Ouellette et al. reading got me thinking that this dichotomy of period vs. cohort effects on mortality can play out differently depending on the exposure or disease that we study. That's why it seems that it is much more informative to study these effects in more nuanced contexts rather than in the context of overall mortality. Disaggregating based on the disease or exposure (e.g. cancers caused by smoking) helps to highlight some cohort effects that are not necessarily seen in the analysis of overall mortality. So, early life factors might play a large role after all.