On August 8, 2018, the Centers for Medicare and Medicaid Services (CMS) released a sweeping proposed rule that, if enacted, will significantly change the Medicare Shared Savings Program (MSSP). This paper is the third in a series of Milliman white papers about the proposed rule and focuses on beneficiary assignment.
In the current program, the track (1, 1+, 2, or 3) chosen by an accountable care organization (ACO) determines the methodology used to assign beneficiaries to that ACO. Under the proposed rule, an ACO will be allowed to select between two beneficiary assignment methods, prospective or retrospective, regardless of track or risk level, and to change its choice annually. In this paper, we explore the differences between the two proposed assignment methodologies and considerations for ACOs as they evaluate their options. We also briefly discuss additional proposed changes to the definition of primary care services used for assignment.
Choice of assignment methodology
Under the proposed rule, CMS will offer the choice of retrospective or prospective beneficiary assignment to ACOs in the BASIC and ENHANCED tracks for agreement periods beginning July 1, 2019, or later. An ACO will be able to choose a beneficiary assignment methodology at the time of entry and can alter this selection prior to the start of each performance year. Under retrospective assignment, an ACO’s assigned population is based on services incurred during the performance year. Under prospective assignment, an ACO’s assigned population is based on services incurred during the 12-month period ending three months prior to the start of the performance year. If an ACO changes its assignment methodology election, its historical benchmark will be updated (consistent with current practice).
There are distinct trade-offs between the two assignment methodologies, and the optimal choice will vary by ACO. Figure 1 highlights key considerations.
Figure 1: Features of retrospective and prospective assignment
|Who is included in the assigned population
||Under retrospective assignment, the assigned population is based on services that happened during the performance year. However, the assigned population is unknown until the final settlement and preliminary estimates can shift dramatically over the course of the year.
Conversely, under prospective assignment the assigned population is largely known going into the performance year.1 However, because assignment does not reflect performance year care patterns the ACO may be responsible for beneficiaries who receive little or no care from ACO participants during the performance year.
|Number of assigned beneficiaries
||Typically, the number of assigned beneficiaries drops by 5% to 10% under prospective assignment as compared to retrospective assignment.2 Larger differences often occur in areas with higher Medicare Advantage penetration.
|Priority of assignment
||Prospective assignment has priority over retrospective assignment. In other words, a beneficiary who is prospectively assigned to an ACO cannot be assigned to another ACO during the performance year.
A few other considerations for ACOs exploring the two options for assignment methodology include:
- If an ACO is near the minimum of 5,000 assigned beneficiaries, it may want to avoid the beneficiary reduction that typically accompanies prospective assignment.
- If there are multiple ACOs in a market, an ACO may consider prospective assignment which has priority over retrospective assignment. It is worth noting that voluntary alignment takes precedence over both prospective and retrospective assignment.
- If there is significant churn (change in assigned beneficiaries from year to year), the ACO may favor retrospective assignment so that performance year expenditures reflect costs for beneficiaries the ACO interacted with during the year.
- If the market is growing and there is a high rate of “age-ins,” the ACO may favor retrospective assignment in order to capture these beneficiaries.
- If the ACO has a high level of success with beneficiary engagement and annual wellness visits, the ACO may favor prospective assignment so that it can proactively identify and manage assigned beneficiaries.
- As stated previously, changes in assignment methodology affect both the performance and benchmark periods. While a change in assignment methodology during a contract period is certainly an interesting option, it can and likely would cause noticeable changes in the benchmark.
There are also several operational items an ACO should consider when choosing prospective or retrospective assignment, including:
- Under retrospective assignment, changes in an ACO’s provider network and practice patterns can have an immediate impact on performance year assignment.
- Under retrospective assignment, the ACO might find itself using limited care management resources across a larger population because it is unsure who will ultimately be assigned. Prospective assignment allows ACOs to focus these resources.
- Prospective assignment allows providers the opportunity to proactively engage assigned beneficiaries.
Addition of new codes used for assignment
Beneficiaries are assigned to an ACO based on having a plurality of primary care services during the assignment window. As described in the proposed rule, CMS plans to add the following codes to the definition of primary care services3 for purposes of beneficiary assignment:
- Advance care planning (Current Procedural Terminology [CPT] codes 99497 and 99498)
- Administration of health risk assessment (CPT codes 96160 and 96161)
- Prolonged evaluation and management or psychotherapy service(s) beyond the typical service time of the primary procedure (CPT codes 99354 and 99355)
- Annual depression screening (Healthcare Common Procedures Coding System [HCPCS] code G0444)
- Alcohol misuse screening (HCPCS code G0442) and alcohol misuse counseling (HCPCS code G0443)
- Additional resource costs, beyond those involved in the base evaluation and management codes, of providing face-to-face primary care services for established patients (HCPCS code GPC1X, a new add-on code)
- Resource costs intended to reflect the complexity inherent to evaluation and management services associated with endocrinology, rheumatology, hematology/oncology, urology, neurology, obstetrics/gynecology, allergy/immunology, otolaryngology, cardiology, and interventional pain management-centered care (HCPCS code GCG0X, a new add-on code)
- Prolonged evaluation and management or psychotherapy services beyond the typical service time of the primary procedure, in the office or other outpatient setting requiring direct patient contact beyond the usual service; 30 minutes (HCPCS code GPRO1)
A reference list of the current primary care services included in assigning beneficiaries to ACOs is detailed on pages 401 and 402 of the proposed rule.4
The codes CMS proposes to add to the definition of primary care services focus on additional services that practitioners who are already managing a beneficiary’s care likely provide. An ACO that uses these newly added services more than other ACOs or non-ACO providers in the market might expect to see an increase in assigned beneficiaries. An ACO that currently makes limited use of these new codes may wish to explore the implications of using more of these services.
Change to how primary care services are considered when a patient is in an SNF
Currently, CPT codes 99304 through 99318 are excluded from the beneficiary assignment logic, if provided in a Skilled Nursing Facility (SNF) (identified using place of service (POS) code 31 on the physician claim). This “POS 31 exclusion” was introduced in 2017 and resulted in some unanticipated impacts due, in part, to inconsistency in the coding of POS 31 (Skilled Nursing Facility) versus POS 32 (Nursing Facility) on claims data.5
The new CMS proposal is to exclude professional service claims billed with CPT codes 99304 through 99318 from use in assignment when there is an SNF facility claim with dates of service that overlap with the professional service. We believe the intent of this change is to more accurately identify, and exclude from assignment, services incurred while an individual was in an SNF. The impact of this change will vary by ACO. Because affected beneficiaries are typically higher cost than an average ACO-assigned beneficiary, it is important that ACOs assess the impact on their assignment.
Evaluating the options by ACO
An ACO currently in Track 1 with retrospective assignment may want to explore how its assigned beneficiary population might change if it switched to prospective assignment in the coming contract period. An ACO with experience in a retrospective track can estimate these numbers using its first preliminary prospective assignment list for a performance year, Quarterly Beneficiary Assignment List Report (QASSGN) files, and the Annual Beneficiary Assignment List Report (HASSGN) file included with the settlement report through performing the following six steps:
1. First, create a unique list of beneficiaries (by Health Insurance Claim Number [HICNO]) that are included in any or all of the reports.
2. Identify and count members who are excluded during the year due to death prior to the performance year, one month of Medicare Part-A-only or Medicare Part-B-only, one month in a group health plan, or residence outside the United States. These members are excluded under both prospective and retrospective assignment methods.
3. Identify and count members who are added due to performance year utilization patterns. These members are added under retrospective assignment only.
4. Identify and count members who are removed due to performance year utilization patterns. These members are removed under retrospective assignment only.
5. Identify and count members who are removed due to participation in other shared savings initiatives. These members are removed under retrospective assignment only.
6. Summarize and compare the total resulting membership after adjusting for each of the above assignment criteria.
This information can provide useful insight into changes that would occur under prospective assignment. Figure 2 provides an illustrative, hypothetical example.
Figure 2: Example of retrospective versus prospective analysis
|Exclusions during the year
|Beneficiaries added due to performance year utilization patterns
|Beneficiaries removed due to performance year utilization patterns
|Beneficiaries removed due to participation in other shared savings initiatives
We see that, in the above example, ultimate prospective assignment ends up about 10% lower (19,000 compared to 20,990) than retrospective assignment because performance year utilization patterns do not affect prospective assignment.
Under retrospective assignment, 8,000 beneficiaries were added based on performance year patterns and 6,000 beneficiaries were excluded because they did not receive sufficient primary care services within the ACO during the performance year to be included in assignment. Under prospective assignment, the ACO will be responsible for the performance of these 6,000 beneficiaries even if they receive little or no care in the ACO. However, the ACO would know of these beneficiaries in advance of the performance year and may have the opportunity to engage with them throughout the year.
Additionally, we see 10 people are not included in retrospective assignment because they were previously assigned to other shared savings initiatives. Note that his exclusion does not apply for prospective assignment. This adjustment has a larger impact in markets with several ACOs competing for the same beneficiaries where some are subject to retrospective assignment and others to prospective assignment.
Under current rules, the beneficiary assignment methodology is determined by the track in which an ACO participates. Under the proposed rule, ACOs will be able to select an assignment methodology independent of their track. While research has not indicated a higher overall likelihood of generated savings for ACOs with prospective versus retrospective assignment, it is important for each ACO to assess how the selection could uniquely affect its underlying population, benchmark, and results in determining which assignment option is optimal for its organization.