There has been an idea floating in the background of health policy to allow for insurers to underwrite policies against medical history but subsidize people based on their medical history.
MVP: If insurers are permitted to risk rate, the premiums for the healthy will fall and more healthy people will voluntarily buy insurance – let’s agree that’s a healthy outcome. The premiums for the high-risk will surely rise. My proposal– let the insurers risk rate but let us subsidize the premiums for the high-risk.
It's not important that when your owned and operated urgent care provider codes a risk adjusting DX that it actually makes it to the claim!
— rebeccastob (@rebeccastob) April 5, 2017
I feel her pain. I want to go through how a visit becomes a risk adjustment score. I’m not going to do a full Six-Sigma analysis but the risk adjustment data system leaks as much as a toddler’s diaper after their first time being watched by their young, inept uncle.
Let’s get started with a simple office visit. There, the patient, Mary sees Dr. Smith for her annual well visit. Dr. Smith listens to Mary’s lungs and heart, reviews her medications and listens to Mary’s verbal symptom history. Dr. Smith reconfirms a diagnosis of asthma and hypertension. Dr Smith also notes a few acute, non-chronic low level conditions. Mary goes home eleven minutes later. Dr. Smith then enters her encounter notes into the electronic medical record and clicks on too many buttons before Dr. Smith sees the next patient.
Asthma and hypertension are both risk adjustable conditions. So that’s it, right?
Not even close.
A biller, either a person or an algorithm matches the encounter notes with the appropriate or at least defensible procedure codes and diagnoses. From there, a claim is built. Large practices might build their claims in house. Smaller practices might send their claims to a clearinghouse where a lot more information is added. Some clearinghouses can accommodate four diagnoses on a claim, others can accommodate twenty five diagnoses on a claim. Some clearinghouses will run edits on their claims to clean them up.
A claim is then submitted to the insurer. That claim can arrive hours after the patient leaves the office or six months after the patient leaves the office. The claim payment system receives the claim and begins to process it. More edits will be applied to tweak the data, it will be reviewed automatically for odd combinations. If it is a big claim, there is a manual review. Odd cases will be held for adjustment where every field on the claim detail lines can change and most of the claim header lines can change as well. Finally, the claim gets paid.
The provider is happy now. But the risk adjustment process is barely half way done.
At the end of the data collection period, all claims in the insurer’s claims warehouse for a given product and period are extracted. They are then re-organized to toss extraneous claims, remove duplicates and clean up retrospective adjustments. This huge claims file is then sent to a central risk adjustment processing center. By now, Mary’s claim could be anywhere from three months old to fifteen months old depending on timing and program design. The central processor performs their black magic on all of their data and spits out a risk score. money is then distributed.
So where are the failure points?
1) The doctor never identifies a clinically relevant diagnosis.
2) The biller does not enter a clinically relevant diagnosis that the clinician identified.
3) The claims clearinghouse does not receive the diagnosis code.
4) The claims clearinghouse does not send to the insurer the appropriate diagnosis code.
5) The insurer does not receive the appropriate diagnosis code from the clearinghouse
6) Insurer does not put diagnosis code into the encounter file to send to the risk adjustment entity.
7) Risk adjustment entity does not receive sent diagnosis code.
There are other failure points concerning eligibility and a few other things but this is a basic road map of common failure points in the risk adjustment process.
Rebecca’s tweet was about failure #5, the insurer not receiving valid codes that were sent to it by a clearinghouse. It brought back memories.
When I worked on risk adjustment revenue optimization, I saw all of these failure scenarios. I was dealing with membership pools with hundreds of thousands of members in a reporting period. Minor errors probably washed out in the end. But if it did not, my paycheck and my insurance premium were not dependent on 100% perfect and timely data. I just had to assist in minimizing the gap between the actual medical status of our membership and the reported for risk adjustment purposes medical status of our membership.
This mostly works well enough often enough for a multi-billion dollar entity with hundreds of thousands of covered lives. If a claim with a $50,000 risk adjustment potential falls through the cracks, that is not good but it is acceptable as at some point the chase cost outweighs the recovery value.
However, if we are in a world of individualized risk adjustment where the individual is responsible to maximize their risk score as they know that their ability to pay for insurance is directly related to their underwritten medical risk, any failure is unacceptable. And how will an individual be able to control or accurately predict their risk score when the current process flow has scoring lagging the actual visit by three to fifteen months and there are multiple administrative failure points that are opaque to the patient and out of their control to correct. This is worse than correcting a credit report when your name is David Anderson.
If there is a way to costlessly verify and update all risk scores every night, then this could start making sense on a practical basis but that would require federal legislation and a five year process rebuild of the entire medical sector for a very small portion of the insured population.