Association of Outpatient Practice-Level Socioeconomic Disadvantage With Quality of Care and Outcomes Among Older Adults With Coronary Artery Disease
What Is Known
-
The Centers for Medicare and Medicaid Services have focused on improving care for Medicare beneficiaries with coronary artery disease through the implementation of outpatient value-based programs.
-
However, outpatient practices that care for socioeconomically disadvantaged patients are more likely to receive financial penalties under value-based programs, as they tend to have worse clinical outcomes than practices that care for more advantaged patients.
-
It is unknown whether the worse clinical outcomes at outpatient practices that serve socioeconomically disadvantaged patients reflect true differences in quality of care delivered, or instead, differences in patient population or community characteristics that are largely beyond clinicians’ control.
What the Study Adds
-
Outpatient practices serving a high proportion of socioeconomically disadvantaged Medicare patients provide similar guideline-recommended care for coronary artery disease but have worse clinical outcomes compared with practices serving a low proportion of disadvantaged patients.
-
Patient-level social factors largely explain these worse clinical outcomes, indicating that outpatient practices that primarily care for socioeconomically disadvantaged patients may fare worse under value-based programs, such as the Merit-Incentive Based Payment System, despite providing similar quality of care as other practices.
Introduction
In the United States, value-based payment (VBP) models are being implemented nationally to encourage improvement in quality of healthcare delivery and have increasingly focused on cardiovascular care.1,2 Outpatient practices that care for socioeconomically disadvantaged patients are more likely to receive penalties under VBP, as they tend to have worse outcomes than practices that care for more advantaged patients.3,4 As a result, physicians and policymakers have voiced concern that because VBP programs do not account for socioeconomic disadvantage, they may be unfairly penalizing practices for serving poor and vulnerable patients rather than for poor quality care.5–7
Understanding whether worse outcomes at outpatient practices that serve poor populations represent true differences in the quality of care delivered, or whether they simply reflect differences in patient population or community characteristics that are largely beyond providers’ control is critically important. If practices serving disadvantaged groups have worse outcomes because they provide low-quality care, VBP could be an important tool to improve quality and outcomes and reduce disparities. On the other hand, if these practices have worse outcomes because of inherent differences in patient population or community characteristics—despite providing high-quality care—then VBP may have the unintended consequence of widening disparities by inappropriately penalizing practices that serve socially disadvantaged patients.
Patients with coronary artery disease (CAD) have been a significant focus of VBP programs administered by the Centers for Medicare and Medicaid Services (CMS). The mandatory Medicare Physician Value-Based Payment Modifier program and its successor, the Merit-based Incentive Payment System (MIPS), financially reward or penalize physician outpatient practices based on outcomes and costs for Medicare patients. Many of the included measures in outpatient VBP, including preventable hospitalizations, readmissions, and per capita spending are explicitly linked to CAD.8 Therefore, examining quality of CAD care in the outpatient setting represents a particular opportunity to understand patterns of care and outcomes.
Prior studies attempting to disentangle the impact of provider quality versus patient/community factors on outcomes have been limited in their ability to distinguish between the two because of a lack of granular data on quality, such as the prescription of guideline-based therapy. The National Cardiovascular Data Registry (NCDR) Practice Innovation and Clinical Excellence (PINNACLE) registry captures detailed data from the electronic health records of outpatient practices across the nation. Prior work with this registry has shown significant variability in quality of care for CAD at these practices, though it is unknown if this is related to the socioeconomic composition or community characteristics of populations that these practices serve.9 This data source thus represents a unique opportunity to combine detailed clinical information with socioeconomic and outcomes data to assess provider quality.
In this study, we aimed to answer three questions. First, are physician outpatient practices that care for a higher proportion of socioeconomically disadvantaged patients less likely to provide guideline-recommended care to Medicare patients age 65 years or older with CAD, a population targeted by VBP programs? Second, do these patients have worse cardiovascular outcomes? Third, if so, is this in part explained by patient-level socioeconomic disadvantage?
Methods
Data Source
We used the NCDR PINNACLE registry, a national, prospective registry for patients seen in outpatient practices in the United States.10 Participating practices collect detailed patient data at the point of care for each outpatient visit. Patient data include demographics, comorbidities, symptoms, vital signs, medications, contraindications to medications, and laboratory values. Data elements are collected either by PINNACLE using paper-based case report forms or by exporting a practice’s electronic health record (EHR) to capture comprehensively the requisite data elements for PINNACLE program participation. Data collection is standardized through written definitions, uniform data entry and transmission requirements, and data quality checks.9,10 The data that support the findings of this study can be requested from the NCDR PINNACLE registry; the analyses code are available from the corresponding author upon reasonable request.
Study Population
We performed a retrospective cohort study using PINNACLE patient and practice data collected during clinic visits of Medicare fee-for-service patients (≥65 years of age) with CAD between 1/1/2010 and 1/1/2015. We only included Medicare patients as this population is the focus of CMS VBP initiatives for outpatient practices, including Physician Value-Based Payment Modifier and MIPS (Figure I in the the Data Supplement). Diagnosis of coronary artery disease was defined by established diagnosis of CAD, or documented history of myocardial infarction (MI), percutaneous coronary intervention, coronary artery bypass grafting, or stable angina.
The area deprivation index (ADI) was used to characterize socioeconomic disadvantage of each Medicare patient and was linked based on the zip code of geographic residence.11 The ADI is a factor-based index, which uses 17 US Census poverty, education, housing, and employment indicators to characterize the relative disadvantage of census-based regions and has been validated in studies of diseases and health outcomes.11–13
The ADI of each PINNACLE practice was characterized, based on zip code location, and compared with non-PINNACLE cardiology practices that were identified using the Physician Compare National dataset.14 Next, physician practices in the PINNACLE registry were categorized into quintiles based on the proportion of all patients with CAD served at each site who were most disadvantaged, defined by ADI score in the highest 20% nationally as others have done previously.15 We used this definition because prior studies have shown that the ADI has a nonlinear relationship with outcomes and is instead associated with a threshold effect—individuals who reside in the most disadvantaged areas have a higher risk of adverse outcomes.15 Practices with fewer than 10 eligible patients with CAD were excluded, as sample sizes that small cannot be sufficiently modeled in regression analyses.
Outcomes
Individual prescription rates of guideline-recommended therapies among eligible patients were evaluated.16,17 Patient eligibility for guideline-recommended medical therapy included antiplatelet therapy for patients with established CAD, β-blocker therapy for those with either a previous MI or a left ventricular ejection fraction <40%, ACE (angiotensin-converting enzyme) inhibitor/angiotensin receptor blocker therapy for those with an left ventricular ejection fraction <40% or diabetes mellitus, statin therapy in those with a low-density lipoprotein concentration ≥100 mg/dL, and cardiac rehabilitation if MI, percutaneous coronary intervention, or coronary artery bypass grafting in past 12 months. Although the criterion for statin use has been expanded to include patients with all levels of low-density lipoprotein, we elected to use the more restrictive definition to conform to the 2011 performance measures guidelines.17 Patients with documented medical, patient, or system reasons for not being prescribed any of the studied medication classes were excluded from analyses for that particular class and clinical encounter. If there were no documented reasons for not prescribing a medication, the patient was considered eligible to receive the medication. Combined prescription rates were also calculated by dividing the number of medications prescribed by the number of medications for which each patient was eligible.9
Clinical outcomes were evaluated by linking CMS fee-for-service claims files to NCDR PINNACLE data by patient and year of encounter. The following clinical outcomes were assessed for patients across practice sites: (1) emergency department presentation for chest pain, (2) hospital admission for unstable angina (UA), (3) hospital admission for acute myocardial infarction (AMI), (4) 30-day readmission after AMI, and (5) 30-day mortality after AMI. We chose to examine measures that quantify incident cardiac events as well as measures that examine outcomes after AMI because differences in outpatient care during the vulnerable period after discharge impact the likelihood of short- and long-term receipt of guideline recommended therapies,18,19 and as a result, risk of adverse clinical outcomes.20,21
Statistical Analysis
Physician outpatient practices were categorized into quintiles of socioeconomic disadvantage, with group 1 representing physician practices serving the lowest proportion of most disadvantaged patients, and group 5 representing practices serving the highest. Patient-level characteristics were compared across the groups of physician practices using Pearson χ2 tests for categorical variables and 1-way ANOVA tests for continuous variables.
To investigate the association between physician practice socioeconomic disadvantage and prescription of guideline-recommended therapies, we constructed hierarchical logistic regression models that accounted for patient clustering, and used the group of physician practices serving the fewest disadvantaged patients as a reference (group 1). Next, we used the same approach to evaluate the association of outpatient practice socioeconomic disadvantage and clinical outcomes, adjusted for patient demographics and clinical characteristics. Because this approach only allowed us to examine practice-level effects, we also adjusted models for patients’ national ADI ranking to understand individual-level effects. Clinical covariates that were included in the multivariable model were age, sex, race (white versus nonwhite), atrial fibrillation/flutter, prior MI, prior percutaneous coronary intervention, prior coronary artery bypass grafting, heart failure, diabetes mellitus, hypertension, hypercholesterolemia, peripheral artery disease, prior stroke/transient ischemic attack, chronic kidney disease, chronic liver disease, and current tobacco use. All models accounted for patient clustering within practices.
As a sensitivity analysis, we fit a hierarchical model with random intercept to account for within-practice correlation, and included both patient-level ADI (most- versus less-disadvantaged) and practice-level ADI (proportion of most-disadvantaged patients on a continuous scale) as key model predictors for each outcome, while adjusting for patient-level characteristics as fixed effects in each model.
Statistical tests were 2-sided and considered significant if P<0.05. Analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC) at the Baim Institute for Clinical Research. Waiver of written informed consent and authorization for this study was granted by Chesapeake Research Review Incorporated.
Results
Study Population
A total of 454 783 patients (mean [SD] age, 75.3 [7.7] years; 39.7% female) with CAD at 271 PINNACLE practices from January 1, 2010 to January 1, 2015 were included in the final study cohort (Figure I in the the Data Supplement). The distribution of ADI for all PINNACLE practices (mean [SD] ADI, 46.6 [22.6], n=271) and non-PINNACLE practices (mean [SD] ADI, 45.4 [23.1], n=9,727) were similar and are shown in Figures II and III in the Data Supplement. Across PINNACLE practice groups, the proportion of all patients with CAD classified as most socioeconomically disadvantaged, defined by patient ADI in the highest 20% nationally, ranged from 1.3% (practice group 1, lowest proportion of disadvantaged patients) to 30.7% (practice group 5, highest proportion of disadvantaged patients) as shown in Figure IV in the Data Supplement.
Patient demographics were generally similar across physician practice groups. Clinical comorbidities, including the proportion of patients with heart failure, diabetes, hypertension, and hypercholesterolemia, were higher at the most disadvantaged-serving compared with the least disadvantaged-serving practices, while rates of prior MI and prior stroke/transient ischemic attack were lower (Table 1).
Practice Group 1 (Least Disadvantaged) | Practice Group 2 | Practice Group 3 | Practice Group 4 | Practice Group 5 (Most Disadvantaged) | |
---|---|---|---|---|---|
No. of practices | 55 | 54 | 54 | 54 | 54 |
No. of patients | 55 474 | 97 706 | 118 322 | 107 474 | 74 807 |
Area deprivation index national rank, mean (SD) | 23.2 (18.0) | 36.3 (21.6) | 48.3 (23.5) | 57.1 (22.2) | 65.3 (21.1) |
Age, y; mean (SD) | 75.8 (7.8) | 75.4 (7.7) | 75.3 (7.6) | 75.2 (7.6) | 74.9 (7.5) |
Sex | |||||
Male, no. (%) | 33 775 (60.9) | 59 339 (60.8) | 73 003 (61.7) | 63 663 (59.2) | 43 680 (58.4) |
Female, no. (%) | 21 696 (39.1) | 38 325 (39.2) | 45 311 (38.3) | 43 811 (40.8) | 31 125 (41.6) |
Race, no. (%) | |||||
White | 49 372 (89.0) | 89 113 (91.2) | 109 797 (92.8) | 100 059 (93.1) | 68 459 (91.5) |
Black | 2897 (5.2) | 5153 (5.3) | 4037 (3.4) | 4573 (4.3) | 4610 (6.2) |
Hispanic | 463 (0.8) | 836 (0.9) | 1046 (0.9) | 557 (0.5) | 469 (0.6) |
Asian | 1153 (2.1) | 852 (0.9) | 1015 (0.9) | 558 (0.5) | 181 (0.2) |
Other | 1132 (2.0) | 1122 (1.1) | 1746 (1.5) | 1227 (1.1) | 891 (1.2) |
Unknown | 457 (0.8) | 630 (0.6) | 681 (0.6) | 500 (0.5) | 197 (0.3) |
Medical conditions, no. (%) | |||||
Atrial fibrillation/flutter | 13 777 (24.8) | 23 960 (24.5) | 28 929 (24.4) | 25 637 (23.9) | 16 002 (21.4) |
Prior MI | 18 746 (33.8) | 30 991 (31.7) | 25 402 (21.5) | 29 021 (27.0) | 16 786 (22.4) |
Prior PCI | 10 312 (18.6) | 17 138 (17.5) | 25 535 (21.6) | 22 723 (21.1) | 15 005 (20.1) |
Prior CABG | 7880 (14.2) | 13 917 (14.2) | 21 185 (17.9) | 20 431 (19.0) | 10 766 (14.4) |
Heart failure | 10 983 (19.8) | 20 817 (21.3) | 27 964 (23.6) | 24 177 (22.5) | 16 003 (21.4) |
Diabetes mellitus | 14 226 (25.6) | 23 597 (24.2) | 29 338 (24.8) | 27 996 (26.0) | 20 669 (27.6) |
Hypertension | 41 537 (74.9) | 77 487 (79.3) | 92 090 (77.8) | 82 002 (76.3) | 62 558 (83.6) |
Hypercholesterolemia | 40 587 (73.2) | 77 840 (79.7) | 91 117 (77.0) | 76 204 (70.9) | 58 844 (78.7) |
PAD | 7629 (13.8) | 10 186 (10.4) | 15 590 (13.2) | 15 293 (14.2) | 10 510 (14.0) |
Prior stroke/TIA | 5575 (10.0) | 10 906 (11.2) | 9905 (8.4) | 10 626 (9.9) | 5295 (7.1) |
CKD | 1686 (3.0) | 1737 (1.8) | 2201 (1.9) | 1904 (1.8) | 1025 (1.4) |
Liver disease | 149 (0.3) | 86 (0.1) | 161 (0.1) | 105 (0.1) | 56 (0.1) |
Current tobacco use | 5925 (10.7) | 9217 (9.4) | 12 396 (10.5) | 14 283 (13.3) | 9700 (13.0) |
Guideline-Recommended Therapies
Observed rates of individual and combined guideline-recommended therapies are shown across physician practice groups in Table 2. After accounting for patient clustering within practices, there was no significant difference in the likelihood of prescription of antiplatelet therapy at the most disadvantaged-serving practices (group 5) compared with the least disadvantaged-serving practices (group 1; odds ratio [OR], 0.94 [95% CI, 0.69–1.27]; Figure 1; Table I in the Data Supplement). These patterns were similar for beta-blocker therapy if prior MI or left ventricular ejection fraction <40% (OR, 0.97 [95% CI, 0.69–1.35]), ACE-I or angiotensin receptor blocker if left ventricular ejection fraction <40% and/or diabetes mellitus (OR, 0.93 [95% CI, 0.74–1.19]), statin therapy if LDL (low-density lipoprotein) >100 (OR, 0.88 [95% CI, 0.68–1.14]), and cardiac rehabilitation if MI, percutaneous coronary intervention, or coronary artery bypass grafting in past 12 months (OR, 0.45 [95% CI, 0.20–1.00]). There was also no significant difference in the likelihood of patients being prescribed all appropriate guideline-recommended therapies at the most disadvantaged-serving compared with the least disadvantaged-serving practices (OR, 0.91 [95% CI, 0.65–1.27])
Practice Group 1 (Least Disadvantaged) | Practice Group 2 | Practice Group 3 | Practice Group 4 | Practice Group 5 (Most Disadvantaged) | |
---|---|---|---|---|---|
Guideline-recommended therapy, no. (%) | |||||
Antiplatelet therapy† | 41 351 (74.5) | 73 456 (75.2) | 92 682 (78.3) | 83 534 (77.7) | 57 935 (77.4) |
β-Blocker if prior MI or LVEF <40% | 13 702 (67.4) | 23 146 (67.1) | 23 295 (77.9) | 24 982 (76.4) | 15 613 (74.5) |
ACE inhibitor or ARB if LVEF <40% and/or DM | 11 440 (69.8) | 18 670 (67.2) | 23 381 (68.1) | 22 335 (68.8) | 16 704 (67.1) |
Statin therapy if LDL >100 | 2723 (69.4) | 3982 (64.5) | 4882 (64.6) | 6017 (65.2) | 3549 (68.7) |
Cardiac rehabilitation‡ | 732 (5.2) | 1271 (5.3) | 914 (3.0) | 3020 (9.9) | 734 (4.1) |
Met all appropriate prescriptions | 27 398 (49.4) | 49 140 (50.3) | 60 087 (50.8) | 52 814 (49.1) | 37 261 (49.8) |
Mean combined prescription rate, mean (SD) | 66.9 (38.4) | 67.2 (38.4) | 68.7 (37.2) | 68.2 (36.9) | 68.1 (37.3) |
Clinical outcomes, no. (%) | |||||
ED use for chest pain | 3198 (5.8) | 6208 (6.4) | 6835 (5.8) | 7374 (6.9) | 5942 (7.9) |
Admission for unstable angina | 1309 (2.4) | 2601 (2.7) | 2878 (2.4) | 3401 (3.2) | 3124 (4.2) |
Admission for acute MI | 2734 (4.9) | 5069 (5.2) | 5265 (4.4) | 5781 (5.4) | 4401 (5.9) |
30-d readmission after acute MI | 684 (25.0) | 1154 (22.8) | 1206 (22.9) | 1342 (23.2) | 1048 (23.8) |
30-d mortality after acute MI | 493 (18.0) | 942 (18.6) | 878 (16.7) | 1073 (18.6) | 786 (17.9) |
Clinical Outcomes
Observed clinical outcomes across physician practice groups are also shown in Table 2. After multivariable adjustment for patient demographics, clinical comorbidities, and patient clustering, there was no significant difference in the likelihood of presenting to an emergency department with chest pain among patients at the most disadvantaged-serving practices (group 5) compared with the least disadvantaged-serving practices (group 1; Figure 2). However, patients at the most disadvantaged-serving practices (group 5) were more likely to be admitted for unstable angina (aOR, 1.46 [95% CI, 1.04–2.05]), a relationship that was attenuated after additional adjustment for individual patients’ ADI (aOR, 1.20 [95% CI, 0.86–1.69]; Table 3). There was no significant difference in the likelihood of admission for AMI (aOR, 1.17 [95% CI, 0.86–1.61]) nor in rates of 30-day readmission after AMI (aOR, 0.85 [95% CI, 0.71–1.02]). However, rates of 30-day mortality after AMI were higher among patients at the most disadvantaged-serving practices (group 5; aOR, 1.31 [95% CI, 1.02–1.68]; Figure 2), a relationship that was also attenuated after additional adjustment for patient ADI (aOR, 1.22 [95% CI, 0.94–1.59]; Table 3). Likelihood ratios to measure model goodness-of-fit after adding patient-level ADI to multivariable models are shown in Table II in the Data Supplement. The findings of the sensitivity analysis that included both patient-level ADI (most- versus less-disadvantaged) and practice-level ADI (proportion of most-disadvantaged patients on a continuous scale) as key model predictors for each outcome were similar to the main analysis and demonstrated that differences in clinical outcomes were primarily driven by within-rather than between-practice ADI effects (Table III in the Data Supplement).
Unadjusted | Multivariable Adjusted* | Multivariable and Patient ADI Adjusted† | |
---|---|---|---|
ED use for chest pain | |||
Group 1 (least disadvantaged) | Reference | Reference | Reference |
Group 2 | 1.12 (0.81–1.56) | 1.18 (0.86–1.63) | 1.11 (0.81–1.53) |
Group 3 | 1.26 (0.91–1.75) | 1.33 (0.97–1.83) | 1.18 (0.86–1.62) |
Group 4 | 1.53 (1.11–2.11) | 1.49 (1.09–2.04) | 1.26 (0.92–1.73) |
Group 5 (most disadvantaged) | 1.11 (0.79–1.56) | 1.09 (0.78–1.52) | 0.88 (0.63–1.23) |
Admission for unstable angina | |||
Group 1 (least disadvantaged) | Reference | Reference | Reference |
Group 2 | 1.20 (0.86–1.66) | 1.22 (0.88–1.70) | 1.15 (0.83–1.60) |
Group 3 | 1.16 (0.84–1.61) | 1.13 (0.82–1.57) | 1.01 (0.73–1.40) |
Group 4 | 1.62 (1.17–2.24) | 1.50 (1.09–2.07) | 1.29 (0.93–1.78) |
Group 5 (most disadvantaged) | 1.53 (1.09–2.15) | 1.46 (1.04–2.05) | 1.20 (0.86–1.69) |
Admission for AMI | |||
Group 1 (least disadvantaged) | Reference | Reference | Reference |
Group 2 | 1.07 (0.79–1.44) | 1.14 (0.84–1.55) | 1.07 (0.79–1.46) |
Group 3 | 1.03 (0.76–1.39) | 1.15 (0.85–1.56) | 1.02 (0.75–1.38) |
Group 4 | 1.32 (0.98–1.78) | 1.29 (0.95–1.74) | 1.10 (0.81–1.49) |
Group 5 (most disadvantaged) | 1.14 (0.84–1.56) | 1.17 (0.86–1.61) | 0.96 (0.70–1.32) |
30-d readmission after AMI | |||
Group 1 (least disadvantaged) | Reference | Reference | Reference |
Group 2 | 0.86 (0.74–1.01) | 0.84 (0.70–1.00) | 0.83 (0.70–0.99) |
Group 3 | 0.89 (0.76–1.04) | 0.85 (0.72–1.01) | 0.83 (0.70–0.99) |
Group 4 | 0.92 (0.79–1.08) | 0.90 (0.76–1.06) | 0.88 (0.73–1.05) |
Group 5 (most disadvantaged) | 0.95 (0.80–1.12) | 0.85 (0.71–1.02) | 0.83 (0.68–1.01) |
30-d mortality after AMI | |||
Group 1 (least disadvantaged) | Reference | Reference | Reference |
Group 2 | 1.09 (0.88–1.35) | 1.09 (0.86–1.38) | 1.06 (0.83–1.34) |
Group 3 | 1.01 (0.82–1.24) | 1.03 (0.81–1.29) | 0.98 (0.77–1.25) |
Group 4 | 1.03 (0.84–1.26) | 1.06 (0.85–1.34) | 1.00 (0.79–1.28) |
Group 5 (most disadvantaged) | 1.15 (0.92–1.43) | 1.31 (1.02–1.68) | 1.22 (0.94–1.59) |
Discussion
In this study of physician practices and Medicare beneficiaries with CAD, delivery of guideline-recommended therapy did not differ between practices serving a high proportion of socioeconomically disadvantaged patients compared with practices serving a low proportion of disadvantaged patients. Despite this lack of difference, patients at physician practices serving the most disadvantaged patients were more likely to be admitted to the hospital for unstable angina. Similarly, 30-day mortality after hospitalization for AMI was higher among patients at the most disadvantaged-serving practices. These patterns were primarily driven by patient-level ADI, suggesting that differences in individual socioeconomic disadvantage may explain variation in some clinical outcomes across practice sites.
As the United States increasingly shifts toward VBP models, there is growing concern that these programs unfairly penalize practices for serving socioeconomically disadvantaged patients, which may exacerbate disparities in care.3–5,7 The mandatory Medicare Physician Value-Based Payment Modifier program, which evolved into MIPS last year, financially penalizes or rewards physician practices based on a number of measures including preventable hospitalizations, total per capita costs for CAD, and Medicare spending per beneficiary (3 days before and 30 days after inpatient hospitalization). While the total per capita cost measure indirectly accounts for social risk by including poverty in its multifactorial risk score, the other measures do not account for social risk in any way in their risk adjustment approach. The program also does not account for variation in socioeconomic disadvantage between practice populations.8 Our findings suggest that most disadvantaged-serving physician practices are likely to perform worse on these measures, given higher rates of hospital utilization and worse outcomes among patients with CAD, despite providing similar guideline-recommended care as other practices. Prior studies have shown that practices serving more socially high-risk patients tend to perform worse on Physician Value-Based Payment Modifier measures.3,4 Our analysis builds upon these findings and suggests that among patients with CAD, differences in performance under VBP may not reflect differences in care delivered.
Though patients at physician practices serving primarily disadvantaged populations were more likely to use hospitals for CAD-related care, this association was primarily related to patient-level ADI. There are several potential explanations for these patterns. First, socioeconomically disadvantaged patients have increased difficulty accessing reliable ambulatory care.22 This may be, at least in part, why these patients perceive hospitals to offer better access and technical care quality.22 Second, though patients seen at disadvantaged-serving practices are as likely to be prescribed guideline-recommended medications, they may face challenges filling prescriptions. Recent evidence suggests that in low-income communities, pharmacies are more likely to have limited supply or be out of stock of medications,23 and in extreme cases, may be pharmacy deserts.24 It is also possible, however, that though guideline-recommended medications are prescribed, they are not titrated to optimal doses on subsequent visits.25 Third, socioeconomically disadvantaged patients have more difficulty adhering to cardiovascular medication treatment plans,18,26–28 which may influence downstream hospital utilization, as may other factors related to socioeconomic disadvantage, including housing instability, limited social support and security, and poor medical literacy.
Patients at the most disadvantaged-serving practices also had higher 30-day mortality rates after AMI compared with the least disadvantaged-serving practices. Early cardiology follow-up after hospitalization for AMI is associated with improved survival,29 potentially due to greater medication adherence,18,19 and though we did not examine this directly, it is possible that practices serving primarily disadvantaged patients may have limited resources and availability to accommodate close postdischarge care. The association between disadvantaged practices and mortality, however, was attenuated after accounting for patient ADI, suggesting that social factors beyond the care provided by physician practices may explain this finding. Individuals living in deprived neighborhoods are exposed to unique environmental and lifestyle risk factors, experience chronic stress, and face other social determinants of health, all of which likely contribute to worse outcomes.30 This is consistent with prior studies that have also shown that patient socioeconomic disadvantage is associated with higher short-term mortality rates after AMI,31,32 and may, at least in part, explain why safety-net hospitals are more likely to be penalized by VBP programs that focus on an episode of acute care.33,34 It is also possible, however, that unmeasured dimensions of practice-level quality of care explain these relationships.
This study extends prior research on how practice-level socioeconomic disadvantage may impact performance in outpatient VBP programs.3,4 The MIPS uses35 will use an array of metrics to measure performance across a broader range of practices. Though quality metrics (eg, β-blocker therapy for prior MI and LVSD, ACE-I or angiotensin receptor blocker therapy for diabetes mellitus or LVSD) are used to assess performance under MIPS, other measures such as inpatient admissions and costs of care will be weighted more over time.36 Thus, physician practices that care for a higher proportion of socioeconomically disadvantaged patients with CAD are likely to fare worse under MIPS, despite providing similar guideline-recommended care as other practices, and reconsideration of these policy approaches by CMS may be required. As MIPS is implemented, it will be important to examine carefully whether measures of performance should be adjusted or stratified for socioeconomic disadvantage. Alternatively, CMS might consider providing support to resource-poor practices that serve vulnerable populations and consistently perform worse in VBP programs to help them address social determinants of health and improve delivery of care in communities, to achieve better health outcomes.37,38
Limitations
This study has some limitations. First, our findings are based on physician practices that voluntarily participate in the NCDR PINNACLE registry, which may select for practices that differ from nonparticipating practices. However, we found that the ADI distribution of PINNACLE practice locations was similar to that of nonparticipating practice locations, suggesting that levels of socioeconomic disadvantage in areas that PINNACLE sites serve did not differ from nonparticipating sites. Second, our cohort included only a sample of Medicare patients with CAD that are seen at PINNACLE practices. Nonetheless, focusing on PINNACLE practices provided granular information regarding patient characteristics and quality of care that would otherwise not be available in claims alone. Third, although the delivery of guideline recommended therapy is an important dimension of quality of care that impacts outcomes, we were unable to evaluate other aspects of care quality, such as care coordination, integration of social workers and community health workers, and communication between physician and patients that likely vary across practice sites. It is possible that practices serving a higher proportion of disadvantaged patients provide suboptimal care on these dimensions of quality that were not captured in our evaluation of guideline-recommended therapies. Fourth, we did not explore lifestyle factors (exercise, weight loss, psychological factors) that are part of CAD guidelines and performance measures.
Conclusions
In summary, outpatient practices serving a high proportion of socioeconomically disadvantaged patients provided similar guideline-recommended care for CAD compared with practices serving a low proportion of disadvantaged patients. Despite this, Medicare patients at the most disadvantaged-serving physician practices were more likely to be admitted to the hospital for unstable angina and had higher 30-day mortality rates after MI. These associations were largely driven by patient-level ADI, indicating that social factors beyond care provided by physician practices may explain worse outcomes. Physician practices that care for a high proportion of socioeconomically disadvantaged patients with CAD are likely to fare worse under VBP programs, such as MIPS, despite providing similar guideline-recommended care as other practices.
Acknowledgments
All authors conceived and designed the study, analyzed and interpreted the data, and critically revised the manuscript for important intellectual content. AT and GD acquired the data. YS and LD carried out the statistical analysis. RKW, KJM, DLB, and RWY drafted the manuscript. KJM, AT, and GD supervised the study and are the guarantors.
Sources of Funding
This research was supported by the American College of Cardiology’s National Cardiovascular Data Registry (NCDR). PINNACLE Registry is an initiative of the American College of Cardiology. Bristol-Myers Squibb and Pfizer, Inc are Founding Sponsors of the PINNACLE Registry.
Disclosures
Dr Wadhera receives research support from the National Heart, Lung, and Blood Institute (grant K23HL148525-1) at the National Institutes of Health. He previously served as a consultant for Regeneron. Dr Bhatt discloses the following relationships—Advisory Board: Cardax, Elsevier Practice Update Cardiology, Medscape Cardiology, Regado Biosciences; Board of Directors: Boston VA Research Institute, Society of Cardiovascular Patient Care, TobeSoft; Chair: American Heart Association Quality Oversight Committee; Data Monitoring Committees: Baim Institute for Clinical Research (formerly Harvard Clinical Research Institute, for the PORTICO trial, funded by St. Jude Medical, now Abbott), Cleveland Clinic, Duke Clinical Research Institute, Mayo Clinic, Mount Sinai School of Medicine (for the ENVISAGE trial, funded by Daiichi Sankyo), Population Health Research Institute; Honoraria: American College of Cardiology (Senior Associate Editor, Clinical Trials and News, ACC.org; Vice-Chair, ACC Accreditation Committee), Baim Institute for Clinical Research (formerly Harvard Clinical Research Institute; RE-DUAL PCI clinical trial steering committee funded by Boehringer Ingelheim), Belvoir Publications (Editor in Chief, Harvard Heart Letter), Duke Clinical Research Institute (clinical trial steering committees), HMP Global (Editor in Chief, Journal of Invasive Cardiology), Journal of the American College of Cardiology (Guest Editor; Associate Editor), Population Health Research Institute (for the COMPASS operations committee, publications committee, steering committee, and USA national co-leader, funded by Bayer), Slack Publications (Chief Medical Editor, Cardiology Today’s Intervention), Society of Cardiovascular Patient Care (Secretary/Treasurer), WebMD (CME steering committees); Other: Clinical Cardiology (Deputy Editor), NCDR-ACTION Registry Steering Committee (Chair), VA CART Research and Publications Committee (Chair); Research Funding: Abbott, Amarin, Amgen, AstraZeneca, Bayer, Boehringer Ingelheim, Bristol-Myers Squibb, Chiesi, Eisai, Ethicon, Forest Laboratories, Idorsia, Ironwood, Ischemix, Lilly, Medtronic, PhaseBio, Pfizer, Regeneron, Roche, Sanofi Aventis, Synaptic, The Medicines Company; Royalties: Elsevier (Editor, Cardiovascular Intervention: A Companion to Braunwald’s Heart Disease); Site Co-Investigator: Biotronik, Boston Scientific, St. Jude Medical (now Abbott), Svelte; Trustee: American College of Cardiology; Unfunded Research: FlowCo, Merck, Novo Nordisk, PLx Pharma, Takeda. Dr Yeh receives research support from the National Heart, Lung and Blood Institute (R01HL136708) and the Richard A. and Susan F. Smith Center for Outcomes Research in Cardiology. Dr Turchin is an advisor for Brio Systems and Monarch Medical Technologies and receives research grants from Eli Lilly. Dr Joynt Maddox receives research support from the National Heart, Lung, and Blood Institute (R01HL143421) and National Institute on Aging (R01AG060935), and previously did contract work for the US Department of Health and Human Services. The other authors report no conflicts.
Footnotes
References
- 1.
Roland M, Campbell S . Successes and failures of pay for performance in the United Kingdom.N Engl J Med. 2014; 370:1944–1949. doi: 10.1056/NEJMhpr1316051CrossrefMedlineGoogle Scholar - 2.
Eijkenaar F . Pay for performance in health care: an international overview of initiatives.Med Care Res Rev. 2012; 69:251–276. doi: 10.1177/1077558711432891CrossrefMedlineGoogle Scholar - 3.
Chen LM, Epstein AM, Orav EJ, Filice CE, Samson LW, Joynt Maddox KE . Association of practice-level social and medical risk with performance in the medicare physician value-based payment modifier program.JAMA. 2017; 318:453–461. doi: 10.1001/jama.2017.9643CrossrefMedlineGoogle Scholar - 4.
Roberts ET, Zaslavsky AM, McWilliams JM . The value-based payment modifier: program outcomes and implications for disparities.Ann Intern Med. 2018; 168:255–265. doi: 10.7326/M17-1740CrossrefMedlineGoogle Scholar - 5.
Frakt AB, Jha AK . Face the facts: we need to change the way we do pay for performance.Ann Intern Med. 2018; 168:291–292. doi: 10.7326/M17-3005CrossrefMedlineGoogle Scholar - 6.
Ryan AM . Will value-based purchasing increase disparities in care?N Engl J Med. 2013; 369:2472–2474. doi: 10.1056/NEJMp1312654CrossrefMedlineGoogle Scholar - 7.
Joynt Maddox KE . Financial incentives and vulnerable populations – will alternative payment models help or hurt?N Engl J Med. 2018; 378:977–979. doi: 10.1056/NEJMp1715455CrossrefMedlineGoogle Scholar - 8.
Hales CM, Fryar CD, Carroll MD, Freedman DS, Ogden CL . Trends in obesity and severe obesity prevalence in US youth and adults by sex and age, 2007-2008 to 2015-2016.JAMA. 2018; 319:1723–1725. doi: 10.1001/jama.2018.3060CrossrefMedlineGoogle Scholar - 9.
Maddox TM, Chan PS, Spertus JA, Tang F, Jones P, Ho PM, Bradley SM, Tsai TT, Bhatt DL, Peterson PN . Variations in coronary artery disease secondary prevention prescriptions among outpatient cardiology practices: insights from the NCDR (National Cardiovascular Data Registry).J Am Coll Cardiol. 2014; 63:539–546. doi: 10.1016/j.jacc.2013.09.053CrossrefMedlineGoogle Scholar - 10.
Chan PS, Oetgen WJ, Buchanan D, Mitchell K, Fiocchi FF, Tang F, Jones PG, Breeding T, Thrutchley D, Rumsfeld JS, Spertus JA . Cardiac performance measure compliance in outpatients: the American College of Cardiology and National Cardiovascular Data Registry’s PINNACLE (Practice Innovation And Clinical Excellence) program.J Am Coll Cardiol. 2010; 56:8–14. doi: 10.1016/j.jacc.2010.03.043CrossrefMedlineGoogle Scholar - 11.
Kind AJH, Buckingham WR . Making neighborhood-disadvantage metrics accessible – the neighborhood atlas.N Engl J Med. 2018; 378:2456–2458. doi: 10.1056/NEJMp1802313CrossrefMedlineGoogle Scholar - 12.
Singh GK . Area deprivation and widening inequalities in US mortality, 1969-1998.Am J Public Health. 2003; 93:1137–1143. doi: 10.2105/ajph.93.7.1137CrossrefMedlineGoogle Scholar - 13.
Hu J, Kind AJH, Nerenz D . Area deprivation index predicts readmission risk at an urban teaching hospital.Am J Med Qual. 2018; 33:493–501. doi: 10.1177/1062860617753063CrossrefMedlineGoogle Scholar - 14. Physician Compare datasets. Centers for Medicare and Medicaid Services.2018. https://data.medicare.gov/data/physician-compare. Accessed November 2, 2018.Google Scholar
- 15.
Kind AJ, Jencks S, Brock J, Yu M, Bartels C, Ehlenbach W, Greenberg C, Smith M . Neighborhood socioeconomic disadvantage and 30-day rehospitalization: a retrospective cohort study.Ann Intern Med. 2014; 161:765–774. doi: 10.7326/M13-2946CrossrefMedlineGoogle Scholar - 16.
Fihn SD, Gardin JM, Abrams J, Berra K, Blankenship JC, Dallas AP, Douglas PS, Foody JM, Gerber TC, Hinderliter AL, King SB, Kligfield PD, Krumholz HM, Kwong RYK, Lim MJ, Linderbaum JA, Mack MJ, Munger MA, Prager RL, Sabik JF, Shaw LJ, Sikkema JD, Smith CR, Smith SC, Spertus JA, Williams SV . 2012 ACCF/AHA/ACP/AATS/PCNA/SCAI/STS guideline for the diagnosis of patients with stable ischemic heart disease. A report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines, and the American College of Physicians, American Association for Thoracic Surgery, Preventive Cardiovascular Nurses Association, Society for Cardiovascular Angiography and Interventions, and Society of Thoracic Surgeons.Circulation. 2012; 60:e44–e164. doi: 10.1161/CIR.0b013e3182776f83Google Scholar - 17.
Drozda J, Messer JV, Spertus J, Abramowitz B, Alexander K, Beam CT, Bonow RO, Burkiewicz JS, Crouch M, Goff DC, Hellman R, James T, King ML, Machado EA, Ortiz E, O’Toole M, Persell SD, Pines JM, Rybicki FJ, Sadwin LB, Sikkema JD, Smith PK, Torcson PJ, Wong JB . ACCF/AHA/AMA–PCPI 2011 performance measures for adults with coronary artery disease and hypertension. A report of the American College of Cardiology Foundation/American Heart Association Task Force on Performance Measures and the American Medical Association–Physician Consortium for Performance Improvement.Circulation. 2011; 58:316–336. doi: 10.1161/CIR.0b013e31821d9ef2Google Scholar - 18.
Faridi KF, Peterson ED, McCoy LA, Thomas L, Enriquez J, Wang TY . Timing of first postdischarge follow-up and medication adherence after acute myocardial infarction.JAMA Cardiol. 2016; 1:147–155. doi: 10.1001/jamacardio.2016.0001CrossrefMedlineGoogle Scholar - 19.
Daugherty SL, Ho PM, Spertus JA, Jones PG, Bach RG, Krumholz HM, Peterson ED, Rumsfeld JS, Masoudi FA . Association of early follow-up after acute myocardial infarction with higher rates of medication use.Arch Intern Med. 2008; 168:485–491; discussion 492. doi: 10.1001/archinte.168.5.485CrossrefMedlineGoogle Scholar - 20.
Ho PM, Spertus JA, Masoudi FA, Reid KJ, Peterson ED, Magid DJ, Krumholz HM, Rumsfeld JS . Impact of medication therapy discontinuation on mortality after myocardial infarction.Arch Intern Med. 2006; 166:1842–1847. doi: 10.1001/archinte.166.17.1842CrossrefMedlineGoogle Scholar - 21.
Ho PM, Magid DJ, Shetterly SM, Olson KL, Maddox TM, Peterson PN, Masoudi FA, Rumsfeld JS . Medication nonadherence is associated with a broad range of adverse outcomes in patients with coronary artery disease.Am Heart J. 2008; 155:772–779. doi: 10.1016/j.ahj.2007.12.011CrossrefMedlineGoogle Scholar - 22.
Kangovi S, Barg FK, Carter T, Long JA, Shannon R, Grande D . Understanding why patients of low socioeconomic status prefer hospitals over ambulatory care.Health Aff (Millwood). 2013; 32:1196–1203. doi: 10.1377/hlthaff.2012.0825CrossrefMedlineGoogle Scholar - 23.
Amstislavski P, Matthews A, Sheffield S, Maroko AR, Weedon J . Medication deserts: survey of neighborhood disparities in availability of prescription medications.Int J Health Geogr. 2012; 11:48. doi: 10.1186/1476-072X-11-48CrossrefMedlineGoogle Scholar - 24.
Qato DM, Daviglus ML, Wilder J, Lee T, Qato D, Lambert B . ‘Pharmacy deserts’ are prevalent in Chicago’s predominantly minority communities, raising medication access concerns.Health Aff (Millwood). 2014; 33:1958–1965. doi: 10.1377/hlthaff.2013.1397CrossrefMedlineGoogle Scholar - 25.
Arnold SV, Spertus JA, Masoudi FA, Daugherty SL, Maddox TM, Li Y, Dodson JA, Chan PS . Beyond medication prescription as performance measures: optimal secondary prevention medication dosing after acute myocardial infarction.J Am Coll Cardiol. 2013; 62:1791–1801. doi: 10.1016/j.jacc.2013.04.102CrossrefMedlineGoogle Scholar - 26.
Kareem H, Shetty PN, Devasia T, Karkala YR, Paramasivam G, Guddattu V, Singh A, Chauhan S . Impact of socioeconomic status on adverse cardiac events after coronary angioplasty: a cohort study.Heart Asia. 2018; 10:e010960. doi: 10.1136/heartasia-2017-010960CrossrefMedlineGoogle Scholar - 27.
Kolandaivelu K, Leiden BB, O’Gara PT, Bhatt DL . Non-adherence to cardiovascular medications.Eur Heart J. 2014; 35:3267–3276. doi: 10.1093/eurheartj/ehu364CrossrefMedlineGoogle Scholar - 28.
Kolandaivelu K, Bhatt DL . Overcoming ‘resistance’ to antiplatelet therapy: targeting the issue of nonadherence.Nat Rev Cardiol. 2010; 7:461–467. doi: 10.1038/nrcardio.2010.71CrossrefMedlineGoogle Scholar - 29.
Ayanian JZ, Landrum MB, Guadagnoli E, Gaccione P . Specialty of ambulatory care physicians and mortality among elderly patients after myocardial infarction.N Engl J Med. 2002; 347:1678–1686. doi: 10.1056/NEJMsa020080CrossrefMedlineGoogle Scholar - 30.
Havranek EP, Mujahid MS, Barr DA, Blair IV, Cohen MS, Cruz-Flores S, Davey-Smith G, Dennison-Himmelfarb CR, Lauer MS, Lockwood DW, Rosal M, Yancy CW ; American Heart Association Council on Quality of Care and Outcomes Research, Council on Epidemiology and Prevention, Council on Cardiovascular and Stroke Nursing, Council on Lifestyle and Cardiometabolic Health, and Stroke Council. Social determinants of risk and outcomes for cardiovascular disease: a scientific statement from the American Heart Association.Circulation. 2015; 132:873–898. doi: 10.1161/CIR.0000000000000228LinkGoogle Scholar - 31.
Udell JA, Desai NR, Li S, Thomas L, de Lemos JA, Wright-Slaughter P, Zhang W, Roe MT, Bhatt DL . Neighborhood socioeconomic disadvantage and care after myocardial infarction in the national cardiovascular data registry.Circ Cardiovasc Qual Outcomes. 2018; 11:e004054. doi: 10.1161/CIRCOUTCOMES.117.004054LinkGoogle Scholar - 32.
Rao SV, Schulman KA, Curtis LH, Gersh BJ, Jollis JG . Socioeconomic status and outcome following acute myocardial infarction in elderly patients.Arch Intern Med. 2004; 164:1128–1133. doi: 10.1001/archinte.164.10.1128CrossrefMedlineGoogle Scholar - 33.
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; 163:427–436. doi: 10.7326/M14-2813CrossrefMedlineGoogle Scholar - 34.
Joynt KE, Jha AK . Characteristics of hospitals receiving penalties under the hospital readmissions reduction program.JAMA. 2013; 309:342–343. doi: 10.1001/jama.2012.94856CrossrefMedlineGoogle Scholar - 35. MIPS Overview. Centers for Medicare and Medicaid Services.2018. Available at: https://qpp.cms.gov/mips/overview. Accessed October 15, 2018.Google Scholar
- 36.
Gilstrap LG, Joynt KE . MACRA and cardiology-the devil is in the details.JAMA Cardiol. 2017; 2:1177–1178. doi: 10.1001/jamacardio.2017.2412CrossrefMedlineGoogle Scholar - 37.
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; 376:510–513. doi: 10.1056/NEJMp1616278CrossrefMedlineGoogle Scholar - 38.
Wadhera RK, Yeh RW, Joynt Maddox KE . The Hospital Readmissions Reduction Program – Time for a Reboot.N Engl J Med. 2019; 380:2289–2291. doi: 10.1056/NEJMp1901225CrossrefMedlineGoogle Scholar