Comment by worstspotgain

Just chiming in to add one factor that you overlooked: selective caloric restriction.

We know caloric restriction can extend lifespan. It's usually tested as a discrete variable, but its statistical effect in the wild is continuous. It can be unintentional, coincidental, income-related, historical event-related, culturally related, and related to the local economy. Multiple categories can apply at the same time, complicating the analysis.

Add fasting to the mix, and the analysis is even harder. Here's a difficult to refute conjecture: areas that underwent episodes of accidental fasting during world wars may have spikes of ultra-centenarians. The keyword is may, because you have to subtract out the negative correlations that come with the event. For instance, during a famine you may end up eating lifespan-reducing foods.