New Briefs on Concentration of Health Spending

We spend a lot of money on health care in this country–where does it all go?   For the care of a relatively small percent of the population is the answer revealed in the AHRQ Statistical Briefs (Nos. 448, 449 and 450) related to the concentration of health spending.   (Stat. Briefs)  This annual look suggests the complex nature of health spending:  it is concentrated on a few individuals, but not usually the same individuals from year-to-year.  The data are from the MEPS survey and reflect years up to 2012 for spending by individuals living in the community (so this excludes nursing home residents, for example).   If you rank the entire community-dwelling population by their health spending, in 2011 the top 1% accounted for 21.5%, with an average expense of $92,825, of the total and in 2012 they accounted for 22.7%, with an average expense of $97,956.  In both 2011 and 2012, the top 5% accounted for about 50% of total health spending, at an average of around $42,000.  In stark contrast, the bottom 50% of the population represented only 2.8% of all spending in 2011 and 2.7% in 2012, with an average spend of only $240 in 2011.  Let that sink in for a minute; it is simply astounding and suggests why forcing people to buy insurance in the form we make them buy it seems like a ludicrous “solution” to a non-problem.

Persistence of spending patterns is very important in thinking about managing high-cost patients, particularly because intensive care management costs money, and there has to be a return on that effort.  Unfortunately, there isn’t a high level of persistence in that top 1%.  For those who were in the top 1% in 2011, only 19.6% retained that rank in 2012.   There is more persistence in the top 5%; for those in this group in 2011, 35% were in the top 5% in 2012.  By the time you get to the top 10%, who represent 65% of all spending, at an average of around $28,000, persistence is 41.5%.  But note that for the 5% and 10% groupings, average spending is probably not large enough to make care management efforts have a significant return on the cost of management.  On the other hand, that bottom 50% in 2011 had a 75% likelihood of being in the bottom 50% in 2012.  Age was a predominant factor in being and staying in a higher spending subset of the population.  Women were more likely to be in higher spending groups as well, but this probably interacts with age.

There is a relationship between insurance status and spending, but cause and effect is difficult to discern.  While 15% of the population under age 65 was uninsured for all of 2012, those individuals represented 24% of people who stayed in the bottom half of spenders from 2011 to 2012.  Do individuals not have insurance because they have and anticipate low health expenses or do they not have health spending because they lack insurance?  Probably some of each.  To some extent, I always find this report disheartening.  It shows our health spending issues are concentrated across a few people (although 1% is 3.5 million persons), which should allow for intensive focus, but because of lack of persistence and some time lag in identifying high spenders, working on one year’s high spenders doesn’t necessarily help you in future years.  If you think about the diagnoses related to high spending, this makes sense.  Some are acute episodes, like a serious car accident.  Some are acute illnesses, like a high-risk pregnancy or a serious cancer, that also are unlikely to recur.  And a lot is for people with chronic illnesses who have a flare-up requiring inpatient and other expensive care, like a heart failure patient who ends up being hospitalized two or three times in a year.  And the predictability of these high-cost cases is not great.  So all the data analytics and predictive modeling in the world isn’t likely to move the needle much.