Understanding Trends in Health Spending

I am fond of papers from the National Bureau of Economic Research, but they take a fair amount of time to read and digest, so I have several stacked up.  This week is an attempt to clear the backlog.  The first paper addresses health spending analysis.

You would think something as large as national health spending in the United States would have well-understood dynamics.  But economists and others are still working to define what the sources of that spending are and how and why it changes over time.  The Bureau of Economic Analysis at the Commerce Department has been laboring for several years to come up with a better decomposition method for trends in health spending, since that group is responsible for preparing most of the national economic statistics.  A new paper at the NBER presents their latest work.   (NBER Paper)  Unfortunately, they apply their analytic framework to quite old data, looking at the period from 2003 to 2007, but they do look at the commercial market, which still covers most Americans.  One fundamental precept for their work is to try to understand spending for diseases, rather that according to the usual division by service category–inpatient, outpatient, drugs, etc.  Over this period per commercially-insured person spending grew by 26%, compared to the 20% increase in nominal GDP over the same time.

Changes in health spending can relate to changes in service prices and/or changes in the amount of services used.  When looked at on a disease basis, changes in disease prevalence and in how diseases are treated, particularly technological “advances” can play a role as well.  These researchers think the best way to understand spending changes is to analyze disease prevalence and the cost of treating a disease over time, with that cost broken down into unit prices and prevalence factors.  They also factor in demographic changes which affect spending, primarily the aging of the population.  First they identify that demographic change-caused spending growth; then they track disease-level spending, finally breaking that down to prevalence and service cost.  The results are informative.  During the study period, the demographic shift accounted for 18% of spending growth.  27% was due to real (inflation-adjusted) price increases.  But 60% was attributable to greater disease prevalence (or at least greater treatment of disease, until people seek treatment you don’t know how much disease there is) and none was due to greater service utilization.  (This again supports my skepticism about large claims of savings from “wasted” health care spending.)

The fastest growing large medical categories were orthopedics, gastroenterology and endocrinology, which were 33% of spending in 2003 but 40% by 2007.  Cardiology is interesting, because while it was 12% of spending in 2003, it accounted for less than 8% of spending growth in the study period.  There was both a decline in disease prevalence and a lowering in service use.  This appears to be driven by a shift from brand to generic drugs and from inpatient to outpatient treatment.  As with some other medical categories, the shift to prevention by early treatment and prevention, such as use of lipid-lowering agents, avoidance of diabetes and treatment of hypertension appears to mean fewer heart attacks and other acute cardiology issues.  Cancer treatment costs rose twice as fast as the overall spending during the study period, largely due to a rise in service prices.  Think of all the new expensive cancer drugs and radiation approaches.  This paper is very informative and well worth a read.