We were fortunate to speak with pop health expert, Kate Casaday, Population Health & Value-Based Lead at CareMount Medical, about best practices and insights on getting started with managing risk-based contracts.
Kate holds a Master’s in Public Health at Columbia University’s Mailman School of Public Health and has direct experience working for a Medicare Advantage Plan. At CareMount Medical, which is the largest independent multi-specialty group in New York State, Kate has many responsibilities; including physician contracting and developing strategies and innovations for CareMount’s Next Generation accountable care organization (ACO). CareMount is 1 of the 41 Next Generation ACOs in the country and received a perfect score on their Initial Readiness Review (IRR) from the CMS.
Question 1: If you were just getting started with a Medicare Shared Savings Program (MSSP) where would you start?
HCC coding. HCC coding is your best place to start. You need to make sure that the premium revenue that’s attributed to your assigned beneficiaries is in-line with their expenses. A risk adjustment methodology is basically a way for CMS to say, “I can’t take all of the Medicare premium and divide it equally among all Medicare beneficiaries, I have to adjust for differences in both demographics and clinical conditions.” Using the clinical component of HCC risk adjustment, you can get an accurate disease prevalence and burden view of your population and risk score and actual premium revenue.
Question 2: If CareMount was only focused on Medicare Advantage and not part of a Next Generation ACO would you set up your operations differently?
No, not at all, they both use the same risk adjustment methodology. The only difference is that with Medicare Advantage plans we get a little extra help on the retrospective submission side. With Medicare Advantage, we can use their RAPS resubmission and feed them information that’s been verified by both our coders and our physicians that we may have missed on a first pass. With the ACO we must resubmit a clerical error reopening request should we need to edit diagnostic information. So, it’s a little bit easier with Medicare Advantage but the concept and the process aside from that is identical.
Question 3: CMS has announced several new value-based models recently, is there one that is more interesting?
I’m really interested in direct contracting. I like any model that puts the primary care provider at the center and allows the primary care provider to function as sort of a general contract holder with ideally hospital and post-acute care settings downstream from that. I see the referral base to be a powerful tool to control the cost of health care.
Question 4: What has CareMount found to be most effective in getting the providers on board with appropriate HCC coding and documentation?
Firstly, we’ve done a few things to make it easier for physicians. We’ve integrated a pick list of historical HCC codes from our integrated data sets directly with our EHR at the point of care, so physicians can quickly scan a curated list of HCC codes to gain an overall picture of that patient. Because, healthcare data is filled with false positives we allow them to tell our analysts if there are false positive or we allow them to directly bill off this list, so, that’s a big tool.
Secondly, we’ve developed an incentive program. I’m a big believer people do what you incentivize them to do. Our value-based initiative incorporates quality, HCC coding, annual wellness visit completion and cost of care initiative into an incentive for both our primary care and some critical specialists. This goes a long way to really demonstrate organizational commitment to this idea and offset some of what physicians feel could be a little bit more tedious or burdensome to their initial ICD-10 coding kind of behaviors.
Question 5: Are you able to accurately calculate individual patient risk scores on a running basis in the current year?
Yes, with a program called SAS (Statistical Analysis Software). CMS releases the HCC models each year in SAS as SAS code, and gives you around 6,000 lines of code that you can replicate how CMS will calculate the risk score. It includes hierarchy and all conditions that apply. Because you get the claims feed on a monthly basis you can see month over month how your risk scores are trending. If you were to map this out across the year your risk scores in the beginning of the year will go down as your sicker patients typically die first. You will see a bit of trending and it takes some experience to cut through the statistical noise if your interventions are working. We use a lot of proxy metrics because actual risk scores depend more on your decedent rate and factors that are a little bit independent of education. We use risk recapture to measure the effectiveness of our physician communication and outreach programs.
