The Role of Social Risk Factors in Dialysis Quality and Patient Outcomes Under a Medicare Quality Incentive Program : Medical Care

Secondary Logo

Journal Logo

Original Articles

The Role of Social Risk Factors in Dialysis Quality and Patient Outcomes Under a Medicare Quality Incentive Program

Breck, Andrew PhD*; Marr, Jeffrey BA*; Turenne, Marc PhD; Esposito, Dominick PhD*

Author Information
doi: 10.1097/MLR.0000000000001750

Abstract

In 2017, over 740,000 people in the United States had end-stage renal disease (ESRD).1 ESRD, which disproportionately afflicts Black and low-income individuals, contributes to poor quality of life and shortened lifespans.2–5 Living with ESRD requires managing impaired kidney functioning with multiple dialysis treatments each week or receiving a kidney transplant. With a shortage of available kidney donors, over 70% of ESRD patients receive maintenance dialysis.6

Over 80% of ESRD patients are covered by Medicare.7 The Centers for Medicare & Medicaid Services (CMS) has implemented several initiatives that align Medicare payments with patient care. One example is the ESRD Quality Incentive Program (QIP). Since 2012, dialysis facilities that do not meet CMS’s performance standards are assessed a penalty of up to 2% of their Medicare payments.8

During earlier years of the program dialysis facilities treating larger proportions of Black and dual-eligible patients were more likely to receive a payment reduction under the ESRD QIP.9–15 It remains unclear whether these differences persisted in more recent performance years.

A key concern with value-based purchasing programs, such as the ESRD QIP, is their potential to exacerbate health disparities if the financial strain of payment reductions negatively affects facilities’ ability to provide quality care.16 The structure of many value-based purchasing programs creates incentives for providers to not treat patients who they think may reduce their performance scores.17 Facing such incentives, patients whose providers perceive to be high risk would face challenges in accessing high-quality care, and facilities that serve patients with social risk factors may be perceived to provide low-quality care. These perceptions could contribute to existing disparities in patient outcomes.16,17

This paper examines the interaction of social risk factors with the ESRD QIP on dialysis quality and patient outcomes. The study answers three research questions: (1) Do ESRD QIP payment reductions vary among facilities on the basis of patient social risk factors? (2) Have differences in quality outcomes between patients with and without social risk factors changed under the ESRD QIP? (3) Do these differences in quality outcomes reflect variations in quality across facilities?

METHODS

Data

Our analyses used Medicare claims and enrollment data from January 2010 to December 2018 and dialysis facility and patient information from the Consolidated Renal Operations in a Web-Enabled Network (CROWNWeb) database from 2012 to 2018. We obtained information about facilities’ payment reductions under the ESRD QIP from the CMS Dialysis Facility Compare Website.

The analytic data include all ESRD patients from 2010 to 2018 and all dialysis facilities eligible for the ESRD QIP in any given calendar year. We created a patient-year-outcome level dataset that included annualized measures for patient outcomes during eligible months of each year. Patients are linked to the first facility they were attributed to when they were first eligible for each measure in a given year.

Our analyses focus on four social risk factors: Black race, Hispanic ethnicity, dual eligibility for Medicare and Medicaid, and rurality. These characteristics have been identified in the literature as risk factors for incidence of ESRD, worse dialysis quality outcomes, and higher healthcare utilization among ESRD patients.9,18–21 We determined the rurality of dialysis facilities on the basis of the Rural Urban Commuting Area (RUCA) designation of each facility’s ZIP Code. Facilities in urban-designated ZIP Codes were coded as 0 and all classified, nonurban ZIP Codes were coded as 1.22,23

We focused on outcomes that reflect the core ESRD QIP facility performance and clinical measures (eg, comprehensive dialysis adequacy, hypercalcemia >10.2, hemoglobin [Hgb] >12 g/dL, urea reduction ratio [URR] >65, fistula use, and catheter use) and related nonESRD QIP measures, including mortality and healthcare utilization (eg, mean monthly all-cause hospital admissions and emergency department visits). We obtained measures of comprehensive dialysis adequacy and hypercalcemia from the CROWNWeb clinical extract and other measures from Medicare claims.

Estimation Strategy

First, we estimated linear probability models to examine differences in performance on ESRD QIP across facilities by the share of patients with social risk factors (model 1). The outcomes of interest included a binary indicator of whether a facility received a payment reduction during the calendar year. The independent variable was the proportion of patients with social risk factors. Regressions adjusted for facility characteristics including ownership (large dialysis organization, other dialysis organization, independent), location (hospital-based, satellite, freestanding), ESRD Network, and other characteristics as of December 31 of each year: number of patients, proportion of the patients by sex, age category, dialysis modality, cause of ESRD, and time with ESRD. Social risk factors were calculated at the facility level and based on the share of patients with Black race, Hispanic ethnicity, dual, and rurality. We determined facility-patient mix on the basis of the characteristics of patients dialyzed for at least 90 days and attributed to the facility by December 31 of each year.

Second, we used an interrupted time series model to examine differences in dialysis quality across patient groups and between facilities by the share of patients with social risk factors. These models included four 3-way interactions of a time trend, an indicator of whether the ESRD QIP measure corresponding to the dependent variable was in the ESRD QIP in a given year, and indicators for each beneficiary-level social risk factor (model 2). The interaction terms in the regressions comparing patients attributed to facilities by the share of patients with social risk factors included an indicator of the proportion of patients attributed to the facility each year with each social risk factor (model 3). Each regression included data from all years for which each outcome was measured.

Third, we examined the relationship of social risk factors and ESRD patient outcomes outside the ESRD QIP measure set (eg, probability of death during calendar year, hospitalizations, and emergency department utilization). Here, the beneficiary-level linear regression models we estimated included 2-way interactions of social risk factors with a time trend (model 4: beneficiary-level social risk factors; model 5: facility-year level indicator of proportion of patients attributed to the facility).

To aid in the interpretation of results from multiple interaction terms used in models 2 through 5, we present differences in regression-adjusted changes in outcomes over time for different patient groups. For measures of dialysis quality, we compare outcomes by patient group for performance years the measure was versus was not in the ESRD QIP measure set (models 2 and 4). Because there was no obvious pre/post comparison for mortality and healthcare utilization outcomes, we estimated differences across groups during the first and last year of the study period. For the comparisons among patients attributed to facilities with different mixes of social risk factors (models 3 and 5), we calculated differences in predicted outcomes at the 80th versus 20th percentile of the overall facility distribution for each social risk factor (appendix tables A3 and A4, Supplemental Digital Content 1, https://links.lww.com/MLR/C488 show the proportion of facilities in top or bottom quintile of multiple social risk factors and the patient-casemix at the 20th and 80th percentiles).

Regression models 2–5 included adjustments for patient characteristics (age, age squared, sex, dialysis modality, cause of ESRD, receipt of nephrology care before ESRD, time with ESRD, and Medicare claims-based indicators for diagnosis of a wide range of chronic conditions using Chronic Conditions Warehouse algorithms applied to claims records from the prior calendar year24) and facility characteristics (characteristics as of December 31 of each year, including number of patients and the proportion of the patient population by sex, age category, dialysis modality, cause of ESRD, and time with ESRD). We included facility fixed effects in models that included the beneficiary-level social risk factors (models 2 and 4). The models that included facility-level social risk factors did not include facility fixed effects but did include additional facility-level control variables, including ownership type, location (hospital based, satellite, freestanding), and ESRD network. Models 2 through 5 were weighted by the number of measure-eligible patient-months, except for regressions on unplanned readmissions, which were weighted by the number of index hospital discharges, and on mortality, which was not weighted. SE were clustered by facility.

RESULTS

Sample Characteristics

Approximately one-third of the analytic sample were Black patients, 17% were Hispanic, 36% were dually eligible for Medicaid, and 9% were attributed to dialysis facilities in rural ZIP Codes (Table 1). Diabetes was a more common cause of ESRD for Hispanic patients, although hypertension was a more common cause for Black patients. Patients receiving care at rural dialysis facilities were more likely to have had pre-ESRD nephrology care. Black and Hispanic patients were less likely to have had pre-ESRD nephrology care.

TABLE 1 - Characteristics of Medicare ESRD Dialysis Patients, Calendar Years 2010−2018
All ESRD Patients Black Hispanic Dual Eligibility Rural
Patient Characteristic Mean Mean Mean Mean Mean
Age
 <18 0.2 0.2 0.4 0.3 0.1
 18−44 14.1 17.1 17.8 19 12.8
 45−64 43.3 48.3 45.9 47 42.5
 64−74 23.5 21.1 22.1 20.4 25
 75+ 18.9 13.3 13.7 13.3 19.7
Female 43.8 46.4 42.1 49.8 44.9
Race
 Black 36 100 2 43.8 27.7
 White 57.1 0 96.5 48.7 64.8
 Multiracial/other 6.9 0 1.5 7.5 7.5
Hispanic or Latino 17.3 1 100 19.5 8.3
Dual eligibility for Medicaid 35.6 43.3 40.1 100 41.8
Treated at rural facility 8.8 6.7 4.2 10.3 100
Primary cause of ESRD
 Diabetes 45.1 39.7 57.8 46.1 48
 Glomerulonephritis 8.9 8.1 8.2 8.6 8.5
 Hypertension 29.2 39.3 20.4 29.3 26.2
 Other/unknown 16.8 12.8 13.6 15.9 17.3
Time with ESRD
 <1 y 23.5 19.2 22.6 16.8 23.5
 1−2 y 28.9 25.9 28.3 26.7 29.8
 3−4 y 17.9 18.2 18.6 19.2 18
 5+ y 29.7 36.7 30.5 37.3 28.8
Modality
 In-center HD 88.3 91.5 90.8 92.1 90.7
 Home HD 1.4 1.1 0.6 1 0.8
 PD 9 6.3 7.5 5.8 7.2
 Other/unknown 1.4 1.2 1.1 1.1 1.3
Prior nephrologist care 53.3 47.3 45.4 45.4 56.8
Patients-years (n) 4,896,259 1,656,154 803,035 1,677,770 440,585
Percentages may not sum to 100 because of rounding or missing values. χ2 test of independence of each subgroup shown in columns from the subgroups not shown indicates subgroups are significantly different from one another on all characteristics at P<0.001.
ESRD indicates end-stage renal disease; HD, hemodialysis; PD, peritoneal dialysis.
Sources: CROWNWeb data, CMS PUF, and Medicare claims.

Unadjusted sample means for the outcome measures reveal several differences across patient groups (Table 2). We found that Black patients were less likely to meet the arteriovenous (AV) fistula use targets (57% vs. 63% for the entire sample) and had lower mortality compared with other patients (12% vs. 14% for the entire sample). Hispanic patients fared better than other patient groups on most outcomes, including mortality (11% vs. 14% for the entire sample) and emergency department utilization (0.24 vs. 0.26 visits per month for the entire sample). There was no clear pattern to differences between ESRD patients in rural versus nonrural areas.

TABLE 2 - Mean of Dialysis Quality Measures, Mortality, and Healthcare Utilization by ESRD Patient Characteristic, Calendar Years 2010 to 2018
All ESRD Patients Black Hispanic Dual Eligibility Rural
Outcomes Mean SD Mean SD Mean SD Mean SD Mean SD
ESRD QIP measures
 AV fistula use 63.0 (45.7) 57.2 (47.3) 69.0 (43.7) 61.2 (46.2) 65.4 (44.8)
 Comprehensive Kt/v 93.5 (16.3) 93.1 (16.4) 93.7 (16.4) 95.0 (13.1) 94.8 (14.0)
 URR >65 96.8 (10.8) 96.2 (11.5) 97.6 (9.2) 96.4 (11.4) 97 (10.5)
 Catheter use 11.4 (28.8) 10.9 (28.2) 9.41 (26.2) 12.1 (29.6) 11.1 (28.0)
 Hgb >12 g/dL 7.8 (13.3) 7.7 (13.1) 8.27 (13.8) 7.85 (13.2) 7.84 (13.5)
 Hypercalcemia >10.2 1.2 (7.1) 1.4 (7.5) 1.07 (6.6) 1.23 (7.0) 1.3 (7.3)
 Unplanned 30-day readmissions 28.4 (30.0) 29.3 (30.1) 26.8 (29.3) 30.4 (30.2) 25.7 (29.4)
 Transfusions 4.61 (13.9) 4.63 (13.4) 3.87 (11.9) 4.7 (13.5) 4.69 (14.3)
Non-ESRD QIP measures
 ED visits, monthly (n) 0.265 (0.442) 0.287 (0.497) 0.244 (0.394) 0.310 (0.512) 0.271 (0.445)
 Inpatient admissions, monthly (n) 0.147 (0.270) 0.148 (0.274) 0.137 (0.256) 0.159 (0.284) 0.128 (0.241)
 Mortality rate, annual 14.3 (35.0) 11.7 (32.2) 1.0 (31.3) 13.9 (34.6) 16.1 (36.8)
AV indicates arteriovenous; ED, emergency department; ESRD, end-stage renal disease; Hgb, hemoglobin; QIP, Quality Improvement Program; URR, urea reduction ratio.
Sources: CROWNWeb data, CMS Dialysis Facility Compare, and Medicare claims.

ESRD QIP Measure Scores and Facility Patient Characteristics

When we compared measure scores for facilities with the 80th versus 20th percentile of the distribution of facility case-mix for different patient characteristics, we found facilities with a higher percentage of Black patients scored worse on most clinical measures (Table 3). For example, across payment years 2012−2020, facilities with the 80th percentile of percent Black patients scored, on average, 1.95 points lower (out of 10) on the AV fistula use measure, compared with facilities with the 20th percentile of percent Black patients. These facilities also scored lower on the comprehensive Kt/V measure (−0.70 points), hypercalcemia (−0.43), standardized readmissions ratio (−0.44), and standardized transfusion ratio (−0.54).

TABLE 3 - Regression-Adjusted Differences in Dialysis Facility Clinical and Reporting Measure Scores Between Facilities With Patient Population at the 80th Versus 20th Percentile of Distribution of Facility-Patient Mix for ESRD Calendar Years 2010−2018
Facility Score Black Hispanic Dual Eligibility Rural
Outcomes Mean Mean Difference Mean Difference Mean Difference Mean Difference
ESRD QIP measure scores (0-10)
 AV fistula use 5.39 −1.95 *** 0.20 *** −0.56 *** 0.41 ***
 Comprehensive Kt/v 7.57 −0.70 *** 0.07 −0.07 0.39 ***
 URR >65 8.98 0.11 0.13 −0.09 −0.15 *
 Catheter use 5.72 0.21 * 0.17 *** −0.50 *** 0.34 ***
 Hgb >12 g/dL 9.6 0.02 −0.01 −0.03 0.02
 Hypercalcemia >10.2 7.46 −0.43 *** −0.01 0.03 0.01
 Unplanned 30-day readmissions 4.85 −0.44 *** −0.02 *** −0.19 *** 0.90 ***
 Transfusions 5.31 −0.54 *** 0.12 0.02 −0.1
Payment reductions (0/1)
 2012 0.31 0.04 −0.02 0.03 * 0.05 *
 2013 0.11 −0.04 ** −0.02 * −0.01 0.01
 2014 0.04 0.03 ** <0.01 0.01 0.01
 2015 0.06 0.03 * <0.01 0.01 −0.01
 2016 0.05 0.04 *** <0.01 0.01 −0.01
 2017 0.2 0.12 *** −0.01 0.06 *** −0.04 **
 2018 0.14 0.11 *** −0.02 ** 0.03 *** −0.01
 2019 0.27 0.16 *** −0.02 0.04 *** −0.06 ***
 2020 0.42 0.16 *** −0.01 0.03 ** −0.11 ***
 All years 0.18 0.08 *** −0.01 * 0.02 *** −0.02 ***
Cells in the ‘Facility Score’ column present the overall pooled sample mean score on each ESRD QIP measure shown. The remaining columns present the difference in linear regression predicted values at the 80th versus 20th percentile of facility patient mix of Black (0.64 vs. 0.04), Hispanic (0.23 vs. 0.0), and dual-eligible (0.47 vs. 0.24) patients dialyzed at least 90 days as of December 31 across payment years 2012−2020. Results in the ‘Rural’ column are based on difference in the predicted outcomes of rural and urban facilities. Controls include ownership, location (hospital-based, satellite, and freestanding), network, facility characteristics as of December 31 of each year, including number of patients and the proportion of the patient population by sex, age category, dialysis modality, cause of ESRD, and time with ESRD. Standard errors were clustered at the facility.
AV indicates arteriovenous; ESRD, end-stage renal disease; Hgb, hemoglobin; QIP, Quality Improvement Program; URR, urea reduction ratio.
*P<0.05.
**P<0.01.
***P<0.001.
Sources: CROWNWeb data, CMS Dialysis Facility Compare, and Medicare claims.

Facilities with a larger share of dual-eligible patients also performed worse on several ESRD QIP clinical measures during payment years 2012−2020. During this period, facilities with the 80th percentile of percent dual-eligible patients had lower measure scores for AV fistula use (−0.56), catheter use (−0.50), and unplanned readmissions (−0.19) compared with facilities with the 20th percentile of percent dual-eligible patients.

When there were differences on ESRD QIP measures, facilities with a higher proportion of Hispanic patients and facilities in rural areas performed better than facilities with no Hispanic patients and urban facilities.

Social Risk Factors and Facility Payment Reductions

ESRD QIP payment reductions were more common at dialysis facilities with a higher proportion of Black patients (8% points [pp]; P<0.001) (Table 3). The disparity in payment reductions was larger in more recent years of the ESRD QIP. Similarly, facilities with larger proportions of dual-eligible patients tended to be more likely to receive a payment reduction, with the largest differences observed in payment years 2017–2020 (6 pp, 3 pp, 4 pp, and 3 pp, respectively; P<0.01). Facilities with a high proportion of Hispanic patients were less likely to receive a payment reduction in certain ESRD QIP payment years (2013 and 2018), whereas rural facilities were less likely to receive a payment reduction in recent payment years (2017, 2019, and 2020).

Dialysis Quality Measures by Patient-level Social Risk Factors

We found disparities in unadjusted dialysis quality measures between patients with and without social risk factors (Supplemental tables Exhibit A1 and A2, Supplemental Digital Content 1, https://links.lww.com/MLR/C488). These differences often existed both before and after related quality measures were included in the ESRD QIP. For example, in the years before inclusion of a AV fistula use measure in the ESRD QIP, 52.5% of Black patients received dialysis through an AV fistula compared with 61.6% of White patients. AV fistula use improved for both groups during the years AV fistula measure was in the ESRD QIP but a disparity remained: 57.9% of Black patients received dialysis through and AV fistula compared with 66.8% of White patients.

For some patient groups on some measures, the historically disadvantaged group had higher quality measures in periods before and after the addition of related measures into the ESRD QIP. For example, Hispanic and patients treated at rural dialysis facilities have higher rates of AV fistula use before and during years the AV fistula measure was included in the ESRD QIP.

Relative Changes in Dialysis Quality Measures by Patient-level Social Risk Factors

We found no evidence that the inclusion of ESRD QIP measures systematically affected relative differences in quality outcomes across patient groups with and without social risk factors (Table 4, panel A). When results from our multivariable regressions did show relative differences in outcomes between groups over time, the change tended to be small and did not consistently favor any particular group of patients.

TABLE 4 - Regression-Adjusted Difference-in-Differences Analysis of the Relative Impact of Including Dialysis Quality Measures in the ESRD QIP Measure Set for Patients With Versus Without Social Risk Factors, Calendar Years 2010−2018
Outcomes Black Hispanic Dual Eligibility Rural
Panel A: Patient characteristics
 AV fistula use 0.31 (0.21) −0.15 (0.27) 1.03 (0.18) ***
 Comprehensive Kt/v 0.00 (0.07) 0.47 (0.10) *** −3.47 (0.07) ***
 URR >65 0.62 (0.06) *** 0.65 (0.07) *** −0.19 (0.04) ***
 Catheter use 0.62 (0.15) *** 1.55 (0.19) *** −0.94 (0.13) ***
 Hgb >12 g/dL 0.01 (0.06) 0.42 (0.09) *** 0.02 (0.04)
 Hypercalcemia >10.2 −0.33 (0.06) *** 0.17 (0.07) * 0.15 (0.04) ***
 Unplanned readmissions >−0.01 (<0.01) >−0.01 (<0.01) >−0.01 (<0.01) *
 Transfusions <0.01 (<0.01) *** <0.01 (<0.01) *** >−0.01 (<0.01) ***
Panel B: Facility patient mix
 AV fistula use 1.05 (0.33) ** 0.15 (0.17) −0.13 (0.30) 0.70 (0.39)
 Comprehensive Kt/v −0.20 (0.15) −0.02 (0.09) −0.60 (0.16) *** −0.78 (0.15) ***
 URR >65 0.87 (0.11) ** 0.28 (0.05) ** −0.28 (0.09) ** 0.22 (0.13)
 Catheter use 0.61 (0.23) ** 0.61 (0.11) *** −0.06 (0.22) 0.23 (0.29)
 Hgb >12 g/dL 0.16 (0.13) 0.24 (0.07) *** 0.12 (0.10) 0.61 (0.16) ***
 Hypercalcemia >10.2 −0.58 (0.11) *** 0.05 (0.04) −0.05 (0.07) −0.53 (0.14) ***
 Unplanned readmissions −0.28 (0.19) 0.13 (0.09) 0.07 (0.15) −0.09 (0.24)
 Transfusions 0.48 (0.05) *** 0.12 (0.03) * −0.14 (0.05) ** 0.05 (0.07)
Each estimate represents the relative difference in predicted outcome between patients with versus without social risk factors (panel A) and attributed to facilities with the 80th percentile versus 20th percentile of the distribution in each social risk factor (panel B) during years each measure was included in ESRD QIP versus not included. Regression estimates adjust for patient characteristics (age, age squared, sex, dialysis modality, cause of ESRD, receipt of nephrology care before ESRD, time with ESRD, and indicators for diagnosis of a wide range of chronic conditions) and facility characteristics (characteristics as of December 31 of each year, including number of patients, and the proportion of the patient population by sex, age category, dialysis modality, cause of ESRD, and time with ESRD). Regression results shown in panel A were estimated with facility fixed effects. Panel B regressions include additional controls for facility ownership type, location, and ESRD Network. ESRD QIP incentives lower values on catheter use, Hgb >12 g/dL, hypercalcemia, unplanned readmissions, and transfusions. All models were weighted by the number of measure-eligible patient-months, except for unplanned readmissions, which were weighted by the number of index hospital discharges. Standard errors (shown in parenthesis) were clustered by facility. Point estimates shown with inequalities are between 0 and the shown value.
AV indicates arteriovenous; ESRD, end-stage renal disease; Hgb, hemoglobin; QIP, Quality Improvement Program; URR, urea reduction ratio.
*P<0.05.
**P<0.01.
***P<0.001.
Sources: CROWNWeb data, CMS Dialysis Facility Compare, and Medicare claims.

The results in panel A of Table 4 show the measured change for performance periods when each measure was in the ESRD QIP minus when it was not in the ESRD QIP for patients with each social risk factor relative to the reference group. For the first three quality measures listed in the table (AV fistula, comprehensive Kt/V, and URR), positive estimates indicate greater improvement in the social risk factor group after the addition of the measure to the ESRD QIP measure set. For the remaining measures (catheter, Hgb >12, hypercalcemia, unplanned readmissions, and transfusions), negative estimates indicate greater improvement in the social risk factor group.

After regression adjustment, we found that Black patients performed worse or similar to White patients on each ESRD clinical measure in each performance period we analyzed. Black patients improved more than White patients on the hypercalcemia measure in periods after that measure was added to the ESRD QIP measure set (0.33 pp larger improvement). In a few other cases, White patients experienced a larger improvement in outcomes compared with Black patients. This finding was true for catheter use and URR (0.62 pp and 0.62 pp larger improvements, respectively; P<0.001 for all comparisons).

Overall, Hispanic patients had better dialysis quality and outcome measures than nonHispanic patients. For some outcomes, such as comprehensive Kt/V and Hgb measures, Hispanic patients experienced greater improvements (0.47 and 0.42 pp larger improvement, respectively) compared with nonHispanic patients but showed less improvement in measures of catheter and URR (1.55 and 0.65 pp less improvement) during periods when each measure was included in the ESRD QIP measures set compared with the periods when it was not included (P<0.001 for all comparisons).

Compared with nondual-eligible patients, dual-eligible patients had lower measured levels of dialysis quality and poorer outcomes. When measures were added to the ESRD QIP, dual-eligible patients experienced larger relative improvements in fistula use, catheter use, hypercalcemia, and URR (1.03, −0.94, 0.15, and −0.19 pp relative improvements, respectively) and lower relative improvement in comprehensive Kt/V (−3.47 pp) during the period after the measure was included in the ESRD QIP measure set (P<0.001 for all comparisons).

Relative Changes in Dialysis Quality Measures by Facility-level Social Risk Factors

The changes in patient outcomes across patients attributed to facilities with a high versus low proportion of patients with social risk factors (Table 4 panel B) mirrored the changes across patient characteristics. For example, patients at facilities with a high versus low proportion of Black patients experienced greater improvement in measures of hypercalcemia (−0.58 pp), fistula use (1.05 pp), and URR (0.87 pp). However, facilities treating larger proportions of Black patients demonstrated less relative improvement in transfusions (0.48 events per 100 patient months), catheter use (0.61 pp) compared with facilities serving a low proportion of Black patients (P<0.01 for all comparisons).

Patients attributed to dialysis facilities in rural areas had similar or better dialysis quality and health outcomes for all observed years of the ESRD QIP. When there were relative differences before and after measures were included in the ESRD QIP, the differences indicated a small convergence in outcomes for the patients at urban and rural facilities (comprehensive Kt/V, Hgb, and hypercalcemia; P<001 for each comparison).

Relative Changes in Measures of Mortality and Healthcare Utilization by Patient-level and Facility-level Social Risk Factors

We found either no change or small relative changes over time across social risk factors in emergency department use, inpatient visits, and mortality since the start of the ESRD QIP (Table 5 panels A and B). Improvements in emergency department utilization, inpatient stays, and mortality were similar for Black and White patients, patients attributed to facilities serving high versus low proportions of Black patients, and patients attributed to rural and urban facilities.

TABLE 5 - Regression Adjusted Difference-in-Differences Analysis of ESRD Patient Mortality and Healthcare Utilization for Patients With Versus Without Social Risk Factors, Calendar Years 2010 to 2018
Black Hispanic Dual Eligible Rural
Outcomes Delta SE Delta SE Delta SE Delta SE
Panel A: Differences by patient characteristic
 ED visits (n) −0.003 (0.002) 0.007 (0.003) ** −0.000 (0.002)
 Inpatient visits (n) 0.000 (0.001) 0.009 (0.002) *** −0.005 (0.001) ***
 Mortality rate −0.002 (0.001) −0.005 (0.002) ** 0.004 (0.001) ***
Panel B: Differences by facility patient mix
 ED visits (n) 0.005 (0.003) 0.011 (0.001) *** 0.003 (0.002) −0.001 (0.004)
 Inpatient visits (n) 0.001 (0.002) 0.008 (0.001) *** −0.005 (0.001) *** −0.001 (0.002)
 Mortality rate −0.001 (0.002) 0.001 (0.001) 0.005 (0.001) *** −0.004 (0.002)
Each cell represents the relative difference in predicted outcome between 2018 and 2010 (ESRD payment years 2020 and 2012) for patients with versus without social risk factors (panel A) and for patients attributed to facilities with the 80th percentile versus 20th percentile of the distribution in each social risk factor (panel B). Regression estimates adjust for patient characteristics (age, age squared, sex, dialysis modality, cause of ESRD, receipt of nephrology care before ESRD, time with ESRD, and indicators for diagnosis of a wide range of chronic conditions) and facility characteristics (characteristics as of December 31 of each year, including number of patients and the proportion of the patient population by sex, age category, dialysis modality, cause of ESRD, and time with ESRD). Regression results shown in panel A were estimated with state fixed effects. Panel B regressions include additional controls for facility ownership type, location, and ESRD Network. The inpatient and ED visit models were weighted by eligible patient-months. SE were clustered by facility.
ED indicates emergency department; ESRD, end-stage renal disease.
*P<0.05.
**P<0.01.
***P<0.001.
Sources: CROWNWeb data, CMS Dialysis Facility Compare, and Medicare claims.

We observed relatively smaller reductions in emergency department and inpatient stays for Hispanic patients compared with nonHispanic patients from payment year 2012−2020. These results reflect a modest convergence in emergency department visits and inpatient admissions between Hispanic and nonHispanic groups because nonHispanic patients had higher average utilization rates over all periods.

Mortality among nondual-eligible patients improved more than among dual-eligible patients from payment year 2012−2020. The relatively large improvements in inpatient and emergency department utilization for dual-eligible patients compared with nondual-eligible patients during the same period represent a reduction in disparity in these outcomes.

DISCUSSION

During the first 9 years of the ESRD QIP facilities with higher proportions of Black and dual-eligible patients received higher rates of ESRD QIP payment reductions. This was especially true in the most recent years of the program. However, we did not find evidence of widening disparities in quality outcomes for these patient groups. Differences in outcomes between patients with and without social risk factors persisted over time. Similarly, differences in outcomes between facilities with high versus low proportions of patients with social risk factors also persisted. The measures added to the ESRD QIP had little effect on existing differences in outcomes, either across patient subgroups or across facilities based on their patient mix with social risk factors.

There may be several explanations for these findings. It is possible that facilities treating patients with high proportions of social risk factors provide lower quality care, which would suggest the need for greater improvement efforts at those facilities. Comparing similar patients across facilities, we found patients treated in facilities with higher proportions of some social risk factors performed worse on measures of dialysis quality than if they were treated in a facility with a low proportion of patients with social risk factors. This finding suggests at least some of the differences in outcomes are attributable to differences across facilities in the level of care they provide.

It is also important to consider whether disparities in the outcomes we evaluated are systematically worse for patients with certain social risk factors for reasons unrelated to dialysis facility quality of care. As with other value-based purchasing programs, such an explanation poses a longer-term risk that the ESRD QIP may establish disincentives for facilities to care for patients with certain social risk factors if facilities perceive that doing so will put them at a disadvantage based on how quality is assessed under the program.

This study found that within the same facility, Black patients and dual-eligible patients tend to have worse dialysis quality than their White and nondual-eligible peers. These findings do not decisively point to a single explanation because they could be the result of facilities providing different levels of care to patients within the facility or the result of factors unrelated to the care provided by the facility. Further research is needed to ascertain whether additional approaches could reduce the risk that facilities are held accountable for factors beyond their control while also preserving incentives for quality improvement among providers that disproportionately care for patients with social risk factors.

This study also identified a pattern of more frequent ESRD QIP payment reductions among facilities treating higher proportions of Black patients and dual-eligible patients. This result appears to be related to the changing mix of measures included in the ESRD QIP, rather than widening disparities over time for existing measures. The proportion of Black and dual-eligible patients at a facility is associated lower readmission and transfusion measure scores.

Although the ESRD QIP was not originally designed with the goal of reducing health disparities, the program is positioned to be a lever that affects health equity. It is important that the ESRD QIP continue to incentivize quality improvements among facilities caring for a disproportionate share of patients with social risk factors. With such incentives, the program has the potential to eventually contribute to the narrowing of gaps in quality. Moreover, last year’s ESRD QIP federal rule included requests for information about possible approaches CMS could implement to improve health equity.25 More targeted approaches to reduce gaps in quality may have stronger potential to reduce disparities among ESRD patients.

CONCLUSIONS

Despite improvements in dialysis quality and mortality under the ESRD QIP, persistent disparities remain for dialysis patients. Dialysis facilities caring for a disproportionate share of patients with certain social risk factors have experienced a recent pattern of more frequent ESRD QIP payment reductions. Although the financial incentives for facilities to improve their performance have the potential to reduce disparities in patient outcomes, this improvement has not occurred as of payment year 2020. Fortunately, however, concerns that value-based purchasing programs could lead to widening disparities in patient outcomes have not borne out under the ESRD QIP.

REFERENCES

1. United States Renal Data System. 2019 USRDS Annual Data Report: Epidemiology of Kidney Disease in the United States; 2019.
2. U.S. Department of Health and Human Services Office of Disease Prevention and Health Promotion. Healthy People 2020: Chronic Kidney Disease. 2014. Available at: https://www.healthypeople.gov/2020/topics-objectives/topic/chronic-kidney-disease. Accessed August 21, 2019.
3. Patzer RE, Mcclellan WM. Influence of race, ethnicity and socioeconomic status on kidney disease. Nat Rev Nephrol. 2012;8:533–541.
4. Young EW, Mauger EA, Jiang KH, et al. Socioeconomic status and end-stage renal disease in the United States. Kidney Int. 1994;45:907–911.
5. Norton JM, Moxey-Mims MM, Eggers PW, et al. Social determinants of racial disparities in CKD. J Am Soc Nephrol. 2016;27:2576–2595.
6. United States Renal Data System. 2018 USRDS annual data report: Epidemiology of kidney disease in the United States, Volume 2: End-stage renal disease in the United States. National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases, Bethesda, MD, 2018.
7. Kirchhoff SM. Medicare coverage of end-stage renal disease (ESRD). Congressional Research Service; 2018. Available at: http://fas.org/sgp/crs/misc/R45290.pdf. Accessed March 8, 2019.
8. Centers for Medicare & Medicaid Services. ESRD Quality Incentive Program. 2018. Available at: https://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/ESRDQIP/index.html. Accessed August 20, 2019.
9. Office of the Assistant Secretary for Planning and Evaluation. Report to Congress: Social Risk Factors and Performance Under Medicare’s Value-Based Payment Programs. 2016. Available at: https://aspe.hhs.gov/sites/default/files/migrated_legacy_files//171041/ASPESESRTCfull.pdf. Accessed August 20, 2019.
10. Almachraki F, Tuffli M, Lee P, et al. Socioeconomic status of counties where dialysis clinics are located is an important factor in comparing dialysis providers. Popul Health Manag. 2016;19:70–76.
11. Zhang Y. The association between dialysis facility quality and facility characteristics, neighborhood demographics, and region. Am J Med Qual. 2016;31:358–363.
12. Saunders MR, Lee H, Chin MH. Early winners and losers in dialysis center pay-for-performance. BMC Health Serv Res. 2017;17:816.
13. Qi AC, Butler AM, Joynt Maddox KE. The role of social risk factors in dialysis facility ratings and penalties under a medicare quality incentive program. Health Aff. 2019;38:1101–1109.
14. Saunders MR, Lee H, Maene C, et al. Proximity does not equal access: racial disparities in access to high quality dialysis facilities. J Racial Ethn Heal Disparities. 2014;1:291–299.
15. Ajmal F, Probst J, Brooks J, et al. Association between freestanding dialysis facility size and medicare quality incentive program performance scores. Am J Nephrol. 2019;49:64–73.
16. Fiscella K, Burstin HR, Nerenz DR. Quality measures and sociodemographic risk factors: to adjust or not to adjust. J Am Med Assoc. 2014;312:2615–2616.
17. Casalino LP, Elster A. Will pay-for-performance and quality reporting affect health care disparities? Health Aff. 2007;26:405–414.
18. Zeng X, Liu J, Tao S, et al. Associations between socioeconomic status and chronic kidney disease: a meta-analysis. J Epidemiol Community Health. 2018;72:270–279.
19. Maripuri S, Arbogast P, Alp Ikizler T, et al. Rural and micropolitan residence and mortality in patients on dialysis. Clin J Am Soc Nephrol. 2012;7:1121–1129.
20. Mehrotra R, Kermah D, Fried L, et al. Racial differences in mortality among those with CKD. J Am Soc Nephrol. 2008;19:1403–1410.
21. Madden EF, Qeadan F. Dialysis hospitalization inequities by hispanic ethnicity and immigration status. J Health Care Poor Underserved. 2017;28:1509–1521.
22. USDA Economic Research Service. Rural-Urban Continuum Codes. 2013. Available at: https://www.ers.usda.gov/data-products/rural-urbancontinuum-codes.aspx. Accessed March 8, 2019.
23. University of Washington Rural Health Research Center. Rural Urban Commuting Area Codes Data V2.0. 2005. Accessed March 8, 2019.
24. Centers for Medicare & Medicaid Services Chronic Conditions Data Warehouse. Condition Categories. 2014. Available at: https://www2.ccwdata.org/web/guest/condition-categories. Accessed July 31, 2019.
25. Department of Health and Human Services Centers of Medicare & Medicaid Services. Medicare Program; End-Stage Renal Disease Prospective Payment System, Payment for Renal Dialysis Services Furnished to Individuals With Acute Kidney Injury, End-Stage Renal Disease Quality Incentive Program, and End-Stage Renal Disease Treatment Choices M. 2021. Available at: https://www.govinfo.gov/content/pkg/FR-2021-07-09/pdf/2021-14250.pdf. Accessed April 17, 2022.
Keywords:

end stage renal disease; quality incentive program; value based purchasing; dialysis; chronic kidney disease

Supplemental Digital Content

Copyright © 2022 Wolters Kluwer Health, Inc. All rights reserved.