overweight. Become a no-go when got a muscle strain for simply going downstairs because a physiotherapist pushed a bit too much and was under muscular fatigue.

I'd prefer to take a bath and have some benefits that fell all that pain again and immobilize my leg and loose what I've earned in the past 4 months.

For those who already like to exercise (or are willing to despite saying "I don't like, just like the benefits", or any kind of wording of this idea), well, this study won't make a difference.

For me, yes, it makes.

I agree with your comments entirely. I'll also add, however, that even though this is cohort data 1) there is A LOT of data, almost all of it in agreement for hundreds and hundreds of studies, over millions of person years and 2) there is pretty strong evidence of a dose-response relationship. Both of those two points help (though don't entirely eliminate) the weaknesses inherent in cohort data.

It doesn't change the fact that if you take approx 150 from each of the activity groups, the cohort study will tell:

Approximately 5 from the "slobs" will die.

Approximately 4 from the "moderate exercise group" will die.

Approximately 3 from the "forrest gump group" will die.

On a personal level, to me these groups are more or less the same in terms of mortality, but somebody marketed the 20% really well.

I understand your point, while not detracting from the benefits of exercise, eating well, having moments of calm, etc,etc, these studies when reduced to a personal level become nonsensical, while one can make choices that at a 'group' level might suggest cause and effect, when the sample size is one, and it's you, the 'odds' are of little consequence, what is happening to you is the reality, knowing some malaise is is common, rare, or influenced by some life factors has no bearing...

that's not the point, the posters point is that the way the math was used added dramatic effect, it did not aid in understanding the meaning of the findings.

Exactly. 20% or 40% sounds really big. In reality, you're maybe lowering your chance to die (during some time) by less than 2% by exercising almost constantly. Now, if you're not enjoying exercise, or feel exhausted after exercising, you actually reduce the amount of "quality life" by exercising.

If exercise was a drug, they'd never allow it on the market.

> If exercise was a drug, they'd never allow it on the market.

Plenty of drugs have much worse efficacy and "numbers needed to treat" figures than exercise.

Current exercise advice is about 30 minutes per session with 3 sessions a week.

Not really. Take your 30 minutes of exercise per week that will reduce the chance of dying by 1% (2% is for the forrest gumps). Let's say you can do that by spending 60 minutes a day (changing clothes, getting dressed / undressed, showered and so on). In a 52-week year that will amount to loss of 156 hours of your life. For 8-hour work days that's 19.5 workdays. A good vacation's worth! No drug is allowed to do that.

And if you break your ankle or break your wrist or elbow when exercising, it'll never be as good as new.

What a silly set of arguments. Exercise is literally empty time for you with no other benefit? Lets put aside the fact that you receive endorphin from almost any exercise you do. You can't go to the beach, rock climb, play with your kids/dog, etc.? No, must be productive all the time. Even if that was the case, it's quality adjusted lifespan that counts for most people.

You may break your wrist sure. In the same way that you may get wrist injury from typing too much. Exercise will reduce the likelihood of other injury.

In real world terms, holding on to the banister when going up and down stairs and sitting down when putting on your trousers on are real life extenders :)

This is the correct answer. Mobility is more important than exercise. You'll lose mobility by breaking bones.

There are are plenty of well understood mechanisms for how exercise can improve health. There are way may plausible but not yet well understood ways it can help.

Clearly any error from the self selection you described does not counter an demonstrable mechanism.

What if the study compared bedridden 100-year-olds as non-exercisers to the college track team? The track team was 20% less likely to die in the subsequent 10 years...

Self-selection can completely override other well-understood mechanisms...

These studies are usually done as cohort studies. Meaning the study group shares a defining trait (such as the same date of birth).

Obviously this is not as great as a double-blind study. But you can't let one group do real exercise and the other one placebo exercise. So the cohort study is probably the best possible study design in this case.

The people doing these studies try very hard to minimize biases (sometimes failing, but ruling out every possible biases is hard).

This is exactly why I can't trust prescriptive conclusions from correlation studies. Didn't Google publish some method on resolving causation in correlation? Something about transfer of noise?

3.84% to 2.08% is a 45% reduction in the amount of people that died. I don't understand your point, there is no obfuscation here, that's just how percentages work.

(3.84 - 2.08) / 3.84 = 45% less people dead

It's not 'my' percentage, nor subjective.

I mean you could argue the study is biased or the sample size too small, but your understanding of percentages seems flawed.

I recommend you take basic university statistics. Especially study efficacy vs effectiveness so you might have an idea how you're being mislead here.