Impact of the 2018 National Assignable Trend Factor on ACO Financials

Executive Summary

  • Most ACOs do not use a sufficiently accurate national assignable trend factor. While it often gets lost among the many inputs to the shared savings calculation, not being aware of its implications could result in unpredictable swings in shared savings estimates.
  • It is not sufficient to use the national trend estimation methodology recommended by CMS. Analysis of the past 3 years demonstrates that doing so would have consistently overestimated shared savings, swinging the benchmark as much as 2.2%. (In the example of an average size ACO, this swing would overestimate shared savings by $1,500,000.)
  • The 2018 national trend is surprisingly high, with growth in the combined trend being about 1% higher than 2017. This increase:
    • Could signal a “bull market” for ACOs to improve their financial position compared to unmanaged FFS in their regions, providing an opportunity to expand networks and grow assigned beneficiaries
    • Is particularly noticeable with ESRD enrollment type, so proactively navigating the new payment models affecting this population is becoming more important
  • Even after recognizing the impact of national trend on shared savings, it is important to recognize that it’s still one of many inputs impacting shared savings. ACOs can improve their financials and mitigate risk exposure by proactively managing these inputs through multi-year scenario simulations.

The Findings

Validate Health has aggregated mid-year benchmark files recently released by CMS in order to derive an estimated 2018 national assignable trend factor. These estimates can be used by ACOs to improve financial projection accuracy for 2018 final settlement prior to its release.

Percentage Increase Year over Year by Enrollment Type

Enrollment Type 2012 2013 2014 2015 2016 2017 2018
Combined -0.2% 0.8% 2.0% 2.9% 1.4% 2.9% 3.8%
Aged/Non Dual -0.2% 1.1% 1.9% 3.2% 1.4% 3.1% 3.6%
Disabled 1.5% 0.1% 3.9% 2.8% 2.4% 3.3% 4.1%
Aged/Dual 1.8% 0.5% 1.5% 3.2% 0.4% 2.5% 3.8%
ESRD 0.2% 0.3% 1.2% -0.2% 1.2% 0.5% 5.0%

These results for 2018 are especially surprising for two reasons. First, the 2018 trend of 3.8% is record-breaking, being the highest for of years since MSSP program inception. Second, the previously estimated 4.3% trend, based on CMS’s suggested estimation methodology, is overly optimistic by 0.5%.

This difference is significant enough to turn an expected shared savings into a loss. (See section The Danger of Relying on CMS Estimates.)

The trend has increased steadily across all enrollment types since the MSSP program inception. ESRD expenditures has a disproportionately large jump among all enrollment types. So the Kidney Care First (KCF) and Comprehensive Kidney Care Contracting (CKCC) payment changes are now particularly relevant. They are an opportunity for ACOs to reduce expenditures below benchmark and capture shared savings for this population over the coming years.

Although the cumulative growth in the disabled population outpaced all other enrollment types, this impact is mitigated by the fact that typically these beneficiaries account for only 12% of the total assigned population.

Disability-focused initiatives should be considered to control this fastest trending sub-population.

   

Real-World Example

  • A hypothetical ACO can be constructed to demonstrate the impact of relying on an inaccurate national trend. This example constructs an ACO that:
  • Is a 2018 starter with 2017 benchmark year 3 (BY3)
  • Has an average enrollment type mix of 80%, 12%, 7%, 1% for aged non-dual, disabled, aged dual, ESRD, respectively
  • Experiences a 2% expenditure increase for its beneficiaries from BY3 2017 to PY 2018
  • Has no changes in risk scores from BY3 2017 to PY 2018
  • Has the same historical benchmark rates as the national assignable FFS rates in the 2017 EXPU
  • Has a median total assigned beneficiaries of 12,156, as identified in the 2017 PUF
  • Specified a 2% minimum savings rate (MSR)

Suppose this ACO overestimated the national trend by 0.5% for 2018. They would have overestimated their gross benchmark by about $680,000. This difference is enough to exceed their 2% MSR, swinging their expected net shared savings to about $1,500,000. But in reality, their final settlement would result in no net savings realized.

Estimate Actual
Historical Benchmark (PBPY) $11.180 $11.180
Growth Increment 4.3% 3.8%
Updated Benchmark (PBPY) $11.658 $11.601
Expenditure (PBPY) $11.404 $11.404
Gross Shared Savings (PBPY) $254 $197
Gross Shared Savings % 2.2% 1.7%
Gross Shared Savings $3,083,062 $2,399,448
Net Shared Savings % 1.1% 0.0%
Net Shared Savings $ $1,541,531 $0

The Danger of Relying on CMS Estimates

CMS recommends using expenditures from the EXPU expenditure and utilization reports to estimate benchmark trend. Backtests on 2016 and 2017 data demonstrate the difference between this estimation approach versus the actual final settlement. While it was reasonably close in 2016, the differences for 2017 final and 2018 preliminary figures become more pronounced. Differing truncation and completion methods are two primary causes for the difference in results.

Applying these actual overestimates to the previously defined hypothetical ACO, shows a historical error of almost $3,000,000 in the expected benchmark.

Year CMS Estimate Trend (1) Actual Final Trend (2) Resulting Error Benchmark Difference (3)
2016 (4) 1.4% 1.2% 0.1% Overestimate $270,000
2017 5.1% 2.9% 2.2% Overestimate $2,990,000
2018 4.3% 3.8%(5) 0.5% Overestimate $680,000
(1) Estimated national assignable trend obtained using CMS’s recommended method
(2) Actual trend based on final settlement
(3) For an ACO with median total assigned beneficiaries of 12,156 (as of 2017 PUF)
(4) The 2016 performance year used National FFS instead of National Assignable FFS trends
(5) Derived by Validate Health in advance of CMS determination. Actual is still pending release by CMS. See “Methodology” section for more.

Based on historical data, ACOs that rely on the CMS suggested trend for financial projections are likely to overestimate shared savings. This causes financial planning issues including:

  • Under-reserving for potential losses for ACOs in downside risk tracks.
  • Overly optimistic expectations set with participants, leading to disappointment when the final settlement is released.
  • Basing future participation decisions on an overly-optimistic outlook, which could lead to undertaking too much risk too fast.

Considerations Beyond National Trend

While these examples demonstrate the impact of national trend factor, when it comes to predicting performance, it is just one of the possible market-wide variables to consider. It is important to generate realistic scenarios that include other key uncontrollable risks, including regional update factor, truncation threshold and renormalization factor. The set of simulated scenarios should be based on a range of values from previously observed events, as well as potential outlier scenarios to account previously unobserved outcomes.

To optimize for annual decisions facing an ACO, the controllable variables should be added to the modeling. These include participant selection, track selection and assignment methodology. More sophisticated multi-year scenarios can further improve the optimization model. It is possible to model 5-year contract decisions, such as MSR/MLR selection and optimal repayment mechanism. While there will always be unpredictable factors, such modeling positions ACOs to achieve the strongest possible financial outcomes on a statistical basis. It also equips the ACO to hedge risk exposure to better lock in expected gains. Hedging can be achieved using traditional financial instruments, such as reinsurance, or through offsetting cashflows from a portfolio of value-based contracts.

Financial Optimization though Scenario Simulation

Uncontrollable Variables
Driven by Market Conditions		 Controllable Variables
Within a Performance Year		 Controllable Variables
Full Multi-Year Planning
National Trend Participant Selection MSR/MLR
Regional Update Factor Track Selection Repayment Machanism
Truncation Threshold Assignment Methodology CACO Bifurcation
Renormalization Factor Reinsurance Network Design & Acquisitions
Entry/Exit for Non-MSSP Programs
Hedging with Offsetting Cash Flows
  • Backtests were performed using 2016 and 2017 actual trends. They compare historical settlements to the mid-year benchmark report to show expected error ranges for the estimated 2018 trend.

    2018 Error Range by Enrollment Type

    Enrollment Type Min Expected Max
    Combined $11,954 $12,016 $12,079
    Aged/Non-Dual $10,557 $10,609 $10,661
    Disabled $11,829 $11,959 $12,089
    Aged/Dual $17,583 $17,646 $17,708
    ESRD $85,792 $85,920 $85,977
  • The 2018 trend rates were derived using data from the legacy MSSP program. Going forward, trend estimates may differ due to methodology changes between MSSP and Pathways to Success.
  • Since the trend rate is sufficiently different for each enrollment type, an ACO’s mix of beneficiary types will affect the combined trend
  • In 2018, most ACOs were under 100% national trend. Under Pathways to Success, ACOs will use a mix of regional and national assignable trend
  • Source for multi-year actual trends from 2011 through 2018 under 2018 rules was “National Assignable FFS Factors Through 2017 for PY18 And Subsequent Years.xlsx” (as of July 2018)
  • EXPU-based estimates under the CMS approach were not estimated for 2015 and earlier since the final BY3 EXPU reports were not available prior to 2016