MedPAC Report on Geographic Medicare Spending Variation

Most of the early research on geographic variation in health care use and spending was done on the Medicare population and that data continues to be a primary source for studies in this area.  The Medicare Payment Advisory Commission issued a report to Congress on the topic a few years and at Congress’ request released an updated analysis in September.   (MedPAC Report)   It is important that these analyses take into account both utilization and total spending.  Prices do vary geographically even for Medicare, and that price variation isn’t as important as the pattern of service use, which should be similar across the country for beneficiaries with the same health status and health needs.  This analysis used data from 2013 and 2014 and used MSAs and non-MSA areas within the same state as the basis for analysis.  For both Part A and Part B, per capita monthly service use had less geographic variation than did per capita monthly spending.  45% of beneficiaries lived in service areas with utilization within 5% of the national average, but only 24% resided were monthly spending was within 5% of the national average.  The highest spending area had per capita spending 117% above the lowest spending one, while the area with the greatest service use was “only” 73% higher than the lowest one.  This indicates that price, even for Medicare, plays a substantial role in spending differences.

The differences in utilization and spending were slightly reduced from the earlier report, but still fairly significant and hard to understand in a world where physicians are constantly driven to provide evidence-based medicine and are subject to ongoing quality measurement and reporting.  It extended to drug use as well, with a 38% higher spending on drugs in the top areas compared to the low ones, and 21% variation in drug use between these two groups.  Since the analyses attempt to adjust for health status, you wonder how the same patients could be prescribed more drugs in one place than in another.  As has been the case in past analyses, post-acute care is the leading source of the service and spending variation.  Average monthly spending ranged from $63 in lowest use area to $357 in the highest.  Inpatient care had a much smaller range from low to high, with monthly averages of $254 and $378, respectively.  In general, service use level by subcategory tended to be correlated, not substitutive.  In other words, if an area ranked higher for inpatient use, it likely also did for ambulatory and post-acute care, and vice versa.  Service use was almost identical in major urban and rural areas and geographic variation occurred in the same proportion for each type of area.  And within a state, multiple urban area will show significant variation, suggesting that the factors underpinning spending variation don’t necessarily occur at the state level, but perhaps more locally.  Interestingly, poster children for high spending Miami and McAllen, Texas, had reduced levels of use in this study, although still above the national average.