Research on the optimal amount of aerobic activity per week

Conversion of MET*h/wk to number of steps

Wiki says that 3m walking is 4.8km/hr: https://en.wikipedia.org/wiki/Metabolic_equivalent_of_task
so 1 MET*h is 1.6km, which is ~2k steps.
so say 30 MET*h/wk is ~60k steps/week, so ~8.5k steps/day

Different articles sometimes differ a bit on how many METs they give for a brisk walk, so the steps may also float, but in general it's +- this.

The papers

The following articles are sorted by the persuasive power and importance of the results, beginning with the best ones.
(0) https://pubmed.ncbi.nlm.nih.gov/25844730/
110k deaths, 6 prospective studies. Sounds good for article.
According to this article, in my opinion, the best is ~35 MET *h/wk, which corresponds to ~9k steps/day.
5-15k steps/day -- also almost ideal. >20k steps/day is grey zone: perhaps little worse, but statistically insignificantly.
It's amazing that their median is just 8 МЕТ *h/wk (suppl.eTable2). Though their average year is 62 (still ~2k steps/day is almost nothing..)

(1) https://pubmed.ncbi.nlm.nih.gov/32100573/ AHA statement. See Table 4, page 18.
A review of papers showing that too much physical activity can be bad.
And their "too much" is not too much. For example, ">75 MET/h" or >3h running/week.

(1) https://pubmed.ncbi.nlm.nih.gov/23139642/ 82k deaths (Table 1), i.e. rather high weight of the article.
Median 8 MET*h/wk, highest quartile starts at 22, mostly US (but also Sweden) -- Table 1
Conclusion of the article: you need >22.5 MET*h/wk (>7k steps per day), and more is probably even better. So the article is almost useless.
All-cause HR improves up to >22.5 MET*h/wk, and the authors didn't explore beyond that (well... that's more than ~6-7k steps/day!!! why would anyone explore activity levels that high! *sarcasm*).
On the one hand, they write (and Table 3 shows) that >22.5 is about to plateau (but not yet).
On the other hand, in suppl.fig.1 they have this plateau starting to improve again after 25 MET*h/wk (however, this is a cubic spline, and who knows what it may show in the distance)
The authors advise to get >22.5 met, ie more than 22.5/33*10000 = 6800 steps per day (besides the steps that are not in a relatively long walk).
Moreover, the benefit of 22.5+ met compared with 15.0-22.4 met in neither table is statistically significant (at least in Table 4, in Table 5 for men, in Table 6 for never smokers and healthy never smokers) - so, just a hint, which should be investigated further. Also the benefit of 22.5+ met compared with 15.0-22.4 met was only found in 4 of the 8 studies used in this meta-analysis. So even 6,000 steps a day (19.8 met) may well have all the effects of aerobic exercise on life extension. At least according to the article under discussion.

(1) https://pubmed.ncbi.nlm.nih.gov/21846575/ only 10,780 deaths, i.e. the average weight of the article.

(2) https://pubmed.ncbi.nlm.nih.gov/32207799/ only 1165 deaths, i.e. low weight for the article.
p.6, Fig.2 -- the more the better for men. Even 14k is noticeably better than 12k (though it's not clear if it's statistically significant). And 16k is slightly better than 14k (again not clear). But this figure doesn't show CI.
Taking CIs into account (but on a common graph for both sexes) judging from Fig. 1, 12k steps is optimal (further HR is almost horizontal, but CIs go up).

(3) https://pubmed.ncbi.nlm.nih.gov/28051177/ only 1267 deaths, i.e. low article weight.
But they research cohort of chinese, 5% of whom have even >75 MET*h/wk. (finally, some people still walk much)
Judging from Fig.1, the optimal phys.activity is somewhere around 90 MET*h/wk (less -- the average HR increases more -- the upper limit of CI begins to increase, while the average HR is already ~horizontal) which is ~23k steps/day.