Predictors of Readmission Among Elderly Survivors of Admission With Heart Failure

Harlan M. Krumholz, MD, Ya-Ting Chen, PhD, Yun Wang, MS, Viola Vaccarino, MD, PhD, Martha J. Radford, MD, Ralph I. Horwitz, MD, Section of Cardiovascular Medicine, Department of Medicine, the Section of Chronic Disease Epidemiology, School of Epidemiology and Public Health, and the Department of Medicine, Yale University School of Medicine, New Haven; Yale-New Haven Hospital Center for Outcomes Research and Evaluation; Qualidigm, Middletown, Conn.

Am Heart J. 2000;139(1) 

In This Article


Study Samples

The study sample of 2176 patients included 1129 patients in the derivation set and 1047 in the validation set. The mean age of the overall sample was 78.9 years, and 42% of the cohort was older than 80 years. The overall sample was 59% female and 89% white. The 6-month all-cause hospital readmission rate was 49%. The 6-month heart failure readmission rate was 23%. The 6-month mortality rate was 17%. A comparison of the derivation set and the validation set shows no substantial differences between the groups (Table I).


In the derivation cohort, 570 patients (50%) were rehospitalized within 6 months after discharge. Of these 570 rehospitalized patients, the majority (n = 287; 50%) were readmitted for heart failure. Other frequent reasons for rehospitalization included pneumonia, myocardial infarction, cardiac dysrhythmia, other ischemic heart disease, acute renal failure, dehydration, respiratory failure, and chest pain.

Table II shows the association between all-cause readmission and potential risk factors. Thirty-two patient and clinical factors were selected, based on clinical judgment and previous literature, as potential risk factors. Among these factors, the strongest bivariate associations with an increased risk of all-cause readmission within 6 months after discharge were prior heart failure, prior renal failure, diabetes, prior admission within 1 year, creatinine at discharge, and blood urea nitrogen level at discharge.

The results of the Cox regression model for all-cause readmission are shown in Table III. The significant predictors were prior admission within 1 year, prior heart failure (at least 1 prior heart failure admission or documented history of heart failure), diabetes, and creatinine level >2.5 mg/dL at discharge. Adding angiotensin-converting enzyme inhibitor and digoxin use at discharge to the model did not change the significance of these clinical predictors.

Table IV shows the 6-month all-cause readmission rate according to number of risk factors in the derivation and validation sets. A greater number of risk predictors was associated with higher risk of adverse outcomes (Cochran-Armitage trend test P < .0001). The readmission rates according to number of risk predictors were similar in the derivation and the validation sets. In the validation cohort, patients with none of the risk factors had a risk of all-cause readmission of 26%, whereas patients with 3 or all of the risk factors had an all-cause readmission rate of 59%. These risk predictors provide a stratification of risk for readmission from heart failure, the combined end point of all-cause readmission, or death.