Association for Public Policy Analysis & Management

Research Objective:  To better understand spending growth in a commercial population from an episode perspective. 

Study Design: We run commercial claims through the ETG episode grouper, allocating spending in any given year to one of 514 episodes.  We adjust spending for demographic changes and then decompose spending growth by episode, identify the episodes most responsible for overall growth in spending.  We further decompose spending growth in to that attributable to growth in the number of episodes and that attributable to increases in the spending per case.  Moreover, we assess the skewness in spending growth by episode (what share of episodes account for 50 or 90% of spending growth).  Finally, we divide the sample into two sub periods:  2002 – 2007 and 2007 – 2011 and assess how stable the drivers of spending growth were over time.

Population Studied:  Commercially insured individuals using the HCCI data base between 2002 and 2011.

Principal Findings: Over the entire period, and particularly during the 2007-2011 period, spending growth was driven by spending per episode as opposed to the number of episodes per capita (which were falling).  Specifically, spending per episode rose almost 3% per year during 2007-2011 while the number of episodes fell by over 2% per year.   Spending growth was driven by a relatively small number of episodes.  Over the entire period, 50% of spending growth was accounted for by roughly 5% of episodes and 90% of spending growth accounted for by less than 25% of episodes.  From 2007 - 2011 spending growth was even more skewed, with about 15% of episodes accounting for 90% of spending growth.   Persistence in spending patterns across episodes was small.  Correlation in spending growth among episodes between the two periods was .08.   Correlation in growth in cost per episode was 0.14 and correlation in growth in the number of episodes per capita was 0.55.  Over the whole period, the episodes that contributed most to spending growth (and the share of spending growth they account for) were: Breast cancer (3.4%), Multiple sclerosis (3.4%) , pregnancy (3.2%), diabetes (3%), and joint degeneration (2.8%).  Yet the contribution of each of these changes over time.  For example, breast cancer, which was the most significant contributor to spending growth overall and for the 2002-2007 period, is not in the top 10 episodes for 2007-2011. 

Conclusions: Health care spending growth is driven by a small share of episodes which change over time.  This emphasized the role of technical progress and idiosyncratic factors in driving spending growth.  While sure there are common components, it is important to recognize that clinical factors are important

Implications for Policy: Updating payment rates in episode based payment models will be complicated by the skewness of spending growth by episode.  Balancing the need for innovation and spending constraint will be crucial to creating a sustainable health care system.