Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2020 Nov 13.
Published in final edited form as: Health Aff (Millwood). 2016 Aug 1;35(8):1461–1470. doi: 10.1377/hlthaff.2015.0394

Accounting For Patients’ Socioeconomic Status Does Not Change Hospital Readmission Rates

Susannah M Bernheim 1, Craig S Parzynski 2, Leora Horwitz 3, Zhenqiu Lin 4, Michael J Araas 5, Joseph S Ross 6, Elizabeth E Drye 7, Lisa G Suter 8, Sharon-Lise T Normand 9, Harlan M Krumholz 10
PMCID: PMC7664840  NIHMSID: NIHMS1642973  PMID: 27503972

Abstract

There is an active public debate about whether patients’ socioeconomic status should be included in the readmission measures used to determine penalties in Medicare’s Hospital Readmissions Reduction Program (HRRP). Using the current Centers for Medicare and Medicaid Services methodology, we compared risk-standardized readmission rates for hospitals caring for high and low proportions of patients of low socioeconomic status (as defined by their Medicaid status or neighborhood income). We then calculated risk-standardized readmission rates after additionally adjusting for patients’ socioeconomic status. Our results demonstrate that hospitals caring for large proportions of patients of low socioeconomic status have readmission rates similar to those of other hospitals. Moreover, readmission rates calculated with and without adjustment for patients’ socioeconomic status are highly correlated. Readmission rates of hospitals caring for patients of low socioeconomic status changed by approximately 0.1 percent with adjustment for patients’ socioeconomic status, and only 3–4 percent fewer such hospitals reached the threshold for payment penalty in Medicare’s HRRP. Overall, adjustment for socioeconomic status does not change hospital results in meaningful ways.

In response to public reporting and payment incentives by the Centers for Medicare and Medicaid Services (CMS),1 the nation’s hospitals are intently focused on reducing hospital readmissions, and hospital readmission rates have been declining for Medicare enrollees nationwide.2–4 However, critics contend that hospitals caring for large numbers of patients of low socioeconomic status might perform poorly on CMS’s readmission measures because such patients have an unavoidably higher risk of readmission.5,6 Based on such concerns, many stakeholders have advocated for changing the readmission measures to include risk adjustment for patients’ socioeconomic status.7,8 The current measures, which do adjust for factors such as a patient’s sex, age, and health status, do not adjust for socioeconomic status. The policy rationale for this is to avoid holding hospitals to different standards for the outcomes of patients of low socioeconomic status.9,10 Controversy surrounding this issue has led to legislative proposals to change Medicare pay-for-performance programs to include socioeconomic status variables in the quality measures and to the allotment of millions of dollars to the Office of the Assistant Secretary for Planning and Evaluation by Congress for further study of the relationship between socioeconomic status and hospital quality measures.11,12

Despite the public debate, little is known about the readmission performance of hospitals caring for patients of low socioeconomic status on the publicly reported readmission measures. Studies have often examined the relationship between socioeconomic status and readmission risk at the patient level but have not always found socioeconomic status to be a risk factor for readmission.13–19 Studies that have examined the association of socioeconomic status with hospital-level readmission performance have shown that hospitals receiving high disproportionate-share hospital payments (payments intended to help cover costs of caring for uninsured patients) or hospitals with high proportions of patients eligible for both Medicare and Medicaid are more likely than others to be penalized, but these studies have not examined the range of performance among hospitals treating patients with low socioeconomic status.20–24 Aside from one evaluation of safety-net hospitals by our group, no national studies have examined the quality of hospitals caring for patients of low socioeconomic status compared with that of other hospitals, using the readmission measures in CMS reporting and payment programs.25–27

The impact of adding risk adjustment for socioeconomic status to the readmission measures is also uncertain. The stated rationale for not including socioeconomic status in current readmission risk-adjustment models used for public reporting is largely a policy decision based on conceptual concerns about setting different standards for the results of care for poor patients. However, it is valuable to understand empirically the actual impact of risk adjustment for socioeconomic status on hospital performance. Prior work assessing the effect of risk adjustment for socioeconomic status has focused on a narrow subset of hospitals and has not used the publicly reported measures.21,26–28 A recent study assessing the effect of socioeconomic status on readmission examined a wide range of factors that are not feasible to incorporate into national measures.29 Given the increasing attention to the impact of readmission policy on safety-net hospitals and the specific call for risk adjustment for patients’ socioeconomic status, it is important to examine the impact of the proposed change on hospital performance using the current reported CMS readmission measures and the measures of socioeconomic status that would likely be used in such a policy.

In this study we examined the readmission performance of hospitals caring for high proportions of patients of low socioeconomic status, to understand the range and limits of their performance compared with that of other hospitals. We then examined the change in hospital results when adjustment for the most feasible measures of socioeconomic status is added to the current CMS measures: one based on the median income of the ZIP code where the patient resides, and another based on patients’ Medicaid eligibility (that is, dual eligibility) status. Specifically, we calculated hospital risk-standardized readmission rates, using the methodology used to derive these rates for acute myocardial infarction, heart failure, and pneumonia in CMS programs (with no socioeconomic status adjustment),30–33 comparing hospitals with low versus high proportions of patients of low socioeconomic status. We then assessed the change in each hospital’s risk-standardized readmission rate after adjusting each of the readmission measures for patient socioeconomic status.

Study Data And Methods

DATA SOURCES

We used Medicare administrative claims data for hospitalizations from July 1, 2007, to June 30, 2010.We identified the cohort and the clinical risk factors using the inpatient and outpatient Standard Analytic Files. We used the Medicare enrollment file to identify Medicaid and Medicare dual eligibility as one marker of low socioeconomic status. We linked data from the 2008–12 American Community Survey, administered by the Census Bureau, to patients’ ZIP codes from the Standard Analytic Files. Using these linked data, we identified patients’ neighborhood median income and also calculated a composite of neighborhood socioeconomic indicators.

COHORT

Consistent with the readmission measures reported by CMS, we included patients ages sixty-five and older enrolled in fee-for-service Medicare who were hospitalized with a principal discharge diagnosis of acute myocardial infarction, heart failure, or pneumonia.30–33 We included patients who had been enrolled in Medicare for one year before their admission, to ensure adequate data for risk adjustment, and who remained enrolled for thirty days after their discharge, to assess for readmissions. We excluded patients who died during the hospitalization or were discharged against medical advice. For patients transferred to another acute care institution, we attributed the hospitalization to the hospital that ultimately discharged the patient to a nonacute setting.

Hospitals with fewer than twenty-five hospitalizations in the measure cohort over the three-year period were included in the analyses but were not included in the reported results; this is consistent with the approach taken for public reporting.33

SOCIOECONOMIC STATUS OF PATIENTS AND HOSPITALS

We used ZIP-code-median household income level as the primary indicator of a patient’s socioeconomic status. We linked each patient to the median income for his or her ZIP code from the 2008–12 American Community Survey.34 Patients whose ZIP codes could not be linked or who were missing a ZIP code were imputed with the median income of the patients from the hospital where they were admitted. This imputation was made for fewer than 2 percent of all discharges.

For each hospital, we determined the proportion of Medicare patients cared for at that hospital whose ZIP-code-median income was below a prespecified cutoff for “low socioeconomic status” of $43,710, which was 300 percent of the federal poverty level for a family of two in 2010. The proportion at each hospital was calculated separately for each measure cohort. We categorized hospitals into three groups using quintiles of the proportion of patients of low socioeconomic status: lowest quintile (the 20 percent of hospitals with the smallest proportion of patients of low socioeconomic status, deemed “highest socioeconomic status” hospitals); quintiles 2–4 (“middle socioeconomic status”); and highest quintile (the 20 percent of hospitals with the largest proportion of patients of low socioeconomic status, referred to as “low socioeconomic status” hospitals).

We replicated all analyses using a patient indicator for Medicaid (dual eligibility) as an alternative indicator of low socioeconomic status. For this analysis, we stratified hospitals into quintiles based on the percentage of patients within each measure who were enrolled in both Medicaid and Medicare. Because the cost of living varies, particularly between rural and urban areas, we also stratified the ZIP code–based analyses by urban and rural status, to evaluate whether the results differed within these subgroups.

Finally, as a sensitivity analysis, we examined additional facets of patients’ socioeconomic status by replicating our results using a validated composite assessment of socioeconomic indicators from the American Community Survey. This indicator includes seven different variables that assess income and poverty, as well as housing, employment, and education levels, within the ZIP code.35

RISK-STANDARDIZED READMISSION RATES

For all three measures, we first calculated risk-standardized readmission rates as currently specified for the CMS public reporting program.33 We assessed unplanned readmission for any cause to an acute care hospital within thirty days of discharge from an eligible admission.

Briefly, the risk-standardized readmission rate for each hospital is computed as the ratio of the number of “predicted” readmissions to the number of “expected” readmissions at a given hospital, multiplied by the national observed readmission rate. Predicted and expected numbers of readmissions are obtained from a two-level hierarchical logistic regression model. The approach simultaneously models data at the patient and hospital levels to account for the variation in patient readmissions within and between hospitals. The method for calculating risk-standardized rates has been previously described in detail.30–32

Each measure was risk-adjusted using covariates derived from publicly available CMS Hierarchical Condition Categories. The acute myocardial infarction model includes patient risk adjustment for twenty-nine clinical characteristics; the heart failure model includes risk adjustment for thirty-five clinical characteristics; and the pneumonia model includes risk adjustment for thirty-nine clinical characteristics.36

STATISTICAL ANALYSES

For each condition, we evaluated the patient characteristics (including comorbidities) and hospital characteristics for hospitals with low, middle, and highest socioeconomic status. We derived hospital characteristics from the 2011 American Hospital Association annual survey database.37

We compared the distributions of the risk-standardized readmission rates of low-socioeconomic-status hospitals to highest-socioeconomic-status hospitals. To examine the impact of adjusting for patients’ socioeconomic status on these rates, we first fitted a model with that status as the only factor, to understand the impact of socioeconomic status without adjustment for patient comorbidities. We then fitted the CMS model of risk-standardized readmission rates and included the patient-level socioeconomic status indicator in the model. For the socioeconomic status indicator for income, we categorized patients based on quartile cutoffs of ZIP-code-median income.We assessed the agreement between the risk-standardized readmission rates estimated by the current CMS model and socioeconomic status–adjusted model using interclass correlation. We calculated the interclass correlation using a one-way random effects model weighted by the inverse variance of each hospital’s estimated risk-standardized readmission rate. When the two risk-standardized readmission rates are close to each other, the ratio approaches 1, which indicates strong agreement. We also assessed the change in risk-standardized readmission rates from the CMS model with both indicators added to the model.

Finally, we estimated the percentage of hospitals that would be penalized under the Hospital Readmissions Reduction Program (HRRP) for failing to meet the quality measures with and without risk adjustment for patients’ socioeconomic status. The HRRP determines payment penalties on the basis of the ratio of predicted readmissions to expected readmissions (excess readmission ratio). Hospitals for which this ratio is greater than 1 face a payment penalty. We examined what proportion of hospitals with low socioeconomic status that had excess readmission ratios greater than 1 change to having ratios less than or equal to 1 when adjustment is made for patients’ socioeconomic status.

All analyses were performed separately for acute myocardial infarction, heart failure, and pneumonia and were conducted using SAS software, version 9.4. The Human Investigation Committee at the Yale University School of Medicine approved an exemption for the authors to use CMS claims and enrollment data for research analyses and publication.

LIMITATIONS

There were a number of limitations to our work. Socioeconomic status is not a singular patient-level characteristic that can be easily measured. We classified hospitals on the basis of their proportion of patients eligible for Medicaid and from ZIP codes with low socioeconomic status, and we adjusted for these patient-level factors. These variables, although widely used, imprecisely reflect each patient’s status; in particular, ZIP code is not as narrowly specified a geographic area as the census tract or census block group. Nevertheless, our primary aim was to simulate a feasible policy response to the current call for risk adjustment and therefore to use patient-level characteristics for risk adjustment that are broadly available.

Additionally, we note that although the risk-standardized readmission rates are reported as a measure of hospital quality, a broader set of factors in addition to hospital quality influence readmission risk. Safety-net hospitals serving patients of low socioeconomic status might face challenges preventing readmission because of factors such as hospital financial status and lack of community supports (for example, availability of postacute providers), which might influence regional patterns of readmissions; the study of these factors, beyond the scope of this work, could yield different results.38

Study Results

COHORT

The study included 526,272 acute myocardial infarction admissions to 4,432 hospitals, 1,278,296 heart failure admissions to 4,733 hospitals, and 1,099,230 pneumonia admissions to 4,773 hospitals.

HOSPITAL CHARACTERISTICS

Hospitals deemed low socioeconomic status, on average, had more than 90 percent of their measured patients living in a ZIP code with a median household income of less than $43,710. Those with the highest socioeconomic status had, on average, fewer than 1 percent of measured patients from low-income ZIP codes admitted for acute myocardial infarction and between 2 percent and 3 percent admitted for heart failure and pneumonia (Exhibit 1).

EXHIBIT 1.

Hospital characteristics for the low and highest quintiles of patient socioeconomic status, based on ZIP-code-median income

Acute myocardial infarction cohort Heart failure cohort Pneumonia cohort
Variable All Highest SES Low SES All Highest SES Low SES All Highest SES Low SES
Number of hospitals (%) 4,432 (100.0%) 888 (20.0%) 892 (20.1%) 4,733 (100.0%) 949 (20.1%) 947 (20.0%) 4,773 (100.0%) 956 (19.9%) 955 (20.0%)
Number of annual admissions for condition (standard deviation) 118.8 (193.9) 75.2 (137.4) 29.5 (65.8) 270.2 (325.0) 259.4 (311.9) 126.2 (143.5) 230.2 (218.3) 231.4 (223.5) 146.2 (127.9)
Mean percent of patients from low-income ZIP codes (SD) 44. 8% (36.1) 0.9% (1.4) 96.9% (4.3) 45.5% (35.0) 2.7% (2.3) 94.5% (3.9) 44.4% (34.5) 3.0% (2.2) 93.4% (4.1)
Cardiac procedure provided by hospital
 Coronary artery bypass graft 36.4% 30.4% 10.0% 34.7% 33.7% 9.7% 34.5% 34.9% 9.3%
 Percutaneous coronary intervention 11.8 12.4 10.8 11.1 13.8 9.4 11.1 14.1 10.4
 Other 51.9 57.2 79.1 54.2 52.5 80.9 54.4 51.0 80.3
Ownership Public 22.3% 22.4% 35.3% 23.8% 19.5% 35.8% 24.0% 20.2% 36.1%
Teaching status
 Member of Council of Teaching Hospitals and Health Systems 6.5% 3.3% 0.8% 6.2% 4.1% 0.8% 6.2% 4.1% 0.9%
 Affiliated with residency or fellowship program 12.3 10.7 3.8 11.8 11.7 3.5 11.8 12.6 3.7
 Nonteaching 81.1 86.0 95.4 82.0 84.2 95.7 82.0 83.4 95.4
Core-Based Statistical Area Rural 24.0% 21.9% 50.5% 25.9% 16.1% 52.1% 25.9% 15.5% 51.6%
Census division
 New England 4.1% 9.1% 0.7% 3.9% 9.3% 0.9% 3.9% 9.4% 0.9%
 Middle Atlantic 9.0 11.7 1.4 8.6 13.6 1.3 8.5 13.8 1.2
 South Atlantic 15.1 7.7 16.8 14.7 8.5 17.9 14.7 8.6 17.9
 East North Central 15.9 15.7 9.2 15.2 15.6 8.1 15.1 14.4 8.9
 East South Central 8.8 1.3 22.7 8.6 0.3 23.3 8.5 0.6 23.5
 West North Central 14.1 21.1 12.9 14.6 18.0 11.3 14.5 17.7 9.9
 West South Central 13.5 6.1 19.6 14.1 6.5 21.1 14.1 6.5 21.2
 Mountain 7.3 8.3 6.8 8.1 7.5 7.8 8.3 7.9 7.5
 Pacific 11.1 18.9 4.8 11.0 20.2 3.8 11.1 20.7 3.9
Open in a new tab

source Authors’ analysis of Medicare administrative claims data for hospitalizations from July 1, 2007, to June 30, 2010, and data from the American Community Survey, 2008–12. note “Highest SES” (socioeconomic status) is quintile 1; “low SES” is quintile 5.

Patients at low-socioeconomic-status hospitals had higher rates of many, but not all, comorbid diseases. For example, in the acute myocardial infarction cohort, low-socioeconomic-status hospitals had 42.1 percent of patients with comorbid congestive heart failure, compared to 35.5 percent at highest-socioeconomic-status hospitals; 47.4 percent with diabetes, compared to 41.7 percent; and 36.4 percent with chronic obstructive pulmonary disease, compared to 26.7 percent (see online Appendix Exhibits A1a–c).36 These patterns were consistent across the different measure conditions.

On average, low-socioeconomic-status hospitals had fewer patients, were less likely to perform cardiac procedures, and were more likely to be public hospitals and nonteaching hospitals compared to the highest-socioeconomic-status hospitals (Exhibit 1). The former were predominantly rural and more likely to be found in the East South Central or West South Central regions of the country.

RISK-STANDARDIZED READMISSION RATES

The distributions of performance on the CMS risk-standardized readmission measures for hospitals with low versus the highest socioeconomic status overlap substantially, regardless of measure (Appendix Exhibits A2a–c).36 However, the distribution of low-socioeconomic-status hospitals was shifted modestly to the right, with slightly higher readmission rates for all three measures. The median risk-standardized readmission rate for low-socioeconomic-status hospitals was an absolute 0.3 percentage point higher for acute myocardial infarction, 0.6 percentage point higher for heart failure, and 0.4 percentage point higher for pneumonia. There were hospitals with readmission rates at the extremes of the range in both groups for all three conditions (Exhibit 2).

EXHIBIT 2.

Distribution of risk-standardized readmission rates (RSRR) by ZIP-code-median income for acute myocardial infarction, heart failure, and pneumonia

Acute myocardial infarction Heart failure Pneumonia
Hospital SES quintile Hospital SES quintile Hospital SES quintile
Distribution of RSRR Highest SES (1) 2–4 Low SES (5) Highest SES (1) 2–4 Low SES (5) Highest SES (1) 2–4 Low SES (5)
Mean 19.9 19.8 20.2 24.7 24.7 25.4 18.2 18.4 18.7
Standard deviation 1.3 1.4 1.1 1.8 2.0 1.9 1.4 1.5 1.5
5th percentile 17.9 17.5 18.5 21.9 21.7 22.7 16.1 16.1 16.5
25th percentile 19.1 18.9 19.4 23.5 23.5 24.0 17.3 17.4 17.6
50th percentile 19.8 19.7 20.1 24.6 24.6 25.2 18.1 18.2 18.5
75th percentile 20.8 20.6 20.9 25.8 25.9 26.5 19.0 19.2 19.6
95th percentile 22.0 22.1 22.1 28.0 28.1 28.7 20.9 21.0 21.6
Open in a new tab

source Authors’ analysis of Medicare administrative claims data for hospitalizations from July 1, 2007, to June 30, 2010, and data from the American Community Survey, 2008–12. notes Hospitals with fewer than twenty-five cases were excluded. SES is socioeconomic status.

When hospitals were stratified by urban versus rural status, the results were similar (data not shown). Among rural hospitals, the difference in median risk-standardized readmission rates between low- and highest-socioeconomic-status hospitals was 0.5 percent for acute myocardial infarction, 0.6 percent for heart failure, and 0.6 percent for pneumonia. Among urban hospitals, the differences were 0.2 percent for acute myocardial infarction, 0.8 percent for heart failure, and 0.5 percent for pneumonia.

EFFECT OF RISK ADJUSTMENT ON RISK-STANDARDIZED READMISSION RATES

When compared with models including patients’ socioeconomic status alone, the coefficient associated with socioeconomic status was substantially lower in the full CMS model that accounts for patient comorbidities. For acute myocardial infarction, the parameter estimate decreased from 0.13 to 0.07; for heart failure, from 0.08 to 0.05; and for pneumonia, from 0.10 to 0.08 (data not shown).

In all models, socioeconomic status was a statistically significant predictor of lesser magnitude than many common comorbidities (Appendix Exhibits A3a–c).36 For the acute myocardial infarction measure, for example, the risk imparted by being from a low-socioeconomic-status ZIP code compared with the highest-socioeconomic-status ZIP code was comparable to the risk for a patient having a history of cerebrovascular disease; in contrast, the risk imparted by conditions such as heart failure and metastatic cancer were more than twice as great as the risk imparted by being from a low-socioeconomic-status ZIP code.

Exhibit 3 shows a scatterplot comparing each hospital’s risk-standardized readmission rate for heart failure calculated as it is for the CMS website Hospital Compare on the x axis, and then the same measure with the addition of patient-level risk adjustment for the patient’s ZIP-code-median income on the y axis. The scatterplots for the other measures are very similar (Appendix Exhibits A4a–b).36 The interclass correlation between the Hospital Compare measures and the same measures’ risk adjustment for patient-level median ZIP-code income was greater than 0.99 for all three measures. After adjustment for socioeconomic status, the median hospital among the low-socioeconomic-status hospitals had a risk-standardized adjusted readmission rate one-tenth of 1 percent lower (0.1 percent change in risk-standardized adjusted readmission rates for acute myocardial infarction, heart failure, and pneumonia; Exhibit 4). Similarly, the median hospital in the highest-socioeconomic-status quintile saw an absolute increase in risk-standardized readmission rate of 0.1 percent for all three conditions. The low-socioeconomic-status hospitals in the top 5th percentile of changes in risk-standardized adjusted readmission rates, when put together, had an average absolute decrease in risk-adjusted readmission rates of 0.3 percent for acute myocardial infarction, 0.4 percent for heart failure, and 0.4 percent for pneumonia. Finally, among the low-socioeconomic-status hospitals, 4 percent of hospitals (10 of 233) would change from having an excess readmission ratio of above 1 to below or equal to 1 for acute myocardial infarction; 3 percent of hospitals (21 of 787) would do so for heart failure, as would 4 percent (34 of 867) for pneumonia (data not shown).

EXHIBIT 3. Correlation of hospitals’ performance on risk-standardized readmission rates with and without adjustment for patients’ socioeconomic status: heart failure cohort.

EXHIBIT 3

Open in a new tab

source Authors’ analysis of Medicare administrative claims data for hospitalizations from July 1, 2007, to June 30, 2010, and data from the 2008–12 American Community Survey. notes Colors show quintiles of hospitals’ proportion of patients of low socioeconomic status (quintile 1 hospitals have the smallest proportion of low-socioeconomic-status patients; quintile 5 hospitals have the largest proportion).

EXHIBIT 4.

Mean hospital risk-standardized readmission rates (RSRR), socioeconomic status–adjusted RSRR, and absolute change in hospital RSRR with socioeconomic status adjustment for acute myocardial infarction, heart failure, and pneumonia readmission measures

Change in RSRR
Variable Original RSRR Standard error SES-adjusted RSRR Standard error 5th percentile Median 95th percentile Interclass correlationa
Acute myocardial infarction 0.99
 Quintile 1 (highest SES) 19.9 0.1 20.0 0.1 0.0 0.1 0.2
 Quintiles 2–4 19.8 0.0 19.8 0.0 −0.1 0.0 0.1
 Quintile 5 (low SES) 20.2 0.1 20.1 0.1 −0.3 −0.1 0.0
Heart failure 0.99
 Quintile 1 (highest SES) 24.7 0.1 24.8 0.1 0.0 0.1 0.2
 Quintiles 2–4 24.7 0.0 24.7 0.0 −0.2 0.0 0.1
 Quintile 5 (low SES) 25.4 0.1 25.2 0.1 −0.4 −0.1 0.0
Pneumonia 0.99
 Quintile 1 (highest SES) 18.2 0.0 18.4 0.0 0.0 0.1 0.3
 Quintiles 2–4 18.4 0.0 18.4 0.0 −0.2 0.0 0.2
 Quintile 5 (low SES) 18.7 0.1 18.5 0.1 −0.4 −0.1 0.0
Open in a new tab

source Authors’ analysis of Medicare administrative claims data for hospitalizations from July 1, 2007, to June 30, 2010, and data from the American Community Survey, 2008–12. notes Hospitals with fewer than twenty-five cases were excluded. SES is socioeconomic status.

a

Interclass correlation is between RSRR and SES-adjusted RSRR.

SECONDARY ANALYSES

When the hospitals were categorized based on the proportion of patients on Medicaid, within the low-socioeconomic-status hospitals, on average, more than half of the Medicare patients in the measure were also eligible for Medicaid, whereas in the highest-socioeconomic-status hospitals, fewer than 10 percent were (Appendix Exhibits A5a–c).36

The hospitals identified as low socioeconomic status on the basis of patients’ Medicaid dual-eligibility status were somewhat different than those identified as low socioeconomic status on the basis of patients’ ZIP-code-median income. Only 35 percent of hospitals deemed low socioeconomic status based on patients’ Medicaid status were rural, and they were spread more evenly across regions (Appendix Exhibits A5a–c).36 However, the performance among low-socioeconomic-status hospitals compared with the highest-socioeconomic-status hospitals showed the same pattern of results, with slightly greater differences in median performance when Medicaid dual eligibility was used as the patient socioeconomic status indicator. The difference between the median risk-standardized readmission rates for low- and highest-socioeconomic-status hospitals was 0.7 percent for acute myocardial infarction, 1.2 percent for heart failure, and 0.6 percent for pneumonia (Appendix Exhibit A6).36 The impact on the hospital risk-standardized readmission rates when adjusted for patient Medicaid status was similar to the impact when using ZIP-code-median income as the socioeconomic status indicator; the interclass correlation of hospital risk-standardized readmission rates between models with and without adjustment for socioeconomic status was greater than 0.99 for all three conditions (Appendix Exhibit A7).36 This was also true when both socioeconomic status indicators (ZIP-code-median income and dual eligibility for Medicare and Medicaid) were added to the model. The percentage of low-socioeconomic-status hospitals that changed from having an excess readmission ratio from above 1 to equal to or below 1 was 4 percent for acute myocardial infarction, 3 percent for heart failure, and 3 percent for pneumonia when Medicaid status was added to the model (data not shown).

Finally, when the composite indicator of socioeconomic status was used, the results remained essentially unchanged.

Discussion

Our study reveals little difference in hospitals’ performance on the publicly reported CMS readmission measures between hospitals with low socioeconomic status (in which 90 percent or more of patients were from low-socioeconomic-status ZIP codes) and those with the highest socioeconomic status (those caring for few patients of low socioeconomic status), regardless of the measure of socioeconomic status used or the condition studied. The range of performance on the readmission measures among low-socioeconomic-status hospitals largely overlaps with the that of the highest-socioeconomic-status hospitals, with median rates 0.3 to 0.6 percentage points higher among the former. Risk adjustment for patients’ socioeconomic status does not substantially change the characterization of hospital performance; in most cases, it lowers the risk-adjustment readmission rates of low-socioeconomic-status hospitals by approximately one-tenth of 1 percent. Moreover, risk adjustment for patients’ socioeconomic status reduces the proportion of low-socioeconomic-status hospitals facing penalties under the HRRP by less than 5 percent on any given measure.

Much of the prior research examining patients’ socioeconomic status and hospital performance in the HRRP has focused on assessing whether the proportion of patients of low socioeconomic status, or safety-net status, is associated with meeting the threshold for penalties under this program. In this study we sought to examine directly the differences in the distribution of hospital performance between hospitals characterized by their proportion of patients of low socioeconomic status on three CMS readmission measures used in the HRRP. Using the same data and methodology as CMS, we found substantial variation in the proportions of such patients served by hospitals, yet the subgroups of hospitals did not have substantially different performance. We also sought to understand the impact of risk adjustment of these measures for patient-level socioeconomic status using variables that are feasible to incorporate into the current national measures.

There are a number of potential explanations for our results and why they might seem at odds with other work examining the relationship between socioeconomic status and readmissions. First, patients of low socioeconomic status often present with greater clinical severity.39–41 The publicly reported measures are adjusted using Medicare claims for differences in patients’ clinical status, including comorbid diseases found in claims for the year prior to hospitalization. This clinical case-mix adjustment captures some of the higher inherent risk for the population with low socioeconomic status. Furthermore, hospitals exert influence on the likelihood of readmission for all of their patients through the quality of the inpatient and transitional care they provide. Because we see substantial variation in performance among hospitals with similar socioeconomic status case-mix and little difference between the overall performance of subgroups of hospitals with low and high socioeconomic status, our findings could suggest that the influence hospitals exert on the likelihood of readmission through quality and use of community resources has more effect on readmission rates than individual patient socioeconomic status does. To the extent that we found small differences between hospitals with low and high socioeconomic status, this might indicate that poorer patients are more likely to receive care in lower-quality hospitals. There is evidence for this hypothesis in the literature on race and quality outcomes.42–44 Alternatively, differences in performance between hospitals with low and high socioeconomic status could reflect inadequate care for patients of low socioeconomic status within hospitals as compared to other patients, or they might reflect remaining inherent patient differences not accounted for by claims-based risk adjusters.

It is this final hypothesis that has led to calls for patient-level risk adjustment for socioeconomic status in the readmission measures. However, consistent with prior studies examining the impact of patient-level risk adjustment,21,26 we found that additional risk adjustment for patients’ socioeconomic status would not substantially change hospital profiling or the proportion of hospitals caring for patients of low socioeconomic status that face penalties. Therefore, this risk adjustment is unlikely to be a strong policy response to protecting safety-net hospitals from pay-for-performance penalties. Other changes to this program, such as setting thresholds by hospital type or adjusting for hospital characteristics, might have more effect on penalties for providers treating more patients of low socioeconomic status.21,45 Such approaches, which would require complex decisions about setting thresholds or identifying appropriate hospital characteristics, are not the focus of our work but might warrant further examination. Ultimately, decisions about how or whether to modify measures or payment programs should be decided based on the intended goal of the programs and potential consequences.

Conclusion

The results of our study help illuminate the important differences between what might be expected based on patient-level studies and the actual results of analysis of hospital profiling. Our study shows that hospitals caring for large proportions of Medicare patients from low socioeconomic backgrounds do not differ substantially in their performance on CMS’s publicly reported thirty-day risk-standardized readmission rates from hospitals with very few patients of low socioeconomic status. Moreover, our results suggest that adding patient-level socioeconomic status risk adjustment would not change hospital results in a meaningful way.

These results are reassuring that hospitals with various mixes of patient socioeconomic status can achieve low readmission rates. However, much remains to be learned about strategies for low-socioeconomic-status hospitals and about whether such hospitals are able to improve at the same rate as others.

Supplementary Material

Appendix

Acknowledgments

The analyses upon which this article is based were performed under Contract No. HHSM-500-2008-0025I/HHSM-500-T0001, Modification No. 000008, titled “Measure and Instrument Development and Support (MIDS)—Development and Re-evaluation of the CMS Hospital Outcomes and Efficiency Measures,” funded by the Centers for Medicare and Medicaid Services, an agency of the Department of Health and Human Services (HHS). The content of this article does not necessarily reflect the views or policies of HHS. The authors assume full responsibility for the accuracy and completeness of the ideas presented. Leora Horwitz is supported by the National Institute on Aging (Grant No. K08 AG038336) and by the American Federation for Aging Research through the Paul B. Beeson Career Development Award Program. Susannah Bernheim, Craig Parzynski, Leora Horwitz, Zhenqiu Lin, Michael Araas, Joseph Ross, Elizabeth Drye, Lisa Suter, Sharon-Lise Normand, and Harlan Krumholz all receive funding under contract with CMS to develop and maintain performance measures. In addition, Krumholz and Ross are the recipients of research contracts from Medtronic and from Johnson & Johnson, through Yale University, to develop methods of clinical trial data sharing, and a contract from Medtronic for a collaborative project with the Food and Drug Administration on device surveillance. Krumholz is chair of a cardiac scientific advisory board for UnitedHealth.

Contributor Information

Susannah M. Bernheim, director of quality measurement at the Center for Outcomes Research and Evaluation (CORE) at Yale-New Haven Hospital and an assistant clinical professor in the Department of Internal Medicine at Yale School of Medicine, both in New Haven, Connecticut.

Craig S. Parzynski, senior statistician at CORE, Yale-New Haven Hospital.

Leora Horwitz, associate professor of internal medicine, population health, at New York University School of Medicine, in New York City..

Zhenqiu Lin, director of analytics at CORE, Yale-New Haven Hospital..

Michael J. Araas, research project manager at CORE, Yale-New Haven Hospital.

Joseph S. Ross, associate professor of medicine in the Department of Internal Medicine at Yale School of Medicine.

Elizabeth E. Drye, director of quality measurement at CORE, Yale-New Haven Hospital.

Lisa G. Suter, associate director of quality measurement at CORE, Yale-New Haven Hospital, and an associate professor of medicine in the Section of Rheumatology at Yale School of Medicine.

Sharon-Lise T. Normand, professor of health care policy and biostatistics at Harvard Medical School and at the Harvard T. H. Chan School of Public Health, both in Boston, Massachusetts.

Harlan M. Krumholz, Harold H. Hines, Jr. Professor of Medicine and Epidemiology and Public Health at Yale School of Medicine.

NOTES

  • 1.Centers for Medicare and Medicaid Services. [Hospital] Readmissions Reduction Program (HRRP) [Internet]. Baltimore (MD): CMS; 2015. [cited 2016 Jun 28]. Available from: http://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/AcuteInpatientPPS/Readmissions-Reduction-Program.html [Google Scholar]
  • 2.Suter LG, Li SX, Grady JN, Lin Z, Wang Y, Bhat KR, et al. National patterns of risk-standardized mortality and readmission after hospitalization for acute myocardial infarction, heart failure, and pneumonia: update on publicly reported outcomes measures based on the 2013 release. J Gen Intern Med. 2014;29(10):1333–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Gerhardt G, Yemane A, Hickman P, Oelschlaeger A, Rollins E, Brennan N. Data shows reduction in Medicare hospital readmission rates during 2012. Medicare Medicaid Res Rev. 2013;3(2):E1–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Zuckerman RB, Sheingold SH, Orav EJ, Ruhter J, Epstein AM. Readmissions, observation, and the Hospital Readmissions Reduction Program. N Engl J Med. 2016;374(16):1543–51. [DOI] [PubMed] [Google Scholar]
  • 5.Joynt KE, Jha AK. Thirty-day readmissions—truth and consequences. N Engl J Med. 2012;366(15):1366–9. [DOI] [PubMed] [Google Scholar]
  • 6.Bhalla R, Kalkut G. Could Medicare readmission policy exacerbate health care system inequity? Ann Intern Med. 2010;152(2):114–7. [DOI] [PubMed] [Google Scholar]
  • 7.American Hospital Association. RE: CMS-1599-P, Medicare program; hospital inpatient prospective payment systems for acute care hospitals and the long-term care hospital prospective payment system and proposed fiscal year 2014 rates; quality reporting requirements for specific providers; hospital conditions of participation; Medicare program; proposed rule (vol. 78, no. 91) [Internet] Chicago (IL): AHA; 2013. May [cited 2016 Jun 28]. Available from: http://www.aha.org/advocacy-issues/letter/2013/130620-cl-cms-1599p.pdf [PubMed] [Google Scholar]
  • 8.Hernandez AF, Curtis LH. Minding the gap between efforts to reduce readmissions and disparities. JAMA. 2011;305(7):715–6. [DOI] [PubMed] [Google Scholar]
  • 9.Centers for Medicare and Medicaid Services. Medicare program; hospital inpatient prospective payment systems for acute care hospitals and the long-term care hospital prospective payment system and fiscal year 2015 rates; quality reporting requirements for specific providers; reasonable compensation equivalents for physician services in excluded hospitals and certain teaching hospitals; provider administrative appeals and judicial review; enforcement provisions for organ transplant centers; and electronic health record (EHR) incentive program. Final rule. Fed Regist 2014;79(163):49853–50449. [PubMed] [Google Scholar]
  • 10.National Quality Forum. Risk adjustment for socioeconomic status or other socioeconomic factors [Internet]. Washington (DC): NQF; 2014. August [cited 2016 Jun 28]. Available from: http://www.qualityforum.org/Publications/2014/08/Risk_Adjustment_for_Socioeconomic_Status_or_Other_Sociodemographic_Factors.aspx [Google Scholar]
  • 11.Eaton S. Hospital readmission fines should reflect patient demographics, says bill by Rep Jim Renacci. Cleveland Plain Dealer [serial on the Internet] 2014. March [cited 2016 Jun 28]. Available from: http://www.cleveland.com/open/index.ssf/2014/03/hospital_readmission_fines_sho.html [Google Scholar]
  • 12.IMPACT Act of 2014, Pub. L No. 113–185; 2014.
  • 13.van Walraven C, Wong J, Forster AJ. Influence of neighborhood household income on early death or urgent hospital readmission. J Hosp Med. 2013;8(5):261–6. [DOI] [PubMed] [Google Scholar]
  • 14.Kind AJ, Jencks S, Brock J, Yu M, Bartels C, Ehlenbach W, et al. Neighborhood socioeconomic disadvantage and 30-day rehospitalization: a retrospective cohort study. Ann Intern Med. 2014;161(11): 765–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Damiani G, Salvatori E, Silvestrini G, Ivanova I, Bojovic L, Iodice L, et al. Influence of socioeconomic factors on hospital readmissions for heart failure and acute myocardial infarction in patients 65 years and older: evidence from a systematic review. Clin Interv Aging. 2015;10:237–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Eapen ZJ, McCoy LA, Fonarow GC, Yancy CW, Miranda ML, Peterson ED, et al. Utility of socioeconomic status in predicting 30-day outcomes after heart failure hospitalization. Circ Heart Fail. 2015;8(3):473–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Calvillo-King L, Arnold D, Eubank KJ, Lo M, Yunyongying P, Stieglitz H, et al. Impact of social factors on risk of readmission or mortality in pneumonia and heart failure: systematic review. J Gen Intern Med. 2013;28(2):269–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Hu J, Gonsahn MD, Nerenz DR. Socioeconomic status and readmissions: evidence from an urban teaching hospital. Health Aff (Millwood). 2014;33(5):778–85. [DOI] [PubMed] [Google Scholar]
  • 19.Joynt KE, Orav EJ, Jha AK. Thirty-day readmission rates for Medicare beneficiaries by race and site of care. JAMA. 2011;305(7):675–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Joynt KE, Jha AK. Characteristics of hospitals receiving penalties under the Hospital Readmissions Reduction Program. JAMA. 2013;309(4): 342–3. [DOI] [PubMed] [Google Scholar]
  • 21.Gu Q, Koenig L, Faerberg J, Steinberg CR, Vaz C, Wheatley MP. The Medicare Hospital Readmissions Reduction Program: potential unintended consequences for hospitals serving vulnerable populations. Health Serv Res. 2014;49(3): 818–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Berenson J, Shih A. Higher readmissions at safety-net hospitals and potential policy solutions. Commonwealth Fund Issue Briefs [serial on the Internet] 2012. December 10 [cited 2016 Jun 28]. Available from: http://www.commonwealthfund.org/publications/issue-briefs/2012/dec/higher-readmissions-and-potential-policy-solutions [PubMed] [Google Scholar]
  • 23.Gilman M, Adams EK, Hockenberry JM, Wilson IB, Milstein AS, Becker ER. California safety-net hospitals likely to be penalized by ACA value, readmission, and meaningful-use programs. Health Aff (Millwood). 2014;33(8):1314–22. [DOI] [PubMed] [Google Scholar]
  • 24.Gilman M, Hockenberry JM, Adams EK, Milstein AS, Wilson IB, Becker ER. The financial effect of value-based purchasing and the Hospital Readmissions Reduction Program on safety-net hospitals in 2014: a cohort study. Ann Intern Med. 2015;163(6):427–36. [DOI] [PubMed] [Google Scholar]
  • 25.Ross JS, Bernheim SM, Lin Z, Drye EE, Chen J, Normand SL, et al. Based on key measures, care quality for Medicare enrollees at safety-net and non-safety-net hospitals was almost equal. Health Aff (Millwood). 2012; 31(8):1739–48. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Blum AB, Egorova NN, Sosunov EA, Gelijns AC, DuPree E, Moskowitz AJ, et al. Impact of socioeconomic status measures on hospital profiling in New York City. Circ Cardiovasc Qual Outcomes. 2014;7(3):391–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Keyhani S, Myers LJ, Cheng E, Hebert P, Williams LS, Bravata DM. Effect of clinical and social risk factors on hospital profiling for stroke readmission: a cohort study. Ann Intern Med. 2014;161(11):775–84. [DOI] [PubMed] [Google Scholar]
  • 28.Nagasako EM, Reidhead M, Waterman B, Dunagan WC. Adding socioeconomic data to hospital readmissions calculations may produce more useful results. Health Aff (Millwood). 2014;33(5):786–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Barnett ML, Hsu J, McWilliams JM. Patient characteristics and differences in hospital readmission rates. JAMA Intern Med. 2015;175(11): 1803–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Keenan PS, Normand SL, Lin Z, Drye EE, Bhat KR, Ross JS, et al. An administrative claims measure suitable for profiling hospital performance on the basis of 30-day all-cause readmission rates among patients with heart failure. Circ Cardiovasc Qual Outcomes. 2008;1(1):29–37. [DOI] [PubMed] [Google Scholar]
  • 31.Krumholz HM, Lin Z, Drye EE, Desai MM, Han LF, Rapp MT, et al. An administrative claims measure suitable for profiling hospital performance based on 30-day all-cause readmission rates among patients with acute myocardial infarction. Circ Cardiovasc Qual Outcomes. 2011; 4(2):243–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Lindenauer PK, Normand SL, Drye EE, Lin Z, Goodrich K, Desai MM, et al. Development, validation, and results of a measure of 30-day readmission following hospitalization for pneumonia. J Hosp Med. 2011; 6(3):142–50. [DOI] [PubMed] [Google Scholar]
  • 33.Centers for Medicare and Medicaid Services. Measure methodology [Internet]. Baltimore (MD): CMS; 2013. [cited 2016 Jun 28]. Available from: http://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/HospitalQualityInits/Measure-Methodology.html [Google Scholar]
  • 34.Census Bureau. American Community Survey (ACS): summary file data, 2008–12 [Internet]. Washington (DC): Census Bureau; 2014. January 30 [cited 2016 Jun 28]. Available from: http://www.census.gov/programs-surveys/acs/data/summary-file.html [Google Scholar]
  • 35.Bonito AJ, Bann C, Eicheldinger C, Carpenter L. Creation of new race-ethnicity codes and socioeconomic status (SES) indicators for Medicare beneficiaries: final report. Rockville (MD): Agency for Healthcare Research and Quality; 2008. January (AHRQ Publication No. 08–0029-EF). [Google Scholar]
  • 36.To access the Appendix, click on the Appendix link in the box to the right of the article online.
  • 37.American Hospital Association. AHA annual survey database. Chicago (IL): Health Forum LLC; 2012. [Google Scholar]
  • 38.Herrin J, St. Andre J, Kenward K, Joshi MS, Audet AM, Hines SC. Community factors and hospital readmission rates. Health Serv Res. 2015;50(1):20–39. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Braveman PA, Cubbin C, Egerter S, Williams DR, Pamuk E. Socioeconomic disparities in health in the United States: what the patterns tell us. J Am Public Health. 2010; 100(Suppl 1):S186–96. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Go AS, Mozaffarian D, Roger VL, Benjamin EJ, Berry JD, Borden WB, et al. Heart disease and stroke statistics—2013 update: a report from the American Heart Association. Circulation. 2013;127(1): e6–245. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Mensah GA, Mokdad AH, Ford ES, Greenlund KJ, Croft JB. State of disparities in cardiovascular health in the United States. Circulation. 2005;111(10):1233–41. [DOI] [PubMed] [Google Scholar]
  • 42.Bradley EH, Herrin J, Wang Y, McNamara RL, Webster TR, Magid DJ, et al. Racial and ethnic differences in time to acute reperfusion therapy for patients hospitalized with myocardial infarction. JAMA. 2004;292(13):1563–72. [DOI] [PubMed] [Google Scholar]
  • 43.Hasnain-Wynia R, Baker DW, Nerenz D, Feinglass J, Beal AC, Landrum MB, et al. Disparities in health care are driven by where minority patients seek care: examination of the hospital quality alliance measures. Arch Intern Med. 2007;167(12): 1233–9. [DOI] [PubMed] [Google Scholar]
  • 44.Spertus JA, Jones PG, Masoudi FA, Rumsfeld JS, Krumholz HM. Factors associated with racial differences in myocardial infarction outcomes. Ann Intern Med. 2009;150(5): 314–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Medicare Payment Advisory Commission. Report to the Congress: Medicare and the health care delivery system [Internet]. Washington (DC): MedPAC; 2013. June [cited 2016 Jun 28]. Available from: http://medpac.gov/documents/reports/jun13_entirereport.pdf?sfvrsn=0 [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Appendix
NIHMS1642973-supplement-Appendix.pdf (274.3KB, pdf)

RESOURCES