Evaluation of the Merit-Based Incentive Payment System and Surgeons Caring for Patients at High Social Risk

Key Points

Question

Is the Merit-Based Incentive Payment System (MIPS) disproportionately penalizing surgeons serving patients at high social risk?

Findings

In this cohort study of 10 252 general surgeons who participated in the first year of MIPS, surgeons with patients in the highest quintile of social risk had a caseload with at least 37% of Medicare patients dual eligible for Medicare and Medicaid. Caring for patients at high social risk was associated with the lowest MIPS scores and a significantly increased risk of a negative payment adjustment.

Meaning

In this study, surgeons caring for patients at highest social risk received lower MIPS scores and had an increased risk of negative payment adjustment, despite ongoing efforts to target surgical disparities.

Abstract

Importance

The latest step in the Centers for Medicaid & Medicare transformation to pay-for-value is the Medicare Merit-Based Incentive Payment System (MIPS). Value-based payment designs often do not account for uncaptured clinical status and social determinants of health in patients at high social risk, and the consequences for clinicians and patients associated with their use have not been explored.

Objective

To evaluate MIPS scoring of surgeons caring for patients at high social risk to determine whether this implementation threatens disadvantaged patients’ access to surgical care.

Design, Setting, and Participants

A retrospective cohort study of US general surgeons participating in MIPS during its first year in outpatient surgical practices across the US and territories. The study was conducted from September 1, 2020, to May 1, 2021. Data were analyzed from November 1, 2020, to March 30, 2021 (although data were collected during the 2017 calendar year and reported ahead of 2019 payment adjustments).

Main Outcomes and Measures

Characteristics of surgeons participating in MIPS, overall MIPS score assigned to clinician. and practice-level disadvantage measures. The MIPS scores can range from 0 to 100. For the first year, a score of less than 3 led to negative payment adjustment; a score of greater than 3 but less than 70 to a positive adjustment; and a score of 70 or higher to the exceptional performance bonus.

Results

Of 20 593 general surgeons, 10 252 participated in the first year of MIPS. Surgeons with complete patient data (n = 9867) were evaluated and a wide range of dual-eligible patient caseloads from 0% to 96% (mean [SD], 27.1% [14.5%]) was identified. Surgeons in the highest quintile of dual eligibility cared for a Medicare patient caseload ranging from 37% to 96% dual eligible for Medicare and Medicaid. Surgeons caring for the patients at highest social risk had the lowest final mean (SD) MIPS score compared with the surgeons caring for the patients at least social risk (66.8 [37.3] vs 71.2 [35.9]; P < .001).

Conclusions and Relevance

Results of this cohort study suggest that implementation of MIPS value-based care reimbursement without adjustment for caseload of patients at high social risk may penalize surgeons who care for patients at highest social risk.

This cohort study evaluates the Merit-Based Incentive Payment System scoring of surgeons caring for patients at high social risk to assess whether this implementation is associated with patient access to surgical care.

Introduction

Teaching hospitals, safety-net hospitals, and large hospitals are most likely to receive penalties under the Medicare Hospital Readmissions Reduction Program.¹ However, research suggests that adjusting for patients with low incomes would decrease the penalty for half of the analyzed hospitals.² This situation will persist in value-based payment initiatives if appropriate risk adjustment is not incorporated. After implementation of the US Centers for Medicare & Medicaid Services (CMS) value-based program, the Merit-Based Payment System (MIPS), a national survey of internal medicine physicians found that only 8% of physicians were familiar with its measures and incentives. On clarification of the metrics, 60% of respondents felt that MIPS would encourage physicians to avoid certain disadvantaged patients to improve metrics.³

To date, few studies focus on the value-based care systems and surgeons and surgical patients.^4,5 Implementation of value-based care measures in the absence of adjustment for social risk has the potential to unfairly penalize surgeons serving populations at high social risk. Of 510 000 clinicians participating in the first year of MIPS implementation, those with patient caseloads in the highest quintile of social risk had a lower MIPS performance score than those in the lowest social risk quintile.⁶ The MIPS scores can range from 0 to 100. For the first year, a score of less than 3 led to negative payment adjustment; a score of greater than 3 but less than 70 to a positive adjustment; and a score of 70 or higher to the exceptional performance bonus. Most of these clinicians were in primary care because many of the initial performance measures were based in primary care. However, there are CMS initiatives under way to develop additional specialty-specific, value-based measures for the future.⁷

This study assesses the surgeon and practice characteristics of surgeons participating in the first year of outpatient MIPS implementation based on Medicare data. Social risk of the patients seeing each surgeon was estimated by the percentage of Medicare patients who are dual eligible for Medicaid and Medicare. The dual-eligible designation has been used as a proxy for patient social risk but has not been explored in surgeons participating in MIPS.⁶

Methods

Data Source

We identified practicing general surgeons by primary specialty in the Physicians Compare National Downloadable File.⁸ National Provider Identifier numbers were used to link data with MIPS participation using the 2017 Clinician Public Reporting Overall MIPS Performance file.⁹ Patient and practice characteristics including dual-eligible patient caseload and practice location were merged from the 2017 Medicare Physician and Other Supplier National Provider Identifier Aggregate Report.¹⁰ Surgeons who did not participate in MIPS and those with insufficient caseload data were excluded. In addition, we converted 9-digit practice zip codes to corresponding Area Deprivation Index (ADI) using Neighborhood Atlas state files.^11,12 The study was conducted from September 1, 2020, to May 1, 2021. Data were analyzed from November 1, 2020, to March 30, 2021 (although data were collected during the 2017 calendar year and reported ahead of 2019 payment adjustments). This study was submitted to the institutional review board at the University of Michigan and received "not regulated" determination as the University of Michigan policy for publicly available data sets is that institutional review board approval is not required. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies.¹³

Outcomes of Interest

Using the merged data set, we summarized participating surgeon demographic data including sex, time since medical school graduation, and ADI of practice location. We also summarized overall Medicare patient caseload characteristics including age and comorbidities associated with surgical intervention outcome. For the first year of participation, we investigated geographic distribution of participating surgeons by state as well as the neighborhood disadvantage of the practices. Maps were created using publicly available Stata code.¹⁴ We then summarized the characteristics of the Medicare patients cared for by participating surgeons. Codes for surgical procedures performed were not available, but important surgical comorbidities (chronic kidney disease, diabetes, and cancer) and Hierarchical Condition Categories (HCC) risk score for the surgeon’s caseload were used as markers of patient clinical status. CMS uses HCC scores for risk adjustment of Medicare payments.¹⁵ Categories of related International Classification of Diseases, Ninth Revision, Clinical Modification diagnosis codes are grouped into categories prioritizing the most severe illness within that group.¹⁶

Statistical Analysis

Surgeons’ social risk caseloads were categorized for analysis by the creation of quintiles of dual-eligible caseload following the approach of Johnston et al.⁶ Surgeons in the highest quintile of dual eligibility were defined as highest social risk surgeons and the lowest quintile defined as lowest social risk. We report the mean (SD) MIPS scores of all participating surgeons to facilitate comparison to published summary statistics for other participating clinicians.¹⁷ Comparisons between the demographic data and outcomes of the highest social risk (fifth quintile) and lowest social risk (first quintile) were performed with χ² and t tests. We estimated the association between high social risk and MIPS score with ordinal least squares multivariable regression. To determine likelihood of payment adjustment based on social risk, we used multivariable logistic regression. Score thresholds for the first year of payment adjustment were used to determine payment consequence of scores.¹⁸ All regression models adjusted for patient age, health status, and practice location. MIPS quality subscores are not publicly reported, so targeted outcome score adjustment was not feasible. Results were considered statistically significant at a 2-sided P < .05. We performed all data merging and analyses in Stata, version 16 (StataCorp).

Results

In the initial year of MIPS payment adjustment, there were 10 252 participating general surgeons. We evaluated the MIPS scores and patient caseloads of surgeons with complete records with the Centers for Medicare & Medicaid public files (n = 9867). Dual-eligible patient caseloads ranged from 0% to 96% (mean [SD], 27.1% [14.5%]). Of participating surgeons, 8226 of 9867 (83%) were men and the mean (SD) time since graduation from medical school was 24.2 (10.6) years (Table 1). The patient caseloads of these surgeons varied, with a mean (SD) of dual-eligible Medicare patients of 27.1% (14.5%). The surgeons in the highest social risk quintile had Medicare caseloads that ranged from 37% to 96% dual-eligible patients. The mean (SD) patient HCC risk score was 2.0 (0.9).

Table 1. Profile of Participating Surgeons, Practice Caseloads, and MIPS Scores.

Variable	Total, mean (SD)
No.	9867
Surgeon sex, No. (%)
Female	1641 (16.6)
Male	8226 (83.4)
Time since graduation from medical school, y	24.2 (10.6)
Dual eligible caseload, %	27.1 (14.5)
Practice ADI	50.5 (27.2)
Patient age, y	71.0 (3.7)
Beneficiaries, %
With cancer	20.5 (12.7)
With chronic kidney disease	43.1 (13.9)
With diabetes	38.7 (10.5)
Mean HCC risk score of beneficiaries	2.0 (0.9)
Quality Category score	70.3 (37.1)
Advancing Care Information score	73.5 (38.9)
Improvement Activities Category score	32.4 (15.3)
Final MIPS Score	72.0 (35.4)

Abbreviations: ADI, Area Deprivation Index; HCC, Hierarchical Condition Categories; MIPS, Merit-Based Incentive Payment System.

The dual eligibility level cut off for the fifth quintile of dual eligibility started at a caseload of 37% dual-eligible patients. This quintile was defined as “highest social risk” with the lowest dual-eligible quintile (0% to 15%) defined as “lowest social risk.” Patients in the highest social risk group were younger and sicker than the lowest social risk cohort with a mean (SD) age of 68.4 (5.2) years vs 73.1 (1.9) years and a mean (SD) HCC risk score of 2.7 (1.4) vs 1.4 (0.4). There were also differences in clinician demographic characteristics (Table 2). Geographic distribution of participating surgeons is depicted in the Figure, with the color of each state determined by the total number of surgeons participating. Five states (Alaska, Hawaii, North Dakota, Vermont, Rhode Island) and the District of Columbia had fewer than 32 surgeons participating in MIPS. Eight surgeons in Guam and Puerto Rico are not represented on this map.

Table 2. Comparison Between Lowest and Highest Social Risk Caseload Quintiles.

Variable	Mean (SD)		P value
	Lowest social risk (n = 1974)	Highest social risk (n = 1973)
Clinician sex, No. (%)
Female	432 (21.9)	303 (15.4)	<.001
Male	1542 (78.1)	1670 (84.6)
Time since graduation from medical school, y	26.8 (9.6)	24.4 (11.6)	<.001
Dual eligible caseload, %	10.5 (3.0)	49.6 (11.5)	<.001
Area Deprivation Index	40.4 (26.2)	52.3 (29.4)	<.001
Mean patient age, y	73.1 (1.9)	68.4 (5.2)	<.001
Beneficiaries, %
With cancer	28.8 (18.3)	15.0 (6.5)	<.001
With chronic kidney disease	32.0 (10.5)	53.0 (13.8)	<.001
With diabetes	30.3 (7.5)	48.7 (10.7)	<.001
Mean HCC risk score of beneficiaries	1.4 (0.4)	2.7 (1.4)	<.001
Quality Category score	69.7 (37.9)	64.4 (38.9)	<.001
Advancing Care Information score	70.4 (40.4)	71.1 (40.5)	.62
Improvement Activities Category score	32.1 (15.7)	30.9 (16.2)	.03
Final MIPS score	71.2 (35.9)	66.8 (37.3)	<.001

Abbreviations: HCC, Hierarchical Condition Categories; MIPS, Merit-Based Incentive Payment System.

Figure. Geographic Distribution of Participating Surgeons.

Maps were created using publicly available Stata code.¹⁴ The polygon visual style was chosen from maptiles code for ease of reference.

Surgeons caring for the patients at highest social risk had the lowest final mean (SD) MIPS scores compared with surgeons caring for the patients at lowest social risk (66.8% [37.3] vs 71.2% [35.9]; P < .001). Subscores differed between the groups for quality category (lowest social risk: 69.7% [37.9] vs highest social risk: 64.4% [38.9]; P < .001) and improvement activities category (lowest social risk: 32.1% [15.7] vs highest social risk: 30.9% [16.2]; P = .03) but did not differ for advancing care information (lowest social risk: 70.4% [40.4] vs highest social risk: 71.1% [40.5]; P = .62) (Table 2). There was no difference in MIPS score based on practice location within the highest social risk quintile. Surgeons caring for patients at high social risk in the most disadvantaged neighborhoods received the same mean (SD) score as their counterparts treating similar patients in other neighborhoods (66.2 [36.6] vs 66.1 [37.9]; P > .05). Only 404 of 1527 surgeons with the highest patient social risk caseload had practice locations in the most disadvantaged neighborhoods.

In adjusted multivariable regression, a decrease in MIPS score for increasing social risk (dual eligibility) was found. In logistic regression, high dual-eligible caseload (fifth quintile) was associated with an increased likelihood of negative payment adjustment (odds ratio, 1.76; 95% CI, 1.29-2.42). There was no significant difference in likelihood of positive payment adjustment for the highest social risk group There was a significant decreased likelihood of receiving an exceptional performance bonus for this group (Table 3). Discrete financial data for these adjustments were not captured. Payment adjustments represent financial consequences at the level of the outpatient surgical practice in the first year of implementation.

Table 3. Association of Highest vs Lowest Social Risk Caseload With MIPS Score and Likelihood of Payment Adjustments.

2019 Highest social risk caseloads MIPS performance payment	Value (95% CI)	P value
Final score	β, −9.33 (−12.59 to −6.07)	<.001
Exceptional performance bonus	OR, 0.64 (0.53 to 0.77)	<.001
Positive payment adjustment	OR, 1.04 (0.75 to 1.45)	.80
Negative payment adjustment	OR, 1.76 (1.29 to 2.42)	<.001

Abbreviations: MIPS, Merit-Based Incentive Payment System; OR, odds ratio.

Discussion

Consistent with overall outpatient clinician MIPS scores, surgeons caring for patients in the highest social risk quintile receive the lowest overall MIPS score.^4,6 Quality of care has been a focus of research for patients at social risk seeking care in safety-net hospitals.^19,20,21 However, the unintended effect of value-based payment systems may complicate protecting and improving the quality of care in safety-net settings. Value-based payment programs penalize hospitals providing needed surgical care to the patients at highest social risk.^22,23 MIPS applies value-based payment at the clinician level to individuals and practices. With two-thirds of surgical procedures occurring as outpatient surgeries, MIPS also puts a focus on value and quality for surgeons practicing in the outpatient setting.²⁴ It is important to understand the interaction between medical complexity, unmeasured social need, and MIPS scores to avoid unintended harm to surgical practices. Surgeons should be aware of these policy implications to advocate for their patients and their profession.

The unadjusted implementation of these value-based care programs risks deterring surgeons from treating patients at social risk because of factors beyond the patient or clinician’s control. These programs as implemented are intended to focus on high-value care but may deprive patients at social risk of necessary care by compromising the surgical safety net.²⁵ This study shows that concerns raised for the negative consequences associated with MIPS payment adjustments to clinicians should be shared by surgeons.^4,6 Surgeons often focus on outcomes data, which contribute to MIPS quality scores but are not publicly reported. The use of pay-for-performance to drive successful surgical outcome improvement has not been demonstrated.²⁶ A clinician’s MIPS performance is determined by weighted quality, cost, interoperability, and improvement activities categories. The weight of each component of MIPS scoring is subject to policy change. Quality measures comprised 60% of the MIPS final score for 2017 but will determine 40% of the final MIPS score for 2021.²⁷ There are 15 specialty-specific quality measures for general surgery in the 2021 performance year. These include advanced care planning, anastomotic leak, sentinel lymph node biopsy, preventative care (body mass index, tobacco, high-blood pressure), and perioperative antibiotics and venous thromboembolism prophylaxis.²⁸ The measures are undergoing revision each year by CMS with input from stakeholders, though only orthopedic clinicians are gaining a risk-adjusted measure for joint replacements.^7,28 Risk adjustment by dual-eligible patient caseload required by the 21st Century Cures Act has resulted in decreased penalties for safety-net hospitals under the Hospital Readmissions Reduction Program.²⁹ However, that adjustment has not been applied to clinician-level measures. In addition, dual eligibility alone may not reflect the complex interaction between social risk factors and access and quality measures given geographic variation.

MIPS scores among the surgeons treating the patients at highest social risk did not change by neighborhood disadvantage. This finding suggests that in this patient population, practice neighborhood setting was not associated with performance metrics. This finding may be owing to underlying similarities in patient catchment despite different practice setting, or may suggest practice setting neighborhood resources do not affect the value of care delivered to these patients at social risk. Practice settings are important to capture because this method of clinician-level value-based care can capture and affect outpatient surgery. Access to outpatient surgical care in an ambulatory surgical center may be associated with race, Medicaid, and rural location. This inequity was identified in patient data collected from 2011 to 2013, before the launch of MIPS.³⁰ However, the findings of this study suggest that value-based disincentives for operating on a high caseload of dual-eligible patients in the form of negative payment adjustment. The first step in preventing practice relocation or patient selection bias is adjusting value-based payments for socioeconomic risk.²⁵

Ongoing policy changes to the weighting of scores without established adjustments for social risk is not the solution for surgical disparities. We will need to distinguish between outcomes from substandard care or those that reflect the challenges of providing equitable care to populations at social risk. Patient, clinician, and hospital factors contribute to these disparities.³¹ Care delivery for these dual-eligible patients is expensive and a focus of redesign efforts.³² However, negative payment adjustments risk perpetuating inequalities in surgical care rather than driving needed change. These concerns were summarized in the report and recommendations of the National Academies’ Committee on Accounting for Socioeconomic Status in Medicare Payment Programs following passage of the Improving Medicare Post-Acute Care Treatment Act which required consideration of social risk in Medicare payments.³³ The most recent report to Congress from the Department of Health and Human Services recognized that addressing social risk will extend beyond risk adjustment to targeting support for clinicians caring for high social risk patients. However, the Department of Health and Human Services is clear to state that any implementation with occur gradually as the equity measures and adjustment approaches have not yet been developed.³⁴ Future development of these equity measures is another opportunity for surgeons to become involved in the process as key stakeholders.

Limitations

This clinician-level study does not account for specifics of underlying patient disease and surgical services provided. Surgeons were classified as general surgeons in the Physicians Compare National Downloadable File without further description of areas of specialization or referral patterns. These data cannot differentiate between preoperative consultations, outpatient operative interventions, or postoperative visits among a surgeon’s patient caseload. As this is a current CMS payment policy, we review the dual-eligibility caseloads but cannot extrapolate value-based care policy consequences for non-Medicare patients with high social needs. Dual-eligible Medicare patients are more likely to die of any cause, be hospitalized, or die as a result of their hospitalization than their Medicare-only peers.³⁵ Dual eligibility is used by CMS as a marker of the patients at highest social risk who receive Medicare but assumes the same dual eligibility across practices and hospitals. It cannot capture the nuances between medical complexity and unmeasured social needs.³⁶ We use ADI determined by practice zip code but recognize that this use of ADI may reflect differences in outpatient practice location selection more than patient catchment area. As patient level geographic data are unavailable, we use the practice location as a surrogate for the neighborhood socioeconomic need. ADI as a measure of socioeconomic need has previously been applied to surgical risk adjustment and burn severity risk prediction.^37,38

Conclusions

Practices located in the most disadvantaged neighborhoods have the potential to mitigate inequities in access. This also applies to the outpatient surgical centers in which many of these surgeons are operating. However, the initial MIPS data do not include data on surgical setting. Beginning in 2021, CMS will use Medicare Claims data to determine if a clinician is hospital based or ambulatory surgery center based (a dichotomy that may not reflect the practice patterns of all surgeons) to redistribute the Promoting Interoperability weight to other score components.³⁹ These changes do not consider the difference in patient complexity between these surgical delivery settings nor do they consider the potential for disparities in patient access to ambulatory surgery. This highlights the difference between the daily variety of a surgery practice from a primary care practice, for whom many of these measures are created. However, both specialties see patients at social risk with complex medical and social needs. Surgeons should participate in ongoing specialty-specific measure development and must advocate for the needs of the sickest patients.

Surgeons caring for the patients at highest social risk are at increased risk of negative payment adjustments under MIPS. As CMS solicits clinician input for the further iterations of value-based payment, it will be important to prioritize the voices of patients at highest social risk and the voices of the surgeons serving them. Inadequate risk adjustment and negative payment adjustment can adversely affect practices that may instead require a more substantial investment to achieve surgical equity. However, the current markers of social need and clinical status need to be investigated to understand the intervenable causes of surgical disparities. The limited patient data available for outcome risk adjustment was highlighted by Birkmeyer in 2006 as a limitation of pay-for-performance.⁴⁰ Dual eligibility is used as a surrogate for patient social risk, but the data reported to MIPS do not capture more granular details of patient social needs. This study cannot draw conclusions about specific approaches to social risk adjustment for outcomes, but supports efforts to consider surgical patient social vulnerability when implementing MIPS payment adjustments. To protect surgical patients at social risk, surgeons should remain engaged as stakeholders and advocates with CMS as further changes are proposed and discussed.