Post-Acute Care and ACOs — Who Will Be Accountable?

Abstract

Objective

To determine how the inclusion of post-acute evaluation and management (E&M) services as primary care affects assignment of Medicare beneficiaries to accountable care organizations (ACOs).

Data Sources

Medicare claims for a random 5 percent sample of 2009 Medicare beneficiaries linked to American Medical Association Group Practice data identifying provider groups sufficiently large to be eligible for ACO program participation.

Study Design

We calculated the fraction of community-dwelling beneficiaries whose assignment shifted, as a consequence of including post-acute E&M services, from the group providing their outpatient primary care to a different group providing their inpatient post-acute care.

Principal Findings

Assignment shifts occurred for 27.6 percent of 25,992 community-dwelling beneficiaries with at least one post-acute skilled nursing facility stay, and they were more common for those incurring higher Medicare spending. Those whose assignment shifted constituted only 1.3 percent of all community-dwelling beneficiaries cared for by large ACO-eligible organizations (n = 535,138), but they accounted for 8.4 percent of total Medicare spending for this population.

Conclusions

Under current Medicare assignment rules, ACOs may not be accountable for an influential group of post-acute patients, suggesting missed opportunities to improve care coordination and reduce inappropriate readmissions.

The Medicare accountable care organization (ACO) programs provide participating organizations with new incentives to lower spending while improving quality of care. These incentives apply to the care of long-term nursing home residents and patients of post-acute nursing facilities only if these groups are included in populations assigned to ACOs for performance evaluation. Thus, the potential for ACO programs to improve care and control spending for these costly Medicare beneficiaries depends in part on the rules governing beneficiary assignment.

In the two principal Medicare ACO programs, the Shared Savings Program (SSP) and the Pioneer program, beneficiaries are assigned to participating ACOs if they receive more primary care services from ACO providers than from any other provider group. Assignments define the population for which an ACO is held accountable, but they do not restrict beneficiaries' choice of providers. For the purpose of assignment, both the SSP and Pioneer program define primary care as specific sets of evaluation and management (E&M) services (Table 1) delivered not only in outpatient settings but also in skilled nursing facilities (SNFs) and nursing homes (Center for Medicare and Medicaid Innovation 2011a; Department of Health and Human Services and Centers for Medicare and Medicaid Services 2011).

Table 1.

Evaluation and Management (E&M) Service Codes Used to Assign Medicare Beneficiaries to Shared Savings Program and Pioneer Accountable Care Organizations

Current Procedural Terminology Codes for E&M Physician Services	Setting/Description of E&M Physician Services
99201–99215	Office or other outpatient services
99304–99318	Nursing facility services
99324–99340	Domiciliary, rest home, or custodial care services
99341–99350	Home services
G0402, G0438, G0439	Wellness visits

This definition ensures that long-term nursing home residents no longer receiving primary care in the community are still eligible for assignment to ACOs that include nursing homes. Bed-hold payments from Medicaid and lucrative post-acute reimbursements from Medicare have created perverse incentives for nursing homes to hospitalize long-term residents, contributing to high rates of potentially avoidable admissions in this population (Grabowski 2007; Grabowski, O'Malley, and Barhydt 2007; Ouslander and Berenson 2011). ACOs may view these hospitalizations as easy targets for cutting costs and seek to partner with nursing homes in shared savings contracts with Medicare. Although nursing facilities are currently included in very few of the 147 SSP and Pioneer ACO network descriptions, including nursing home E&M services in the assignment process appropriately allows for ACO contracts to apply to long-term nursing home residents if such partnerships should develop (Center for Medicare and Medicaid Innovation 2011b; Centers for Medicare and Medicaid Services 2012).

For community-dwelling beneficiaries, however, counting E&M services provided in SNFs as primary care services could transfer the locus of accountability from primary to post-acute care providers as an unintended consequence. Specifically among beneficiaries who receive both outpatient primary care and short-term post-acute care, the assignment rules may selectively assign the sickest patients requiring the most post-acute care away from community primary care providers, and thus away from ACOs that do not include SNFs in their contracting networks. Shifting assignment of these beneficiaries from ACOs to non-ACO SNF providers could result in missed opportunities for ACOs to improve continuity during care transitions and reduce inappropriate transfers and readmissions for a costly and vulnerable population (Saliba et al. 2000; Mor et al. 2010). Likewise, physician organizations and hospitals best able to manage these medically complex patients and coordinate their care across acute, post-acute, and outpatient settings may be denied the chance to share in greater savings.

The assignment rules are used not only to determine populations for which ACOs are held accountable during performance years of a contract (when quality and spending are assessed relative to performance targets) but also to identify baseline populations whose costs of care in the years preceding the performance period can be extrapolated to establish spending targets. Thus, rules that limit assignment of especially costly beneficiaries to ACOs could lower spending targets significantly. Due to the differences between programs in the timing of assignment relative to performance assessment, the financial implications of these effects on spending targets may differ for SSP and Pioneer ACOs (Table 2).

Table 2.

Summary of Medicare ACO Program Rules for Beneficiary Assignment and Spending Target Calculations

	Shared Savings Program (SSP) ACOs	Pioneer ACOs
Assignment algorithm	Beneficiary assigned to group of providers accounting for more allowed charges for primary care services provided by PCPs than any other group. If no primary care services received from PCPs, beneficiary assigned to group accounting for more allowed charges for primary care services provided by specialists, NPs, or PAs than any other group. If no primary care services, not eligible for assignment.	If ≥10% of primary care services provided by PCPs, NPs, or PAs, beneficiary assigned to group of providers accounting for more allowed charges for primary care services provided by PCPs, NPs, or PAs than any other group. If <10% of primary care services provided by PCPs, NPs, or PAs, beneficiary assigned to group accounting for more allowed charges for primary care services provided by specialists than any other group. If no primary care services, not eligible for assignment.
Timing of assignment (for performance assessment)	Retrospective: assignment based on primary care use during performance year	Prospective: assignment based on primary care use in 3 years preceding performance year
Baseline spending determination	Based on spending for beneficiaries who would have been assigned to the ACO in any of the three baseline years (e.g., 2009–2011 for a performance period of 2012–2014). (Population assigned for performance years may differ from baseline population used to establish baseline spending)	Based on spending in the three baseline years for the prospectively assigned population. (Population assigned for performance years generally the same as baseline population used to establish baseline spending)
Spending target for performance year Y	ACO-specific baseline spending + average absolute growth in Medicare spending ($) from baseline to year Y	ACO-specific baseline spending + 50% (average absolute growth in Medicare spending from baseline to year Y) + 50% (ACO-specific baseline spending × average relative growth in Medicare spending (%) from baseline to year Y)
Predicted effects of assigning costly post-acute patients away from ACOs	Offsetting decreases in spending target and spending during performance years → no financial gain or loss for ACOs	Spending target decreased by greater amount than spending during performance years → financial loss for ACOs

ACO, accountable care organization; NP, nurse practitioner; PA, physician assistant; PCP, primary care physician.

Using 2009 Medicare claims linked to the American Medical Association (AMA) Group Practice File, we quantified the potential assignment problem for community-dwelling beneficiaries caused by the consideration of post-acute E&M services as primary care and discuss its implications for the two Medicare ACO programs.

Methods

Data Sources

For physicians in the AMA Physician Masterfile practicing in groups of three or more, the AMA Group Practice File identifies their practice site(s) and parent organization if part of a larger group. Physician rosters and practice information in the Group Practice File are verified and updated every 9–12 months by the AMA via telephone and fax communications with practice managers and from provider group websites. For groups identified as members of larger organizations, we used the highest level of organization to which they were connected (through up to five levels of hierarchical affiliations) for analyses. We linked AMA Group Practice data to 2009 Medicare claims via physicians' National Provider Identifiers (NPIs) (see Appendix for details of linkage).

Study Population

From a random 5 percent sample of traditional fee-for-service Medicare beneficiaries in 2009, we identified those who would have been assigned to provider groups sufficiently large to participate as ACOs in the SSP (eligibility threshold: ≥5,000 assigned beneficiaries or ≥250 in the 5 percent sample). Using Part B claims for physician services and following the SSP assignment rules (Table 2), we assigned each beneficiary to the provider group accounting for the most allowed charges for primary care services among all groups providing primary care to the beneficiary (Department of Health and Human Services and Centers for Medicare and Medicaid Services 2011).

To identify provider groups, we used both the AMA Group Practice File and tax identification numbers (TINs) indicating billing entities in Medicare claims. We used both of these sources of group identifiers because they were complementary in identifying groups large enough to participate in the SSP. Physicians identified in the Group Practice File as members of a single large networked group, for example, might bill under multiple TINs. Conversely, multiple smaller groups identified in the Group Practice File might bill under a single TIN if owned by a larger organization. Therefore, we assigned each beneficiary to both a TIN and a provider group in the Group Practice File by applying the SSP assignment algorithm to each beneficiary twice, using each source of group identifiers (TINs or Group Practice File) independently. We considered beneficiaries to be assigned to ACO-eligible groups if they were assigned to either a TIN or group in the Group Practice File that met the size threshold for SSP eligibility.

Our study sample included 535,138 traditional Medicare beneficiaries who were assigned to ACO-eligible groups and were not long-term residents of nursing homes according to a validated claims-based algorithm (Yun et al. 2010). We focused in particular on 25,992 of these beneficiaries who had at least one post-acute SNF stay, because only for these community-dwelling beneficiaries receiving both outpatient primary care and inpatient post-acute care could assignment to ACO-eligible groups be influenced by the consideration of post-acute services in the assignment process.

Assignment Shifts

We first assigned each of these 25,992 beneficiaries to a TIN and a Group Practice File group using the E&M codes delineated in the Pioneer and SSP rules, but excluding services delivered in nursing facilities (Center for Medicare and Medicaid Innovation 2011a; Department of Health and Human Services and Centers for Medicare and Medicaid Services 2011). We then reassigned these beneficiaries after expanding the definition of primary care to include nursing facility E&M services. We calculated the fraction whose assignment consequently shifted from the group providing their outpatient primary care to a different group providing their inpatient post-acute care.

We considered a change in either the assigned TIN or the assigned Group Practice File group as an assignment shift. Because Medicare ACOs define themselves for contracting purposes as sets of TINs (in the SSP) or sets of TIN-NPI combinations (in the Pioneer program) and may choose to include or exclude affiliated post-acute care providers, changes in either the assigned TIN or Group Practice File group (a group of NPIs) indicated a shift in assignment to a group of post-acute care providers that could potentially be excluded from an ACO's contracting network.1 Our estimates of assignment shifts provide an upper bound for this potential assignment problem, as we assume that no post-acute care providers are included in ACO networks. This assumption is nevertheless consistent with the absence of nursing facilities in the overwhelming majority of network descriptions for ACOs currently participating in the SSP and Pioneer program (Center for Medicare and Medicaid Innovation 2011b; Centers for Medicare and Medicaid Services 2012).

Association between Medicare Spending and Assignment Shifts

For each beneficiary, we assessed total Medicare spending on all services covered by Part A and B, spending on inpatient facility care, spending on post-acute SNF care, and number of hospitalizations in 2009. We characterized the extent to which acute and post-acute care needs predicted assignment shifts to post-acute providers by comparing the frequency of these shifts across deciles of per-beneficiary spending. We also compared spending for beneficiaries whose assignment shifted versus those in the larger sample of community-dwelling beneficiaries who were assigned to SSP-eligible groups but had no post-acute SNF stays (n = 509,146).

Our study was approved by the Harvard Medical School Committee on Human Studies and the Centers for Medicare and Medicaid Services Privacy Board.

Results

AMA Group Practice data and TINs were complementary in identifying groups sufficiently large to participate in the SSP. Among the 25,992 community-dwelling beneficiaries with at least one SNF stay, 93.7 percent were assigned to ACO-eligible groups via Group Practice File group assignments and 61.7 percent via TIN assignments (100 percent via either assignment). Thus, relative to the use of TINs alone, the use of AMA Group Practice data substantially increased the number of beneficiaries assigned to ACO-eligible groups, suggesting many large provider groups bill at organizational levels (TINs) below the highest levels identified in the Group Practice File.

Including nursing facility E&M services in the definition of primary care caused a shift in assignment for 27.6 percent of community-dwelling beneficiaries with at least one post-acute SNF stay. As expected, assignment shifts were more common for those in higher deciles of total Part A and B spending (Figure 1, Panel A), inpatient spending (Panel B), and SNF spending in particular (Panel C). From the lowest to highest decile of post-acute spending on SNF care, the fraction with assignment shifts increased from 13 to 44 percent.

Figure 1.

Assignment Shifts from Primary to Post-acute Care Providers by Decile of Annual Medicare Spending on (A) All Part A and B Services, (B) Inpatient Care, and (C) SNF care. (Among 25,992 community-dwelling Medicare beneficiaries with at least one post-acute SNF stay who were assigned to ACO-eligible provider groups, we determined the percentage whose assignment shifted from outpatient primary care to inpatient post-acute care providers as a consequence of expanding the definition of primary care to include nursing facility E&M services.)

Among all 535,138 community-dwelling beneficiaries assigned to ACO-eligible groups, mean total Medicare spending in 2009 was $55,184 higher for those whose assignment shifted than for those with no post-acute SNF stays (Table 3). Spending on inpatient and SNF care and number of hospitalizations were also substantially higher. Thus, although beneficiaries whose assignment shifted constituted only 1.3 percent of all community-dwelling beneficiaries assigned to ACO-eligible groups, they accounted for 8.4 percent of total Medicare spending for this population.

Table 3.

Medicare Spending and Hospitalizations by Use of Post-Acute Care and Assignment Shifts

Measure of Annual Spending or Utilization in 2009	No Post-Acute Care ( N = 509,146)	Post-Acute Care, No Assignment Shift ( N = 18,828)	Post-Acute Care, Assignment Shift ( N = 7,164)
Total Medicare Part A and B spending, mean (95% CI)	7,858 (7,812, 7,905)	50,322 (49,777, 50,867)	63,042 (61,915, 64,168)
Medicare Part A spending on acute inpatient care, mean (95% CI)	2,890 (2,860, 2,920)	24,003 (23,604, 24,401)	30,481 (29,644, 31,318)
Medicare Part A spending on post-acute inpatient care, mean (95% CI)	–	11,652 (11,510, 11,794)	16,317 (16,043, 16,590)
Hospitalizations, mean (95% CI)	0.300 (0.298, 0.303)	2.196 (2.172, 2.219)	2.431 (2.388, 2.475)

Discussion

These findings suggest that an influential subgroup of Medicare beneficiaries receiving post-acute care may be selectively excluded from populations assigned to ACOs. Under Medicare's current definition of primary care services, ACOs may not be held accountable for nearly 30 percent of their PCPs' patients who receive post-acute SNF care, and for an even higher fraction of those with multiple hospitalizations and high post-acute costs. Thus, assignment algorithms that count post-acute physician visits in SNFs as primary care services may diminish incentives for ACOs to coordinate care and lower spending for a group of patients whose care is often fragmented and particularly costly.

The financial implications of these findings differ for SSP and Pioneer ACOs because of different methods of assigning beneficiaries for performance years of contracts (Table 2). In both programs, an ACO's global spending target for a performance year is based on spending for beneficiaries receiving a plurality of their primary care from the ACO during a baseline period, trended forward by an update factor. For a Pioneer ACO, the population assigned in each performance year is generally the same as this baseline population, because Pioneer ACOs are assigned cohorts of beneficiaries prospectively.

Accordingly, for Pioneer ACOs that do not include SNFs, our findings suggest that a group of beneficiaries with substantial post-acute care costs in the baseline period will be systematically excluded from prospectively assigned cohorts, and thus their high costs excluded from spending target calculations as well. In performance years, however, some assigned cohort members will require substantial post-acute care as a result of new illnesses and hospitalizations, and none of these patients (or their associated post-acute costs) will be removed from the cohort because they were assigned prospectively. Thus, the selective exclusion of costly post-acute patients at baseline will generate spending targets for assigned cohorts that are systematically lower than their expected costs in performance years. Including post-acute E&M services in the definition of primary care could therefore result in financial losses for Pioneer ACOs. Pioneer organizations could mitigate these losses by including more SNFs in their ACO networks.

Conversely, populations assigned to SSP ACOs in each performance year will be determined by primary care use during that year, not the baseline period, while spending targets will be based on analogous populations that would have been assigned in the baseline period. Therefore, high users of post-acute care will be consistently excluded from determinations of both spending targets and performance-year spending, posing no losses to SSP ACOs on average. SSP ACOs could gain financially in performance years, however, by transferring particularly sick inpatients with long expected post-acute stays to SNFs not included in their ACO networks. Moreover, SSP ACOs may choose not to partner with SNFs when contracting with Medicare to avoid accountability for this small, but high-risk group, whose costs are likely to vary substantially and unpredictably from year to year. Thus, the inclusion of post-acute E&M services in the definition of primary care could limit the scale of efforts by SSP ACOs in particular to coordinate with post-acute care providers and prevent avoidable readmissions from post-acute settings.

To better align incentives for ACOs with other strategies to control spending for care spanning acute, post-acute, and long-term settings, such as bundled payments, value-based purchasing, and Medicare Special Needs Plans (Mor et al. 2010; Ouslander and Berenson 2011) the Medicare ACO programs could consider the following modifications to beneficiary assignment. First, the current definition of primary care E&M services (including those provided in nursing facilities) could be used to assign long-term nursing home residents, who can be reliably identified by existing algorithms (Yun et al. 2010) to their long-term care providers as the locus of potential accountability. Thus, nursing homes partnered with large provider organizations in ACO contracts could be held jointly accountable for long-term residents no longer receiving primary care in the community. Second, for the assignment of community-dwelling beneficiaries, E&M services provided in nursing facilities could be removed from the definition of primary care. Thus, community-dwelling patients of an ACO's primary care providers would not be assigned away from the ACO if they received substantial post-acute care from SNFs not included in the ACO's contract. These changes could extend the influence of Medicare ACO programs to a particularly costly group of patients, enhance efforts to coordinate care across acute and post-acute settings, and potentially generate greater savings to Medicare.

SUPPORTING INFORMATION

Additional supporting information may be found in the online version of this article:

Appendix SA1: Author Matrix.

Appendix SA2: Linkage to American Medical Association (AMA) Group Practice File.