Using Financial Benchmarking to Build a Successful Foundation for APM Performance

Historical Expenditure

The next input required to build a benchmark is your aligned population’s historical expenditure. In a Total Cost of Care (TCoC) model like MSSP or ACO REACH, primary care and specialty care expenditure will be used to calculate the historical benchmark.

Having access to 100% of Medicare Part A and Part B claims for your aligned population will be critical in ensuring you have an accurate calculation of historical expenditure.

We’ll couple this with risk adjustment in the next step to get a risk-adjusted expenditure for your population so we can make appropriate comparisons of expenditure between populations.

The example below demonstrates TCoC for the ACO’s aligned population during their benchmark years. For MSSP ACOs, the per capita expenditures are calculated for the four Medicare eligibility categories (Aged/dual, Aged/Non-Dual, Disabled, End Stage Renal Disease).

Risk Adjustment

Risk adjustment is one of the most critical steps to success in value-based care. Risk adjustment is based on the central concept that more medically complex patients should have higher expenditure than a less medically complex patient. A risk adjustment factor (RAF) score uses demographic and disease factors to normalize spend based on the patient or population’s relative risk.

CMS alternative payment models use the Hierarchical Condition Categories (HCC) risk model. The HCC model follows prospective attribution of chronic conditions, where diagnoses from the previous year are factored into the patient’s HCC score for the current year. Each condition is assigned a different weight, with certain interacting conditions being weighted more heavily. In the example below, we can interpret this Medicare beneficiary’s 1.323 HCC score to mean that due to their comorbidities and demographic factors, this patient is expected to cost 1.323 more than the average beneficiary in the performance year.

Different HCC models use a different lookup table to assign weights to each diagnosis – for example MSSP uses a blend of v.22 and v.24, while ACO REACH uses only v.24. In 2024, MA plans will begin incorporating HCC v.28, given historical precedence we can assume v.28 logic will begin to be incorporated into other CMS models in the future.

The first three inputs (Alignment, Historical Expenditure, and Risk Adjustment) will be combined to come up with a risk-adjusted, per-beneficiary per-year (PBPY) expenditure for your aligned population. A few adjusters will also be applied:

• Inflation adjustment
• Geographic Adjustment Factor (GAF) to account for geographic variations in healthcare expenditure

CareJourney is hosting a webinar Tuesday, August 8 at 1pm ET to discuss a beneficiary cohort comparison approach to support accurate comorbidity coding and documentation. Register here.

Regional Expenditure

The final core component to generating a benchmark is the regional expenditure, or the average expenditure of the Medicare FFS cohort living in the same geography as your population. The definition of the region will vary by payment model. In calculating the final benchmark for your population, the risk-adjusted PBPY expenditure for your aligned population will be blended with the risk-adjusted PBPY regional expenditure. This blend benefits participants who are already providing more efficient care compared to the regional average.

In order to evaluate an APM, an organization should compare their claims-aligned PBPY expenditure against the regional PBPY expenditure. If their historical experience is more efficient than the region, they have a “tailwind” – all else being equal in the performance year, they are likely to earn savings. If their historical expense is higher than the region , they likely have a “headwind” and may need to change their current care models to head in the direction of savings. Some CareJourney members seek out providers with a headwind as it is a good indicator of an opportunity for improvement.

Our example ACO has a risk-adjusted historical PMPY of $11,251 once the MSSP weighted blend is applied to the three benchmark years. The region’s risk-adjusted historical PMPY is $10,859.

Due to this headwind, a conservative organization may elect not to participate in MSSP with this group of providers. An organization that is bullish on its ability to reduce unnecessary expenditures may find this benchmark full of opportunity. Their next step would be to further drill into their historic performance to identify areas of improvement. They may also consider participating in an upside only risk track to start.

Putting It All Together

The sections above describe the four core elements of an APM benchmark: 1) aligned population 2) historical expenditure 3) risk adjustment and 4) regional expenditure. In order to calculate the regionally-adjusted historical benchmark, the ACO’s claims aligned historical expenditures are blended with regional expenditures. Blend ratios are dependent on APM, performance year and efficiency relative to region.

For an MSSP ACO in their first performance year, they would be subject to the following blends, depending on their relative efficiency.

As a reminder, our example ACO has a headwind. In other words, they are regionally inefficient. Their claims-aligned risk-adjusted historical PMPY ($11,251) will account for 85% of their benchmark in PY1 and the region’s risk-adjusted historical PMPY ($10,859) will account for 15%. Their regionally adjusted historical benchmark will be $11,193. As this MSSP ACO continues through their agreement period, the region’s performance will progressively have a greater influence on their benchmark. It will be a 50/50 split in PY4, encouraging the ACO to achieve efficiencies to “catch up” to the rest of their region.

Next Steps

Now that we have our final Benchmark for our model(s) of interest, we can use this value to inform our critical participation decisions, including:

• Which provider(s) (for ACO REACH) or provider groups (for MSSP and ACO REACH) should I go into the model with?
• How can I optimize my roster for this performance year?
• What are my projected savings for a given performance year?
• Should my organization participate in MSSP or ACO REACH in future performance years?
• Which model track is best for my organization in future performance years?
• How should I structure sub-capitation or other gainsharing agreements?

Conclusion

Accurate financial benchmarks are critical in evaluating the potential impact and feasibility of joining or taking on more risk in a VBC model. Having a robust historical claims experience ensures that your financial benchmarking predictions are accurate and actionable, steering your organization to the model that is the best fit for your patients and providers.

How CareJourney Can Help

No matter where you are on in your value-based care journey, there are several ways in which CareJourney can support your organization, including:

• Calculating financial benchmarks for your participating providers to inform your model go/no-go decision and model track selection
• Simulating success with a given roster of providers, including evaluating new groups, to optimize your model participant list
• Understanding utilization of services from an attributed population and cultivating strong preferred provider and facility networks.
• Understanding provider performance and providing the insights needed for performance management conversations.
• Leveraging CareJourney claims ingestion and transformation Population Insights tool to track care interventions in real time.

Many CareJourney members are top-performing ACOs, including Rush Health, PBACO, Privia, Silver State, DVACO, and South Texas ACO Clinical Partners. If you are currently a member and interested in how CareJourney can help you gain insights into designing and managing a value-based organization, please reach out to your Member Services representative for more information. If you are not a CareJourney member, please request a meeting with our APM expert or email us at info@carejourney.com.