Political Calculations has a post with an interesting table of life expectancy in OECD countries. In addition to the standard life expectancy numbers, there is an additional set that is standardized to eliminate differences in a category of deaths that is roughly described as accidents and homicide (those least likely to be connected to healthcare problems).
I haven’t found an online explanation of how they were standardized (it’s apparently explained in the book The Business of Health: The Role of Competition, Markets, and Regulation by Robert Ohsfeldt and John Schneider, which I haven’t checked), and I can’t evaluate the extent to which their desire to promote the U.S. medical system has biased their methods.
What surprised me most was that it implies that the differences in what we normally think of as health and healthcare explain a surprisingly small part of the difference between national life expectancies. The actual life expectancy shows a difference of 3.6 years between the highest (Japan) and lowest (Denmark), but the standardized life expectancy shows a difference of 1.2 years between the highest (U.S.) and the lowest (U.K.).
This implies that national difference in traffic accidents, homicides, and some similar (poorly identified) causes of death are a good deal more important than the following differences: healthcare systems, diet, serious vitamin D deficiencies (which I expect to vary by latitude), FDA rules, and litigation of medical outcomes.
On a loosely related note, the book A Farewell to Alms mentions a report that 16th century Japan had an unusual absence of disease (but no indication whether it’s possible to get any quantitative evidence of this). This made me think of the alleged high Cuban life expectancy. Could relatively isolated islands be healthier due to lower influx of disease? Not that this would make isolation nice, especially since it might mean increased vulnerability to disease when contact with the outside increases.