| The local minima for diabetes is likely not the overall minima for death rate. I would be surprised if the minima for diabetes diagnosis was NOT slightly below the minima for overall death rate. The famous JAMA article from 2013 that everyone likes to cite, including in comments below, showed no significant increase in death rate for grade 1 obesity and the effects really kicked in strongly around grade 2 and 3 obesity. The more recent BMJ article from 2016 that no one wants to cite, showed minimum death rate in the 20-24 BMI range depending on smoking history. That paper reported the most reliable looking studies of 'non-smokers followed up for over 20 years' had a minimum total death rate at a BMI around 20-22, but that does not support the "Healthy at Every Size" narrative so its memoryholed. I try to keep up to date on diet and supplement journal articles; there's probably journal articles newer than 2016 thats not in my notes yet. Something EVERY study seems to agree on is the death-curve looks very U shaped kind of like computer chip hardware failure rates. The point being that studies disagree on the exact minima death rate vs BMI which is only relevant for large scale population goals, however they all agree that going from, perhaps, 22 to 23 will have an effect that although possibly measurable if across enough people, will tiny and be deep in the decimal places, whereas going from "twenties" to "forties" for BMI means the patient is unquestionably going to die very young, although EXACTLY how young may vary from study to study. The problem with BMI of course is it was originally a screening criteria to "find the worst quartile and counsel them" but as happens with all metrics over time eventually the rough and imprecise low resolution screening criteria turned into an "optimize for its own sake" metric and people getting very weird and hyperfocused about their personal metric calculated to five sig figs at least. |