Federal policy options to realize the potential of APCDs

Fig 1
Editor's note:

This analysis is part of the USC-Brookings Schaeffer Initiative for Health Policy, which is a partnership between Economic Studies at Brookings and the University of Southern California Schaeffer Center for Health Policy & Economics. The Initiative aims to inform the national health care debate with rigorous, evidence-based analysis leading to practical recommendations using the collaborative strengths of USC and Brookings.

An all-payer claims database (APCD) is a system that collects health care claims and related data from all (or nearly all) entities that pay for health care services in a geographic area, including private and public health plans. Today, 23 states have APCDs, and they are valuable tools that virtually anyone with a stake in the health care system—including consumers, employers, health care providers, health insurers, researchers, and policymakers—can use to better understand the system and find ways to improve it. Indeed, in the states that have them, APCDs can provide a comprehensive picture of health care spending, health care delivery, and health insurance enrollment, and they offer insights that no other data can replicate, particularly with respect to private health insurance markets.

Unfortunately, today’s APCDs have important limitations. Most prominently, a 2016 Supreme Court decision, Gobeille v. Liberty Mutual Insurance Co., held that states may not require data collection from non-governmental self-insured group health plans. Because self-insured plans represent 61% of enrollment in employer coverage—and about one-third of all covered people—this decision left a large gap in state APCDs.

Current APCDs also have limitations that predate the Gobeille decision and arise from the fact that they are state entities that operate independently of one another. Because each state has its own data submission protocols and data access procedures, combining data from multiple states is challenging, which can frustrate efforts to compare experiences in different states. Further, because APCDs are state-controlled, federal policymakers have no automatic access to APCD data. Relying on a patchwork of state APCDs also forfeits potential economies of scale, which increases administrative costs for both payers and APCDs themselves and may prevent APCDs from making valuable investments in public reporting or data quality. Further, half the U.S. population lives in a state without an APCD.

Federal policymakers have multiple options to address the limitations of current APCDs:

  • Enable state collection of self-insured data: Federal policymakers can directly restore state APCDs’ ability to collect data from self-insured plans. Congress could enact simple legislation that would achieve this objective. Alternatively, the Supreme Court’s decision in Gobeille described a legal pathway by which the federal government could authorize states to collect these data on its behalf without new legislation. In either case, policymakers could choose to limit the ability to collect data from self-insured plans to state APCDs that meet certain conditions, like collecting data in a uniform format to reduce administrative burden on payers or providing adequate researcher access to the data.
  • Create a national APCD: The federal government could also build a national APCD that would collect data from all payers in all states. The federal government and outside researchers could then access this rich national dataset, subject to appropriate privacy safeguards. Data from a national APCD could also be shared with state APCDs, which might need to adapt their systems to accept it, but would then be freed from the burden of collecting data themselves and could focus their limited resources on high-value efforts to support broader use of the data.

In designing such a system, policymakers would need to pay careful attention to privacy and data security concerns. A national APCD could adopt security procedures like those that govern existing federal health care databases. It could also implement data access procedures similar to those that govern Medicare claims data today; in particular, it could bar users from disclosing anything other than aggregated results, require them to abide by rigorous data security practices, limit their access to only those data elements required for their planned analyses, and require them to access and analyze data in a secure computing environment controlled by the federal government. Policymakers could also bar an APCD from holding direct individual identifiers, although such restrictions would need to be crafted carefully to avoid reducing the APCD’s capabilities.

The federal government already has the authority to collect these data and could do so without new legislation, although such an initiative is more likely to be pursued and completed if Congress directs and funds the work. Policymakers could also consider housing this initiative within a non-profit, rather than a federal agency, as proposed in legislation recently considered in the Senate Health, Education, Labor, and Pensions Committee, although such an approach poses some governance concerns and would likely require new legislation.

  • Expand state APCD coverage and harmonize state APCDs: Policymakers could also pursue a hybrid approach that would expand, improve, and harmonize the existing network of state APCDs. In this model, policymakers would take three steps: (1) authorize state APCDs to collect self-insured data; (2) provide grants to states to encourage creation of new APCDs; and (3) require state APCDs to collect data through a standardized process and share their data with a federal “clearinghouse,” similar to the federal clearinghouse that currently exists for state hospital encounter data. The clearinghouse could support research projects that require data from multiple states and facilitate use of these data by federal policymakers, while the grants might encourage more states to create APCDs. However, some states would likely continue to lack APCDs under this approach, and each state APCD would still need to invest in its own infrastructure for collecting and maintaining claims data. The federal government likely already has the authority to pursue this approach, but as with creating a national APCD, such an initiative is more likely to be undertaken with Congressional support.

We believe that creating a national APCD is the best of these approaches. With a small investment of resources relative to total US health care spending, policymakers could create a tool that offers a comprehensive picture of the health care system that no existing data source can come close to matching, thereby accelerating efforts by a wide range of public and private actors to better understand and ultimately improve American health care. While efforts to expand the number of state APCDs and harmonize existing APCDs could also improve on the status quo (and the pre-Gobeille status quo) and might encounter somewhat less stakeholder opposition, such a project would achieve less than creating a truly national APCD, and we view it as a decidedly second-best alternative.

If the political will cannot be summoned to pursue these larger projects, either Congress or federal agencies should act swiftly to at least restore states’ ability to collect self-insured data. While not cost free, this is a fairly simple undertaking that will make existing APCDs more useful and comprehensive. It would allow state APCDs to maximize their own potential, hopefully building a constituency for creating a better national infrastructure over the longer-term.

Stakeholders might raise objections to the policy approaches we recommend here, but policymakers should not be deterred from moving forward. Payers might raise concerns about the burden of reporting data to APCDs, but since payers already provide similar data to many entities, including state APCDs, under the status quo, the incremental burden associated with these proposals would likely be modest. There is also an ongoing debate about whether disclosure of payers’ negotiated prices, including through an APCD, could put upward pressure on prices. While the evidence on this question is mixed, policymakers could prevent disclosure of negotiated prices if they wished, albeit not without reducing the utility of APCD data in important respects. Privacy and security concerns are also sometimes raised in response to options that involve the federal government holding claims data. However, APCDs do not appear to present any novel privacy or security issues relative to other government undertakings, and these concerns can be addressed using privacy and security standards similar to those that have proven successful in protecting other sensitive data, like Medicare claims data. We also note that some stakeholders sell data products that would become less valuable if APCD data became more robust and widely available, which may lead them to oppose APCD expansion, but this concern likely should not factor into policymakers’ decisions.

The full the paper is available here.

The authors thank Kathleen Hannick and Spoorthi Kamepalli for excellent research assistance and Brieanna Nicker for excellent editorial assistance.