MSSP ACOs Didn’t Reduce Income-Based Quality Gaps

Key Takeaways

• Across Medicare Shared Savings Program (MSSP) accountable care organizations (ACOs) in U.S. traditional Medicare, ACO formation was not associated with significant differential improvement in potentially avoidable emergency department visits, ambulatory care–sensitive admissions, or 30-day readmissions for beneficiaries with low vs higher income after 3 years.
• Before ACO formation, beneficiaries with low income had unadjusted rates 1.6 to 3 times higher across preventable ED visits, ambulatory care–sensitive admissions, and readmissions than beneficiaries with higher income.
• Exploratory analyses suggested organizational heterogeneity: 76 ACOs (13.0%) were associated with improved relative rates on at least 1 quality measure while maintaining or improving quality for higher-income beneficiaries, whereas 68 ACOs (11.6%) were associated with worsened relative rates under the source’s predefined criteria; the remaining ACOs had mixed or no relative associations.
• ACOs linked to reduced disparities had larger baseline disparity levels, while other measured organizational characteristics were similar.

Whether value-based payment in traditional Medicare has narrowed income-based quality gaps for beneficiaries attributed to Medicare Shared Savings Program (MSSP) accountable care organizations (ACOs) remains unsettled. The question is especially relevant for utilization-based outcomes that can reflect care coordination, access to ambulatory care, and postdischarge stability among beneficiaries with low income. Investigators compared attributed beneficiaries at ACO and non-ACO organizations to assess whether those gaps changed after MSSP entry.

A Bond et al cohort study in JAMA Network Open used 2010 to 2019 Medicare fee-for-service claims, Medicare Data on Provider Practice and Specialty files, and 2013 to 2019 MSSP ACO Provider files to evaluate 585 ACOs and 35 025 903 traditional Medicare patients. Beneficiaries were classified as having low income if they were ever eligible for the Medicare Part D Low-Income Subsidy (LIS) during the study period, and all others were classified as having higher income. Patients were attributed to the Taxpayer Identification Number (TIN) or CMS Certification Number (CCN) providing a plurality of outpatient primary care in the prior year, and investigators used a triple difference-in-differences design within hospital-referral regions. Primary effects were estimated 3 years after ACO formation, representing 1 MSSP contract period, for potentially avoidable emergency department (ED) visits, ambulatory care–sensitive admissions (ACSAs), and 30-day unplanned hospital readmissions. Exploratory analyses examined ACO size, rurality, hospital inclusion, the share of patients with low income, federally qualified health center presence, and baseline disparity level, and sensitivity analyses were directionally similar.

Adjusted analyses showed no significant differential change for beneficiaries with low vs higher income in any of the 3 quality measures at ACOs compared with non-ACO organizations 3 years after formation. In secondary unadjusted analyses, the disparity in ACSAs increased by 1.8 per 1000 patients (95% CI 0.8-2.8; P < .001), a 5.6% widening. Exploratory analyses found that 76 ACOs (13%) were associated with reduced disparities on at least 1 quality measure and 68 ACOs (11.6%) with worsened disparities. ACOs linked to reduced disparities had a 0.36 SD higher baseline disparity z score than ACOs linked to worsened disparities (95% CI 0.17-0.56; P < .001), while other measured ACO characteristics were similar.

The authors noted several factors that could affect interpretation, including the voluntary nature of MSSP participation and the possibility of unobserved time-varying differences between organizations. They also described possible patient or practice selection despite retaining practices that entered during the first 2 years of ACO formation, overlap with other Medicare payment reforms, the pre-COVID end of the study period, possible differential coding intensity, and the use of LIS eligibility as a proxy that may not capture income changes over time. Within those constraints, the study documents no overall disparity reduction on these utilization-based quality measures during the program’s first decade, even though exploratory organization-level trajectories varied.

The authors concluded that MSSP ACO formation was not associated with overall reductions in income-based disparities across these utilization-based measures, although some organizations appeared to narrow gaps. They also noted that the association between larger baseline disparities and later disparity reduction does not establish causation and may reflect regression to the mean. The reported finding was an overall absence of disparity reduction at the program level alongside variation across individual organizations.

Clinician Questions

How was low income defined in the MSSP ACO disparity analysis?

Beneficiaries were classified as having low income if they were ever eligible for the Medicare Part D Low-Income Subsidy (LIS) during the study period, while all other traditional Medicare beneficiaries were classified as having higher income. The investigators used LIS because, unlike dual eligibility, it does not vary by state, and the authors described it as a proxy that may miss income changes over time.

Which Medicare beneficiaries were included in the comparison of MSSP ACOs and non-ACO organizations?

The cohort included traditional Medicare beneficiaries living in hospital-referral regions where at least 1 MSSP ACO formed between 2012 and 2017. Beneficiaries were excluded in any year if they lacked full Part A or Part B coverage, enrolled in Medicare Advantage, or lived outside the United States, and comparison beneficiaries were attributed to non-ACO organizations in the same regions.

How were improved and worsened disparity changes defined for individual MSSP ACOs?

An MSSP ACO was classified as having improved disparities when relative outcomes for beneficiaries with low income improved on at least 1 quality measure while quality for beneficiaries with higher income on that measure and the other measures was maintained or improved. The mirror definition was used for worsened disparities, and the remaining MSSP ACOs had mixed or no relative associations.

How did the exploratory analysis define small and rural ACOs?

Small ACOs were defined as having fewer than 10 000 attributed patients, and rural ACOs were defined as having at least 25% of practices or clinics in nonmetropolitan areas using Rural-Urban Commuting Area codes. The study used those categories for organizational comparisons rather than as the main program-level exposure.