The three R’s of Obamacare are Reinsurance, Risk Corridors and Risk Adjustment. These three steps are attempts to stabilize the individual on-Exchange marketplace by changing the incentive structure of insurance companies. Reinsurance and Risk Corrdidors are short term plans while Risk Adjustment is a long term program. We’ll talk a lot about Risk Adjustment today, but let’s go over the other two first.
Reinsurance is a back-end payment made by the federal government to insurers that have “shock” or “catastrophic” claims. For 2014, reinsurance was in play for claims between $45,000 and $250,000. The federal government would pick up 80% of the costs of these claims for a maximum federal exposure of $164,000. The average pay-out will be significantly less. Reinsurance is financed by a $63 per member year fee for people with major medical insurance. The goal of reinsurance is to remove some of the tail risk of big claims. 2015 will have a $44 reinsurance fee. 2016 has not been announced yet.
The goal of risk corridors is to create a three year window of forgiving rate estimation for insurers if one is not a cynical bastard. If one is a cynical bastard, the risk corridors create a three year window of loss leader membership building. In either case, it provides a transitional time frame for insurers to figure out what the Exchange population looks like while segmenting the market. Pricing for a given actuarially value, network scope and buggering thy neighbors in races to the bottom on sick people avoidance schemes, should converge into tight clusters. During that time, the Federal government takes on some of the downside of losses based on too optimistic actuarial assumptions while also taking on the upside of too pessimistic actuarial assumptions. If a company’s medical expenses are more than 3% but less than 8% (a 5% range) greater than expected, the Feds kicked in a 50% payment of the gap (up to 2.5%). Losses above 8% saw the Feds pick up 80% of the excessive loss. The converse is true on gains. The Feds would get a slice of the gains.
Federal reinsurance and PPACA risk corridors are temporary programs. Risk adjustment is a long term program.
The basic goal of risk adjustment is to discourage insurance companies from cherry picking only healthy members. Risk adjustment uses claims history at the individual level to assess the projected medical risk (general health) of an insured population. Each insurer receives their own aggregate per member risk score, and then money is moved from low risk/healthy groups to high risk/sick groups to compensate the sicker groups for their increased risk of medical expenses in the upcoming year. Risk adjustment has been a feature of American healthcare for decades, and it is especially notable in Medicare Advantage as well as Medicaid managed care.
So how does this happen?
There are a couple of major risk adjustment models out there. The one that I’ve used has nineteen categories of chronic conditions with up to five variations within each category. The model produces an estimated risk factor for a dozen demographic groups and two major payer groups, so we’re looking at 200 independent potential risk scores. An individual’s claim and diagnosis history is examined and their history buckets people into different groups. For instance, someone with a heart transplant would be expected to have very high claims compared to the average person in their demographic group. This person could have a Cardiology risk factor of 5.52 which would mean, that all else being equal, that person would have 552% more claims just for Cardiology than their typical peer. Another person could have an irregular heartbeat, which is an elevated but minor risk factor, so their Cardiology risk score could be 1.07. Each condition category is then added up, so a total risk score would be Cardiology plus Endocrinology plus Psychiatry plus a bunch of other buckets.
This per member individual risk score is than averaged across the entire unit of analysis. This is a raw risk. If the risk adjustment program is revenue neutral, this is only first stage. Every other insurer in the region of analysis is doing the same thing. Mayhew Insurance may have an unadjusted risk score of 1.08, Big Blue might have an unadjusted risk score of 1.05 and Those Other Guys might have a risk score of 1.12. Those Other Guys, assuming everyone is doing honest data work, have the sickest population. The scores will be enrollment adjusted and then normalized at 1.0 for the entire population. Relative risk scores are then handed out. Mayhew Insurance would get a score of 1.027, Those Other Guys could see a score of 1.031 and Big Blue could have a score of .975. This means Mayhew and Those Other Guys have sicker than average populations and Big Blue has a slightly healthier population in the area of interest and analysis. Big Blue would kick some money into a pot, and the other two insurers would get a big check from that shared pool of money.
Assuming perfect and complete data, this is a good system. However claims data seldom is as complete as we want it to be for this. Most providers will bill sufficient data to get paid. That means providers will bill what they did but not always all of the non-presenting conditions. For instance, someone with a heart condition who goes to their PCP for their annual might not have their heart condition noted on the claim, so their risk score is too low. If they did not go to their cardiologist in the time period being examined, the insurer would not have a claim for the disease. This produces an incentive for systemic waste.
Medicare Advantage plans aggressively chase after providers to bill every diagnosis (they get paid for the procedure, not the diagnosis) in order to improve their relative risk scores against insurers who don’t chase after their providers. What happens is significant resources are devoted to increase the appearance of illness in the claims history without actually doing a damn thing to make a patient better off. Medicare Advantage members are coded as 10% to 15% sicker than Fee for Service members. That entire gap is due to chasing down documentation. The problem is insurance companies that elect not to chase their providers for higher diagnosis coding intensity will see their risk scores look comparatively lower compared to high intensity coding companies, and thus they’ll lose money. So everyone has to do it, or no one has to do it. Right now we are in a stable, socially negative equilibrium of high intensity claim coding for risk score optimization.
The PPACA risk adjustment models will suffer from the same problem. It will be partially compensated by the medical loss ratio restrictions of non-claims spending of 15% or 20% of revenue, but it is a source of waste. If we wanted to eliminate this minor source of waste, the options are either single (regional) payer, totally random locked in assignment or going to a non-risk adjusted system which would lead to every insurance company even more actively screwing the sick over. Since I don’t think single payer is on the horizon, and random locked in assignment destroys the competitive nature of the Exchanges, keeping risk adjustment with the acceptance that there is some pointless Red Queen racing going on in the background is far better than racing to the bottom.