Correlation Between Medicare and Commercial Spending

Geographic variation in Medicare spending is well-established, although the extent, especially on an adjusted basis, and the causes, are hotly debated.  Less well-studied is geographic variation in commercial health spending.  A new study examines whether the two types of coverage have spending that correlates in a particular geography and, if so or if not, why.   (AJMC Article)    The authors used data from a commercial health insurance database and Medicare data.  The research looked only at inpatient hospital spending in 2007.  Other categories of spending may vary differently, but hospital inpatient is the largest single sector of spending.  They examined both utilization of services and prices, and attempted to adjust prices by the Medicare wage index.

As others have, they found significant variation in Medicare spending across regions, even after adjusting for input cost differences and their effect on local prices.  The commercial insurance geographic variation was only slightly lower than Medicare’s.  The quantity of services delivered to each population was strongly positively correlated, but total inpatient spending was only weakly positively correlated.  This is because there is a negative correlation between utilization in one type of coverage and prices in the other.  Medicare volume was low where private insurance prices were high and commercial utilization was low where Medicare prices were relatively higher.

The wage index in a particular region is strongly correlated with prices for both types of coverage and is negatively correlated with quantity of services.  So the lowest Medicare spending is found in high wage areas and the higher Medicare spending is found in the low wage areas.  The suggestion is that lower wages allow for higher margins for hospitals and they are therefore motivated to see more patients.  It also appears that as private patients become less profitable to hospitals due to lower prices, they see fewer of them and more Medicare patients.  It is unclear what mechanisms hospitals might use to achieve this end.  The authors did find that in areas where hospitals have more market power, they appear to achieve higher private insurance margins, making those patients more attractive than Medicare beneficiaries.  It is also likely that some of the wage level effects are more generally related to the socioeconomic characteristics of a population in an area, which might affect relative utilization.