Abstract 15557: Comorbid Illness Profiles Predict Greater Length of Stay for Women: an Analysis of 93,242 Heart Failure Admissions

Abstract

Introduction: There is significant heterogeneity in length-of-stay (LOS) associated with hospitalizations for heart failure (HF). We hypothesized that comorbid illness profiles identified among women admitted for HF would be predictive of significant differences in LOS and explain some of this heterogeneity.

Methods: We analyzed all adult female hospitalizations for HF (primary diagnosis) from the 2009 Agency for Healthcare Research and Quality (AHRQ) National Inpatient Sample (n=93,242). Latent class mixture modeling was used to identify distinct profiles among 32 complementary AHRQ and Deyo-Charlson comorbidity measures, and profiles were labeled according to dominant characteristics. Negative binomial regression was used to quantify differences in inpatient LOS by profile adjusting for age, race, payer, weekend admission, and regional income, as well as hospital size, control, urban/rural location, and teaching status.

Results: Four distinct comorbid illness profiles were identified (model entropy = 0.72; LMRT=8,563, p<1x10-7). The largest profile, “common variations” (50.0%), included women with few and diffuse comorbidities. The “lifestyle/COPD” profile (19.8%) had a greater percentage of women with uncomplicated diabetes, hypertension, obesity and chronic pulmonary disorders than other profiles (all p<1x10-7). The “renal/endocrine” profile (25.8%) had a greater percentage of women with renal disease, complicated diabetes, fluid/electrolyte imbalances, hypothyroidism, and coagulopathy than other profiles (all p<1x10-7). The “vascular/ischemic” profile (4.4%) had a greater percentage of women with cerebrovascular disease, paralysis, myocardial infarction, peripheral vascular disease, neurological disorders, depression and dementia than other profiles (all p<1x10-7). Relative to patients fitting the “common variations” profile, patients within the latter profiles all had greater LOS: “lifestyle/COPD” = 0.63 days (13.4%) longer, “renal/endocrine” = 1.13 days (24.3%) longer, and “vascular/ischemic” = 1.26 days (27.7%) longer LOS (all p<1x10-7).

Conclusions: Recognizing comorbid illness profiles among women admitted for HF may be a helpful mechanism to identify those at higher risk for greater LOS.