Abstract 16486: Comorbid Illness Profiles Predict Inpatient Death in Older Adults with Heart Failure: an Analysis of 236,109 Heart Failure Admissions
Introduction: There is significant variability in inpatient mortality of older adults admitted for heart failure (HF). Given the prevalence of concomitant comorbidities in this population, we hypothesized that identifying common patterns of comorbid illness would be predictive of inpatient death.
Methods: We analyzed all HF hospitalizations of adults (≥65 years of age) from the 2009 Agency for Healthcare Research and Quality (AHRQ) National Inpatient Sample (n=236,109). Using latent-class mixture modeling, we identified 4 distinct profiles of comorbid conditions among 32 complementary AHRQ and Deyo-Charlson measures, and labeled these profiles according to dominant comorbidities. Multilevel logistic modeling was used to quantify inpatient mortality risk by profile, while adjusting for age, gender, race, median income by region, payer, weekend admission, hospital size and control, urban/rural location, and teaching status.
Results: Four comorbid profiles were identified (model entropy=0.68, LMRT=22,836, all p<0.0001). The largest profile (48.2%), “common variations,” contained patients with few and diffuse comorbidities. A “renal/endocrine” profile (28.4%) had the greatest percentages of patients with renal disease, anemia, electrolyte imbalances, complicated diabetes, and hypothyroidism; a “lifestyle/COPD” profile (17.8%) had the greatest percentages with uncomplicated diabetes, hypertension, chronic lung disease, and obesity; and a “vascular/ischemic/neuro” (5.6%) profile had the greatest percentages with cerebrovascular disease, paralysis, MI, peripheral vascular disease, and neurological disorders. In comparison to patients within the “common variations” profile, patients within the remaining profiles had significantly higher odds of inpatient mortality: “renal/endocrine” OR = 1.30 (95%CI=1.24-1.36), “lifestyle/COPD” OR = 1.10 (95%CI=1.03-1.17), and “vascular/ischemic/neuro” OR = 1.43 (95%CI=1.31-1.55) (all p<.01).
Conclusions: In older adults admitted for HF, comorbid illness profiles are predictive of inpatient mortality. Classifying patients by profile may help identify those who are at higher risk of inpatient death.
- © 2012 by American Heart Association, Inc.