I am noodling around on some risk adjustment thoughts. These are not fully formed and I am putting them out here just to get some clarity imposed in the process of writing.
Risk adjustment is needed in guarantee issue, community rated plans. It can minimize the pay-off for cherry picking and risk dodging strategies. Without risk adjustment or some other mean of funding high cost care for sick individuals, no carrier will even think about looking to cover sick people. Instead they will devote incredible amounts of effort to avoid potentially sick people and only cover people who will never file a claim.
I want to play with two different risk adjustment models and shake through some incentive structures. The first is a revenue neutral model where carriers with proportionally lower coded as sick people transfer funds to insurers that have proportionally higher number of people coded as sick. A dollar that Insurer A gains means Insurer B loses a dollar. What are the incentives for the following two states: single carrier in a risk transfer region and multiple carriers in a risk transfer region?
In a revenue neutral system, a single carrier in a risk adjustment region has no incentive to care much about the coding practices of their providers. There is no other carrier that can give the sole carrier revenue or take revenue away. The solo carrier has already priced in the full risk of the population into their premiums. The only incentive for an insurer in this scenario to care about coding is if the coding is so bad it is missing clinically relevant information that could have saved money and improved patient outcomes. This is a weak constraint.
Now if there are multiple carriers in the rating region, the incentive changes. The carriers are locked in a zero sum competition. They are red queen racing. As soon as one carrier engages in aggressive coding maximization, every carrier has a strong incentive to engage in aggressive coding maximization. This behavior can be legal and legitimate. An individual with diabetes and nerve pain in the feet should be coded as diabetic with peripheral neuropathy which is a higher level of risk than just diabetes without complications. An individual with a left arm amputation should have that information in their chart every year. If there is no limit on the amount of revenue that changes hands, the incentive to maximize coded risk score is very strong. If there is a limited amount of premium that can move for risk adjustment, there is a limited incremental gain for risk adjustment optimization.
The other funding model is an externally funded widget payment model. There are two variants. The first is open ended like Medicare Advantage and the other is a fixed sum proportional system. Medicare advantage pays a fixed sum for each diagnosis category found on claims or retrospective chart review. Insurers are not competing for a fixed pot of money. As long as an incremental coder can bring in more incremental revenue than their costs, it makes sense to chase diagnoses very aggressively. Feed back is often given to doctors on how to maximize the coding revenue from a menu of medically similar coding choices. Doctors frequently receive incentives to code aggressively in a risk adjustment revenue optimization direction. There is a difference between aggresssive coding and fraudulent coding but the money and more importantly the direct attribution of money to a particular code does provide an incentive to cross the line between aggressive but medically defensible coding to fraud. Medicare Advantage upcoding drives billions of dollars in incremental costs. Every carrier has a strong incentive to push upcoding as it is free money.
The same logic of revenue neutral multi-insurer behavior applies if there is a capped pool of external money. Every new diagnosis brought in increases the relative risk score of a carrier which means more money for that carrier. This is where red queen racing works in society’s favor as aggressive coding practices will cancel each other out.
One of the side notes that I am fussing with is the impact of Medical Loss Ratio regulations. Current regulation requires Medicare Advantage plans to have an MLR of 85%. The MLR is claims expense plus quality improvement expenses divided by the sum of premium revenue and net risk adjustment revenue. Coding does not drive claims expense. Adding a Z-code to a claim does not increase the reimbursement value of a claim. Adding a Z-code to a claim will increase the risk adjustment revenue. The numerator stays constant while the denominator gets bigger which means the MLR incrementally gets smaller. The 85% MLR places a limit on the value of chasing additional diagnoses. Are we seeing any impact of MLR on risk adjustment revenue optimization behaviors?