Premium
Development, validation, and results of a risk‐standardized measure of hospital 30‐day mortality for patients with exacerbation of chronic obstructive pulmonary disease
Author(s) -
Lindenauer Peter K.,
Grosso Laura M.,
Wang Changqin,
Wang Yun,
Krishnan Jerry A.,
Lee Todd A.,
Au David H.,
Mularski Richard A.,
Bernheim Susannah M.,
Drye Elizabeth E.
Publication year - 2013
Publication title -
journal of hospital medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.128
H-Index - 65
eISSN - 1553-5606
pISSN - 1553-5592
DOI - 10.1002/jhm.2066
Subject(s) - medicine , copd , exacerbation , emergency medicine , mortality rate , cohort , receiver operating characteristic , decile , logistic regression , risk of mortality , hospital medicine , cohort study , intensive care medicine , mathematics , statistics
BACKGROUND Despite its large clinical and economic significance, measuring and improving the outcomes of patients hospitalized for chronic obstructive pulmonary disease (COPD) is only beginning to emerge as a national priority for policy makers and payers. OBJECTIVE To facilitate the public reporting of hospital outcomes, we developed a risk‐standardized measure of hospital 30‐day mortality for patients admitted with exacerbation of COPD. DESIGN Hierarchical logistic regression model. SETTING/PATIENTS Medicare Part A and Part B claims in a random sample of half of all admissions for patients admitted to acute care hospitals in 2008 (development cohort) and remaining 2008 admissions (validation cohort). We also assessed model performance and predictive ability in 2007 and 2009 data. MEASUREMENTS Hospital risk‐standardized 30‐day mortality rates. RESULTS The model development sample consisted of 150,035 admissions at 4537 nonfederal acute care US hospitals, with a mean unadjusted hospital 30‐day mortality rate of 8.6%. The mean risk‐standardized mortality rate was 8.6% and ranged from 5.9% to 13.5%. The development and validation models had good discrimination (areas under the receiver operating characteristic curve 0.72 and 0.72, respectively) and predictive ability (predicted mortality at the 1st and 10th deciles 1.5%, 23.7%, and 1.6%, 23.8%, respectively) and showed no evidence of over‐fitting. CONCLUSIONS A 30‐day mortality model based on administrative claims had similar discrimination to other public reporting models and can be used to compare risk‐adjusted outcomes for patients with exacerbations of COPD and to track changes in outcomes over time. The high mortality and variation in rates across institutions suggest opportunities to improve quality of care. Journal of Hospital Medicine 2013;8:428–435. © 2013 Society of Hospital Medicine