[Skip to Content]
[Skip to Content Landing]
Figure.
Performance on Quality and Costs, 2013
Performance on Quality and Costs, 2013

Two quality domains (patient and family engagement and efficient use of health care resources) applied to fewer than 10 practices and thus were not included in the performance analyses. P values are for comparisons across groups. Higher z scores reflect better performance on quality; lower scores, better performance on costs. Error bars indicate 95% CIs.

Table 1.  
Patient, Clinician, and Physician Practice Characteristics, 2013
Patient, Clinician, and Physician Practice Characteristics, 2013
Table 2.  
Mean Practice Performance on Mandatory Quality and Cost Measures by Physician Practice Risk Categories, 2013
Mean Practice Performance on Mandatory Quality and Cost Measures by Physician Practice Risk Categories, 2013
Table 3.  
Actual Payments With Physician Value-Based Payment Modifier Program Payment Adjustments in 2015 (N = 899 Physician Practices)a
Actual Payments With Physician Value-Based Payment Modifier Program Payment Adjustments in 2015 (N = 899 Physician Practices)a
Table 4.  
Simulated Performance-Based Payments With Physician Value-Based Payment Modifier Program Payment Adjustments in 2015 (N = 899 Physician Practices)a
Simulated Performance-Based Payments With Physician Value-Based Payment Modifier Program Payment Adjustments in 2015 (N = 899 Physician Practices)a
1.
Coughlin  TA, Waidmann  TA, Phadera  L.  Among dual eligibles, identifying the highest-cost individuals could help in crafting more targeted and effective responses.  Health Aff (Millwood). 2012;31(5):1083-1091.PubMedGoogle ScholarCrossref
2.
Bennett  KJ, Probst  JC.  Thirty-day readmission rates among dual-eligible beneficiaries.  J Rural Health. 2016;32(2):188-195.PubMedGoogle ScholarCrossref
3.
Centers for Medicare & Medicaid Services. Detailed methodology for the 2013 Quality and Resource Use Reports and 2015 Value-Based Payment Modifier. https://www.cms.gov/medicare/medicare-fee-for-service-payment/physicianfeedbackprogram/downloads/2013-detailed-methodology.pdf. Accessed April 19, 2017.
4.
Min  LC, Wenger  NS, Fung  C,  et al.  Multimorbidity is associated with better quality of care among vulnerable elders.  Med Care. 2007;45(6):480-488.PubMedGoogle ScholarCrossref
5.
Higashi  T, Wenger  NS, Adams  JL,  et al.  Relationship between number of medical conditions and quality of care.  N Engl J Med. 2007;356(24):2496-2504.PubMedGoogle ScholarCrossref
6.
Bae  S, Rosenthal  MB.  Patients with multiple chronic conditions do not receive lower quality of preventive care.  J Gen Intern Med. 2008;23(12):1933-1939.PubMedGoogle ScholarCrossref
7.
Min  L, Kerr  EA, Blaum  CS, Reuben  D, Cigolle  C, Wenger  N.  Contrasting effects of geriatric versus general medical multimorbidity on quality of ambulatory care.  J Am Geriatr Soc. 2014;62(9):1714-1721.PubMedGoogle ScholarCrossref
8.
Streit  S, da Costa  BR, Bauer  DC,  et al.  Multimorbidity and quality of preventive care in Swiss university primary care cohorts.  PLoS One. 2014;9(4):e96142.PubMedGoogle ScholarCrossref
9.
Lehnert  T, Heider  D, Leicht  H,  et al.  Review: health care utilization and costs of elderly persons with multiple chronic conditions.  Med Care Res Rev. 2011;68(4):387-420.PubMedGoogle ScholarCrossref
10.
Wolff  JL, Starfield  B, Anderson  G.  Prevalence, expenditures, and complications of multiple chronic conditions in the elderly.  Arch Intern Med. 2002;162(20):2269-2276.PubMedGoogle ScholarCrossref
11.
Yoon  J, Zulman  D, Scott  JY, Maciejewski  ML.  Costs associated with multimorbidity among VA patients.  Med Care. 2014;52(suppl 3):S31-S36.PubMedGoogle ScholarCrossref
12.
Joynt  KE, Jha  AK.  Characteristics of hospitals receiving penalties under the Hospital Readmissions Reduction Program.  JAMA. 2013;309(4):342-343.PubMedGoogle ScholarCrossref
13.
Gilman  M, Adams  EK, Hockenberry  JM, Milstein  AS, Wilson  IB, Becker  ER.  Safety-net hospitals more likely than other hospitals to fare poorly under Medicare’s value-based purchasing.  Health Aff (Millwood). 2015;34(3):398-405.PubMedGoogle ScholarCrossref
14.
Gilman  M, Hockenberry  JM, Adams  EK, Milstein  AS, Wilson  IB, Becker  ER.  The financial effect of value-based purchasing and the Hospital Readmissions Reduction Program on safety-net hospitals in 2014: a cohort study.  Ann Intern Med. 2015;163(6):427-436.PubMedGoogle ScholarCrossref
15.
Medicare Payment Advisory Commission.  March 2015 Report to the Congress: Medicare Payment Policy, Chapter 13: The Medicare Advantage Program: Status Report. Washington, DC: Medicare Payment Advisory Commission; 2015.
17.
Allen  SM, Piette  ER, Mor  V.  The adverse consequences of unmet need among older persons living in the community: dual-eligible versus Medicare-only beneficiaries.  J Gerontol B Psychol Sci Soc Sci. 2014;69(suppl 1):S51-S58.PubMedGoogle ScholarCrossref
18.
Komisar  HL, Feder  J, Kasper  JD.  Unmet long-term care needs: an analysis of Medicare-Medicaid dual eligibles.  Inquiry. 2005;42(2):171-182.PubMedGoogle Scholar
19.
Trivedi  AN, Zaslavsky  AM, Schneider  EC, Ayanian  JZ.  Relationship between quality of care and racial disparities in Medicare health plans.  JAMA. 2006;296(16):1998-2004.PubMedGoogle ScholarCrossref
20.
Deswal  A, Petersen  NJ, Urbauer  DL, Wright  SM, Beyth  R.  Racial variations in quality of care and outcomes in an ambulatory heart failure cohort.  Am Heart J. 2006;152(2):348-354.PubMedGoogle ScholarCrossref
21.
Hausmann  LR, Gao  S, Mor  MK, Schaefer  JH  Jr, Fine  MJ.  Understanding racial and ethnic differences in patient experiences with outpatient health care in Veterans Affairs Medical Centers.  Med Care. 2013;51(6):532-539.PubMedGoogle ScholarCrossref
22.
Fiscella  K, Sanders  MR.  Racial and ethnic disparities in the quality of health care.  Annu Rev Public Health. 2016;37:375-394.PubMedGoogle ScholarCrossref
23.
Heisler  M, Smith  DM, Hayward  RA, Krein  SL, Kerr  EA.  Racial disparities in diabetes care processes, outcomes, and treatment intensity.  Med Care. 2003;41(11):1221-1232.PubMedGoogle ScholarCrossref
24.
Sequist  TD, Fitzmaurice  GM, Marshall  R, Shaykevich  S, Safran  DG, Ayanian  JZ.  Physician performance and racial disparities in diabetes mellitus care.  Arch Intern Med. 2008;168(11):1145-1151.PubMedGoogle ScholarCrossref
25.
Hayes  SL, Salzberg  CA, McCarthy  D,  et al.  High-Need, High-Cost Patients: Who Are They and How Do They Use Health Care—A Population-Based Comparison of Demographics, Health Care Use, and Expenditures. New York, NY: The Commonwealth Fund; 2016.
26.
Ryan  AM, Krinsky  S, Kontopantelis  E, Doran  T.  Long-term evidence for the effect of pay-for-performance in primary care on mortality in the UK: a population study.  Lancet. 2016;388(10041):268-274.PubMedGoogle ScholarCrossref
27.
Dale  SB, Ghosh  A, Peikes  DN,  et al.  Two-year costs and quality in the Comprehensive Primary Care Initiative.  N Engl J Med. 2016;374(24):2345-2356.PubMedGoogle ScholarCrossref
28.
Friedberg  MW, Schneider  EC, Rosenthal  MB, Volpp  KG, Werner  RM.  Association between participation in a multipayer medical home intervention and changes in quality, utilization, and costs of care.  JAMA. 2014;311(8):815-825.PubMedGoogle ScholarCrossref
29.
Bardach  NS, Wang  JJ, De Leon  SF,  et al.  Effect of pay-for-performance incentives on quality of care in small practices with electronic health records: a randomized trial.  JAMA. 2013;310(10):1051-1059.PubMedGoogle ScholarCrossref
30.
Petersen  LA, Simpson  K, Pietz  K,  et al.  Effects of individual physician-level and practice-level financial incentives on hypertension care: a randomized trial.  JAMA. 2013;310(10):1042-1050.PubMedGoogle ScholarCrossref
31.
Scott  A, Sivey  P, Ait Ouakrim  D,  et al.  The effect of financial incentives on the quality of health care provided by primary care physicians.  Cochrane Database Syst Rev. 2011;(9):CD008451.PubMedGoogle Scholar
32.
Campbell  SM, Reeves  D, Kontopantelis  E, Sibbald  B, Roland  M.  Effects of pay for performance on the quality of primary care in England.  N Engl J Med. 2009;361(4):368-378.PubMedGoogle ScholarCrossref
33.
Campbell  S, Reeves  D, Kontopantelis  E, Middleton  E, Sibbald  B, Roland  M.  Quality of primary care in England with the introduction of pay for performance.  N Engl J Med. 2007;357(2):181-190.PubMedGoogle ScholarCrossref
34.
Serumaga  B, Ross-Degnan  D, Avery  AJ,  et al.  Effect of pay for performance on the management and outcomes of hypertension in the United Kingdom: interrupted time series study.  BMJ. 2011;342:d108.PubMedGoogle ScholarCrossref
35.
Tao  W, Agerholm  J, Burström  B.  The impact of reimbursement systems on equity in access and quality of primary care: a systematic literature review.  BMC Health Serv Res. 2016;16(1):542.PubMedGoogle ScholarCrossref
36.
Alshamsan  R, Majeed  A, Ashworth  M, Car  J, Millett  C.  Impact of pay for performance on inequalities in health care: systematic review.  J Health Serv Res Policy. 2010;15(3):178-184.PubMedGoogle ScholarCrossref
37.
Centers for Medicare & Medicaid Services. 2015 Value-Based Payment Modifier Program experience report. June 16, 2015. https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/PhysicianFeedbackProgram/Downloads/2015-VM-Program-Experience-Rpt.pdf. Accessed January 16, 2017.
38.
Centers for Medicare & Medicaid Services (CMS), HHS.  Medicare Program; Merit-based Incentive Payment System (MIPS) and Alternative Payment Model (APM) incentive under the physician fee schedule, and criteria for physician-focused payment models: final rule with comment period.  Fed Regist. 2016;81(214):77008-77831.PubMedGoogle Scholar
39.
Joynt  KE, De Lew  N, Sheingold  SH, Conway  PH, Goodrich  K, Epstein  AM.  Should Medicare Value-Based Purchasing take social risk into account?  N Engl J Med. 2017;376(6):510-513.PubMedGoogle ScholarCrossref
Original Investigation
August 1, 2017

Association of Practice-Level Social and Medical Risk With Performance in the Medicare Physician Value-Based Payment Modifier Program

Author Affiliations
  • 1Division of General Medicine, Department of Internal Medicine, University of Michigan, Ann Arbor
  • 2Institute for Healthcare Policy and Innovation, University of Michigan, Ann Arbor
  • 3Center for Healthcare Outcomes and Policy, University of Michigan, Ann Arbor
  • 4Office of the Assistant Secretary for Planning and Evaluation, US Department of Health and Human Services, Washington, DC
  • 5Department of Medicine, Brigham and Women’s Hospital, Boston, Massachusetts
  • 6Department of Health Policy and Management, Harvard T. H. Chan School of Public Health, Boston, Massachusetts
  • 7Atrius Health, Newton, Massachusetts
  • 8Now with Commonwealth Medicine, University of Massachusetts Medical School, Shrewsbury
  • 9Now with Washington University School of Medicine, St Louis, Missouri
JAMA. 2017;318(5):453-461. doi:10.1001/jama.2017.9643
Key Points

Question  Was there an association between the social or medical risk of patients treated at physician practices and performance during the first year of the Medicare Physician Value-Based Payment Modifier Program?

Findings  Practices that served more socially high-risk patients had lower quality and lower costs, and practices that served more medically high-risk patients had lower quality and higher costs. These patterns were associated with fewer bonuses and more penalties for high-risk practices.

Meaning  As value-based payment programs continue to increase in size and scope, practices that disproportionately serve high-risk patients may be at particular risk of receiving financial penalties.

Abstract

Importance  Medicare recently launched the Physician Value-Based Payment Modifier (PVBM) Program, a mandatory pay-for-performance program for physician practices. Little is known about performance by practices that serve socially or medically high-risk patients.

Objective  To compare performance in the PVBM Program by practice characteristics.

Design, Setting, and Participants  Cross-sectional observational study using PVBM Program data for payments made in 2015 based on performance of large US physician practices caring for fee-for-service Medicare beneficiaries in 2013.

Exposures  High social risk (defined as practices in the top quartile of proportion of patients dually eligible for Medicare and Medicaid) and high medical risk (defined as practices in the top quartile of mean Hierarchical Condition Category risk score among fee-for-service beneficiaries).

Main Outcomes and Measures  Quality and cost z scores based on a composite of individual measures. Higher z scores reflect better performance on quality; lower scores, better performance on costs.

Results  Among 899 physician practices with 5 189 880 beneficiaries, 547 practices were categorized as low risk (neither high social nor high medical risk) (mean, 7909 beneficiaries; mean, 320 clinicians), 128 were high medical risk only (mean, 3675 beneficiaries; mean, 370 clinicians), 102 were high social risk only (mean, 1635 beneficiaries; mean, 284 clinicians), and 122 were high medical and social risk (mean, 1858 beneficiaries; mean, 269 clinicians). Practices categorized as low risk performed the best on the composite quality score (z score, 0.18 [95% CI, 0.09 to 0.28]) compared with each of the practices categorized as high risk (high medical risk only: z score, −0.55 [95% CI, −0.77 to −0.32]; high social risk only: z score, −0.86 [95% CI, −1.17 to −0.54]; and high medical and social risk: −0.78 [95% CI, −1.04 to −0.51]) (P < .001 across groups). Practices categorized as high social risk only performed the best on the composite cost score (z score, −0.52 [95% CI, −0.71 to −0.33]), low risk had the next best cost score (z score, −0.18 [95% CI, −0.25 to −0.10]), then high medical and social risk (z score, 0.40 [95% CI, 0.23 to 0.57]), and then high medical risk only (z score, 0.82 [95% CI, 0.65 to 0.99]) (P < .001 across groups). Total per capita costs were $9506 for practices categorized as low risk, $13 683 for high medical risk only, $8214 for high social risk only, and $11 692 for high medical and social risk. These patterns were associated with fewer bonuses and more penalties for high-risk practices.

Conclusions and Relevance  During the first year of the Medicare Physician Value-Based Payment Modifier Program, physician practices that served more socially high-risk patients had lower quality and lower costs, and practices that served more medically high-risk patients had lower quality and higher costs.

Introduction

Ambulatory pay-for-performance programs provide incentives for physician practices to improve the care they deliver. The Medicare Physician Value-Based Payment Modifier (PVBM) Program, which launched in 2015, will be the largest mandatory pay-for-performance program for physicians when fully phased in. Under this program, physician practices receive penalties or bonuses (from −1% to 10% of Medicare payments in 2015) based on the quality and costs of care. The PVBM Program serves as a precursor to the Medicare Quality Payment Program, which will launch in 2019, apply to clinicians and practices, and measure performance across a broader array of metrics. Clinicians eligible in the PVBM Program include physicians, nurse practitioners, and physician assistants.

Despite the growth of ambulatory pay-for-performance programs, there is concern about unintended consequences, including disproportionately penalizing practices that care for complex patients. Prior studies have shown that patients with high levels of medical risk as well as patients with social risk factors, such as those dually enrolled in Medicare and Medicaid, have worse quality outcomes.1,2 Thus, it is possible that physician practices that care for these high-risk populations will fare poorly in pay-for-performance programs. The PVBM Program has instituted safeguards such as risk adjustment, eliminating measures with extreme performance values, and using conservative performance thresholds for bonuses and penalties to mitigate potential risks faced by participating practices.

However, there are no prior studies of the PVBM Program and little is known about performance patterns. Therefore, this study sought to answer 3 questions. What are the patient, practice, and clinician characteristics of large physician practices that serve a disproportionate share of medically or socially high-risk patients? How did large practices perform on quality and cost performance metrics included in the PVBM Program? What implications did any performance differences have on payment?

Methods
Data Sources and Study Population

We performed a cross-sectional observational study using PVBM Program data for payments made in 2015 based on performance of large US physician practices caring for fee-for-service Medicare beneficiaries in 2013. To characterize physician practices, we used the Provider Enrollment, Chain, and Ownership System, PVBM Program data, and the Medicare Electronic Health Record Incentive Program Eligible Professionals Public Use File.

Performance metrics are based on a core set of claims-based mandatory measures as well as more than 200 elective measures from a variety of sources (eg, registries, claims, medical record–based data submitted to the Centers for Medicare & Medicaid Services [CMS] via a web interface). Policy research at the US Department of Health and Human Services that uses secondary, administrative, and deidentified data does not require approval by an institutional review board or informed consent.

The study sample consisted of physician practices that were eligible for the PVBM Program during its first year (ie, practices that were not participating in the Medicare Shared Savings Program or the Comprehensive Primary Care Initiative and had ≥100 physicians or other clinicians), and had at least 1 attributed fee-for-service Medicare beneficiary. These data were linked to Medicare claims from 2013 to identify patients and physicians. Fee-for-service Medicare beneficiaries were attributed to practices using the PVBM algorithm, which is based on the setting where a beneficiary received the plurality of primary care services.3

Primary Exposures

Following PVBM Program parameters,3 physician practices with a mean Hierarchical Condition Category (HCC) risk score greater than the 75th percentile among all fee-for-service beneficiaries were categorized as practices with high medical risk (eFigure 1 in the Supplement). Similarly, physician practices in the top quartile of the proportion of attributed beneficiaries who were dually eligible for Medicare and Medicaid (defined as enrollment in Medicaid in January 2013) were categorized as practices with high social risk (eFigure 2 in the Supplement). From these 2 primary exposures, we created 4 mutually exclusive groups: (1) low risk (neither high social nor high medical risk); (2) high medical risk only; (3) high social risk only; and (4) high medical and social risk.

Primary Outcome Measures

The primary outcome measures were the PVBM Program quality composite score and cost composite score.3 Each practice’s composite scores are based on a combination of mandatory and elective measures of their choosing. The quality score is reported as a z score and is based on performance across 6 domains: clinical process and effectiveness, patient and family engagement, population and public health, patient safety, care coordination, and efficient use of health care resources (eAppendix and eTable 1 in the Supplement). For a given domain or measure, a z score was created by taking the practice’s performance and subtracting the average performance of comparison practices. This difference was then divided by the measure’s standard deviation among all comparison practices, producing a z score. For a given measure, the comparison practices were the subset of practices with data for that measure. The cost score is also reported as a z score and is composed of the 2 domains of total per capita costs for all attributed beneficiaries and per capita costs for beneficiaries with specific conditions, which were averaged to create a cost composite for each practice.

The secondary outcome measures for quality were individual metrics, including the following mandatory, claims-based quality measures used in the PVBM Program: (1) all-cause readmissions (risk-adjustment model included age and a number of clinical comorbidities as specified by the CMS), (2) admissions for acute ambulatory care–sensitive conditions (adjusted for age and sex), and (3) admissions for chronic ambulatory care–sensitive conditions (adjusted for age and sex).

The secondary outcome measures for costs were the following mandatory, claims-based metrics calculated by the CMS for the PVBM Program: (1) total per capita costs and per capita costs for (2) heart failure, (3) diabetes mellitus, (4) chronic obstructive pulmonary disease, and (5) coronary artery disease. The CMS payment standardizes the per capita cost measures to account for geographic differences in prices, and includes risk adjustment to account for patient case mix. The risk-adjustment model controls for HCC risk score, HCC risk score squared, and end-stage renal disease. The HCC risk score includes 70 categories corresponding to specific groups of International Classification of Diseases, Ninth Revision, diagnosis codes and age, sex, Medicaid, and original reason for Medicare qualification (ie, age or disability). Measure performance was reported as a z score and as an absolute value (eg, $1000).

The other secondary outcomes were payment adjustments (ie, performance-based bonus, no adjustment, penalty for not reporting, performance-based penalty). These were determined by following program parameters from the CMS. Physician practices that failed to successfully register for the program and report a minimum number of measures received an automatic penalty for not reporting (nonparticipation).

Practices were categorized as high or low quality (or high or low cost) only if their z score for quality (or cost) was more than 1 SD from the peer group mean and the difference was statistically significant. Physician practices that elected to tie performance to payment received a penalty if they had (1) low quality and high or average cost or (2) average quality and high cost. Physician practices received an upward payment adjustment if they had (1) high quality and low or average cost or (2) average quality and low cost (eTable 2 in the Supplement). All other practices received no adjustment.

Statistical Analysis

The characteristics of the 4 types of physician practices were compared using χ2 tests. To compare differences in performance on the outcomes listed earlier, we used quality and cost metrics calculated by the CMS and z scores for domain and composite scores. Using the CMS’ performance metrics, we created unadjusted ordinary least-squares regression models with practice type as the primary predictor. For the practices that did not have a sufficient number of cases for individual measures, we did not impute missing values. This mirrors the PVBM Program’s approach.

We also described how payments differed across practice types. In our first set of analyses, we reported on actual bonuses and penalties, only assigning performance-based penalties to those practices that elected this option. In a set of exploratory analyses, we used performance data to simulate bonuses and penalties for all participating practices because performance-based payment is mandatory starting in 2016 for large physician practices. Ordered logit regression models were used to test for statistically significant differences in payments between the 4 physician practice risk categories.

We used Stata statistical software version 13 (StataCorp) and SAS version 9.4 (SAS Institute Inc). Two-sided P < .01 was considered statistically significant to account for multiple testing.

Results
Patient, Physician, and Practice Characteristics

Among 899 physician practices with 5 189 880 attributed beneficiaries, 547 practices were categorized as low risk (neither high social risk nor high medical risk) (mean, 7909 beneficiaries; mean, 320 clinicians), 128 were high medical risk only (mean, 3675 beneficiaries; mean, 370 clinicians), 102 were high social risk only (mean, 1635 beneficiaries; mean, 284 clinicians), and 122 were high medical and social risk (mean, 1858 beneficiaries; mean, 269 clinicians) (Table 1). Although 79.2% of low-risk practices successfully registered and reported data to the program (thus avoiding the automatic reporting penalty), this was done in only 69.5% of practices with high medical risk only, 47.1% of practices with high social risk only, and 54.1% of practices with high medical and social risk.

High-risk practices served patient populations that were generally younger and more often of ethnic minority, dually enrolled in Medicare and Medicaid, and disabled (Table 1). For example, 8% of patients treated at low-risk practices were black, 16% at practices with high medical risk only, 24% at practices with high social risk only, and 31% at practices with both high medical and social risk (P < .001). There were few major differences in physician specialty or practice size across practice types (Table 1). However, stage 1 meaningful use (defined as the adoption of certified electronic health records across a number of dimensions as specified by the CMS) was higher in practices categorized as low risk (31%) and high medical risk only (29%) compared with practices categorized as high social risk only (8%) and high medical and social risk (14%) (P < .001).

Performance on Domain and Composite Scores

Practices categorized as low risk performed the best on the composite quality score (z score, 0.18 [95% CI, 0.09 to 0.28]) compared with each of the practices categorized as high risk (high medical risk only: z score, −0.55 [95% CI, −0.77 to −0.32]; high social risk only: z score, −0.86 [95% CI, −1.17 to −0.54]; and high medical and social risk: z score, −0.78 [95% CI, −1.04 to −0.51]) (part A in the Figure and eTable 3 in the Supplement) (P < .001 across groups). Patterns were generally similar for each of the quality domains, with low-risk practices outperforming the higher-risk practice groups on every domain with the exception of population and public health.

However, patterns for costs differed. On the composite cost score, practices categorized as high social risk only performed best (z score, −0.52 [95% CI, −0.71 to −0.33]). Low-risk practices (z score, −0.18 [95% CI, −0.25 to −0.10]) performed next best, whereas practices categorized as high medical risk only (z score, 0.82 [95% CI, 0.65 to 0.99]) and high medical and social risk (z score, 0.40 [95% CI, 0.23 to 0.57]) had the highest costs (part B in the Figure and eTable 3 in the Supplement) (P < .001 across groups). Patterns were similar for both cost domains.

Performance on PVBM Program Quality and Cost Measures

Risk-adjusted readmission rates were lowest at practices categorized as low risk (15.3%) compared with practices categorized as high medical risk only (16.3%; difference from low risk, 1.01% [95% CI, 0.79%-1.23%]; P < .001), high social risk only (15.9%; difference from low risk, 0.58% [95% CI, 0.31%-0.85%]; P < .001), and high medical and social risk (16.6%; difference from low risk, 1.32% [95% CI, 1.08%-1.55%]; P < .001; Table 2). Compared with low-risk practices, admissions for acute ambulatory care–sensitive conditions were higher in the practices categorized as high medical risk only and high medical and social risk practices. In contrast, admissions for chronic ambulatory care–sensitive conditions were similar across groups.

The total per capita costs were $9506 for practices categorized as low risk, $13 683 for high medical risk only (difference from low risk, $4177 [95% CI, $3437 to $4917]; P < .001), $8214 for high social risk only (difference from low risk, −$1292 [95% CI, −$2105 to −$480]; P = .002), and $11 692 for high medical and social risk (difference from low risk, $2186 [95% CI, $1432 to $2940]; P < .001) (Table 2). Patterns were similar for the condition-specific cost measures.

Penalties and Bonuses

Only 112 physician practices opted to receive performance-based payment during the first year of the PVBM Program (when performance-based bonuses and penalties were optional). Thus, actual penalties were largely driven by failure to register and report data (Table 3 and eTables 4 and 5 in the Supplement). Such penalties were more common in practices categorized as high social risk only (20.8% of low-risk practices were penalized for failure to register and report data, 30.5% of high medical risk only, 52.9% of high social risk only, and 45.9% of high medical and social risk). In simulations in which performance-based bonuses and penalties were applied to all practices with sufficient data, practices categorized as high medical risk only or high medical and social risk had a higher likelihood of receiving a performance-based penalty (3.7% of low-risk practices were penalized for poor performance, 18.0% of high medical risk only, 9.8% of high social risk only, and 13.1% of high medical and social risk) (Table 4 and eTables 6 and 7 in the Supplement).

Discussion

In this analysis of the first year of Medicare’s new physician pay-for-performance program based on 899 practices with 5 189 880 attributed Medicare beneficiaries, performance patterns differed by physician practice type. Compared with practices categorized as low risk, practices with high medical risk only and with high medical and social risk had lower quality and higher costs, whereas practices with high social risk only had lower quality but lower costs. Because of a high frequency of failure to register and report data as well as these differences in performance, physician practices that served a disproportionate share of medically and socially high-risk patients were more likely to receive a penalty compared with other practices.

The largest driver of penalties during the first year of the PVBM Program was failure to successfully register and report. Even though some percentage of practices failing to participate may reflect an active choice, some may reflect a lack of infrastructure or technology that makes reporting more difficult, particularly among high-risk practices that may lack access to electronic health records and other supporting factors. This may be an important area for technical support as the PVBM Program expands to include a broader range of practices and as penalties for failure to participate grow with the advent of the Merit-Based Incentive Payment System (MIPS) in 2019.

Medical risk was associated with both worse quality and higher costs. The mechanism underlying these relationships is unclear. The findings from prior studies have been mixed, with some studies finding that patients with multiple chronic conditions are more likely to receive high-quality care,4-6 and others finding that the relationship between multimorbidity and quality is neutral or depends on the type of comorbidity present.7,8 Research is less mixed for costs; many prior studies show that patients with more severe illness use more medical care and require more resources,9-11 which is the basis for risk adjustment of costs under this and other programs that assess clinicians’ performance on costs of care. Without patient-level analyses, it is difficult to determine whether these higher costs are driven by inadequate risk adjustment or by a high-intensity practice style, and this warrants future research.

The finding that practices categorized as high social risk only performed significantly worse on quality metrics is similar to previously described patterns in other care settings also subject to differential payment based on quality, such as safety-net hospitals under Medicare’s value-based payment programs12-14 and Medicare Advantage plans with high proportions of patients dually enrolled in Medicare and Medicaid under the Quality Star Rating Program.15,16 Patients treated at practices categorized as high social risk only may face basic challenges such as transportation, food, housing, and security, which are unmeasured in Medicare claims but may be associated with outcomes. In some cases, fewer resources may also make it more difficult for practices who serve these patients to attract qualified clinicians.

With regard to costs of care, practices categorized as high social risk only in the PVBM Program were less expensive than practices that served a lower-risk population. These findings may in part be explained by the PVBM Program’s risk-adjustment model, which accounts for medical risk by modeling cost as a function of HCC risk score and includes Medicaid enrollment status as a covariate. Alternatively, practices categorized as high social risk only may treat both high- and low-risk patients in a less expensive style. Previous studies have documented significant unmet health care needs for poor patients,17,18 which may make these patients less expensive to treat than predicted by risk models.

This study extends prior research describing disparities in care for both socially and medically complex patients.19-25 Research on ambulatory pay-for-performance programs (including certain patient-centered medical home models and the UK’s Quality and Outcomes Framework) has evaluated the effects on quality and use26-34 or disparities in care,35,36 but not the effects of such programs on clinicians who serve vulnerable populations. The CMS report on the first year of the PVBM Program did not examine the performance of all PVBM Program–eligible practices, nor did it describe practice performance by social risk.37

Better understanding of the disparities in ambulatory programs may become increasingly important under the MIPS, which is modeled after parts of the PVBM Program and replaces it.38 Findings from this study suggest that if current performance patterns persist, practices that serve a high proportion of socially or medically complex patients may fare poorly under the MIPS. There are a number of options to address disparities, including development of health equity measures.39 This latter option would be similar to the extra bonus provided to practices that perform well in the PVBM Program and serve medically complex patients.

Limitations

The study has several limitations. First, data for this analysis were from the first year of the PVBM Program when only large practices were eligible to participate in the program and performance-based payment was optional. Although these were the most recent data available, patterns of performance may change as a broader range of practices are included and as the program continues to evolve. Second, dual enrollment in Medicare and Medicaid was used as a marker of social risk. Because Medicaid provides additional benefits, dual enrollment may underestimate the true effect of caring for patients with high social risk absent any support from government programs. Third, this study examined Medicare patients only. Studies examining other dimensions of social risk or other populations might reveal different patterns.

Conclusions

During the first year of the Medicare Physician Value-Based Payment Modifier Program, physician practices that served more socially high-risk patients had lower quality and lower costs, and practices that served more medically high-risk patients had lower quality and higher costs.

Back to top
Article Information

Corresponding Author: Lena M. Chen, MD, MS, University of Michigan Division of General Medicine, North Campus Research Complex, 2800 Plymouth Rd, Bldg 16, Room 407E, Ann Arbor, MI 48109 (lenac@umich.edu).

Accepted for Publication: July 3, 2017.

Author Contributions: Drs Chen and Joynt Maddox had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Chen, Epstein, Filice, Samson, Joynt Maddox.

Acquisition, analysis, or interpretation of data: Chen, Epstein, Orav, Filice, Joynt Maddox.

Drafting of the manuscript: Chen, Joynt Maddox.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Chen, Orav, Filice.

Administrative, technical, or material support: Epstein, Filice, Samson.

Supervision: Epstein, Joynt Maddox.

Conflict of Interest Disclosures: The authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Chen reported receiving an honorarium from the National Institutes of Health; grants from Blue Cross Blue Shield of Michigan Foundation, the American Heart Association, and the University of Michigan MCubed Program; being part of the Michigan Value Collaborative with Blue Cross Blue Shield of Michigan; and serving as an advisor at the US Department of Health and Human Services. Dr Joynt Maddox reported serving as an advisor at the US Department of Health and Human Services while work on this study was ongoing; and currently being a contractor for the US Department of Health and Human Services. No other disclosures were reported.

Funding/Support: This study was supported by the US Department of Health and Human Services. Dr Chen is currently supported by grant P01 AG019783 from the National Institute on Aging and grant R01 HS024698 from the Agency for Healthcare Research and Quality. Dr Joynt Maddox is supported by Career Development Grant Award K23-HL109177-03 from the National Heart, Lung, and Blood Institute.

Role of the Funder/Sponsor: The US Department of Health and Human Services had a role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The views expressed in this article are those of the authors and do not necessarily reflect the official views of the US Department of Health and Human Services. Dr Joynt Maddox, a JAMA Associate Editor, was not involved in the review of or decision to accept this article.

References
1.
Coughlin  TA, Waidmann  TA, Phadera  L.  Among dual eligibles, identifying the highest-cost individuals could help in crafting more targeted and effective responses.  Health Aff (Millwood). 2012;31(5):1083-1091.PubMedGoogle ScholarCrossref
2.
Bennett  KJ, Probst  JC.  Thirty-day readmission rates among dual-eligible beneficiaries.  J Rural Health. 2016;32(2):188-195.PubMedGoogle ScholarCrossref
3.
Centers for Medicare & Medicaid Services. Detailed methodology for the 2013 Quality and Resource Use Reports and 2015 Value-Based Payment Modifier. https://www.cms.gov/medicare/medicare-fee-for-service-payment/physicianfeedbackprogram/downloads/2013-detailed-methodology.pdf. Accessed April 19, 2017.
4.
Min  LC, Wenger  NS, Fung  C,  et al.  Multimorbidity is associated with better quality of care among vulnerable elders.  Med Care. 2007;45(6):480-488.PubMedGoogle ScholarCrossref
5.
Higashi  T, Wenger  NS, Adams  JL,  et al.  Relationship between number of medical conditions and quality of care.  N Engl J Med. 2007;356(24):2496-2504.PubMedGoogle ScholarCrossref
6.
Bae  S, Rosenthal  MB.  Patients with multiple chronic conditions do not receive lower quality of preventive care.  J Gen Intern Med. 2008;23(12):1933-1939.PubMedGoogle ScholarCrossref
7.
Min  L, Kerr  EA, Blaum  CS, Reuben  D, Cigolle  C, Wenger  N.  Contrasting effects of geriatric versus general medical multimorbidity on quality of ambulatory care.  J Am Geriatr Soc. 2014;62(9):1714-1721.PubMedGoogle ScholarCrossref
8.
Streit  S, da Costa  BR, Bauer  DC,  et al.  Multimorbidity and quality of preventive care in Swiss university primary care cohorts.  PLoS One. 2014;9(4):e96142.PubMedGoogle ScholarCrossref
9.
Lehnert  T, Heider  D, Leicht  H,  et al.  Review: health care utilization and costs of elderly persons with multiple chronic conditions.  Med Care Res Rev. 2011;68(4):387-420.PubMedGoogle ScholarCrossref
10.
Wolff  JL, Starfield  B, Anderson  G.  Prevalence, expenditures, and complications of multiple chronic conditions in the elderly.  Arch Intern Med. 2002;162(20):2269-2276.PubMedGoogle ScholarCrossref
11.
Yoon  J, Zulman  D, Scott  JY, Maciejewski  ML.  Costs associated with multimorbidity among VA patients.  Med Care. 2014;52(suppl 3):S31-S36.PubMedGoogle ScholarCrossref
12.
Joynt  KE, Jha  AK.  Characteristics of hospitals receiving penalties under the Hospital Readmissions Reduction Program.  JAMA. 2013;309(4):342-343.PubMedGoogle ScholarCrossref
13.
Gilman  M, Adams  EK, Hockenberry  JM, Milstein  AS, Wilson  IB, Becker  ER.  Safety-net hospitals more likely than other hospitals to fare poorly under Medicare’s value-based purchasing.  Health Aff (Millwood). 2015;34(3):398-405.PubMedGoogle ScholarCrossref
14.
Gilman  M, Hockenberry  JM, Adams  EK, Milstein  AS, Wilson  IB, Becker  ER.  The financial effect of value-based purchasing and the Hospital Readmissions Reduction Program on safety-net hospitals in 2014: a cohort study.  Ann Intern Med. 2015;163(6):427-436.PubMedGoogle ScholarCrossref
15.
Medicare Payment Advisory Commission.  March 2015 Report to the Congress: Medicare Payment Policy, Chapter 13: The Medicare Advantage Program: Status Report. Washington, DC: Medicare Payment Advisory Commission; 2015.
17.
Allen  SM, Piette  ER, Mor  V.  The adverse consequences of unmet need among older persons living in the community: dual-eligible versus Medicare-only beneficiaries.  J Gerontol B Psychol Sci Soc Sci. 2014;69(suppl 1):S51-S58.PubMedGoogle ScholarCrossref
18.
Komisar  HL, Feder  J, Kasper  JD.  Unmet long-term care needs: an analysis of Medicare-Medicaid dual eligibles.  Inquiry. 2005;42(2):171-182.PubMedGoogle Scholar
19.
Trivedi  AN, Zaslavsky  AM, Schneider  EC, Ayanian  JZ.  Relationship between quality of care and racial disparities in Medicare health plans.  JAMA. 2006;296(16):1998-2004.PubMedGoogle ScholarCrossref
20.
Deswal  A, Petersen  NJ, Urbauer  DL, Wright  SM, Beyth  R.  Racial variations in quality of care and outcomes in an ambulatory heart failure cohort.  Am Heart J. 2006;152(2):348-354.PubMedGoogle ScholarCrossref
21.
Hausmann  LR, Gao  S, Mor  MK, Schaefer  JH  Jr, Fine  MJ.  Understanding racial and ethnic differences in patient experiences with outpatient health care in Veterans Affairs Medical Centers.  Med Care. 2013;51(6):532-539.PubMedGoogle ScholarCrossref
22.
Fiscella  K, Sanders  MR.  Racial and ethnic disparities in the quality of health care.  Annu Rev Public Health. 2016;37:375-394.PubMedGoogle ScholarCrossref
23.
Heisler  M, Smith  DM, Hayward  RA, Krein  SL, Kerr  EA.  Racial disparities in diabetes care processes, outcomes, and treatment intensity.  Med Care. 2003;41(11):1221-1232.PubMedGoogle ScholarCrossref
24.
Sequist  TD, Fitzmaurice  GM, Marshall  R, Shaykevich  S, Safran  DG, Ayanian  JZ.  Physician performance and racial disparities in diabetes mellitus care.  Arch Intern Med. 2008;168(11):1145-1151.PubMedGoogle ScholarCrossref
25.
Hayes  SL, Salzberg  CA, McCarthy  D,  et al.  High-Need, High-Cost Patients: Who Are They and How Do They Use Health Care—A Population-Based Comparison of Demographics, Health Care Use, and Expenditures. New York, NY: The Commonwealth Fund; 2016.
26.
Ryan  AM, Krinsky  S, Kontopantelis  E, Doran  T.  Long-term evidence for the effect of pay-for-performance in primary care on mortality in the UK: a population study.  Lancet. 2016;388(10041):268-274.PubMedGoogle ScholarCrossref
27.
Dale  SB, Ghosh  A, Peikes  DN,  et al.  Two-year costs and quality in the Comprehensive Primary Care Initiative.  N Engl J Med. 2016;374(24):2345-2356.PubMedGoogle ScholarCrossref
28.
Friedberg  MW, Schneider  EC, Rosenthal  MB, Volpp  KG, Werner  RM.  Association between participation in a multipayer medical home intervention and changes in quality, utilization, and costs of care.  JAMA. 2014;311(8):815-825.PubMedGoogle ScholarCrossref
29.
Bardach  NS, Wang  JJ, De Leon  SF,  et al.  Effect of pay-for-performance incentives on quality of care in small practices with electronic health records: a randomized trial.  JAMA. 2013;310(10):1051-1059.PubMedGoogle ScholarCrossref
30.
Petersen  LA, Simpson  K, Pietz  K,  et al.  Effects of individual physician-level and practice-level financial incentives on hypertension care: a randomized trial.  JAMA. 2013;310(10):1042-1050.PubMedGoogle ScholarCrossref
31.
Scott  A, Sivey  P, Ait Ouakrim  D,  et al.  The effect of financial incentives on the quality of health care provided by primary care physicians.  Cochrane Database Syst Rev. 2011;(9):CD008451.PubMedGoogle Scholar
32.
Campbell  SM, Reeves  D, Kontopantelis  E, Sibbald  B, Roland  M.  Effects of pay for performance on the quality of primary care in England.  N Engl J Med. 2009;361(4):368-378.PubMedGoogle ScholarCrossref
33.
Campbell  S, Reeves  D, Kontopantelis  E, Middleton  E, Sibbald  B, Roland  M.  Quality of primary care in England with the introduction of pay for performance.  N Engl J Med. 2007;357(2):181-190.PubMedGoogle ScholarCrossref
34.
Serumaga  B, Ross-Degnan  D, Avery  AJ,  et al.  Effect of pay for performance on the management and outcomes of hypertension in the United Kingdom: interrupted time series study.  BMJ. 2011;342:d108.PubMedGoogle ScholarCrossref
35.
Tao  W, Agerholm  J, Burström  B.  The impact of reimbursement systems on equity in access and quality of primary care: a systematic literature review.  BMC Health Serv Res. 2016;16(1):542.PubMedGoogle ScholarCrossref
36.
Alshamsan  R, Majeed  A, Ashworth  M, Car  J, Millett  C.  Impact of pay for performance on inequalities in health care: systematic review.  J Health Serv Res Policy. 2010;15(3):178-184.PubMedGoogle ScholarCrossref
37.
Centers for Medicare & Medicaid Services. 2015 Value-Based Payment Modifier Program experience report. June 16, 2015. https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/PhysicianFeedbackProgram/Downloads/2015-VM-Program-Experience-Rpt.pdf. Accessed January 16, 2017.
38.
Centers for Medicare & Medicaid Services (CMS), HHS.  Medicare Program; Merit-based Incentive Payment System (MIPS) and Alternative Payment Model (APM) incentive under the physician fee schedule, and criteria for physician-focused payment models: final rule with comment period.  Fed Regist. 2016;81(214):77008-77831.PubMedGoogle Scholar
39.
Joynt  KE, De Lew  N, Sheingold  SH, Conway  PH, Goodrich  K, Epstein  AM.  Should Medicare Value-Based Purchasing take social risk into account?  N Engl J Med. 2017;376(6):510-513.PubMedGoogle ScholarCrossref
×