Section 3022 of the Affordable Care Act (ACA) provides for the development of the Medicare shared savings program for accountable care organizations (ACOs). The ACO program is intended to improve care for individuals and health for populations, as well as slow the rate of cost growth.
The proposed rule recently released by the Centers for Medicare & Medicaid Services (CMS) for the ACO shared savings program introduces a variety of innovative concepts that will guide Medicare providers and suppliers of services as they become ACOs. ACOs will be accountable for a designated patient population, but will continue to receive traditional fee-for-service Medicare payments under Parts A and B, and will be responsible for Part D services as well. Under this arrangement, beneficiaries will retain their freedom of choice of which providers to receive services from.
This regulatory framework will create new ways for ACO participants to acquire, handle, analyze, and integrate Medicare claims data. While the ACO proposed rule points to numerous uses of data and information (e.g., public reporting on quality measures and beneficiary surveys; collecting marketing information; using evidenced-based protocols; distributing clinical information for coordinating care; and providing information to beneficiaries on how ACO participation will affect them), the formation of ACOs creates a need for the gathering, processing, and interpreting Medicare beneficiary identifiable claims data.
The proposed rule indicates that both aggregate and person-level data will be made available to ACOs to “assist ACOs in the overall redesign of the care process and coordination of care in their assigned beneficiary populations.” We emphasize the unique aspects of person-level indentified data below.

What CMS Will Provide
CMS will provide a standard set of claims data to ACOs for the beneficiaries who received the majority of their care from ACO-affiliated primary care physicians. Thus, the information will be highly targeted to ACOs as the primary users.
Monthly CMS claims files will contain Parts A and B data, including procedure code, diagnosis code, beneficiary ID, date of birth, gender, date of death, service dates, provider or supplier IDs, and claims payment type. Part D data will include beneficiary ID, prescriber ID, drug service date, drug product data, and formulary indication. No substance abuse data will be available.
How ACOs Can Use Personally Identifiable Claims data
ACOs will use personal-level claims data for a variety of purposes:

  • Population-based health – promoting and tracking population-based care over time: The transition from treating individual patients to being responsible for a patient population will create a new perspective for many ACO providers/suppliers. Medicare claims data can be linked over time to form “patient pathways” (e.g. the sequence of physician, inpatient and outpatient hospital, emergency room, post-acute, and hospital readmission care). The availability of these types of data will play a pivotal role in informing new approaches to population-based health. For instance, ACOs will be able to track patient care and identify key beneficiary subgroups most in need of coordinated care. Tracking this information at the person and group levels will allow ACOs to determine performance on at least two levels: 1) ACO participating providers, and 2) performance on population-based measures.
  • Freedom of choice: ACOs will also need to track the care provided to their assigned beneficiary population, including non-ACO providers to determine total care expenditures. The only way to accomplish this type of tracking is with Medicare claims files.
  • Quality measurement and improvement: CMS is requiring ACOs to report data on 65 quality measures. Many of these can be recreated and tracked with CMS claims data.
  • Shared savings: ACOs will need to track per capita or per-member per-month (PMPM) expenditures against CMS expenditure targets (benchmarks) in order to determine if they will incur savings or losses. Furthermore, the Medicare statistics will need to be risk adjusted to match the CMS benchmark per capita expenditure targets. Identifiable beneficiary claims data will be an important component in developing PMPM expenditure estimates.

Considerations
In order to obtain these data, ACOs will need to sign a data use agreement (DUA) with CMS. This agreement primarily relates to uses of the data and data confidentiality. Under the DUA, the ACO will be a business associate (as defined by HIPAA) of their participants and suppliers. Individual beneficiaries will be able to opt out of the data gathering process, and their claims will not be shared with the ACO.
CMS believes that the use of these data “would promote coordinated care and a better understanding of the population served by the ACO with real-time positive impact – both quality and efficiency of care delivered.” While the use of person-level identifiable claims files will be most valuable to ACOs without fully developed information systems, these data will be helpful to all ACOs regardless of their current health information technology infrastructure.
Due to the freedom of choice clause, wherein each beneficiary can choose to use a non-ACO provider, all ACOs will need to develop ways to efficiently access and analyze the CMS claims files. The bundling of Part D data to the claims files is, as of yet, relatively uncharted territory. ACOs will need to produce a series of descriptive reports that would enable their management teams to track expenditure targets, subpopulations of care (e.g. diabetics), and ensure that the care they provide is coordinated, costs are managed, and at the same time quality of care is improved.