Evaluating Risk for Psychiatric Re-Hospitalization a Recurrent Event History Analysis

Thesis (Ph.D.), Washington State UniversityBetween 1960 and 2010, the number of psychiatric treatment beds in the United States dropped ten-fold—from 300 to 30 beds per 100,000 persons. The vision for this planned deinstitutionalization was to replace substandard institutional care with community based treatment. In reality, the community care system has failed to meet the needs of many patients with serious mental illness. Approximately one in five patients hospitalized with schizophrenia, bipolar disorder or major depression disorder are re-hospitalized within 30 days. Furthermore, many high-risk patients in this population are recurrently hospitalized and may cycle between emergency departments, jail and homelessness. This study addresses two questions related to recurrent psychiatric hospitalizations: 1) how do risk factors for re-hospitalization vary between the initial and subsequent hospital episodes, and 2) is length of inpatient treatment related to a reduced risk of re-hospitalization in latter episodes? This retrospective, observational cohort study utilizes administrative records from nineteen community hospitals in Washington State to follow patients for up to two years post-discharge. The study cohort of 27,858 adult patients were first hospitalized between 2012 and 2015. A repeated-event Cox proportional hazard model was used to test for factors associated with the first, second and third psychiatric re-hospitalization. This analysis differs from existing studies on this topic in two important ways. First, patients in the study population have no psychiatric hospitalizations within the last three years. Second, risk for concurrent psychiatric re-hospitalizations are examined, as opposed to a single event. This longitudinal approach permits an assessment of the initiation and progression of re-hospitalization risk. The findings illustrate a set of demographic, clinical and financial factors that are strongly predictive of re-hospitalization following an initial psychiatric hospitalization (c=0.81). In latter episodes, results indicate that length of inpatient treatment has a small, but significant association with reduced re-hospitalization risk. Currently, there are no standardized, validated instruments in the United States to assess for psychiatric re-hospitalization risk. This research lays the groundwork for the early identification of high risk patients and eventual targeting of treatment resources to prevent the ongoing cycle of repeated psychiatric hospitalizations