Allow me to respond (your comments in bold):
" N-of-1 approach is not perfect because it is impossible to control all other factors"
There is a slight misunderstanding about the concept of controlling in relation to a properly conducted N-of-1: By definition, in SCEs we only change one factor but leave all others unchanged. Controlling is not issue then.
"Perhaps you under-estimate the amount of salt in something you occasionally eat." That's why we have suitably long baseline and intervention phases during which we collect enough data points for the SCED statsitics to recognize changes even when there are "ocasional" outliers or missing data.
As you recognize yourself: "everything would have to be exactly the same except the factor you are testing" That's exactly what we do in SCED while accounting for the outliers.
"And when it comes to BP testing, you really need to do it at the same time of day and with under the same circumstances" You are preaching to the choir. I have addressed this issue thoroughly in an earlier post (https://medium.com/read-or-die-hq/your-hypertension-diagnosis-why-its-possibly-wrong-and-the-101-of-getting-it-right-c7066423734c).
" Finally, if you measure lifespan from birth, changes are often mostly just measuring infant mortality rates". Valid point. Though, if you look at the the risk of dying against age, what you get is a J-shaped curve (risk of death is initially greatest directly after birth, then declines until adolescence before it begins to rise again), no matter at which cohort you look (from 1800 to today).
Across the ages from 1800 to today, the whole curve has shifted downwards, i.e. the annual rates of death have declined across all age groups, which is reflected in the longer lifespans observed. Also, the shape of the curve has remained intact, which would not be the case if changes in infant mortality had more than marginally affected lifespan extension. Another reason why the gap between lifespan and health span has been widening is the dramatic increase in survival rates of first cardiovascular events (heart attacks and strokes) since the 60s/70s. You survive an infarct, you automatically get membership in the alive-but-diseased group.
