Respecting Patient Choice in Attribution Methodologies: An Example from Medicare’s Comprehensive Primary Care Plus Model Fang He, PhD; Yan Tang, PhD; Kristen Henretty, BA; Chris Beadles, PhD, MD RTI International June 3, 2019 Funder: This research was funded by the Centers for Medicare and Medicaid Services under contract number HHSM-500-2014- 00037I / 75FCMC18F0001. The contents of this publication are those of the authors and do not necessarily reflect the views or policies of the Centers for Medicare and Medicaid Services. www.rti.org RTI International is a registered trademark and a trade name of Research Triangle Institute.
Outline ▪ Background ▪ Study Objectives ▪ Data ▪ Methodology ▪ Results ▪ Conclusion ▪ Policy Implications
Background: Alternative payment models ▪ Medicare and other payers are transitioning from fee-for-service to alternative payment models – Since 2010, the Centers for Medicare & Medicaid Services (CMS) has piloted more than 20 new payment models – By March 2016, Medicare had achieved its goal of making 30% of its payments for providers through alternative payment models – More broadly, 34% of total U.S. health care payments were tied to alternative payment models in 2017, an increase from 23% in 2015, according to the Health Care Payment Learning & Action Network
Background: Attribution ▪ A key feature in designing alternative payment models is the method of attributing patients to providers held accountable for their care ▪ Accurate payment and quality of care calculations hinge on timely, accurate attribution – Otherwise providers may be held responsible for patients they do not treat, which could discourage providers from pursuing population health management
Background: Using claims in attribution Advantages • Readily available source of information • Does not impose any burden on providers • Reliable, as claims are used to pay providers Disadvantages • May be difficult for patients and providers to understand • Potential misalignment between the provider to which the patient is attributed and the patient’s chosen provider • May exclude newly enrolled patients or patients with gaps in enrollment • Patients are not actively engaged in the attribution process
Background: Prioritizing patient choice in attribution ▪ CMS is now prioritizing a patient’s active choice of primary provider within attribution methodologies for several alternative payment models – Next Generation Accountable Care Organizations model (started 2017) – Medicare Shared Savings Program (started 2018) – Comprehensive Primary Care Plus (CPC+) Initiative (started 2019) ▪ Attribution via patient choice enhances beneficiary attribution
Background: How patients choose their primary provider
Background: CMS’s CPC+ Initiative ▪ The nation’s largest ever multi -payer initiative to improve primary care – Began January 2017 – 18 regions across the country – 2,879 primary care practices – 14,810 practitioners – ≈15 million patients, including more than 2 million Medicare patients ▪ Primary care practices are responsible for the care of attributed patients – From January 2017 through December 2018, based on attribution results, Medicare paid $1.12 billion to practices for ▪ Managing their patients’ care ▪ Meeting annual performance targets ▪ Lump sum prospective payments aimed at giving flexibility to practices beyond traditional fee-for-service payments
Background: Attribution in CPC+ The patient is not eligible for attribution The patient makes their STOP choice via MyMedicare.gov No Does the patient meet the CPC+ eligibility requirements? Yes Does the chosen provider meet the provider eligibility requirements? Yes No The patient is attributed to Claims-based the chosen provider attribution
Study Objectives ▪ How many patients have been attributed to CPC+ practices using patient choice? ▪ How do the patients’ choices compare with results from claims - based attribution? ▪ How do the characteristics of attributed patients making a provider choice compare to those attributed via traditional claims-based attribution?
Data ▪ CMS administrative data containing patients’ choices of provider ▪ Medicare claims and enrollment data – Including January 2017 – December 2018 physician and outpatient claims used in CPC+ claims-based attribution ▪ Data on characteristics of Medicare patients – 2017 – 2019 Master Beneficiary Summary File – 2010 – 2014 American Community Survey 5-year estimate
Methodology ▪ We used data on patients’ choices and Medicare claims and enrollment data to attribute patients to CPC+ practices – We examined the number of patients attributed to CPC+ practices using patient choice – We compared the patients’ choices and the results from the CPC+ claims - based attribution method – We compared the characteristics of patients attributed to CPC+ practices using patient choice and patients attributed using claims
Results: CPC+ patients attributed via patient choice increased from Q1 to Q2 in 2019 ▪ In 2019, the number of patients attributed to CPC+ practices using patient choice grew 62.6% from 2,334 in Quarter 1 to 3,796 in Quarter 2 ▪ The number of patients attributed to CPC+ practices using patient choice was small relative to the 2,006,982 patients attributed to CPC+ practices via claims-based attribution in 2019 Quarter 2 Number of Patients Attributed to CPC+ Practices using Patient Choice 4,000 3,500 3,000 2,500 2,000 1,500 1,000 500 - 2019 Quarter 1 2019 Quarter 2 Source: RTI International analysis of CMS administrative and enrollment data.
Results: CPC+ patients attributed via patient choice were concentrated in a few CPC+ practices, 2019 Quarter 2 Distribution of Patients Attributed to CPC+ Practices using Patient Choice by CPC+ Region, 2019 Quarter 2 Distribution of CPC+ Attributed Patients via Patient Choice, 2019 Quarter 2 700 579 600 Number of CPC+ Practices 500 5 CPC+ practices accounted 400 for 30% of patients attributed to CPC+ practices using 300 216 patient choice 200 130 100 42 5 0 1 Patient 2 Patients 3-5 Patients 6-99 Patients 100-335 Patients Source: RTI International analysis of CMS administrative and enrollment data.
Results: High agreement between patient choice and claims-based attribution in 2019 Quarter 2 Claims-Based Attribution Results, Among Patients Attributed to CPC+ Practices using Patient Choice 240, 6% 319, 9% Patients attributed to the same CPC+ practice Patients attributed to another CPC+ practice or to a non-CPC+ practice Patients not attributed by claims- based attribution 3,237, 85% Source: RTI International analysis of CMS administrative, claims, and enrollment data.
Results: CPC+ patients attributed via patient choice differed from claims-based attributed patients in 2019 Quarter 2 Difference: Patients’ Patients attributed using Patients choices minus claims-based attributed using claims Characteristics method patient choice (standard error) Total Number of Patients 2,006,982 3,796 -1.03*** Age 72.77 71.74 (0.17) 0.07*** Male 0.42 0.49 (0.01) -0.06*** White, non-Hispanic 0.87 0.81 (0.01) -0.04*** Eligibility for Full Medicaid Benefits 0.08 0.03 (0.00) Agency for Healthcare Research and 1.43*** Quality (AHRQ) Socioeconomic Status 54.38 55.81 (0.07) Index of Patient’s Zip Code Number of Chronic Conditions through -0.55*** 6.43 5.88 2018 (0.06) Medicare Payments for Physician Services $87.41*** $507.34 $594.75 in 2017 (8.95) -$1,173.35*** Total Medicare Payments in 2017 $8,220.81 $7,047.46 (314.18) Source: RTI International analysis of CMS administrative, claims, and enrollment data (including 2017-2019 Master Beneficiary Summary File) and 2010-2014 American Community Survey 5-year estimate. Note: *** p < 0.001
Conclusion ▪ The number of patients attributed to CPC+ practices using patient choice has increased ▪ Few patients were attributed to CPC+ practices using patient choice relative to number of patients attributed to CPC+ practices via claims-based attribution ▪ High agreement between patients’ choices and claims -based attribution method results, increasing our confidence in both methods ▪ Compared to patients attributed to CPC+ practices using claims, those attributed using patient choice were: – Younger – Fewer chronic conditions – More likely to be male – Higher Medicare payments and non-white for physician services – Higher socioeconomic status – Lower total Medicare payments
Policy Implications ▪ Alternative payment models could consider supplementing their existing claims-based attribution methods with a process to ask the patient to confirm their provider ▪ This simple to understand method may better engage providers and patients, and improve the accuracy of attribution ▪ Studies have shown that more engaged patients (which includes being able to select providers based on performance or quality) have better health outcomes and care experiences (Hibbard and Greene, 2013)
Recommend
More recommend
