[analyticascent]

An interesting piece, this is an important discussion point when it comes to international healthcare policy comparisons. I like collecting things like these for a NLP project that outputs variables and methodologies used in a given study.

However, the title appears to be misleading.

If I'm reading that article correctly, it seems to say that we can't predict whether someone is receiving expenditures that are in the last year of someone's life - the general claim that roughly 25% or so of Medicare spending takes place in the last year of people's lives is still accurate.

[stmw]

Well, the statistical trouble is that if some ER treatment costs $100K and has a 50% chance of saving someone's life (and prolonging it by 15 years), or else not making a difference, the end result is that 50% of spending takes place in the last 10 minutes of people's lives.

[analyticascent]

Not only that, but using such a short time period would fail to distinguish Medicare expenditure trends and any other method of payment - which is why people look at the last few years of someone's life and whether that jump in costs is making any difference.

To extend your ER scenario, would that $100K expenditure be reasonable if it only prolonged someone's life by a year or so?