Listen to this manuscript’s

audio summary by

JACC Editor-in-Chief

Dr. Valentin Fuster.

J O U R N A L O F T H E AM E R I C A N C O L L E G E O F C A R D I O L O G Y V O L . 6 7 , N O . 8 , 2 0 1 6

ª 2 0 1 6 B Y T H E AM E R I C A N C O L L E G E O F C A R D I O L O G Y F O UN DA T I O N I S S N 0 7 3 5 - 1 0 9 7 / $ 3 6 . 0 0

P U B L I S H E D B Y E L S E V I E R h t t p : / / d x . d o i . o r g / 1 0 . 1 0 1 6 / j . j a c c . 2 0 1 5 . 1 2 . 0 3 7

Has Public Reporting ofHospital Readmission RatesAffected Patient Outcomes?

Analysis of Medicare Claims Data

Adam D. DeVore, MD,a,b Bradley G. Hammill, DRPH,a N. Chantelle Hardy, MPH,a Zubin J. Eapen, MD, MHS,a,b

Eric D. Peterson, MD, MPH,a,b Adrian F. Hernandez, MD, MHSa,b

ABSTRACT

Fro

Me

He

the

con

su

Bo

res

Po

Co

Co

Or

rep

Ma

BACKGROUND In 2009, the Centers for Medicare & Medicaid Services (CMS) began publicly reporting 30-day hospital

readmission rates for patients discharged with acute myocardial infarction (MI), heart failure (HF), or pneumonia.

OBJECTIVES This study assessed trends of 30-day readmission rates and post-discharge care since the implementation

of CMS public reporting.

METHODS We analyzed Medicare claims data from 2006 to 2012 for patients discharged after a hospitalization for MI,

HF, or pneumonia. For each diagnosis, we estimated trends in 30-day all-cause readmissions and post-discharge care

(emergency department visits and observation stays) by using hospitalization-level regression models. We modeled

adjusted trends before and after the implementation of public reporting. To assess for a change in trend, we tested the

difference between the slope before implementation and the slope after implementation.

RESULTS We analyzed 37,829 hospitalizations for MI, 100,189 for HF, and 79,076 for pneumonia from >4,100 hos-

pitals. When considering only recent trends (i.e., since 2009), we found improvements in adjusted readmission rates for

MI (�2.3%), HF (�1.8%), and pneumonia (�2.0%), but when comparing the trend before public reporting with the trend

after reporting, there was no difference for MI (p ¼ 0.72), HF (p ¼ 0.19), or pneumonia (p ¼ 0.21). There were no changes

in trends for 30-day post-discharge care for MI or pneumonia; however, the trend decreased for HF emergency

department visits from 2.3% to �0.8% (p ¼ 0.007) and for observation stays from 15.1% to 4.1% (p ¼ 0.04).

CONCLUSIONS The release of the CMS public reporting of hospital readmission rates was not associated with any

measurable change in 30-day readmission trends for MI, HF, or pneumonia, but it was associated with less hospital-based

acute care for HF. (J Am Coll Cardiol 2016;67:963–72) © 2016 by the American College of Cardiology Foundation.

R educing hospital readmissions has become anational priority for patients, providers, andpolicy makers. Nearly 1 in 5 Medicare benefi-

ciaries discharged from a hospital is readmitted

m the aDuke Clinical Research Institute, Durham, North Carolina; and the

dicine, Durham, North Carolina. This project was supported in part by

althcare Research and Quality (AHRQ). The content is solely the responsibi

official views of the AHRQ. Dr. Eapen has membership on the advisory

sultant for Amgen and SHL Telemedicine; and has received an honorar

pport from the American College of Cardiology, American Heart Associatio

ehringer Ingelheim, Merck, Valeant, Sanofi, AstraZeneca, Janssen, Regen

earch funding from Amgen, Inc., AstraZeneca, Bayer Corporation US, Brist

rtola Pharmaceutical (all significant); has received personal income for con

rporation US; and has received personal income for consulting or other non

rporation US, Eli Lilly & Co., Gilead Sciences, Inc., GlaxoSmithKline, Ja

tho-McNeil-Janssen Pharmaceuticals, Inc., Pfizer, Pluristem Therapeutics,

orted that they have no relationships relevant to the contents of this pap

nuscript received June 16, 2015; revised manuscript received November

within 30 days, and the associated estimated costsof unplanned readmissions is >$17 billion annually(1). In response to this high number of readmissions,the Centers for Medicare & Medicaid Services (CMS)

bDepartment of Medicine, Duke University School of

grant number U19HS021092 from the Agency for

lity of the authors and does not necessarily represent

boards of Novartis, Cytokinetics, and Amgen; is a

ium from Janssen. Dr. Peterson has received grant

n, and Janssen; and has been a consultant for Bayer,

eron, and Genentech. Dr. Hernandez has received

ol-Myers Squibb, GlaxoSmithKline, Merck & Co., and

sulting or other services (including CME) from Bayer

-CME services from Amgen, Inc., AstraZeneca, Bayer

nssen, Merck & Co., Novartis Pharmaceutical Co.,

Inc., Sensible, and Myocardia. All other authors have

er to disclose.

4, 2015, accepted December 1, 2015.

ABBR EV I A T I ON S

AND ACRONYMS

CMS = Centers for Medicare &

Medicaid Services

COPD = chronic obstructive

pulmonary disease

ED = emergency department

HF = heart failure

ICD-9-CM = International

Classification of Disease-9th

Revision-Clinical Modification

MI = myocardial infarction

DeVore et al. J A C C V O L . 6 7 , N O . 8 , 2 0 1 6

Public Reporting of Hospital Readmission Rates M A R C H 1 , 2 0 1 6 : 9 6 3 – 7 2

964

adopted a number of policy changes aimed atimproving patients’ outcomes. One signifi-cant change occurred in June of 2009 (2),when the CMS began to report risk-standardized hospital readmission rates pub-licly for patients discharged with acutemyocardial infarction (MI), heart failure(HF), or pneumonia on the Hospital Comparewebsite (3). The reasons the CMS opted forpublic reporting were to increase transpar-ency for consumers and to provide an addi-tional incentive for hospitals to improve

transitional care and reduce hospital readmissions.

SEE PAGE 973

Although increased transparency was anticipatedto improve outcomes, limited data are available toassess the impact of this policy objectively (4–6). Ourstudy was designed to assess for temporal changes in30-day readmission rates and to evaluate whetherthese outcomes improved after the implementationof public reporting. We also evaluated for improve-ments in post-discharge care by assessing changes inoutpatient visits, emergency department (ED) visits,and observation stays without readmission. Finally,we evaluated the potentially unintended impacts ofthis policy by assessing for temporal changes in post-discharge mortality rates.

METHODS

DATA SOURCE. We used research-identifiableadministrative claims data from a 5% nationallyrepresentative sample of Medicare beneficiaries from2005 to 2012. These data include inpatient claims,outpatient claims, carrier claims, and the associateddenominator files. Inpatient files contain institutionalclaims submitted for facility costs associated withinpatient stays. Outpatient files contain similarinstitutional claims for outpatient services. Carrierfiles contain noninstitutional provider and otherprofessional claims for services across all settings.Denominator files include beneficiary identifiers, dateof birth, sex, race or ethnicity, date of death (if pre-sent), and information about program eligibility andenrollment.

STUDY SAMPLE. We analyzed Medicare-enrolledpatients $65 years of age who were dischargedhome from a hospitalization for acute MI, HF, pneu-monia, chronic obstructive pulmonary disease(COPD), or diabetes between July 1, 2006 and June30, 2012. These dates were chosen to permit analysis3 years before and after the public reporting ofrisk-standardized hospital readmission rates in June

2009 (2). Because the Medicare public reportingprogram did not include COPD or diabetes, thesehospitalizations served as comparator conditions.Both COPD and diabetes were selected as comparatorconditions because neither disorder was directlyaffected by the impact of public reporting during thistime period and both are common causes of 30-dayall-cause readmissions among Medicare benefi-ciaries (1). Patients were allowed to be representedmultiple times in the analysis if subsequent hospi-talizations fit the inclusion and exclusion criteria andoccurred >30 days after another hospitalization ofthe same type.

Hospitalizations of interest were identified onthe basis of their primary discharge diagnoses:acute MI (International Classification of Disease-9thRevision-Clinical Modification [ICD-9-CM] codes410.x0, 410.x1), HF (402.x1, 404.x1, 404.x3, 428.x),pneumonia (480.x, 481, 482.x, 483.x, 485, 486, 487.0,488.11), COPD (491.21, 491.22, 491.8, 491.9, 492.8,493.2x, 496, or a primary ICD-9-CM code of 518.81,518.82, 518.84, 799.1 combined with a secondary ICD-9-CM code of acute exacerbation of COPD: 491.21,491.22, 493.21, or 493.22), and diabetes (250.x). Weexcluded planned hospitalizations for all studygroups and hospitalizations for acute MI that resultedin same-day discharges. We also excluded hospitali-zations for patients who did not have fee-for-serviceMedicare coverage for $12 months before admission,to allow characterization of medical history, and forthe 30 days after discharge, to allow complete ascer-tainment of outcomes.OUTCOME MEASURES. The primary outcome ofinterest consisted of unplanned all-cause 30-dayhospital readmissions. These readmissions wereidentified using inpatient claims data. Planned read-missions were defined similar to previous analyses(5,6). Other assessments of interest included all-causemortality, ED visits, outpatient evaluation and man-agement visits, and observation stays without read-mission in the 30 days after discharge. Mortality wasdetermined from the denominator file. ED visits wereidentified from inpatient and outpatient claimshaving any of the following revenue center codes:0450 (emergency room – general classification), 0451(emergency room – Emergency Medical Treatment &Labor Act emergency medical screening services),0452 (emergency room – emergency room beyondEmergency Medical Treatment & Labor Actscreening), 0456 (emergency room – urgent care),0459 (emergency room – other), or 0981 (professionalfees – emergency room). Outpatient evaluationand management visits were identified from carrierclaims classified into any of the following

TABLE 1 Observed Outcomes

Time Period*

Study Group

Acute MI HF Pneumonia COPD Diabetes

30-day readmission rates, readmissions/hospitalizations

2006–2007 1,126/6,427 (17.5) 3,824/17,774 (21.5) 2,305/14,726 (15.7) 2,299/12,055 (19.1) 526/3,173 (16.6)

2007–2008 1,151/6,664 (17.3) 3788/17,368 (21.8) 2,302/14,530 (15.8) 2,812/14,066 (20.0) 562/3,127 (18.0)

2008–2009 1,099/6,332 (17.4) 3,751/16,932 (22.2) 2,040/12,723 (16.0) 2,758/13,668 (20.2) 499/3,059 (16.3)

2009–2010 1,003/6,103 (16.4) 3,730/16,827 (22.2) 1,944/12,517 (15.5) 2,685/13,705 (19.6) 502/2,902 (17.3)

2010–2011 1,018/6,239 (16.3) 3,591/16,251 (22.1) 2,110/12,644 (16.7) 2,650/13,946 (19.0) 464/2,970 (15.6)

2011–2012 974/6,064 (16.1) 3,378/15,037 (22.5) 1,889/11,936 (15.8) 2,461/12,651 (19.5) 473/2,676 (17.7)

30-day post-discharge mortality rates, deaths/hospitalizations

2006–2007 242/6,427 (3.8) 631/17,774 (3.6) 429/14,726 (2.9) 346/12,055 (2.9) 63/3,173 (2.0)

2007–2008 225/6,664 (3.4) 605/17,368 (3.5) 432/14,530 (3.0) 388/14,066 (2.8) 74/3,127 (2.4)

2008–2009 200/6,332 (3.2) 617/16,932 (3.6) 372/12,723 (2.9) 345/13,668 (2.5) 68/3,059 (2.2)

2009–2010 207/6,103 (3.4) 613/16,827 (3.6) 409/12,517 (3.3) 368/13,705 (2.7) 61/2,902 (2.1)

2010–2011 197/6,239 (3.2) 583/16,251 (3.6) 421/12,644 (3.3) 353/13,946 (2.5) 62/2,970 (2.1)

2011–2012 178/6,064 (2.9) 614/15,037 (4.1) 417/11,936 (3.5) 315/12,651 (2.5) 58/2,676 (2.2)

Values are n/N (%). *The time period for each study year began on July 1 and ended on June 30 of the following year.

COPD ¼ chronic obstructive pulmonary disease; HF ¼ heart failure; MI ¼ myocardial infarction.

TABLE 2 Relative Change in Trends in 30-Day Readmission Rates

Relative Change in Trends

Before Public Reporting After Public Reporting D p Value

Unadjusted 30-day all-cause readmission, %

MI �2.1 (�4.7 to 0.5) �1.7 (�4.5 to 1.3) 0.85

HF 1.2 (�0.3 to 2.7) 0.0 (�1.6 to 1.7) 0.43

Pneumonia 1.0 (�1.0 to 3.0) �0.7 (�2.9 to 1.6) 0.40

COPD 1.0 (�0.9 to 2.9) �1.9 (�3.9 to 0.1) 0.10

Diabetes �1.4 (�5.2 to 2.6) 1.7 (�2.7 to 6.2) 0.43

Adjusted* 30-day all-cause readmission, %

MI �3.1 (�5.6 to �0.6) �2.3 (�5.1 to 0.6) 0.72

HF 0.0 (�1.4 to 1.5) �1.8 (�3.3 to �0.2) 0.19

Pneumonia 0.4 (�1.5 to 2.4) �2.0 (�4.1 to 0.2) 0.21

COPD 0.1 (�1.7 to 1.9) �2.6 (�4.5 to �0.7) 0.11

Diabetes �1.9 (�5.7 to 1.9) 0.1 (�4.1 to 4.5) 0.58

Values are mean (95% confidence interval). *Adjusted for patient demographics (age, sex, race) and comorbidconditions (cancer, COPD, cerebrovascular disease, dementia, diabetes, HF, hypertension, ischemic heart disease,liver disease, peptic ulcer disease, peripheral vascular disease, renal disease, rheumatic disease, valvular heartdisease), and accounts for clustering of patients within hospitals.

Abbreviations as in Table 1.

J A C C V O L . 6 7 , N O . 8 , 2 0 1 6 DeVore et al.M A R C H 1 , 2 0 1 6 : 9 6 3 – 7 2 Public Reporting of Hospital Readmission Rates

965

Berenson-Eggers Type of Service (BETOS) codes: M1Aoffice visits – new, M1B office visits – established,M4A home visit, M4B nursing home visit, M5Aspecialist – pathology, M5B specialist – psychiatry,M5C specialist – ophthalmology, M5D specialist –

other, and M6 consultations (excluded codes from thiscategory include M2x hospital and M3 ED codes).Observations stays were identified from outpatientclaims having a revenue center code of 0762 (obser-vation room). Only observation stays that did not leadto an inpatient stay were counted.

STATISTICAL METHODS. For each population (acuteMI, HF, pneumonia, COPD, diabetes), we firstsummarized the annual observed outcomes andpost-discharge care. From these data, we usedregression models to estimate the unadjusted andadjusted trends for all-cause 30-day readmission.Because care may not have changed immediately af-ter the implementation of public reporting ofrisk-standardized readmission rates, the slope of thetrend line was allowed to vary in the model beforeand after implementation of public reporting. Thetrends are reported as relative changes per year. Forexample, if the pre-intervention readmission ratewas 22.0%, then a �2.0% estimated change in-dicates a 1-year improvement of 0.44% (21.6%)in absolute terms. The pre-implementation andpost-implementation trend estimates were comparedusing a Wald test. These models were specified withlog link and Poisson errors, which allowed us to es-timate the relative change in each outcome over time.We used generalized estimating equation methods to

allow for the calculation of robust standard errors,which accounted for clustering of records withinhospitals.

We plotted the risk-standardized quarterly read-mission rates against the adjusted estimated trendsfor each population. In these plots, we show apredicted trend line, which is an extension of thepre-reporting trend to the end of the study period. Weadjusted for the patients’ demographics (age, sex,race) and comorbid conditions (cancer, COPD, cere-brovascular disease, dementia, diabetes, HF, hyper-tension, ischemic heart disease, liver disease, peptic

CENTRAL ILLUS

Trends for Acut

DeVore, A.D. et al. J Am

CMS ¼ Centers for M

DeVore et al. J A C C V O L . 6 7 , N O . 8 , 2 0 1 6

Public Reporting of Hospital Readmission Rates M A R C H 1 , 2 0 1 6 : 9 6 3 – 7 2

966

ulcer disease, peripheral vascular disease, renal dis-ease, rheumatic disease, valvular heart disease) byusing validated algorithms (7,8). We plotted similardata for post-discharge mortality for each studypopulation. We used the outpatient and carrierfiles to assess patients’ comorbid conditions in

TRATION Impact of CMS Public Reporting Policies for Hospite MI 30-Day Readmission Rates

Coll Cardiol. 2016; 67(8):963–72.

edicare and Medicaid Services; IPPS ¼ inpatient prospective payment system

the adjustment models. For all comparisons,p values <0.05 were considered statistically signifi-cant. Data were analyzed using SAS version 9.2 (SASInstitute, Cary, North Carolina). The institutional re-view board of the Duke University Health System(Durham, North Carolina) approved this study.

al Readmission:

; MI ¼ myocardial infarction.

J A C C V O L . 6 7 , N O . 8 , 2 0 1 6 DeVore et al.M A R C H 1 , 2 0 1 6 : 9 6 3 – 7 2 Public Reporting of Hospital Readmission Rates

967

RESULTS

Between July 1, 2006 and June 30, 2012, at >4,100hospitals in the United States, there were thefollowing hospitalizations: 37,829 for acute MI;100,189 for HF; 79,076 for pneumonia; 80,091 forCOPD; and 17,907 for diabetes (Table 1). The 30-dayreadmission rates after acute MI declined slightly

FIGURE 1 Changes in Adjusted Trends for 30-Day Readmission Rate

0.30

0.25

0.20

0.15

0.10

0.05

0.00-12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12

Quarter Relative to Reporting

Read

miss

ion

Rate

0.30

0.25

0.20

0.15

0.10

0.05

0.00-12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12

Quarter Relative to Reporting

Read

miss

ion

Rate

0.30

0.25

0.20

0.15

0.10

0.05

0.00-12 -10 -8 -6 -4 -2

Quarter Rela

Read

miss

ion

Rate

P=0.72

P=0.21

Acute MI

Pneumonia

Diabetes

A

C

E

The expected (solid lines) and observed (dotted lines) adjusted trends

public reporting for each study group: (A) acute myocardial infarction (M

disease (COPD); and (E) diabetes. These trends are adjusted for patient d

cerebrovascular disease, dementia, diabetes, heart failure, hypertension,

vascular disease, renal disease, rheumatic disease, valvular heart disease

from 17.5% in year 1 to 16.1% in year 6. The 30-dayreadmission rates for HF, pneumonia, COPD,and diabetes were unchanged over time. The 30-daypost-discharge mortality rates also decreased foracute MI from 3.8% in year 1 to 2.9% in year 6. Therewere small changes in the 30-day post-dischargemortality rate for HF, pneumonia, COPD, and dia-betes, but no apparent trends.

s by Study Group

0.30

0.25

0.20

0.15

0.10

0.05

0.00-12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12

Quarter Relative to Reporting

Read

miss

ion

Rate

0.30

0.25

0.20

0.15

0.10

0.05

0.00-12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12

Quarter Relative to Reporting

Read

miss

ion

Rate

0 2 4 6 8 10 12tive to Reporting

P=0.19

P=0.11

P=0.58

Heart Failure

COPD

B

D

for 30-day readmission rates before and after implementation of

I); (B) heart failure; (C) pneumonia; (D) chronic obstructive pulmonary

emographics (age, sex, race) and comorbid conditions (cancer, COPD,

ischemic heart disease, liver disease, peptic ulcer disease, peripheral

) and account for clustering of patients within hospitals.

TABLE 3

Unadjuste

MI

HF

Pneumo

COPD

Diabete

Adjusted*

MI

HF

Pneumo

COPD

Diabete

Values are mand comorhypertensioease, renalpatients wit

Abbreviat

DeVore et al. J A C C V O L . 6 7 , N O . 8 , 2 0 1 6

Public Reporting of Hospital Readmission Rates M A R C H 1 , 2 0 1 6 : 9 6 3 – 7 2

968

Unadjusted and adjusted trends in 30-day all-causereadmission are displayed in Table 2, CentralIllustration, and Figure 1. After adjustments forpatient-related and hospital factors, we observed thefollowing improvements post-implementation ofpublic reporting in adjusted readmission rates: forMI, �2.3% (95% confidence interval [CI]: �5.1 to 0.6);for HF, �1.8% (95% CI: �3.3 to �0.2); and forpneumonia, �2.0% (95% CI: �4.1 to 0.2). In contrast,patients discharged with COPD had the largest rela-tive improvement (�2.6%, 95% CI: �4.5% to �0.7%),and patients discharged with diabetes had noimprovement (0.1%, 95% CI: �4.1 to 4.5). Neverthe-less, when comparing the trend before publicreporting with the trend after public reporting, therewere no differences for MI (p ¼ 0.72), HF (p ¼ 0.19), orpneumonia (p ¼ 0.21).

Unadjusted and adjusted trends in 30-day all-causepost-discharge mortality are displayed in Table 3 andFigure 2. In adjusted analyses, the trend in 30-daypost-discharge mortality for patients with HFwas �2.4 (95% CI �6.2% to 1.6%) before the imple-mentation of public reporting and 3.1% (95% CI �1.3%to 7.6%) after implementation, although there was nostatistically significant difference in the change intrend (p ¼ 0.15). We noted no difference in the trendbefore and after the implementation of publicreporting for the other study groups.

The observed rates of post-discharge care are dis-played in Table 4. For all study groups, there was anincrease during the entire 6-year study period in any30-day post-discharge observation stays withoutreadmission. For patients hospitalized with acute MI,

Relative Change in Trends in 30-Day Post-Discharge Mortality

Relative Change in Trends

Before Public Reporting After Public Reporting D p Value

d 30-day all-cause mortality, %

�4.3 (�10.3 to 2.2) �3.0 (�9.7 to 4.3) 0.83

�0.5 (�4.4 to 3.6) 5.2 (0.8 to 9.8) 0.15

nia 2.7 (�2.3 to 8.0) 4.1 (�1.2 to 9.7) 0.78

�3.6 (�8.6 to 1.7) �1.1 (�6.7 to 4.9) 0.61

s 2.1 (�9.3 to 15.0) �3.3 (�16.0 to 11.3) 0.64

30-day all-cause mortality, %

�5.6 (�11.5 to 0.7) �3.7 (�10.3 to 3.5) 0.75

�2.4 (�6.2 to 1.6) 3.1 (�1.3 to 7.6) 0.15

nia 1.8 (�3.1 to 7.0) 2.6 (�2.6 to 8.2) 0.86

�4.5 (�9.4 to 0.6) �1.6 (�7.1 to 4.3) 0.54

s 1.6 (�9.7 to 14.2) �5.4 (�17.9 to 8.8) 0.54

ean (95% confidence interval). *Adjusted for patient demographics (age, sex, race)bid conditions (cancer, COPD, cerebrovascular disease, dementia, diabetes, HF,n, ischemic heart disease, liver disease, peptic ulcer disease, peripheral vascular dis-disease, rheumatic disease, valvular heart disease), and accounts for clustering ofhin hospitals.

ions as in Table 1.

this rate increased from 2.1% to 3.2%; for patientshospitalized with HF, this rate increased from 1.6% to2.6%; for patients hospitalized with pneumonia, thisrate increased from 1.3% to 2.3%; for patients hospi-talized with COPD, this rate increased from 1.4% to2.1%; and for patients hospitalized with diabetes, thisrate increased from 1.6% to 2.5%. There were alsoincreases in the number of any 30-day post-dischargeED visits for HF and diabetes, but not for the otherstudy groups. The rates of any 30-day post-dischargeoutpatient visits for all study groups were unchangedover time.

Adjusted trends of post-discharge care are dis-played in Table 5. We observed an increase in thetrend for 30-day post-discharge outpatient visits forpneumonia from �0.2% (95% CI: �0.7 to 0.2) to 0.9%(95% CI: 0.3 to 1.4; p ¼ 0.01), but not for the otherstudy groups. We also found a decrease in the trendfor HF-related ED visits from 2.3% (95% CI: 1.1 to 3.6)to �0.8% (95% CI: �2.1 to 0.5; p ¼ 0.007) and obser-vation stays from 15.1% (95% CI: 9.0 to 21.5) to 4.1%(95% CI: �1.6 to 10.0; p ¼ 0.04), but not for the otherstudy groups.

DISCUSSION

We analyzed Medicare claims data to assess trends of30-day outcomes before and after the implementa-tion of CMS public reporting and found that thisintervention was not associated with any measurablechange in 30-day readmission or 30-day post-discharge mortality trends. However, we did noteassociations with increased post-discharge outpatientvisits for pneumonia and decreased ED visits andobservation stays without readmission for HF. Theseindirect assessments of transitional care suggestsome improvements related to the policy change, butthey were not able to translate into actual improve-ments in patients’ outcomes. Our study is the first tospecifically assess the impact of the 2009 CMS policychange on public reporting of readmissions, and itadds to the evidence base on the use of publicreporting as a quality improvement tool.

Our findings build on previous national reportsof readmission and post-discharge mortality rates(4–6,9,10). Data from a peer-reviewed analysis (5)and the 2014 Medicare Hospital Quality Chartbook(6), using Medicare claims data combined withVeterans Health Administration data, suggest recentimprovements in national unplanned 30-day read-mission rates for patients hospitalized with acute MI,HF, and pneumonia after implementation of publicreporting (i.e., after June 2009). Our analysis extendsthese results by considering previous secular trends

FIGURE 2 Changes in Adjusted Trends for 30-Day Post-Discharge Mortality by Study Group

-12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12Quarter Relative to Reporting

Mor

talit

y Ra

te

0.05

0.04

0.03

0.02

0.01

0.00-12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12

Quarter Relative to ReportingM

orta

lity

Rate

0.05

0.04

0.03

0.02

0.01

0.00

-12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12Quarter Relative to Reporting

Mor

talit

y Ra

te

0.05

0.04

0.03

0.02

0.01

0.00-12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12

Quarter Relative to Reporting

Mor

talit

y Ra

te

0.05

0.04

0.03

0.02

0.01

0.00

-12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12Quarter Relative to Reporting

Mor

talit

y Ra

te

0.05

0.04

0.03

0.02

0.01

0.00

Acute MI Heart Failure

Pneumonia COPD

Diabetes

A B

C D

E

P=0.75 P=0.15

P=0.86 P=0.54

P=0.54

The expected (solid lines) and observed (dotted lines) adjusted trends for 30-day post-discharge mortality before and after implementation

of public reporting for each study group: (A) acute myocardial infarction (MI); (B) heart failure; (C) pneumonia; (D) chronic obstructive

pulmonary disease (COPD); and (E) diabetes. Adjusted for patient demographics (age, sex, race) and comorbid conditions (cancer, COPD,

cerebrovascular disease, dementia, diabetes, heart failure, hypertension, ischemic heart disease, liver disease, peptic ulcer disease, peripheral

vascular disease, renal disease, rheumatic disease, valvular heart disease), and accounts for clustering of patients within hospitals.

J A C C V O L . 6 7 , N O . 8 , 2 0 1 6 DeVore et al.M A R C H 1 , 2 0 1 6 : 9 6 3 – 7 2 Public Reporting of Hospital Readmission Rates

969

in care (i.e., before June 2009) to specifically assessthe impact of the 2009 policy change. Similarly, pre-vious data suggest modest increases in post-discharge mortality rates for patients with HF, nochange for patients discharged with pneumonia, andmodest improvements for patients discharged withacute MI (5,6,9,10). By considering previous secular

trends, we observed similar patterns but again notedno change in trends after the implementation ofpublic reporting.

Our findings suggest that the 2009 CMS policy de-cision to report hospital readmission rates publiclyon the Hospital Compare website was not associatedwith improvements in outcomes. These findings have

TABLE 4 Observed Post-Discharge Care

Time Period*

Study Group

Acute MI HF Pneumonia COPD Diabetes

Any 30-day post-discharge outpatient visits, visits/hospitalizations

2006–2007 5,124/6,427 (79.7) 14,073/17,774 (79.2) 11,520/14,726 (78.2) 9,212/12,055 (76.4) 2,370/3,173 (74.7)

2007–2008 5,355/6,664 (80.4) 13,699/17,368 (78.9) 11,242/14,530 (77.4) 10,786/14,066 (76.7) 2,342/3,127 (74.9)

2008–2009 5,186/6,332 (81.9) 13,468/16,932 (79.5) 9,918/12,723 (78.0) 10,489/13,668 (76.7) 2,372/3,059 (77.5)

2009–2010 4,996/6,103 (81.9) 13,479/16,827 (80.1) 9,695/12,517 (77.5) 10,495/13,705 (76.6) 2,228/2,902 (76.8)

2010–2011 5,117/6,239 (82.0) 13,185/16,251 (81.1) 9,948/12,644 (78.7) 10,777/13,946 (77.3) 2,289/2,970 (77.1)

2011–2012 5,026/6,064 (82.9) 12,184/15,037 (81.0) 9,469/11,936 (79.3) 9,894/12,651 (78.2) 2,130/2,676 (79.6)

Any 30-day post-discharge ED visits, ED visits/hospitalizations

2006–2007 1,567/6,427 (24.4) 4,639/17,774 (26.1) 3,061/14,726 (20.8) 3,056/12,055 (25.4) 704/3,173 (22.2)

2007–2008 1,650/6,664 (24.8) 4,725/17,368 (27.2) 2,988/14,530 (20.6) 3,611/14,066 (25.7) 757/3,127 (24.2)

2008–2009 1,584/6,332 (25.0) 4,741/16,932 (28.0) 2,738/12,723 (21.5) 3,484/13,668 (25.5) 709/3,059 (23.2)

2009–2010 1,519/6,103 (24.9) 4,830/16,827 (28.7) 2,658/12,517 (21.2) 3,609/13,705 (26.3) 696/2,902 (24.0)

2010–2011 1,550/6,239 (24.8) 4,640/16,251 (28.6) 2,851/12,644 (22.5) 3,606/13,946 (25.9) 708/2,970 (23.8)

2011–2012 1,511/6,064 (24.9) 4,384/15,037 (29.2) 2,709/11,936 (22.7) 3,409/12,651 (26.9) 715/2,676 (26.7)

Any 30-day post-discharge observation stays without readmission, observation stays/hospitalizations

2006–2007 138/6,427 (2.1) 285/17,774 (1.6) 186/14,726 (1.3) 166/12,055 (1.4) 51/3,173 (1.6)

2007–2008 167/6,664 (2.5) 306/17,368 (1.8) 183/14,530 (1.3) 192/14,066 (1.4) 53/3,127 (1.7)

2008–2009 165/6,332 (2.6) 378/16,932 (2.2) 186/12,723 (1.5) 192/13,688 (1.4) 41/3,059 (1.3)

2009–2010 189/6,103 (3.1) 405/16,827 (2.4) 210/12,517 (1.7) 259/13,705 (1.9) 59/2,902 (2.0)

2010–2011 179/6,239 (2.9) 376/16,251 (2.3) 239/12,644 (1.9) 322/13,946 (2.3) 62/2,970 (2.1)

2011–2012 196/6,064 (3.2) 396/15,037 (2.6) 272/11,936 (2.3) 267/12,651 (2.1) 66/2,676 (2.5)

Values are n/N (%). *The time period for each study year began on July 1 and ended on June 30 of the following year.

ED ¼ emergency department; other abbreviations as in Table 1.

TABLE 5 Adjusted Relative Change in Trends in Post-Discharge Care

Relative Change in Trends

Before Public Reporting After Public Reporting D p Value

Adjusted* any 30-day post-discharge outpatient visits, %

MI 1.1 (0.5 to 1.7) 0.5 (�0.2 to 1.1) 0.30

HF 0.4 (0.0 to 0.9) 0.8 (0.4 to 1.2) 0.36

Pneumonia �0.2 (�0.7 to 0.2) 0.9 (0.3 to 1.4) 0.01

COPD �0.1 (�0.6 to 0.4) 0.7 (0.1 to 1.2) 0.11

Diabetes 1.0 (0.0 to 2.1) 1.2 (0.1 to 2.3) 0.88

Adjusted* any post-discharge ED visits, %

MI �0.1 (�2.2 to 2.0) �0.5 (�2.8 to 1.9) 0.85

HF 2.3 (1.1 to 3.6) �0.8 (�2.1 to 0.5) 0.007

Pneumonia 0.9 (�0.8 to 2.5) 0.9 (�0.9 to 2.7) 0.99

COPD �0.1 (�1.6 to 1.5) 0.8 (�0.8 to 2.4) 0.55

Diabetes 0.7 (�2.4 to 3.9) 3.3 (�0.3 to 7.1) 0.40

Adjusted* any 30-day post-discharge observation stays without readmission, %

MI 10.7 (2.5 to 19.6) 4.5 (�3.4 to 13.1) 0.42

HF 15.1 (9.0 to 21.5) 4.1 (�1.6 to 10.0) 0.04

Pneumonia 10.2 (2.6 to 18.3) 15.3 (7.6 to 23.7) 0.48

COPD 11.9 (4.2 to 20.2) 11.8 (5.0 to 18.9) 0.98

Diabetes 1.5 (�11.7 to 16.7) 19.3 (3.7 to 37.3) 0.21

Values are mean (95% confidence interval). *Adjusted for patient demographics (age, sex, race) and comorbidconditions (cancer, COPD, cerebrovascular disease, dementia, diabetes, HF, hypertension, ischemic heart disease,liver disease, peptic ulcer disease, peripheral vascular disease, renal disease, rheumatic disease, valvular heartdisease), and accounts for clustering of patients within hospitals.

Abbreviations as in Table 1.

DeVore et al. J A C C V O L . 6 7 , N O . 8 , 2 0 1 6

Public Reporting of Hospital Readmission Rates M A R C H 1 , 2 0 1 6 : 9 6 3 – 7 2

970

important policy implications and suggest thatwell-conducted trials of quality improvement in-terventions should be performed before widespreaddissemination of these interventions. This is especiallytrue considering that almost all hospitals in the UnitedStates participate in the Hospital Compare program;therefore, there are a limited number of availablecontrol groups for the evaluation of quality improve-ment interventions after implementation. Impor-tantly, we must remain aware that the current CMSpolicy on public reporting for acute MI, HF, andpneumonia is only one early example of publicreporting at the federal level. Readmission ratesrelated to other conditions and procedures are alsonow reported on the Hospital Compare website andinclude COPD, stroke, coronary artery bypass graftsurgery, hip or knee surgery, and overall readmissionrates (3). Public reporting is also being extended to theprovider level with the Physician Compare website(11), which is a newer program that will include infor-mation on physician-level quality metrics (12).

Previous retrospective observational data on theimpact of public reporting of hospital performancehave yielded conflicting results (13–19). Previousrandomized trial data are limited to the EFFECT(Enhanced Feedback for Effective Cardiac Treatment)

PERSPECTIVES

COMPETENCY IN SYSTEMS-BASED PRACTICE: Public

reporting of hospital readmission rates for acute MI, HF, and

pneumonia was not associated with changes in readmission rates

or post-discharge mortality rates.

TRANSITIONAL OUTLOOK: More work is needed to evaluate

the potential impact of policy interventions aimed at reducing

hospital readmissions before these interventions are

disseminated.

J A C C V O L . 6 7 , N O . 8 , 2 0 1 6 DeVore et al.M A R C H 1 , 2 0 1 6 : 9 6 3 – 7 2 Public Reporting of Hospital Readmission Rates

971

study, which assessed whether publicly releasedhospital report cards could improve hospital qualityof care for patients hospitalized with acute MI or HFin Ontario, Canada (20). The investigators found noimprovement in composite scores of performancemeasures for hospitals randomized to public reportcards compared with controls. The reasons that pub-lic reporting may not actually improve readmissionrates are not apparent from these earlier studies orfrom our data. Regardless, the benefit of publicreporting of readmissions should be balanced by po-tential unintended consequences, including divertingattention and resources away from other qualityimprovement efforts.

Our results suggest that changes in post-dischargecare are associated with this policy change, specif-ically, increases in outpatient visits and decreases inhospital-based acute care (i.e., ED visits and obser-vation stays without readmission); these measuresare indirect assessments of transitional care. Forexample, the transition of care from the hospital tooutpatient setting may be improved by ensuringoutpatient clinic follow-up after discharge, yet moststudies have focused on follow-up early afterdischarge (i.e., within 7 days of discharge) (21). Thereis also growing interest in using post-discharge EDvisits as a quality metric for transitional care (22), andour data suggest decreased ED visits for HF. Finally,earlier studies noted increases in observation staysduring this time period (23,24), but our data suggest adecrease in the trend for post-discharge observationstays without readmission for HF. What is unclear iswhich interventions were used by hospitals to ach-ieve these potential benefits in post-discharge care.

STUDY LIMITATIONS. Our study has several limita-tions, in addition to those inherent in a retrospectiveanalysis of claims data. First, we assessed for changesin trends in the month after public reporting ofreadmissions. Interventions aimed at reducing hos-pital readmissions implemented as a consequence ofthis policy decision may have taken effect beforepublic reporting or in the months after this change.Second, our analysis focused on public reporting asan isolated intervention, but it is part of a largerpolicy agenda implemented over multiple years,which includes the recent extension of this policy to

impose cuts in total Medicare reimbursements forhigher-than-expected readmission rates (25). Theimpact of this policy change has yet to be evaluated,but it may be high, given the financial penalties.Finally, our analysis considers nationwide hospitalperformance, but it does not evaluate improvementsat individual hospitals associated with implementa-tion of public reporting.

CONCLUSIONS

In this analysis of Medicare claims data, we found noassociation between the 2009 CMS policy decision toreport hospital readmission rates and changes intrends for readmission rates publicly, nor did we findchanges in trends for post-discharge mortality rates.We did observe some improvements in post-discharge care. The current focus on readmissions asa quality measure is unlikely to change in the nearfuture, and policy interventions aimed at reducinghospital readmissions should be evaluated beforewidespread dissemination.

ACKNOWLEDGMENT The authors thank Erin Hanley,MS, for her editorial contributions to this manuscript.Ms. Hanley did not receive compensation for hercontributions, apart from her employment at theinstitution where this study was conducted.

REPRINT REQUESTS AND CORRESPONDENCE: Dr.Adam D. DeVore, Duke Clinical Research Institute,P.O. Box 17969, Durham, North Carolina 27715.E-mail: adam.devore@duke.edu.

RE F E RENCE S

1. Jencks SF, Williams MV, Coleman EA. Rehospi-talizations among patients in the Medicare Fee-for-Service Program. N Engl J Med 2009;360:1418–28.

2. Axon RN, Williams MV. Hospital readmission asan accountability measure. JAMA 2011;305:504–5.

3. Centers for Medicare & Medicaid Services.Medicare.gov. Hospital Compare. Medicare.govwebsite. Available at: http://www.medicare.gov/hospitalcompare/search.html. Accessed June 16, 2015.

4. Gerhardt G, Yemane A, Hickman P, et al.Medicare readmission rates showed meaningful

decline in 2012. Medicare Medicaid Res Rev 2013;3:003.02.b01.

5. Suter LG, Li SX, Grady JN, et al. National patternsof risk-standardized mortality and readmission af-ter hospitalization for acute myocardial infarction,heart failure, and pneumonia: update on publicly

DeVore et al. J A C C V O L . 6 7 , N O . 8 , 2 0 1 6

Public Reporting of Hospital Readmission Rates M A R C H 1 , 2 0 1 6 : 9 6 3 – 7 2

972

reported outcomes measures based on the 2013release. J Gen Intern Med 2014;29:1333–40.

6. Yale New Haven Health Services CorporationCenter for Outcomes Research and Evaluation.Medicare Hospital Quality Chartbook. performancereport on outcomemeasures. September 2014. Cen-ters for Medicare & Medicaid Services. Available at:http://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/HospitalQualityInits/Downloads/Medicare-Hospital-Quality-Chartbook-2014.pdf. Accessed March 11, 2015.

7. Birman-Deych E, Waterman AD, Yan Y, et al.Accuracy of ICD-9-CM codes for identifying car-diovascular and stroke risk factors. Med Care2005;43:480–5.

8. Quan H, Sundararajan V, Halfon P, et al. Codingalgorithms for defining comorbidities in ICD-9-CMand ICD-10 administrative data. Med Care 2005;43:1130–9.

9. Bernheim SM, Grady JN, Lin Z, et al.National patterns of risk-standardized mortalityand readmission for acute myocardial infarctionand heart failure: update on publicly reportedoutcomes measures based on the 2010 release.Circ Cardiovasc Qual Outcomes 2010;3:459–67.

10. Lindenauer PK, Bernheim SM, Grady JN, et al.The performance of US hospitals as reflected inrisk-standardized 30-day mortality and read-mission rates for Medicare beneficiaries withpneumonia. J Hosp Med 2010;5:E12–8.

11. Centers for Medicare & Medicaid Services.Medicare.gov: Physician Compare. Medicare.govwebsite. Available at: http://www.medicare.gov/

physiciancompare/search.html. Accessed October29, 2015.

12. Joynt KE. Health policy and cardiovascularmedicine: rapid changes, immense opportunities.Circulation 2015;131:1098–105.

13. Ghali WA, Ash AS, Hall RE, Moskowitz MA.Statewide quality improvement initiatives andmortality after cardiac surgery. JAMA 1997;277:379–82.

14. Peterson ED, DeLong ER, Jollis JG,Muhlbaier LH, Mark DB. The effects of New York’sbypass surgery provider profiling on access to careand patient outcomes in the elderly. J Am CollCardiol 1998;32:993–9.

15. Baker DW, Einstadter D, Thomas CL, et al.Mortality trends during a program that publiclyreported hospital performance. Med Care 2002;40:879–90.

16. Werner RM, Bradlow ET. Relationship betweenMedicare’s hospital compare performancemeasures and mortality rates. JAMA 2006;296:2694–702.

17. Ryan AM, Nallamothu BK, Dimick JB. Medi-care’s public reporting initiative on hospital qualityhad modest or no impact on mortality from threekey conditions. Health Aff (Millwood) 2012;31:585–92.

18. Joynt KE, Blumenthal DM, Orav EJ, et al.Association of public reporting for percuta-neous coronary intervention with utilization andoutcomes among Medicare beneficiaries withacute myocardial infarction. JAMA 2012;308:1460–8.

19. Waldo SW, McCabe JM, O’Brien C, et al. As-sociation between public reporting of outcomeswith procedural management and mortality forpatients with acute myocardial infarction. J AmColl Cardiol 2015;65:1119–26.

20. Tu JV, Donovan LR, Lee DS, et al. Effective-ness of public report cards for improving thequality of cardiac care: the EFFECT study: a ran-domized trial. JAMA 2009;302:2330–7.

21. Hernandez AF, Greiner MA, Fonarow GC, et al.Relationship between early physician follow-upand 30-day readmission among Medicare benefi-ciaries hospitalized for heart failure. JAMA 2010;303:1716–22.

22. Vashi AA, Fox JP, Carr BG, et al. Use of hospital-based acute care among patients recently dis-charged from thehospital. JAMA2013;309:364–71.

23. Feng Z, Wright B, Mor V. Sharp rise in Medicareenrollees being held in hospitals for observationraises concerns about causes and consequences.Health Aff (Millwood) 2012;31:1251–9.

24. Daughtridge GW, Archibald T, Conway PH.Quality improvement of care transitions and thetrend of composite hospital care. JAMA 2014;311:1013–4.

25. Brown JR, Sox HC, Goodman DC. Financial in-centives to improve quality: skating to the puckor avoiding the penalty box? JAMA 2014;311:1009–10.

KEY WORDS Centers for Medicare &Medicaid Services, heart failure, myocardialinfarction, quality