IOM’s Final Report of Geographic Spending Variation

Following up on its interim report earlier this spring, the Institute of Medicine has released its final report on geographic health spending variation.  (IOM Report)   Drawing on a review of existing research as well as fresh analyses using large databases, the committee charged with examining the issue finds that while geographic variation clearly exists, the pattern is different across benefit plan types and even within geographic regions.  Looking not just at FFS Medicare, but Medicare Advantage and commercial insurance, the committee found that geographic variation in commercial insurance, for example, is driven largely by differences in the prices providers charge, not in utilization differences.  In Medicare, on the other hand, much geographic variation is unexplained by any analytic factor, including health status and market level factors.  While total spending variation is not well-correlated between Medicare and commercial populations, utilization is fairly correlated.  As units of geography get smaller, variation continues to exist between units and within units, which is consistent with the fact that medical decision-making basically occurs at the individual provider and patient level.  And quality is not substantially correlated with spending, high or low.  Based on its research, the committee recommended first, of course, that CMS should continue to organize large, all-payer databases to continue to study variation.  The committee opposed the suggestion that Medicare in essence punish high-spending regions by lowering their reimbursements, noting that individual providers who are low cost in a region would be unfairly punished by this approach.  The committee further suggested continued use and testing of new payment methods that might enhance use of appropriate care while improving care coordination and quality outcomes.  Basically, the committee is saying that individual provider practice patterns are where the issue lies and addressing those practice patterns is much harder than attempting regional solutions.