Potentially avoidable hospitalizations in patients with advanced cancer.

Abstract

e18275 Background: Cancer patients and their clinicians often wish to avoid preventable hospital admissions, but efforts to understand the predictors of avoidable hospitalizations are lacking. We sought to examine reasons for hospital admissions in patients with advanced cancer, identify potentially avoidable hospitalizations (PAH), and explore predictors of PAH. Methods: We prospectively enrolled hospitalized patients with advanced cancer from 9/2014 - 11/2014 as part of a longitudinal data repository to define symptom burden in this population. Upon admission, we assessed patients’ symptom burden (Edmonton Symptom Assessment System [ESAS]; scored 0-10). We created a summated ESAS physical symptom variable. We used consensus-driven medical record review to identify the primary reason for each hospital admission and categorize it as PAH or not based on of an adaptation of Graham’s criteria for PAH. We used mixed multivariable logistic regression analyses to identify predictors of PAH. Results: We assessed 477 hospital admissions in 200 consecutively admitted patients (mean age = 64.6; 47% female; 67% married). Over half of admissions came through the emergency department (56%). The most common reasons for admissions were fever/infection (30%), symptoms (26%), and planned admission for chemotherapy or procedure (10%). We identified 149 (31%) as PAH. Among these PAH, 45 (30%) were readmissions due to failure of timely outpatient follow-up (within 7 days of discharge) and 44 (30%) were due to premature discharge from prior hospitalization. In a mixed logistic regression model, being married (odds ratio [OR] 0.48 [0.28-0.81]; p < 0.01) was associated with lower likelihood of PAH, while higher physical symptom burden (OR 1.02 [1.00-1.04]; p = 0.04) was associated with greater likelihood of PAH. Conclusions: We identified that a substantial proportion of hospitalizations in patients with advanced cancer are potentially avoidable, often related to failure of timely outpatient follow-up and premature hospital discharge. Our results demonstrate that patients’ symptom burden predicts PAH, thus underscoring the need to address patients’ symptoms in order to reduce preventable hospital admissions.