Distributional impacts of different cost share methods

There is a really good set of discussions about plan design and benefit configuration going on. Andrew Sprung at Xpostfactoid has a good series of posts on Ambetter and donut hole design, Austin Frakt at Incidental Economist has a good overview on the trend in out of pocket costs and Health Affairs blog also looks at trends.

One of the things that I have not seen in this discussion is the distributional impacts of the different types of consumer cost sharing. This matters a lot as to what cost sharing types we should emphasize as different values will produce different “right” answers.

Let’s create four sample individuals:

No Use Nora — she uses her essential health benefit PCP and OBGyn appointment and gets a flu shot in October every year at work. Other than that, she has not used any medical services in the past three years.

Low Use Larry — He also uses his EHB, but he’ll go to the urgent care twice a year and he went to the orthopedist once this year as he tweaked his ankle hard. He had 5 physical therapy appointments and three cheap generic prescriptions.

Medium use Maura — She has well controlled Type 2 diabetes and her blood pressure is a bit higher than it should be. She’ll see her PCP for six sick visits per year and then a specialist once a quarter. Her precriptions are a combination of low cost generics and three brand name drugs.

High use Harry — The past two years have not been good for Harry. He is still fighting cancer and the combination of chemo drugs and being worn down has led to a series of opportunistic infections. Every morning he wakes up, goes to the bathroom and then swallows eleven pills. His Friday’s are his doctor days, and he’ll often talk with his case manager (a really nice nurse who visits on Tuesdays) to help him manage his complex conditions.

These four individuals are all in a Silver plan with no cost sharing assistance. The individuals in the entire risk pools are expected to cover roughly 30% of the expected pooled cost of care. The question of benefit design is who pays for that 30% and how is that paid for.

So let’s take each major component of cost sharing (deductible, co-pays, co-insurance) in isolation and see how costs are distributed to the four types of people in the pool.

Deductible only with a $3,000 individual deductible.

Nora pays nothing besides her premiums as she is using no deductible services.

Larry is paying the contract rate on his two urgent care visits ($125/visit) and the orthopedist visit ($250/visit). The 5 PT appointments ($100/visit) are also applied to the deductible. The three generics cost $50 total for the year. He’ll pay out $1050 towards his dedctible.

Maura has six PCP appointments ($100/visit) and four specialist visits ($250/visit) plus $200 in generic medicine expenses and $1,000 in brand prescription expenses. she runs up a $2,800 deductible.

Harry burns through his deductible by the end of the third week of the year. He owes $3,000.

Total amount applied to the deductible is $6,850.  This is the 30% pool amount that is not covered by insurance.

Co-pay only, $50 PCP, $75 specialist $10 generic drug, $75 brand drug

Nora again pays nothing beyond her premiums.

Larry has three specialist visits (ortho and 2 urgent cares), and 5 PT visits treated as PCP visits for benefit design purposes. The prescriptions cost him $30. His total out of pocket $505.

Maura has the six PCP visits and four specialist visits. 24 prescription months worth of generics leads to $240 in generic costs, and then another $1,000 in brand prescriptions (as co-pays can not be higher than contract rate, so they are treated like deductibles). This leads to her paying $1840 OOP.

Harry is needed to balance the rest of the risk pool’s contribution. He sees his PCP 15 times during the year for $750, and between his oncologist, gastrologist, neurologist and pain management specialist, he is seeing a specialist twice a month that leads to $1,800 in co-pays. Finally, his medications are mostly on brand and that leads to another $2,000 in co-pays.

Coinsurance Only at 40%

Nora pays nothing beyond her premiums.

Larry pays $420 for his utilization

Maura pays $1,120 for her medication and treatment.

The cheap portion of the risk pool is kicking in $1,540 of the $6,850 needed to cover the pool’s 30%. That means Harry is paying the rest.

His treatment costs $4,000 a month, so he’ll be paying $1,600 per month in co-insurance until April.

These are illustrative examples and if you play around with the numbers and assumptions, you’ll get slightly different results. The important thing here is the insight.

Deductible plans favor the sickest people as the low utilizers pay for almost all of their care via deductible cash. That means the proportion of the pool’s individual responsibility amount is borne by healthy people.

Co-pay only plans favor people who use highly concentrated cost services. A co-pay does not differentiate between a specialist visit with a contract expense of $200 and a specialist visit with a contract expense of $600. It is the same fee. So people who use very costly services but only rarely are best off. People who use a lot of fairly low costs services on a regular basis pay more proportionally.

Co-insurance only plans favor low cost utilizers. They are not paying full price via their deductible, and unlike co-pays, the individual cost per unit matters.  Finally, No Use Nora is extremely valuable to the insurance company and the rest of the pool as she is fully cross subsidizing everyone else for this time period no matter how her benefits are built.

Almost no insurance plans are 100% Deductible or 100% co-pay or 100% co-insurance, the benefit design will blend.

HHS is favoring high deductible, low co-insurance plans in their new 2017 proposed standard design. I think this is for two reasons. First, deductibles are simpler to explain. Secondly, it is a distributional issue. HHS is prioritizing minimization total actual out of pocket costs for the sickest individuals over other objectives.  I am morally fine with that decision. Other people can reasonably prioritize other optimization functions and get other ideal benefit configuration matrixes.

The other key insight of this exercise is that the non-covered actuarial value of a plan will get paid somehow by someone. The question is who pays and how much?  If the objection to a deductible level is the size of the level, the problem is not the deductible, the problem is the low actuarial value of the plan.