Current Risk Adjustment and Comorbidity Index Underperformance in Predicting Post-Acute Utilization and Hospital Readmissions After Joint Replacements

Implications for Comprehensive Care for Joint Replacement Model

Amit Kumar, Amol Karmarkar, Brian Downer, Amit Vashist, Deepak Adhikari, Soham Al Snih al snih, Kenneth Ottenbacher

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

Objective: To compare the performances of 3 comorbidity indices, the Charlson Comorbidity Index, the Elixhauser Comorbidity Index, and the Centers for Medicare & Medicaid Services (CMS) risk adjustment model, Hierarchical Condition Category (HCC), in predicting post-acute discharge settings and hospital readmission for patients after joint replacement. Methods: A retrospective study of Medicare beneficiaries with total knee replacement (TKR) or total hip replacement (THR) discharged from hospitals in 2009-2011 (n = 607,349) was performed. Study outcomes were post-acute discharge setting and unplanned 30-, 60-, and 90-day hospital readmissions. Logistic regression models were built to compare the performance of the 3 comorbidity indices using C statistics. The base model included patient demographics and hospital use. Subsequent models included 1 of the 3 comorbidity indices. Additional multivariable logistic regression models were built to identify individual comorbid conditions associated with high risk of hospital readmissions. Results: The 30-, 60-, and 90-day unplanned hospital readmission rates were 5.3%, 7.2%, and 8.5%, respectively. Patients were most frequently discharged to home health (46.3%), followed by skilled nursing facility (40.9%) and inpatient rehabilitation facility (12.7%). The C statistics for the base model in predicting post-acute discharge setting and 30-, 60-, and 90-day readmission in TKR and THR were between 0.63 and 0.67. Adding the Charlson Comorbidity Index, the Elixhauser Comorbidity Index, or HCC increased the C statistic minimally from the base model for predicting both discharge settings and hospital readmission. The health conditions most frequently associated with hospital readmission were diabetes mellitus, pulmonary disease, arrhythmias, and heart disease. Conclusion: The comorbidity indices and CMS-HCC demonstrated weak discriminatory ability to predict post-acute discharge settings and hospital readmission following joint replacement.

Original languageEnglish (US)
JournalArthritis Care and Research
DOIs
StateAccepted/In press - 2017

Fingerprint

Replacement Arthroplasties
Risk Adjustment
Patient Readmission
Comorbidity
Logistic Models
Knee Replacement Arthroplasties
Hip Replacement Arthroplasties
Medicare
Skilled Nursing Facilities
Medicaid
Health
Lung Diseases
Cardiac Arrhythmias
Inpatients
Heart Diseases
Diabetes Mellitus
Rehabilitation
Retrospective Studies
Demography
Outcome Assessment (Health Care)

ASJC Scopus subject areas

  • Rheumatology

Cite this

@article{390921e27cf943e49460b1281d3e142a,
title = "Current Risk Adjustment and Comorbidity Index Underperformance in Predicting Post-Acute Utilization and Hospital Readmissions After Joint Replacements: Implications for Comprehensive Care for Joint Replacement Model",
abstract = "Objective: To compare the performances of 3 comorbidity indices, the Charlson Comorbidity Index, the Elixhauser Comorbidity Index, and the Centers for Medicare & Medicaid Services (CMS) risk adjustment model, Hierarchical Condition Category (HCC), in predicting post-acute discharge settings and hospital readmission for patients after joint replacement. Methods: A retrospective study of Medicare beneficiaries with total knee replacement (TKR) or total hip replacement (THR) discharged from hospitals in 2009-2011 (n = 607,349) was performed. Study outcomes were post-acute discharge setting and unplanned 30-, 60-, and 90-day hospital readmissions. Logistic regression models were built to compare the performance of the 3 comorbidity indices using C statistics. The base model included patient demographics and hospital use. Subsequent models included 1 of the 3 comorbidity indices. Additional multivariable logistic regression models were built to identify individual comorbid conditions associated with high risk of hospital readmissions. Results: The 30-, 60-, and 90-day unplanned hospital readmission rates were 5.3{\%}, 7.2{\%}, and 8.5{\%}, respectively. Patients were most frequently discharged to home health (46.3{\%}), followed by skilled nursing facility (40.9{\%}) and inpatient rehabilitation facility (12.7{\%}). The C statistics for the base model in predicting post-acute discharge setting and 30-, 60-, and 90-day readmission in TKR and THR were between 0.63 and 0.67. Adding the Charlson Comorbidity Index, the Elixhauser Comorbidity Index, or HCC increased the C statistic minimally from the base model for predicting both discharge settings and hospital readmission. The health conditions most frequently associated with hospital readmission were diabetes mellitus, pulmonary disease, arrhythmias, and heart disease. Conclusion: The comorbidity indices and CMS-HCC demonstrated weak discriminatory ability to predict post-acute discharge settings and hospital readmission following joint replacement.",
author = "Amit Kumar and Amol Karmarkar and Brian Downer and Amit Vashist and Deepak Adhikari and {Al Snih al snih}, Soham and Kenneth Ottenbacher",
year = "2017",
doi = "10.1002/acr.23195",
language = "English (US)",
journal = "Arthritis and Rheumatology",
issn = "2326-5191",
publisher = "John Wiley and Sons Ltd",

}

TY - JOUR

T1 - Current Risk Adjustment and Comorbidity Index Underperformance in Predicting Post-Acute Utilization and Hospital Readmissions After Joint Replacements

T2 - Implications for Comprehensive Care for Joint Replacement Model

AU - Kumar, Amit

AU - Karmarkar, Amol

AU - Downer, Brian

AU - Vashist, Amit

AU - Adhikari, Deepak

AU - Al Snih al snih, Soham

AU - Ottenbacher, Kenneth

PY - 2017

Y1 - 2017

N2 - Objective: To compare the performances of 3 comorbidity indices, the Charlson Comorbidity Index, the Elixhauser Comorbidity Index, and the Centers for Medicare & Medicaid Services (CMS) risk adjustment model, Hierarchical Condition Category (HCC), in predicting post-acute discharge settings and hospital readmission for patients after joint replacement. Methods: A retrospective study of Medicare beneficiaries with total knee replacement (TKR) or total hip replacement (THR) discharged from hospitals in 2009-2011 (n = 607,349) was performed. Study outcomes were post-acute discharge setting and unplanned 30-, 60-, and 90-day hospital readmissions. Logistic regression models were built to compare the performance of the 3 comorbidity indices using C statistics. The base model included patient demographics and hospital use. Subsequent models included 1 of the 3 comorbidity indices. Additional multivariable logistic regression models were built to identify individual comorbid conditions associated with high risk of hospital readmissions. Results: The 30-, 60-, and 90-day unplanned hospital readmission rates were 5.3%, 7.2%, and 8.5%, respectively. Patients were most frequently discharged to home health (46.3%), followed by skilled nursing facility (40.9%) and inpatient rehabilitation facility (12.7%). The C statistics for the base model in predicting post-acute discharge setting and 30-, 60-, and 90-day readmission in TKR and THR were between 0.63 and 0.67. Adding the Charlson Comorbidity Index, the Elixhauser Comorbidity Index, or HCC increased the C statistic minimally from the base model for predicting both discharge settings and hospital readmission. The health conditions most frequently associated with hospital readmission were diabetes mellitus, pulmonary disease, arrhythmias, and heart disease. Conclusion: The comorbidity indices and CMS-HCC demonstrated weak discriminatory ability to predict post-acute discharge settings and hospital readmission following joint replacement.

AB - Objective: To compare the performances of 3 comorbidity indices, the Charlson Comorbidity Index, the Elixhauser Comorbidity Index, and the Centers for Medicare & Medicaid Services (CMS) risk adjustment model, Hierarchical Condition Category (HCC), in predicting post-acute discharge settings and hospital readmission for patients after joint replacement. Methods: A retrospective study of Medicare beneficiaries with total knee replacement (TKR) or total hip replacement (THR) discharged from hospitals in 2009-2011 (n = 607,349) was performed. Study outcomes were post-acute discharge setting and unplanned 30-, 60-, and 90-day hospital readmissions. Logistic regression models were built to compare the performance of the 3 comorbidity indices using C statistics. The base model included patient demographics and hospital use. Subsequent models included 1 of the 3 comorbidity indices. Additional multivariable logistic regression models were built to identify individual comorbid conditions associated with high risk of hospital readmissions. Results: The 30-, 60-, and 90-day unplanned hospital readmission rates were 5.3%, 7.2%, and 8.5%, respectively. Patients were most frequently discharged to home health (46.3%), followed by skilled nursing facility (40.9%) and inpatient rehabilitation facility (12.7%). The C statistics for the base model in predicting post-acute discharge setting and 30-, 60-, and 90-day readmission in TKR and THR were between 0.63 and 0.67. Adding the Charlson Comorbidity Index, the Elixhauser Comorbidity Index, or HCC increased the C statistic minimally from the base model for predicting both discharge settings and hospital readmission. The health conditions most frequently associated with hospital readmission were diabetes mellitus, pulmonary disease, arrhythmias, and heart disease. Conclusion: The comorbidity indices and CMS-HCC demonstrated weak discriminatory ability to predict post-acute discharge settings and hospital readmission following joint replacement.

UR - http://www.scopus.com/inward/record.url?scp=85030631401&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85030631401&partnerID=8YFLogxK

U2 - 10.1002/acr.23195

DO - 10.1002/acr.23195

M3 - Article

JO - Arthritis and Rheumatology

JF - Arthritis and Rheumatology

SN - 2326-5191

ER -