HC Statistics

Estimating Price increases for Medicare using Bundling

One question that plagues health economists is estimating the price inflation of medical care.  The output measured in these indices should be “health,” but this is often difficult to measure. Typically, economists simply look at the price increase of different inputs (e.g., the cost of a doctor’s visit, the cost of pharmaceuticals, the cost of different procedures)

In a novel approach, Aizcorbe and Nestoriak (2008) use cost of a bundle of treatments for a given disease as the unit of measurement.  This way, the authors can track how the cost to treat a specific disease changes over time. Even if the input costs for treating a disease increase over time, the mix of inputs used to treat a disease may also shift.  Further, if a new treatment arises, this will be included in the treatment bundles, eliminating the “new goods problem.”  The authors find the following results:

” While the price of treating diseases grew an average of 12% over this period [2003-2005], costs would have risen even faster, 17%, if the mix of treatments in 2005 had been the same as that in 2003…For example, in the treatment of depression, there  has been a shift away from talk-therapy and towards (the lower cost) drug therapy that has reduced the cost of treating depression…[also, there have been] shifts from care at hospitals towards care at ambulatory surgical centers for orthopedic and gastroenterological conditions…

In order for this estimation to be valid, a few things must hold.  First, disease severity must be constant over time.   Second, treatment outcomes must be constant over time.  If increase spending leads to a better outcome, this may not be price inflation, but simply a better treatment bundle.  Bundling will not correct this problem.  Third, the period of observation is only over 3 years and the data lacks information on Medicare claims.  Nevertheless, the bundling of treatments provides an attractive option to constructing medical care price indices.

Leave a Reply

Your email address will not be published. Required fields are marked *