Provider Attribution & CMS-0057-F

Summary: An 11% return on investment (ROI) is achievable through Fee-for-Service (FFS) alternative payment models (APMs), provided that both provider-attribution and results-attribution models undergo testing and integration into clinical workflows. Attribution modeling entails continual reassessment of Value-Based Payment (VBP) terms and rules in light of evolving treatment delivery practices, while also considering patients' clinical and social risk factors. This evaluation is crucial in determining the beneficiaries and potential drawbacks of APMs.

Fortunately, the forthcoming CMS-0057 exchange regulations will facilitate the breakdown of healthcare silos, enabling collaboration between payers and providers. New use cases will emerge, leveraging historical and real-time data to address prevention and efficiency objectives effectively.

Alternative Payment Model ROI

An 11% return on investment (ROI) is feasible through Fee-for-Service (FFS) alternative payment models, also known as APMs (Song, 2019). The Council for Affordable Quality Healthcare (2019) outlines the four most common Medicaid Value-Based Payment (VBP) approaches. It is noteworthy that all Medicaid attribution methods are approved by the Centers for Medicare & Medicaid Services (CMS). As part of its efforts to enhance cost containment, the federal government aims to have 50% of State Medicaid programs operating on value-based payment (VBP) contracts by 2025. Achieving this goal necessitates the implementation of provider-to-patient attribution systems and advanced modeling techniques to identify treatment relationships and attribute measurement results, such as HEDIS quality outcomes.

Definitions and context

In a managed care capitated contract, patients have the option to select a Primary Care Provider (PCP), failing which they are automatically assigned a provider based on factors such as the provider's acceptance of new patients and proximity to the patient's address. Conversely, in a fee-for-service (FFS) arrangement, patients are typically not obligated to designate a PCP. However, in FFS value-based payment programs, patient selection of a provider becomes essential for accountability purposes. In cases where patient self-selection is not feasible, payers employ algorithms to identify the most likely PCP based on various factors such as claims data, encounters, and prescribing patterns.

Attribution, as defined by the National Quality Forum (2018), is a methodology used to assign patients, encounters, or episodes of care to healthcare providers or practitioners for accountability and remuneration purposes. In FFS, payers assess patient self-selection of providers, integrating claims data such as Evaluation & Management Healthcare Common Procedure Coding System (HCPCS) codes for annual well visits and preventive care. For instance, Geisinger (2022) employs eligible provider National Provider Identifier (NPI) specialty taxonomy codes and HCPCS procedure codes to identify engaged providers associated with a patient. If a patient does not have these visits, payers may consider case management and care coordination codes from nurse practitioners and other providers. Subsequently, payers may examine prescribing provider claims or claims from specialty providers (e.g., internists, family medicine practitioners, pediatricians, geriatricians). These rules collectively constitute visit-based attribution (Harker and Olsen, 2018). Other types of attribution include episode-based or bundling attribution. In cases of conflicting claims, the tiebreaker is usually the provider with the most recent visit. The conceptual model depicted below illustrates the various components involved in designing an attribution model.

There isn't a universally accepted standard for designing an attribution model; however, guidance is available from reputable sources such as the National Quality Forum (NQF, 2017) and joint resources from the American Health Insurance Plans (AHIP), American Medical Association (AMA), and the National Association of ACOs (NAACOS) (AHIP, AMA, NAACOS, 2024). For a comprehensive understanding of attribution model design, consulting these documents is recommended.

It's important to note that no single attribution model can fulfill all the requirements of both providers and payers. Rather, the specific design elements of an attribution model dictate the beneficiaries and potential drawbacks within Value-Based Payment (VBP) programs. Research conducted by Ryan, Linden, Maurer, Werner, and Nallamothu (2016) identified 171 known attribution models, with 89% being retrospective and 6% prospective. Of these models, 45.6% focus on all care, while 39.2% concentrate on episodes. Additionally, 77.8% of attribution approaches are mapped to a single provider, whereas 19.3% involve multiple providers.

Constant testing and re-modeling

Over the span of several years, research indicates that most attribution models demonstrate an accuracy rate ranging from 20% to 69% (McCoy et al., 2018). Some attribution designs identified by McCoy et al. tend to favor younger, healthier patients over older, sicker ones, posing challenges for providers unless continuous testing and remodeling are undertaken. Given that attribution is primarily based on claims data, patients who are attributed typically exhibit higher utilization rates compared to younger, healthier patients with minimal or no claims history, effectively attributing only those who are unwell (Wong & Fox, 2020). Consequently, attribution experts recommend ongoing testing to accommodate evolving patient and provider experiences (AHIP, AMA, NAACOS, 2024; National Quality Forum, 2017).

Design choices in attribution significantly influence the outcomes of Value-Based Payment (VBP) programs (National Quality Forum, 2018). Research by Higuera and Carlin (2017) reveals that patient leakage—where patients receive care from non-attributed providers—ranges from 40% to 60%, indicating deviations from the anticipated care patterns outlined by payers with attributed Fee-for-Service (FFS) contracted providers. In cases involving complex patients seeking care from multiple providers for various conditions, it's imperative to assess whether a Primary Care Provider (PCP) can effectively influence outcomes in a fragmented delivery system, known as the locus of control. Moreover, under attribution methodologies, full credit is typically assigned to a provider or group, rather than partial credit, even when multiple physicians contribute to complex treatment (Cantor, 2020), prompting proposals for weighted multi-attribution models (WMAM).

While most clinical risk adjustment methods do not account for social risk factors, vulnerable and marginalized communities often have limited access to healthcare, resulting in minimal claims experience for risk adjustment. To address this gap, payers can utilize tools such as the Area Deprivation Index (ADI) or Social Determinants of Health (SDoH) as proxies for assessing the risk of adverse events (Zhu et al., 2022). For instance, factors like homelessness, disability, and foster status are well-known risk indicators for patients with chronic conditions.

Attribution exchange (CMS-0057-F)

The federal government recently enacted a new interoperability regulation aimed at enhancing healthcare data accessibility (Department of Health and Human Services, 2024). This regulation mandates the creation of patient access APIs, provider access APIs, payer-to-payer APIs, and prior authorization APIs. In-network providers with established treatment relationships will gain access to payer claims, encounters, and USCDI clinical datasets. The USCDI encompasses a wide array of clinical data, including clinical notes, ADT encounter information, health status assessments, medication adherence records, problems, vital signs, demographics, and more (Office of the National Coordinator for Health IT, 2024).

Payers are tasked with maintaining an attribution process that links patients to in-network or enrolled providers with whom they share a treatment relationship. Payers have the flexibility to define treatment relationships using various criteria and methodologies, such as claims history, patient election, planned appointments, enrollment status, or ACO contracts. While the federal guidelines do not dictate specific attribution implementation methods, payers are encouraged to design attribution treatment relationships prospectively. In instances where new patients lack claims history, it falls upon the payer to establish attribution criteria. Additionally, payers are not obligated to share patient data concerning out-of-network or unenrolled providers, nor are they required to disclose data pertaining to unattributed patients.

CMS-0057 Prior Authorization Metrics

The recent sub-regulation on prior authorization facilitates the exchange of information between providers and payers regarding covered services, documentation requirements, approval and denial statuses, turnaround time metrics, and rationales for decisions. CMS mandates that payers annually report prior authorization metrics on their websites. These regulations apply to various Medicaid and CHIP programs, including Medicaid FFS, Medicaid MCA, CHIP FFS, and CHIP MCA. Organizations operating under Medicaid and CHIP FFS are obligated to report at the state level, while those under Medicaid managed care are required to report at the plan level. Compliance reporting is set to commence on January 1, 2026.

Summary

To extract value from a Value-Based Payment (VBP) program, the meticulous design of attribution rules is paramount. This necessitates continual testing against evolving patient and provider datasets, while also considering clinical and social risk factors that determine the program's success or failure. Fortunately, the forthcoming CMS-0057 exchange regulations will facilitate collaboration between payers and providers, breaking down existing silos.

Emerging applications will leverage historical and real-time data to achieve preventive and efficiency objectives. However, challenges remain in utilizing attribution mapping to address bottlenecks and pain points in key operational areas such as portals, call centers, and core processing functions like enrollment, provider contracting, prior authorization, prescription management, and coordination of benefits. Payers must also dynamically reassess clinical and financial models upon integrating historical and real-time USCDI data. The potential of CMS-0057 lies in its ability to dismantle treatment silos and curtail unnecessary expenditures, paving the way for more streamlined and effective healthcare delivery.

Sources

AHIP, AMA, and NAACOS. (2024). Creating a sustainable future for value-based care: a playbook for VBC payment arrangements. America’s Health Insurance Plans. Downloaded from https://www.ahip.org/news/press-releases/ahip-the-american-medical-association-and-the-national-association-of-acos-release-playbook-with-recommendations-toward-a-sustainable-future-for-value-based-care-payment

Cantor, M. (2020). Modernizing medical attribution. Journal of General Internal Medicine, 35, 12, 3691-3693. ?Downloaded from https://pubmed.ncbi.nlm.nih.gov/32323134/

Council for Affordable Quality Healthcare (2019). CAQH core webinar state Medicaid experiences with value-based payments with Center for Health Care Strategies and Minnesota Medicaid. CAQH . Downloaded from https://www.caqh.org/events/state-medicaid-experiences-value-based-payments-center-health-care-strategies-and-minnesota

Department of Health and Human Services. (2024). CMS-0057-F rule in Federal Register Volume 89, Number 27, February 8, 2024, pages 8758-8988. DHHS. Downloaded from https://www.govinfo.gov/content/pkg/FR-2024-02-08/pdf/2024-00895.pdf

Geisinger (2022). Primary care attribution guide with eligible attribution taxonomy codes and eligible attribution procedure codes. Geisinger Health System. Downloaded from https://www.geisinger.org/-/media/onegeisinger/files/pdfs/provider/navinet/education%20and%20resources/amp_attribution_infosheet_0122#:~:text=Geisinger%20Health%20Plan%20uses%20Milliman's,with%20the%20plurality%20of%20care .

Harker, M., Olson, A. (2018). Patient attribution: the basis for all value-based care. Society of Actuaries. Downloaded from https://www.soa.org/globalassets/assets/files/resources/research-report/2018/patient-attribution.pdf

Higuera, L., and Carlin, C. (2017). A comparison of retrospective attribution rules. American Journal of Managed Care, 23, 6, e180-e185. Downloaded from https://www.ajmc.com/view/a-comparison-of-retrospective-attribution-rules

McCoy, R.G., Bunkers, K.S., Ramar, P., Meier, S.K., Benetti, L.L., Nesse, R.E., and Naessens, J.M. (2018). Patient attribution: why the methods matter. American Journal of Managed Care, 24, 12, 596-603. Downloaded from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6549236/

National Quality Forum (2017). Accurately assigning accountability for patient care: NQF’s guide to selecting an attribution model. National Quality Forum. Downloaded from https://www.qualityforum.org/Attribution_2015-2016.aspx

National Quality Forum (2018). Improving attribution models, final report. National Quality Forum. Downloaded from https://www.qualityforum.org/Improving_Attribution_Models.aspx

Office of the National Coordinator for Health IT (2024). United States Core Data for Interoperability (USCDI) version 5, summary of data classes and data elements. Office of the National Coordinator for Health IT. Downloaded from https://www.healthit.gov/isa/sites/isa/files/2024-01/Draft-USCDI-Version-5-January-2024-Final.pdf

Ryan, A., Linden, A., Maurer, K., Werner, R., and Nallamothu. (2016). Attribution methods and implications for measuring performance in healthcare. National Quality Forum. Downloaded from https://www.qualityforum.org/Projects/a-b/Attribution_2015-2016/Commissioned_Paper.aspx and located in National Quality Forum (2016). Attribution: principles and approaches, final report. National Quality Forum. Downloaded from https://www.qualityforum.org/Publications/2016/12/Attribution_-_Principles_and_Approaches.aspx

Song, Z., Ji, Y., Safran, D., and Chernew, M. (2019). Health care spending, utilization and quality 8 years into global payment. The New England Journal of Medicine, 381, 3, 252-255. https://www.nejm.org/doi/full/10.1056/nejmsa1813621

Wong, C. and Fox, W. (2020). Milliman commercial physician attribution research. Milliman MedInsight.

Zhu, M., Japinga, M., Saunders, R., and McClellan, M. (2022). The future of risk adjustment: supporting equitable, comprehensive healthcare. Duke Margolis Center for Health Policy. Downloaded from https://healthpolicy.duke.edu/publications/future-risk-adjustment-supporting-equitable-comprehensive-health-care

?