Table A2.

Attribution for the Intervention and Comparison Group

	Intervention Group	Comparison Group
ACOs
Maine	Beneficiaries assigned to providers associated with an ACO. Beneficiary attribution occurred on an annual basis, and to be eligible members had to have been continuously enrolled for at least six months or nine months noncontinuously enrolled in the year of attribution. Beneficiaries were attributed either through being assigned to a health home (Maine Medicaid's primary care medical homes for individuals with chronic conditions) that was a part of an ACO, having a plurality of primary care visits to an ACO provider, or having 3 or more ED visits to a hospital that was part of an ACO. Medicare‐Medicaid dually eligible beneficiaries were included.	Beneficiaries had to be (1) continuously enrolled for at least 6 months or noncontinuously enrolled for 9 months during the year; AND (2a) enrolled in a health home practice that was not a part of an ACO, OR (2b) have had a plurality of primary care visits at a practice not affiliated with an ACO, OR (2c) have had 3 or more ED visits at a hospital not part of an ACO. Medicare‐Medicaid dually eligible beneficiaries were included and comprise 18% of the sample.
Minnesota	Beneficiaries, identified by the state, who were attributed to an ACO if a provider within that ACO supplied a health care home service, or if the beneficiary received the plurality of primary care services or specialty care services from an ACO provider. The beneficiary had to be continuously enrolled in Medicaid for at least 6 months or have	Beneficiaries not ever attributed between 2013 and 2016 to an ACO but who were eligible. Beneficiaries had to meet the same 6‐month/9‐month enrollment criteria as the intervention group and had to have a health care home service or a plurality of primary care services or specialty care services from a non‐ACO
	a total of 9 or more months of noncontinuous enrollment. For the expenditure analysis using the Minnesota All Payer Claims Database, we used the state's algorithm to attribute beneficiaries to an ACO. Medicare‐Medicaid dually eligible beneficiaries were excluded.	provider. Medicare‐Medicaid dually eligible beneficiaries were excluded.
Vermont	Beneficiaries, identified by the state, who were assigned to providers associated with an ACO participating in the Medicaid ACO. Beneficiary attribution occurred on an annual basis, and to be eligible, members had to have been enrolled for at least 10 months in the year of attribution. Attribution occurred either through a claims‐based algorithm or through the affiliation of the beneficiary's assigned primary care provider. Medicare‐Medicaid dually eligible beneficiaries were excluded.	Beneficiaries, identified by the state, who met the same 10‐month enrollment criteria as the intervention group and were assigned to a provider not affiliated with an ACO or a provider associated with an ACO participating in Vermont's 1 commercial ACO only. Medicare‐Medicaid dually eligible beneficiaries were excluded.
BHH
Maine	Beneficiaries, identified by BHH providers, who met diagnostic and function criteria and agreed to participate in the program. Medicare‐Medicaid dually eligible beneficiaries were included.	Not applicable