Skip to main content

Commercial and Medicare Spending Variation

By December 22, 2011Commentary

Geographic variation in health spending for Medicare beneficiaries is often cited as proof of an inefficient and wasteful system, but much of the research is based on correlation, not clear cause and effect, and it often does not fully adjust for a variety of factors such as health status or sociodemography.  Newer research has begun to focus on variation in spending on commercially insured members, to see if it has a similar pattern.  A study published in the American Journal of Managed Care continues that research thread.   (AJMC Article)   The researchers divided Texas into 32 geographic areas and compared Medicare spending and hospital utilization with that for Blue Cross of Texas members.  Due to data limitations, the researchers were not able to do extensive adjustment for health status or sociodemographic factors.

Average spending for Medicare was $850 per month per person and it was $258 for a commercially covered person.  In most of the geographic regions, 22 out of 32, the variation in spending went in the same direction for the two groups.  The variation in each program across the state was relatively narrow, about .11 for Medicare, unadjusted for McAllen/El Paso, and .09 for the commercial group.   The correlation in spending variation is low and negative including all regions.  Consistent with an earlier study by these researchers, the area of McAllen/El Paso, made famous by Atul Gawande, is an outlier in terms of Medicare spending, which may largely be due to home health care spending.  McAllen has high Medicare spending, El Paso, next door, does not and there is little commercial spending difference between the two.  When the area is excluded from the analysis, the correlation between Medicare and commercial spending rises significantly to .54.  Because most Medicare spending has the same price across the country, most of the spending variation in that program is likely due to utilization.  In the commercial world, at least some of the variation could be due to price differences, but the authors did not have data to analyze that.  Really comprehensive analysis of spending variation awaits a really good database of all payers and needs to be conducted not just at a geographic but at an individual physician level.

Leave a comment