Nothing seems to captivate health researchers more than the mystery of why health spending for the same conditions seems to vary so much across the country or even within a geographic region. There is a proliferation of studies on the subject, which we have faithfully reported. Yet another version is found in a new National Bureau of Economic Research paper. (NBER Paper) The researchers used a clever approach–comparing the health care utilization of Medicare beneficiaries who moved from high utilization/high cost regions to lower cost ones and vice versa. If the people moving from high utilization areas remain high utilizing, that would suggest that patient demand factors are primarily responsible for the variation. On the other hand, if their utilization reduces to the lower utilizing area they moved to, that would suggest that provider or supply-side factors are primarily at work. The same logic can be applied to beneficiaries going from a lower to a higher utilizing region. The data came from 1998 to 2008 and is adjusted for geographic Medicare price level differences.
The authors find that about 47% of variation is attributable to patient demand factors and this result is the same for moves both to and from high and low utilizing regions, and it is similar across all spending quartiles or deciles. The largest changes in utilization were for preventive care and ER visits, care categories in which patient discretion can play a major role. Less change was found for inpatient care and diagnostic and imaging tests, all of which are subject to physician ordering. Patient demographics appear to have little to do with the variation. Habit formation also appears to play a limited role, for example, older patients actually change their utilization more when they move than do younger ones. This seems to suggest that provider practice patterns may play a significant role in non-patient demand factors affecting variation. In their analysis, patient health differences account for a significant 22% to 37% of utilization variation. This paper is a valuable contribution to understanding there is health spending variation and provides guidance on steps that may help in reducing it.