Association of Hospital Performance Based on 30-Day Risk-Standardized Mortality Rate With Long-term Survival After Heart Failure Hospitalization: An Analysis of the Get With The Guidelines–Heart Failure Registry | Cardiology | JAMA Cardiology | JAMA Network
[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
Figure.  Long-term Mortality Risk and Median Survival Across Hospitals Stratified By 30-Day Risk-Standardized Mortality Rates (RSMRs)
Long-term Mortality Risk and Median Survival Across Hospitals Stratified By 30-Day Risk-Standardized Mortality Rates (RSMRs)

A, Cumulative incidence of mortality in the overall population by quartile. Quartiles ranged from low (Q1) to high (Q4) 30-day RSMR. B, Median survival according to 30-day RSMR among 30-day survivors. Error bars indicate 95% CIs.

Table 1.  Hospital Characteristics According to 30-Day Risk-Standardized Mortality Rate (RSMR) Quartilea
Hospital Characteristics According to 30-Day Risk-Standardized Mortality Rate (RSMR) Quartilea
Table 2.  Patient Characteristics According to Hospital 30-Day Risk-Standardized Mortality Rate Quartilea
Patient Characteristics According to Hospital 30-Day Risk-Standardized Mortality Rate Quartilea
Table 3.  Adherence to Get With The Guidelines–Heart Failure Registry Achievement and Quality Metrics Among Patients According to Hospital 30-Day Risk-Standardized Mortality Rate Quartilea
Adherence to Get With The Guidelines–Heart Failure Registry Achievement and Quality Metrics Among Patients According to Hospital 30-Day Risk-Standardized Mortality Rate Quartilea
Table 4.  Long-term Mortality Risk Among Hospitalized Patients With Heart Failure According to 30-Day Risk-Standardized Mortality Rate–Based Hospital Quartile Groupsa
Long-term Mortality Risk Among Hospitalized Patients With Heart Failure According to 30-Day Risk-Standardized Mortality Rate–Based Hospital Quartile Groupsa
Supplement.

eMethods 1. Get With The Guidelines–Heart Failure cohort description.

eMethods 2. Covariates used in the model to determine the 30-day risk-standardized mortality rates for each participating center.

eMethods 3. Hospital characteristics used in the model to determine the association between 30-day risk-standardized mortality rate–based hospital groups and long-term mortality.

eFigure 1. CONSORT study flow diagram.

eFigure 2. Histogram of hospital 30-day risk-standardized mortality rate.

eTable 1. Multivariable-adjusted association between 30-day risk-standardized mortality rate–based hospital groups and long-term outcomes according to heart failure subtypes (HFpEF vs HFrEF) in the overall population.

eTable 2. Long-term mortality risk among hospitalized patients with heart failure according to 30-day risk-standardized mortality rate—early period (2005-2009; 206 hospitals and 43 410 patients).

eTable 3. Long-term mortality risk among hospitalized patients with heart failure according to 30-day risk-standardized mortality rate—late period (2010-2013; 249 hospitals and 62 894 patients).

eTable 4. Long-term mortality risk among hospitalized patients with heart failure according to 30-day risk-standardized mortality rate—addition of calendar year in the adjustment.

eTable 5. Long-term mortality risk among hospitalized patients with heart failure according to 30-day risk-standardized mortality rate—only continuous participating hospitals (hospitals with at least 1 patient entered in 2005-2007 period and at least 1 patient each calendar year thereafter; 56 hospitals and 41 898 patients).

1.
Mozaffarian  D, Benjamin  EJ, Go  AS,  et al; Writing Group Members; American Heart Association Statistics Committee; Stroke Statistics Subcommittee.  Executive summary: heart disease and stroke statistics—2016 update: a report from the American Heart Association.  Circulation. 2016;133(4):447-454.PubMedGoogle ScholarCrossref
2.
US Centers for Medicare and Medicaid Services. Hospital quality initiative: outcome measures. https://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/HospitalQualityInits/OutcomeMeasures.html. Accessed November 18, 2017.
3.
 Medicare Program; hospital inpatient prospective payment systems for acute care hospitals and the long-term care hospital prospective payment system and fiscal year 2014 rates; quality reporting requirements for specific providers; hospital conditions of participation; payment policies related to patient status.  Fed Regist. 2013;78(160):50676-50729.Google Scholar
4.
 Medicare Program; hospital inpatient prospective payment systems for acute care hospitals and the long-term care hospital prospective payment system and fiscal year 2014 rates; quality reporting requirements for specific providers; hospital conditions of participation; payment policies related to patient status.  Fed Regist. 2013;78(160):50649-50676.Google Scholar
5.
Bucholz  EM, Butala  NM, Ma  S, Normand  ST, Krumholz  HM.  Life expectancy after myocardial infarction, according to hospital performance.  N Engl J Med. 2016;375(14):1332-1342.PubMedGoogle ScholarCrossref
6.
Hong  Y, LaBresh  KA.  Overview of the American Heart Association “Get With The Guidelines” programs: coronary heart disease, stroke, and heart failure.  Crit Pathw Cardiol. 2006;5(4):179-186.PubMedGoogle ScholarCrossref
7.
Smaha  LA; American Heart Association.  The American Heart Association Get With The Guidelines program.  Am Heart J. 2004;148(5, suppl):S46-S48.PubMedGoogle ScholarCrossref
8.
Hammill  BG, Hernandez  AF, Peterson  ED, Fonarow  GC, Schulman  KA, Curtis  LH.  Linking inpatient clinical registry data to Medicare claims data using indirect identifiers.  Am Heart J. 2009;157(6):995-1000.PubMedGoogle ScholarCrossref
9.
Drye  EE, Normand  SL, Wang  Y,  et al.  Comparison of hospital risk-standardized mortality rates calculated by using in-hospital and 30-day models: an observational study with implications for hospital profiling.  Ann Intern Med. 2012;156(1, pt 1):19-26.PubMedGoogle ScholarCrossref
10.
Dharmarajan  K, Wang  Y, Lin  Z,  et al.  Association of changing hospital readmission rates with mortality rates after hospital discharge.  JAMA. 2017;318(3):270-278.PubMedGoogle ScholarCrossref
11.
Fonarow  GC, Konstam  MA, Yancy  CW.  The Hospital Readmission Reduction Program is associated with fewer readmissions, more deaths: time to reconsider.  J Am Coll Cardiol. 2017;70(15):1931-1934.PubMedGoogle ScholarCrossref
12.
Gupta  A, Allen  LA, Bhatt  DL,  et al.  Association of the Hospital Readmissions Reduction Program implementation with readmission and mortality outcomes in heart failure.  JAMA Cardiol. 2018;3(1):44-53.PubMedGoogle ScholarCrossref
13.
Pandey  A, Golwala  H, Xu  H,  et al.  Association of 30-day readmission metric for heart failure under the Hospital Readmissions Reduction Program with quality of care and outcomes.  JACC Heart Fail. 2016;4(12):935-946.PubMedGoogle ScholarCrossref
14.
Joynt  KE, Jha  AK.  Thirty-day readmissions: truth and consequences.  N Engl J Med. 2012;366(15):1366-1369.PubMedGoogle ScholarCrossref
15.
Press  MJ, Scanlon  DP, Ryan  AM,  et al.  Limits of readmission rates in measuring hospital quality suggest the need for added metrics.  Health Aff (Millwood). 2013;32(6):1083-1091.PubMedGoogle ScholarCrossref
16.
Pandey  A, Golwala  H, Hall  HM,  et al.  Association of US Centers for Medicare and Medicaid Services hospital 30-day risk-standardized readmission metric with care quality and outcomes after acute myocardial infarction: findings from the National Cardiovascular Data Registry/Acute Coronary Treatment and Intervention Outcomes Network Registry–Get With The Guidelines.  JAMA Cardiol. 2017;2(7):723-731.PubMedGoogle ScholarCrossref
17.
Abdul-Aziz  AA, Hayward  RA, Aaronson  KD, Hummel  SL.  Association between Medicare hospital readmission penalties and 30-day combined excess readmission and mortality.  JAMA Cardiol. 2017;2(2):200-203.PubMedGoogle ScholarCrossref
18.
Ibrahim  AM, Dimick  JB, Sinha  SS, Hollingsworth  JM, Nuliyalu  U, Ryan  AM.  Association of coded severity with readmission reduction after the Hospital Readmissions Reduction Program.  JAMA Intern Med. 2018;178(2):290-292.PubMedGoogle ScholarCrossref
19.
Konstam  MA, Upshaw  J.  Sisyphus and 30-day heart failure readmissions: futility in predicting a flawed outcome metric.  JACC Heart Fail. 2016;4(1):21-23.PubMedGoogle ScholarCrossref
20.
Curtis  LH, Greiner  MA, Hammill  BG,  et al.  Representativeness of a national heart failure quality-of-care registry: comparison of OPTIMIZE-HF and non-OPTIMIZE-HF Medicare patients.  Circ Cardiovasc Qual Outcomes. 2009;2(4):377-384.PubMedGoogle ScholarCrossref
Original Investigation
June 2018

Association of Hospital Performance Based on 30-Day Risk-Standardized Mortality Rate With Long-term Survival After Heart Failure Hospitalization: An Analysis of the Get With The Guidelines–Heart Failure Registry

Author Affiliations
  • 1Division of Cardiology, University of Texas Southwestern Medical Center, Dallas
  • 2Duke Clinical Research Institute, Durham, North Carolina
  • 3Brigham and Women’s Hospital Heart & Vascular Center and Harvard Medical School, Boston, Massachusetts
  • 4Division of Cardiology, Northwestern University Feinberg School of Medicine, Chicago, Illinois
  • 5Deputy Editor, JAMA Cardiology
  • 6Associate Editor, JAMA Cardiology
  • 7Division of Cardiology, Stanford University, Palo Alto, California
  • 8Ahmanson-UCLA Cardiomyopathy Center, Ronald Reagan UCLA Medical Center, University of California, Los Angeles
  • 9Associate Editor for Health Care Quality and Guidelines, JAMA Cardiology
JAMA Cardiol. 2018;3(6):489-497. doi:10.1001/jamacardio.2018.0579
Key Points

Question  What is the association of hospital performance based on 30-day risk-standardized mortality rate (RSMR) with long-term survival?

Findings  In this longitudinal observational study of 106 304 patients hospitalized with heart failure in the Get With The Guidelines–Heart Failure registry linked to Medicare data, there was a graded inverse association of 30-day RSMR with 1-year, 3-year, and 5-year survival. Compared with patients admitted to hospitals with low 30-day RSMRs, patients at hospitals with high 30-day RSMRs had 14% higher risk-adjusted relative risk of 5-year mortality.

Meaning  Lower hospital-level 30-day RSMR is associated with better long-term survival for patients hospitalized with heart failure independent of mortality differences within the first 30 days.

Abstract

Importance  Among patients hospitalized with heart failure (HF), the long-term clinical implications of hospitalization at hospitals based on 30-day risk-standardized mortality rates (RSMRs) is not known.

Objective  To evaluate the association of hospital-specific 30-day RSMR with long-term survival among patients hospitalized with HF in the American Heart Association Get With The Guidelines–HF registry.

Design, Setting, and Participants  The longitudinal observational study included 106 304 patients with HF who were admitted to 317 centers participating in the Get With The Guidelines–HF registry from January 1, 2005, to December 31, 2013, and had Medicare-linked follow-up data. Hospital-specific 30-day RSMR was calculated using a hierarchical logistic regression model. In the model, 30-day mortality rate was a binary outcome, patient baseline characteristics were included as covariates, and the hospitals were treated as random effects. The association of 30-day RSMR-based hospital groups (low to high 30-day RSMR: quartile 1 [Q1] to Q4) with long-term (1-year, 3-year, and 5-year) mortality was assessed using adjusted Cox models. Data analysis took place from June 29, 2017, to February 19, 2018.

Exposures  Thirty-day RSMR for participating hospitals.

Main Outcomes and Measures  One-year, 3-year, and 5-year mortality rates.

Results  Of the 106 304 patients included in the analysis, 57 552 (54.1%) were women and 84 595 (79.6%) were white, and the median (interquartile range) age was 81 (74-87) years. The 30-day RSMR ranged from 8.6% (Q1) to 10.7% (Q4). Hospitals in the low 30-day RSMR group had greater availability of advanced HF therapies, cardiac surgery, and percutaneous coronary interventions. In the primary landmarked analyses among 30-day survivors, there was a graded inverse association between 30-day RSMR and long-term mortality (Q1 vs Q4: 5-year mortality, 73.7% vs 76.8%). In adjusted analysis, patients admitted to hospitals in the high 30-day RSMR group had 14% (95% CI, 10-18) higher relative hazards of 5-year mortality compared with those admitted to hospitals in the low 30-day RSMR group. Similar findings were observed in analyses of survival from admission, with 22% (95% CI, 18-26) higher relative hazards of 5-year mortality for patients admitted to Q4 vs Q1 hospitals.

Conclusions and Relevance  Lower hospital-level 30-day RSMR is associated with greater 1-year, 3-year, and 5-year survival for patients with HF. These differences in 30-day survival continued to accrue beyond 30 days and persisted long term, suggesting that 30-day RSMR may be a useful HF performance metric to incentivize quality care and improve long-term outcomes.

Introduction

Heart failure (HF) is a major cause of morbidity, mortality, and health care expenditure in the United States.1 Heart failure hospitalizations have been the focus of recent health policy initiatives launched by the US Centers for Medicare and Medicaid Services (CMS) to incentivize hospitals to improve acute hospitalization–associated care quality and cost of care. Among these initiatives, CMS mandated public reporting of short-term hospital performance metrics, such as 30-day risk-standardized readmission rates (RSRRs) and 30-day risk-standardized mortality rates (RSMRs) for common conditions, such as HF, acute myocardial infarction (AMI), and pneumonia, in 2007 to 2008.2 Subsequently, CMS implemented the Hospital Readmissions Reduction Program and the Hospital Value-Based Purchasing Program to incentivize hospitals by levying financial penalties based on performance on the 30-day RSRR and 30-day RSMR quality metrics, respectively, for these common conditions.3,4

Evaluating whether patients admitted to hospitals with better short-term outcomes have better long-term outcomes can provide insights into whether care quality for a given disease state is associated with long-term outcomes. In 2016, an analysis of hospital performance based on 30-day RSMRs for AMI showed that the early mortality differences were enduring.5 To our knowledge, the effect of 30-day RSMR on long-term survival in patients hospitalized with HF is not known. Thus, it is not clear whether patients admitted to better-performing hospitals based on 30-day mortality rate have improved long-term survival or whether any short-term gains rapidly dissipate. We used data from the American Heart Association Get With The Guidelines (GWTG)–HF registry, an ongoing prospective performance improvement registry linked to Medicare data, to evaluate hospital performance indexed by 30-day RSMR and its association with intermediate and long-term mortality outcomes along with median survival.

Methods
Data Source

We analyzed data from the GWTG-HF registry and the linked Medicare Part A inpatient fee-for-service claims files for this study. Details about the GWTG-HF program have been published previously6,7 and are described in eMethods 1 of the Supplement. The Duke Clinical Research Institute is the data analysis center and analyzed the aggregate deidentified data for research. The centers participating in the GWTG-HF program are required to obtain institutional review board approval for the GWTG-HF protocol and are granted a waiver for informed consent under the common rule. The Duke Clinical Research Institute has institutional review board approval for analyses of deidentified GWTG-HF data for research purposes. Data quality and completeness was monitored independently for quality assurance.

Data on postdischarge mortality on follow-up were obtained from the linked Medicare claims files. The inpatient claims files contain information from hospital claims submitted for facility costs related to the services provided during the in-hospital stay. The denominator files contain data on Medicare enrollment and mortality. Patients in the GWTG-HF registry were identified in the Medicare claims files by linking registry hospitalizations to Medicare claims files using admission and discharge dates, hospital, date of birth, and sex, as reported previously.8

Study Population

We included patients 65 years and older who were hospitalized in centers participating in the GWTG-HF program with a principal diagnosis of HF between January 1, 2005, and December 31, 2013, and had CMS-linked data available. Hospitals in the GWTG-HF registry with less than 75% complete medical history data or with less than 30 patients hospitalized with HF during the study period were excluded. For patients with multiple admissions, the first hospitalization meeting the above criteria was considered the index hospitalization.

Outcomes of Interest

The primary outcome of interest was mortality at 1 year, 3 years, and 5 years among the study participants, as determined from the CMS-linked denominator files. The CMS-linked data on mortality outcomes were available among study participants from January 1, 2005, through December 31, 2014. Time to death was defined as the number of days from index admission to death date. All patients had at least 1 year of follow-up; therefore, there is no censoring for 1-year mortality. For 3-year and 5-year mortality, the time to death outcome was censored by the end of the study period (ie, December 31, 2014).

Primary Exposure Variable

Multivariable hierarchical logistic models were constructed to calculate hospitals’ 30-day RSMR. In the model, 30-day mortality was a binary outcome, patient baseline characteristics were included as covariates, and the hospitals were treated as random effects. The patient covariates in the adjusted model included demographic characteristics, socioeconomic status, medical history, left ventricular ejection fraction (EF), vital signs and laboratory test results at admission, baseline medications, number of all-cause and HF hospitalizations within 6 months prior to index hospitalization (at non–GWTG-HF centers or prior to enrollment in the GWTG-HF registry), and hierarchical condition category (eMethods 2 in the Supplement). Thirty-day RSMR for a given hospital was defined as the ratio of the number of predicted to expected deaths multiplied by the patient-level average mortality rate9: 30-day RSMR = (No. of predicted deaths using hospital-specific intercept) / (No. of expected deaths using average of all hospital-specific intercepts) × average mortality rate. The predicted number of deaths for each hospital was calculated by summing over the predicted mortality risks for all patients in the hospital using the hospital-specific intercept, and the expected number of deaths for each hospital was calculated by summing over the expected mortality risks for all patients in the hospital using the average of all hospital-specific intercepts.

Statistical Analysis

Hospitals were classified into quartiles based on hospitals’ calculated 30-day RSMR. Baseline hospital-level and patient-level characteristics across the 30-day RSMR-based hospital groups (quartile 1 [Q1] to Q4) were described using proportions for categorical variables and medians with interquartile ranges for continuous variables. Categorical variables were compared with χ2 test and continuous variables were compared with Kruskal-Wallis test across the study groups. Cumulative incidence of 1-year, 3-year, and 5-year mortality across the study groups was determined using Kaplan-Meier estimates, and the between-group differences in long-term mortality were tested using log-rank tests. The adjusted association of 30-day RSMR-based hospital groups with long-term mortality was assessed using a multivariable Cox proportional hazard model, with the lowest 30-day RSMR group as the referent (Q2, Q3, and Q4 vs Q1 [referent]). The robust standard errors in Cox models were used to account for the clustering of patients by hospital. The proportional hazard assumption was assessed using graphical method and statistical tests. These models were adjusted for patient-level covariates (eMethods 2 in the Supplement) and hospital-level covariates (eMethods 3 in the Supplement), such as region, number of beds, teaching hospital (yes vs no), and rural vs urban location.

To account for the potential effect of mortality during the first 30 days on long-term outcomes, the primary analysis was limited to 30-day survivors only. The associations of 30-day RSMR-based hospital groups with long-term outcomes in the overall population, including deaths within the first 30 days, were assessed as secondary analysis. Stratified analysis was also performed to examine the association of 30-day RSMR-based hospital groups with long-term mortality in subgroups of patients with HF with reduced EF (HFrEF), defined as an EF less than 40%, and patients with HF with preserved EF (HFpEF), defined as an EF of 40% or greater. Interactions were tested in the stratified analysis to determine whether the association of 30-day RSMR-based hospital groups with long-term morality was modified by HF subtype (HFrEF vs HFpEF).

To evaluate whether the results were influenced by the period of the study, we conducted additional sensitivity analyses that divided the cohort into 2 study periods (2005-2009 and 2010-2013) and in separate analyses further adjusting for calendar year. Sensitivity analysis was also performed to evaluate the association of 30-day RSMR-based hospital groups (high vs low) with risk of long-term mortality among hospitals that participated continuously in the GWTG-HF program during the study period. All statistical tests were 2-sided, with a P value less than .05 indicating statistical significance. Data were analyzed using SAS, version 9.4 (SAS Institute).

Results
Hospital and Patient Characteristics

From January 2005 to December 2013, 317 hospitals with 106 304 patients hospitalized with HF were included in our study (eFigure 1 in the Supplement). The median (range) 30-day RSMR for all hospitals included in the study was 9.7% (6.4-13.6) (eFigure 2 in the Supplement). The hospitals in the low 30-day RSMR group (Q1) were more likely to be larger, urban centers and had greater availability of percutaneous coronary interventions, cardiac surgery, and advanced HF therapies compared with those in the high 30-day RSMR group (Q4) (Table 1). Among patient-level characteristics, those admitted to hospitals in the low 30-day RSMR group had higher socioeconomic status compared with patients admitted to hospitals in the high 30-day RSMR group. There were no clinically meaningful differences in other demographic characteristics, prevalence of medical comorbidities, and disease severity across the 30-day RSMR-based hospital groups (Table 2). Among process of care measures during the index hospitalization, hospitals in the low 30-day RSMR group had significantly higher use of implantable cardioverter defibrillators and cardiac resynchronization therapies and greater postdischarge follow-up. Use of evidence-based medications was not meaningfully different across the hospital groups despite statistically significant differences (Table 3).

Hospital-Level 30-Day RSMR and Long-term Mortality Risk Among Patients With HF

The cumulative incidence of mortality and median survival over long-term follow-up across different 30-day RSMR-based hospital groups (Q1 to Q4) are shown in Table 4 and the Figure. In the primary landmarked analyses of 30-day survivors, there was a significant graded association between hospital-level 30-day RSMR and the risk of long-term mortality. The 5-year mortality rates were 73.7% in low 30-day RSMR hospitals compared with 76.8% in high 30-day RSMR hospitals (Table 4). In the landmark analysis, patients admitted to hospitals in the low 30-day RSMR group had higher median survival (832 days; 95% CI, 815-852) compared with those admitted to hospitals in the high 30-day RSMR group (759 days; 95% CI, 742-779) (Figure). In multivariable-adjusted analysis, higher 30-day RSMR was significantly associated with higher long-term mortality among patients hospitalized with HF. Compared with the low 30-day RSMR group, patients admitted to hospitals in the high 30-day RSMR group had 14% higher relative hazards of 3-year and 5-year mortality (Table 4). Similar findings were observed in analyses that included all patients and counted deaths within 30 days (Table 4). A significant interaction was observed between HF subtype and 30-day RSMR-based hospital group for the outcomes of 3-year and 5-year mortality in the overall population. Thus, the association between 30-day RSMR-based hospital group (high vs low) and risk of long-term mortality was stronger among patients with HFrEF than those with HFpEF (eTable 1 in the Supplement).

Sensitivity Analysis

To evaluate whether the observed study findings were influenced by the period of the study, sensitivity analyses were conducted by dividing the cohort into 2 study periods (2005-2009 and 2010-2013) and in separate analyses further adjusting for calendar year. The results were similar in each study period and to those overall (eTables 2 and 3 in the Supplement) as well as after adjusting for calendar year (eTable 4 in the Supplement). Furthermore, in sensitivity analysis limited to hospitals with continuous participation in the GWTG-HF program throughout the study period, the higher risk of long-term mortality in the high 30-day RSMR group vs the low 30-day RSMR group was consistent with that observed in the primary analysis (eTable 5 in the Supplement).

Discussion

We observed several important findings in this study. First, we observed significant differences in 1-year, 3-year, and 5-year survival among patients hospitalized at centers with high vs low 30-day RSMR. Compared with patients with HF admitted to hospitals with high 30-day RSMR, those treated at hospitals with low 30-day RSMR had a 14% higher relative 5-year survival after adjustment for patient case-mix and other potential confounders and after excluding deaths within 30 days. Second, there were significant differences in adherence to certain process of care measures, with higher use of implantable cardioverter defibrillators and cardiac resynchronization therapies, greater postdischarge follow-up, and greater use of advanced HF therapies at hospitals with low vs high 30-day RSMR, highlighting the potential role of quality of care in contributing to differences in long-term survival across these hospitals. Third, the association between 30-day RSMR and risk of long-term mortality was stronger for patients with HFrEF vs HFpEF. Taken together, our study is the first, to our knowledge, to demonstrate a long-term survival advantage associated with care at centers with lower 30-day RSMR for patients hospitalized with HF.

We observed a significant graded association between hospital-level 30-day RSMR and long-term mortality among patients hospitalized with HF. This association was independent of hospital-level and patient-level confounders and highlights the potential value of 30-day RSMR in evaluating hospital performance, as better performance on this metric is strongly associated with long-term survival and life expectancy. A 2016 study by Bucholz et al5 demonstrated a survival advantage for patients with AMI hospitalized at high-performing centers based on 30-day RSMR for AMI. However, this survival advantage in the AMI study was largely driven by differences in mortality rates within the first 30 days and not noted among 30-day survivors. This suggests that the early benefits from hospital performance in AMI care persist but beyond 30 days no longer continue to accrue. In contrast, in patients hospitalized with HF in the present study, the differences in care quality that may account for differences in 30-day mortality appear to lead to survival gains that are further enhanced over time. A key strength of our analysis was the availability of patient-level data from the index hospitalization in the GWTG-HF registry that allowed for better adjustment of disease severity, patient case-mix, socioeconomic status, and race/ethnicity in the calculation of the 30-day RSMR, which had not been previously performed in studies using claims-based data only.5,10

Several factors may underlie the observed usefulness of 30-day RSMR-based hospital performance in predicting long-term outcomes. There was significantly higher adherence to some key HF process of care measures in low vs high 30-day RSMR-based hospital groups. For example, a greater proportion of patients with HF admitted to hospitals in the low 30-day RSMR group received implantable cardioverter defibrillator placement and cardiac resynchronization therapy prescription and had postdischarge follow-up appointments. Furthermore, there was up to a 4-fold difference in availability of advanced HF therapy options at hospitals in the low vs high 30-day RSMR-based groups. Taken together, these findings suggest that greater use of evidence-based, life-prolonging therapies and greater postdischarge follow-up care may contribute, at least in part, to better long-term survival at hospitals with low 30-day RSMRs. This notion is also supported by the stronger association between hospital-level 30-day RSMR and long-term outcomes in patients with HFrEF who can benefit from life-prolonging therapies vs those with HFpEF and limited mortality-improving treatment options.

The long-term prognostic value of hospital performance based on 30-day RSMR has important health policy implications, particularly when viewed in context of the limited utility of 30-day RSRR, the primary quality metric used by CMS to financially incentivize hospitals.11-15 In recent studies,13,16 hospital performance based on 30-day RSRR was not associated with in-hospital care quality and long-term clinical outcomes in patients with HF or AMI. Furthermore, there was a trend toward higher mortality for patients with HF in centers with low vs high 30-day RSRR.13 Despite these limitations, the current financial incentive program used by CMS is heavily skewed in favor of 30-day RSRR-based hospital performance, with up to 15-fold greater penalties for excess readmission rates (3%) than excess mortality rates (0.2%) in financial year 2016.17 As a result, there is an increasing drive to invest hospital resources in programs focused on reducing readmissions. While such programs may have led to a decline in readmission rates, there has been a significant and steady increase in risk-adjusted mortality rates in HF over the same period.10,12 This raises the concern that excessive focus on 30-day RSRR as an actionable quality metric may have inadvertently encouraged inappropriate care practices.11,18,19 Findings from our study suggest that hospital-level 30-day RSMR may be a useful metric for hospital performance and should potentially be weighted more in CMS financial incentive programs.

Limitations

Several limitations to our analysis are noteworthy. First, the findings from our study may not be generalizable to all hospitals across the United States because this study focused on hospitals participating voluntarily in the GWTG-HF quality improvement program. However, the GWTG-HF registry captures participating sites from across the United States with significant representation of small and large, academic and nonacademic, and rural and urban hospitals. Furthermore, previous studies have demonstrated that Medicare beneficiaries enrolled in the GWTG-HF registry are nationally representative.20 Second, patient-level data in GWTG-HF were collected by medical record review and could be prone to inaccuracies. Third, because of the observational nature of our study, there is a potential for residual measured and unmeasured confounding. As such, we cannot establish causation between hospital performance–based 30-day RSMR and long-term survival.

Conclusions

In conclusion, long-term survival in patients discharged after HF hospitalizations varied according to hospital-level 30-day RSMR for HF. The survival advantage associated with treatment at hospitals with lower 30-day RSMR was durable over time independent of mortality differences within the first 30 days. Lower 30-day RSMR was associated with better adherence of key process of care measures, greater availability of advanced HF therapies, and better 1-year, 3-year, and 5-year survival. Taken together, these findings highlight the need to increase the focus on 30-day RSMR as a performance metric to incentivize quality care and to improve long-term clinical outcomes for patients with HF.

Back to top
Article Information

Accepted for Publication: February 20, 2018.

Corresponding Author: Gregg C. Fonarow, MD, Ahmanson-UCLA Cardiomyopathy Center, Ronald Reagan UCLA Medical Center, University of California, Los Angeles, 10833 LeConte Ave, Room 47-123 CHS, Los Angeles, CA 90095-1679 (gfonarow@mednet.ucla.edu).

Published Online: March 12, 2018. doi:10.1001/jamacardio.2018.0579

Author Contributions: Dr Fonarow had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Pandey, Patel, Yancy, Fonarow.

Acquisition, analysis, or interpretation of data: Patel, Liang, DeVore, Matsouaka, Bhatt, Hernandez, Heidenreich, de Lemos, Fonarow.

Drafting of the manuscript: Pandey, Patel.

Critical revision of the manuscript for important intellectual content: Patel, Liang, DeVore, Matsouaka, Bhatt, Yancy, Hernandez, Heidenreich, de Lemos, Fonarow.

Statistical analysis: Pandey, Liang, DeVore, Matsouaka.

Obtained funding: Fonarow.

Administrative, technical, or material support: Hernandez, Fonarow.

Study supervision: Matsouaka, Yancy, Fonarow.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr DeVore has received research support from Amgen, the American Heart Association, the National Heart, Lung, and Blood Institute, and Novartis and has consulted with Novartis. Dr Bhatt has served on the advisory boards of Cardax, Elsevier Practice Update Cardiology, Medscape Cardiology, and Regado Biosciences; has served on the board of directors of Boston VA Research Institute and Society of Cardiovascular Patient Care; has served as the chair of the American Heart Association Quality Oversight Committee; has served on data monitoring committees of Cleveland Clinic, Duke Clinical Research Institute, Harvard Clinical Research Institute, Mayo Clinic, Mount Sinai School of Medicine, and Population Health Research Institute; has received honoraria from American College of Cardiology for serving as a senior associate editor of clinical trials and news, Belvoir Publications for serving as editor in chief of Harvard Heart Letter, Duke Clinical Research Institute for serving on clinical trial steering committees, Harvard Clinical Research Institute for serving on a clinical trial steering committee, HMP Communications for serving as editor in chief of the Journal of Invasive Cardiology, Journal of the American College of Cardiology for serving as guest editor and associate editor, Population Health Research Institute for serving on a clinical trial steering committee, Slack Publications for serving as chief medical editor of Cardiology Today’s Intervention, Society of Cardiovascular Patient Care for serving as secretary/treasurer, WebMD for serving on Continuing Medical Education steering committees, Clinical Cardiology for serving as deputy editor, National Cardiovascular Data Registry Acute Coronary Treatment and Intervention Outcomes Network Registry Steering Committee for serving as chair, and VA Clinical Assessment Reporting and Tracking Program Research and Publications Committee for serving as chair; has received research funding from Amarin, Amgen, AstraZeneca, Bristol-Myers Squibb, Chiesi, Eisai, Ethicon, Forest Laboratories, Ironwood, Ischemix, Lilly, Medtronic, Pfizer, Roche, Sanofi Aventis, and The Medicines Company; has received royalties from Elsevier for serving as editor of Cardiovascular Intervention: A Companion to Braunwald’s Heart Disease; has served as site coinvestigator for Biotronik, Boston Scientific, and St Jude Medical (now Abbott); is a trustee for American College of Cardiology; and has performed unfunded research with FlowCo, Merck, PLx Pharma, and Takeda. Dr Hernandez has received research funding from Janssen, Novartis, Portola, and Bristol-Myers Squibb and has consulted for Bristol-Myers Squibb, Gilead, Boston Scientific, Janssen, and Novartis. Dr Fonarow has received research funding from National Institutes of Health and has consulted for Amgen, Bayer, Janssen, Novartis, Medtronic, and St Jude Medical. No other disclosures were reported.

Funding/Support: The American Heart Association supports the Get With The Guidelines–Heart Failure program. The Get With The Guidelines–Heart Failure program has been previously funded through support from Medtronic, GlaxoSmithKline, Ortho-McNeil Pharmaceutical, and the American Heart Association Pharmaceutical Roundtable.

Role of the Funder/Sponsor: The industry funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: Dr Yancy is a Deputy Editor, Dr Hernandez is an Associate Editor, and Dr Fonarow is the Associate Editor for Health Care Quality and Guidelines of JAMA Cardiology. They were not involved in the evaluation or decision to accept this article for publication.

Meeting Presentation: This article was presented at the American College of Cardiology Annual Meeting, 2018; March 12, 2018; Orlando, Florida.

References
1.
Mozaffarian  D, Benjamin  EJ, Go  AS,  et al; Writing Group Members; American Heart Association Statistics Committee; Stroke Statistics Subcommittee.  Executive summary: heart disease and stroke statistics—2016 update: a report from the American Heart Association.  Circulation. 2016;133(4):447-454.PubMedGoogle ScholarCrossref
2.
US Centers for Medicare and Medicaid Services. Hospital quality initiative: outcome measures. https://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/HospitalQualityInits/OutcomeMeasures.html. Accessed November 18, 2017.
3.
 Medicare Program; hospital inpatient prospective payment systems for acute care hospitals and the long-term care hospital prospective payment system and fiscal year 2014 rates; quality reporting requirements for specific providers; hospital conditions of participation; payment policies related to patient status.  Fed Regist. 2013;78(160):50676-50729.Google Scholar
4.
 Medicare Program; hospital inpatient prospective payment systems for acute care hospitals and the long-term care hospital prospective payment system and fiscal year 2014 rates; quality reporting requirements for specific providers; hospital conditions of participation; payment policies related to patient status.  Fed Regist. 2013;78(160):50649-50676.Google Scholar
5.
Bucholz  EM, Butala  NM, Ma  S, Normand  ST, Krumholz  HM.  Life expectancy after myocardial infarction, according to hospital performance.  N Engl J Med. 2016;375(14):1332-1342.PubMedGoogle ScholarCrossref
6.
Hong  Y, LaBresh  KA.  Overview of the American Heart Association “Get With The Guidelines” programs: coronary heart disease, stroke, and heart failure.  Crit Pathw Cardiol. 2006;5(4):179-186.PubMedGoogle ScholarCrossref
7.
Smaha  LA; American Heart Association.  The American Heart Association Get With The Guidelines program.  Am Heart J. 2004;148(5, suppl):S46-S48.PubMedGoogle ScholarCrossref
8.
Hammill  BG, Hernandez  AF, Peterson  ED, Fonarow  GC, Schulman  KA, Curtis  LH.  Linking inpatient clinical registry data to Medicare claims data using indirect identifiers.  Am Heart J. 2009;157(6):995-1000.PubMedGoogle ScholarCrossref
9.
Drye  EE, Normand  SL, Wang  Y,  et al.  Comparison of hospital risk-standardized mortality rates calculated by using in-hospital and 30-day models: an observational study with implications for hospital profiling.  Ann Intern Med. 2012;156(1, pt 1):19-26.PubMedGoogle ScholarCrossref
10.
Dharmarajan  K, Wang  Y, Lin  Z,  et al.  Association of changing hospital readmission rates with mortality rates after hospital discharge.  JAMA. 2017;318(3):270-278.PubMedGoogle ScholarCrossref
11.
Fonarow  GC, Konstam  MA, Yancy  CW.  The Hospital Readmission Reduction Program is associated with fewer readmissions, more deaths: time to reconsider.  J Am Coll Cardiol. 2017;70(15):1931-1934.PubMedGoogle ScholarCrossref
12.
Gupta  A, Allen  LA, Bhatt  DL,  et al.  Association of the Hospital Readmissions Reduction Program implementation with readmission and mortality outcomes in heart failure.  JAMA Cardiol. 2018;3(1):44-53.PubMedGoogle ScholarCrossref
13.
Pandey  A, Golwala  H, Xu  H,  et al.  Association of 30-day readmission metric for heart failure under the Hospital Readmissions Reduction Program with quality of care and outcomes.  JACC Heart Fail. 2016;4(12):935-946.PubMedGoogle ScholarCrossref
14.
Joynt  KE, Jha  AK.  Thirty-day readmissions: truth and consequences.  N Engl J Med. 2012;366(15):1366-1369.PubMedGoogle ScholarCrossref
15.
Press  MJ, Scanlon  DP, Ryan  AM,  et al.  Limits of readmission rates in measuring hospital quality suggest the need for added metrics.  Health Aff (Millwood). 2013;32(6):1083-1091.PubMedGoogle ScholarCrossref
16.
Pandey  A, Golwala  H, Hall  HM,  et al.  Association of US Centers for Medicare and Medicaid Services hospital 30-day risk-standardized readmission metric with care quality and outcomes after acute myocardial infarction: findings from the National Cardiovascular Data Registry/Acute Coronary Treatment and Intervention Outcomes Network Registry–Get With The Guidelines.  JAMA Cardiol. 2017;2(7):723-731.PubMedGoogle ScholarCrossref
17.
Abdul-Aziz  AA, Hayward  RA, Aaronson  KD, Hummel  SL.  Association between Medicare hospital readmission penalties and 30-day combined excess readmission and mortality.  JAMA Cardiol. 2017;2(2):200-203.PubMedGoogle ScholarCrossref
18.
Ibrahim  AM, Dimick  JB, Sinha  SS, Hollingsworth  JM, Nuliyalu  U, Ryan  AM.  Association of coded severity with readmission reduction after the Hospital Readmissions Reduction Program.  JAMA Intern Med. 2018;178(2):290-292.PubMedGoogle ScholarCrossref
19.
Konstam  MA, Upshaw  J.  Sisyphus and 30-day heart failure readmissions: futility in predicting a flawed outcome metric.  JACC Heart Fail. 2016;4(1):21-23.PubMedGoogle ScholarCrossref
20.
Curtis  LH, Greiner  MA, Hammill  BG,  et al.  Representativeness of a national heart failure quality-of-care registry: comparison of OPTIMIZE-HF and non-OPTIMIZE-HF Medicare patients.  Circ Cardiovasc Qual Outcomes. 2009;2(4):377-384.PubMedGoogle ScholarCrossref
×