Document Type


Publication Date



Maine Medical Center, Medical Education, Maine Medical Center Research Institute

MeSH Headings

Humans, Patient Discharge, Hospitalization, Hospitals, Learning, Computer Systems


Purpose: Describe the population and hospitalization characteristics of patients served by the Preble Street Learning Collaborative (PSLC), a program aimed at serving unmet needs of the homeless population in Portland, Maine, who leave the hospital with self-directed discharge.

Background: At least 500,000 patients in the United States leave hospitalizations against medical advice each year. Leaving a hospitalization with self-directed discharge (against medical advice) is associated with higher readmission rates and higher mortality. Self-directed discharges are frequently associated with co-morbid homelessness and substance use disorders. However, little is understood about the PSLC patient and hospitalization characteristics leading to self-directed discharge. The aim of this study is to examine all hospitalizations of PSLC patients in 2018, describing the differences between patients with a self-directed discharge vs. those without. Specifically, we will examine the use of behavioral contracts, addiction medicine team engagement, co-morbid mental health diagnoses, substance use disorder, and discharge disposition.

Methods/Approach: We performed a retrospective chart review of patients who have had at least one interaction with a provider at PSLC and at least one hospitalization at Maine Medical Center between January 1, 2018 and December 31, 2018. We extracted demographic and specific hospitalization-level information from the patient’s medical record. After initially describing the patient population, we stratified data for both the patients and their admissions by discharge characteristics (self-directed vs. non-self-directed).

Results: The cohort included 249 patients with a total of 438 inpatient hospitalizations. Among patients, 35/249 (14.1%) had at least one self-directed discharge, with 13/295 (5.2%) having more than one. Patients with at least one self-directed discharge had a mean age of 43.1 (SD 10.9) years and 21/35 (60.0%) identified as male. History of substance use disorder and co-existing psychiatric disorder were both more frequent among patients with at least one self-directed discharge (30/35 (85.7%) and 29/35 (82.9%) respectively) compared with patients with no self-directed discharges (158/214 (73.8%) and 152/214 (71.0%), respectively). Overall, 60/438 (13.7%) PSLC patient hospitalizations resulted in self-directed discharge. The reason for self-directed discharge was documented for 39/60 (65.0%) of these admissions, and the most frequent reason was altercation with staff (13/39, 33.3%). Behavioral contracts were documented in 12/60 (20.0%) of admissions with self-directed discharge, compared with 39/374 (10.4%) of admissions without. Similarly, 21/60 (35.0%) admissions with self-directed discharge had addiction medicine team engagement during the hospitalization, compared with 82/375 (21.9%) admissions without self-directed discharge. Of hospitalizations ending in self-directed discharge, 31/60 (51.7%) resulted in hospital readmission within 30 days.

Conclusions: This study is the first to describe the characteristics of PSLC patients with self-directed discharge. Our findings that substance use and psychiatric disorders were highly prevalent among these patients are consistent with previous studies. The differing characteristics of patients with self-directed discharge compared to those without provide us with an understanding of specific interventions to help patients stay in the hospital until they are medically prepared for discharge. Further work is needed to improve providers’ documentation of reasons for self-directed discharge and to develop strategies to reduce the rate of self-directed discharge in this population.


2020 Costas T. Lambrew Research Retreat