Largest Medicare ACOs in 2026, by participant names

2026 Medicare Shared Savings Program roster · CMS files retrieved September 23, 2026

The largest Medicare ACOs in the 2026 Shared Savings Program roster, when size means the number of distinct reported participant legal business names, are PBACO Holding, LLC (822), SOMOS ACO (308), and Caravan Collaborative Pathways (271). This is a ranking of participant-name strings in a CMS snapshot. It does not rank patient volume, clinicians, revenue, or verified legal entities. CMS publishes the ACO roster and participant roster separately.

The top 25, including the tie at the cutoff

The table has 26 ACOs because two share rank 25 with 104 distinct names. Equal counts share a competition rank, so rank 14 is followed by rank 16. The top row's 822 names arise from 823 participant rows: one exact name repeats. The 2026 participant file has 15,370 rows, all of which joined to an ACO ID in the 511-row 2026 ACO file; none had a blank Par_LBN in this extraction. The complete 511-ACO ranking is saved with the editorial evidence.

What the count actually measures

An accountable care organization, or ACO, is a group of health care providers that participates in a Medicare payment arrangement. For this article, participant means a record in CMS's Shared Savings Program ACO participant file. The field Par_LBN is the participant's reported legal business name; it is a name field, not a count of individual physicians. CMS's participant data dictionary documents the field.

We join each 2026 participant row to the 2026 ACO roster by ACO_ID. Within each ACO, we remove spaces at the beginning or end of Par_LBN, exclude empty values, and count distinct remaining strings. We preserve letter case and internal punctuation. Thus “ABC Health” and “ABC HEALTH” count separately. We do not infer common ownership or merge similar names. ACOs in the 2026 organization file define the eligible population; this is a roster definition of “active,” not a live membership check on the day of reading.

This distinction matters because a reported name can appear more than once for the same ACO, and two differently written names may refer to one entity. Conversely, a shared string need not prove that two records represent the same legal entity. These counts are a reproducible measure of reported participant-name breadth. They are not an audited legal-entity census. They also cannot establish referral relationships, the number of NPIs or physicians in a network, beneficiary assignment, or the number of counties served.

What stands out in the 2026 list

PBACO Holding has a large lead under this definition: 822 distinct names, compared with 308 for SOMOS ACO and 271 for Caravan Collaborative Pathways. That gap is a property of the published participant roster, not evidence of a similar gap in beneficiaries or clinical capacity. The full ranking should be used before describing an ACO as “largest,” because a different size metric may reorder the list.

The middle of the table is close enough for the counting rule to matter. Optimus Healthcare Partners and Mass General Brigham ACO both have 128 distinct strings and share rank 14. At the cutoff, Trinity Health Integrated Care and QCIPN both have 104 and share rank 25. Excluding one because of an arbitrary alphabetical or ID tiebreak would turn an equal measurement into a misleading ordering.

We checked exact repetition and obvious case or punctuation variants in the displayed rows. The top 26 include a case variant in Caravan Collaborative Pathways' participant names: “Metroplex Care Group” and “METROPLEX CARE GROUP.” If those are the same entity, a normalized-name count for that ACO would be 270 rather than 271. It would remain third. We did not change the published count because a case-insensitive merge is an inference, not the brief's stated exact-string method. This targeted check cannot rule out every ambiguous business name across the file.

There is another source of ambiguity: a legal business name is a label recorded on a participation row, whereas an organization can have subsidiaries, trade names, or differently structured participation arrangements. A raw row count can overstate the number of distinct reported names if a participant appears more than once. In the top group, the difference is visible for PBACO Holding (823 rows versus 822 names) and Piedmont Clinic ACO (121 rows versus 120 names). For most other displayed ACOs, the row and distinct-name counts happen to be equal. That equality still does not validate the legal identity behind each string.

How to use the ranking

For network research, start with the ACO ID, then inspect its participant-name records. Compare organizations using the same performance-year file and definition. If the business question is provider reach, join to clinician or NPI data with an explicit, validated relationship; a participant-name count cannot answer it. If the question is financial performance, use CMS's separate, later-released performance results. An ACO near the top of this roster is not necessarily near the top for earned shared savings.

A useful next step is to compare a short list of ACOs by two columns: this article's participant-name count and the roster's reported participant rows. A large difference flags repeated names worth reviewing; a small difference tells you only that exact strings rarely repeat. To answer whether the same businesses participate in several ACOs, compare their participant records across ACO IDs and verify identity outside the name string. To answer whether a specific provider belongs to an ACO, use the record and identifier relevant to that provider and source year. The top list is a screening tool, not a substitute for that lookup.

The two 2026 inputs came from different CMS file releases: the participant CSV is under CMS's January 2026 release path, and the ACO organization CSV is under its April 2026 path. Both are labeled performance year 2026. CMS may revise these files, so the result should be read as the extracted file versions, retrieved September 23, 2026, rather than a perpetually current count.

Methodology

The eligible universe is the 511 unique ACO_ID values in the 2026 CMS ACO organization file. We grouped the 15,370 rows in the 2026 participant file by matching ACO_ID, excluded blank Par_LBN values (zero in the retrieved file), and counted distinct outer-trimmed strings per ACO. All participant rows matched the 2026 ACO IDs. We ranked counts descending using competition ranks and displayed equal counts in ACO ID order. The table includes all rows at the 25th-position count. Source file names, retrieval times, and checks are recorded in the saved editorial evidence manifest.

Explore ACO Networks in QOPE

For an ACO-level lookup beyond this list, explore ACO Networks data in QOPE and verify the displayed record and source year against the CMS file before using it in an analysis.

Sources

• CMS Medicare Shared Savings Program Organizations Dataset (PY2026)
• CMS Medicare Shared Savings Program Participants Dataset (PY2026)
• CMS Medicare Shared Savings Program ACO Participants Data Dictionary
• CMS Medicare Shared Savings Program (MSSP) Overview
• CMS Data Catalog: Accountable Care Organizations