That was mentioned at the very start of the 2nd link (see "superior good") and it's true virtually everywhere in the long run (especially in developed countries with reasonably reliable growth in disposable incomes and where we have pretty reliable data). If we want to identify systems that might allow us to contain costs on the margins knowing who spends least or the most is not informative in and of itself. Rather we should be interested in residuals with the best model, as in, who is furthest below trend (as a function of comprehensive consumption or disposable income), what plausibly explains this, and whether are we actually willing and able to implement those sorts of features in our system....

Incidentally these patterns are not only found in health care, the basket of goods has changed and continues to change as countries get richer: https://randomcriticalanalysis.wordpress.com/2017/05/09/towa...