Symptom burden to predict health care utilization in hospitalized patients with incurable cancer.

98 Background: Patients with incurable cancer are often hospitalized and have frequent readmissions after discharge. Considering the high physical and psychological symptom burden in this population, we sought to investigate symptoms as predictors of hospital length of stay (LOS) and time to first unplanned readmission.

METHODS: We consecutively enrolled incurable cancer patients with unplanned hospital admissions from 9/2014-4/2016. Within the first 5 days of admission, we assessed physical (Edmonton Symptom Assessment System [ESAS]; scored 0-10) and mood symptoms (Patient Health Questionnaire 4 [PHQ-4]; scored categorically). We created summated ESAS total and physical symptom variables. To identify predictors of LOS we used linear regression and for time to readmission we used Cox regression, with all models adjusted for age, sex, marital status, comorbidity, education, cancer type and time since incurable diagnosis.

RESULTS: We enrolled 1,000 of 1,227 (81%) eligible patients (mean age = 63.4; 50% female; 66% married). Gastrointestinal (33%) and lung (18%) cancers were the most common. Mean hospital LOS was 6.2 days and 30-day readmission rate was 25%. Over half of patients reported moderate/severe fatigue, drowsiness, lack of appetite, pain and poor well-being. Over one-fourth screened positive for PHQ depression and anxiety. All physical and mood symptoms individually predicted for longer LOS. Pain, nausea, poor well-being, ESAS total, ESAS physical and PHQ anxiety predicted for shorter time to readmission.

CONCLUSIONS: Hospitalized patients with incurable cancer experience a high symptom burden, which correlates with their health care utilization. Both physical and psychological symptoms predict for longer hospital LOS and shorter time to readmission. These findings can inform interventions targeting patients' symptoms during hospital admissions in an effort to improve health care delivery and utilization. [Table: see text].