If I want my plan to fail (pt.2):

“I won’t question nationally published cost statistics and wonder if these “industry averages” apply to my member population.  When I read an article stating an average family of four incurs health claims in excess of $20,000, my mind closes”.  http://www.marketplace.org/topics/your-money/makin-money/family-health-care-costs-breach-20000

What does “family of four” typically denote? Most would think, mom, dad and two kids? Look outside of your house, condo/apartment and notice a family on a stroll. How many think the average family walking by incurs $20K per year in claims? So, how could Milliman publish such a high average, or high “Milliman Medical Index, MMI”?  

From a pure mathematical perspective I’d never question the validity of their result. They took millions of values (family claims) came to a total and then divided by the number of values to obtain the average or mean. We should be asking, “is this average a reasonable representation of the “central tendency” or essence of the data”? Another way to think of the Milliman average is what does this number/average tell us about health care risk?

When thinking of families of four, I find it hard to comprehend the average claims expense is in excess of $20 thousand dollars, annually? Could both be correct? The answer is yes, however I hope to convince the audience the Milliman average is unrepresentative of the “central tendency” of the values (family of 4 health claims) as this average is obviously calculated including huge claim values (high) outliers.  Here’s a very clear video explaining the manner outliers change our perception of data sets.  http://www.statisticslectures.com/topics/outliereffects/#video

Huge average claim values tend to “breed” stagnation concerning intelligent plan management actions!

When looking at “average per member per year plan cost” why let consultants define average in a different manner than the financing of the plan? Most plans re-insure specific member claims to a “specific” deductible amount and receive reimbursements for amounts over this deductible. However, when computing the annual per member/employee cost, these amounts reimbursed are included grossly inflating the average cost! In fact the top 1% of members, by claims cost, incur approximately 30% of all plan assets. Removing these as outliers would provide managers a more accurate “central tendency” of their claims data.