End of Life Inpatient Care: An Opportunity for Improved Care Transitions
Background: End-of-life acute inpatient hospital care is common and costly. Between 32% and 38% of U.S. deaths occur in hospital settings. Costs and lengths of stay are much higher for hospital stays ending in death compared to patients discharged alive among patients served by all payers. Factors associated with high inpatient utilization at end of life may inform care transition decisions.
Methods: This retrospective cohort study included adult persons with terminal hospitalizations in non-federal hospitals in the state of Washington in 2012 (N=17,688). Group Based Trajectory Modeling (GBTM) was used to identify groups of patients with distinct trajectories of inpatient utilization. Modeling of each trajectory includes a review and selection of the best fit (linear, quadratic or cubic) by use of the maximum likelihood method and an assessment of the model Bayesian Information Criteria (BIC) score. The final number of trajectory groups is determined by evaluating successive models according to improvement in BIC score (2(ΔBIC) >2), average posterior probability of group assignment exceeding 70%, and a minimum group assignment including at least 5% of the study sample.
Results: GBTM yielded a 3 class solution, in which sixty-two percent (n=10,934) of the study population was classified in the low hospital days trajectory and was labeled as the ‘persistently low’ cohort. Twenty-two percent (n=3,906) of the study population was classified by initial low hospital day utilization and increasing hospitalization days in the last 6 months before terminal hospitalization; this cohort was labeled as ‘escalating’. Sixteen percent (n=2,848) of the study population was characterized by increasing hospital day utilization throughout the last 2 years of life and was labeled as the ‘persistently high’ cohort.
Bivariate cohort differences were observed in patient and clinical characteristics at terminal hospitalization. Younger terminal patients (< 64 years of age), the long term disabled (dual enrollees in Medicare and Medicaid), those on Medicaid only, and patients with a race other than White had higher than expected counts in the persistently high cohort. Patients in the persistently high cohort had the highest rates of comorbid heart failure, chronic lung disease and kidney failure while patients in the escalating cohort had the highest rates of comorbid metastatic cancer. Patients in both the escalating and persistently high cohorts were more likely to be hospitalized for infectious diseases at terminal hospitalization. Lengths of stay and estimated costs of the terminal hospitalization were similar across all three cohorts.
Implications: Few studies have examined health care utilization trajectories before death. Increases in healthcare utilization have been shown to be dominated by increases in hospital use. A contemporary examination of inpatient utilization at end-of-life among patients of all ages and all payers helps identify opportunities for changes in care, care transitions and opportunities for advance care planning and palliative care.