This study was conducted at the University of Utah Medical Center Emergency Department, the primary medical facility of the University of Utah Neuropsychiatric Institute and a unique research environment due to its expansive geographical catchment area, which includes Utah, Nevada, Wyoming, Idaho and western Colorado. The ED evaluates over 35,000 patients per year, and serves as the primary screening site for admission to the region’s largest psychiatric facility, which has 90 inpatient beds and approximately 3,000 inpatient admissions per year. The study was a retrospective chart review using the University of Utah Medical Center electronic medical record database, and was approved by the Institutional Review Board at the University of Utah on January 29, 2008.
The study group was comprised of all patients who presented between January and February 2007 with a psychiatric complaint for which an evaluation by a licensed clinical social worker (LCSW) was requested by an attending emergency physician (EP). The historical control population consisted of 300 randomly selected patients who presented to the ED with a non-psychiatric complaint during the months of January and February 2007. This number of control patients was selected based on power calculation to detect a 30% difference in admission rates and return visits between groups, assuming 80% power and alpha of 0.05. Historical controls were selected from the ED patient log beginning January 1, which was the first day of the study period, and were selected from across the study period using a random number generator. Those who were evaluated by a LCSW at any point during their stay were excluded from the control population. Psychiatric admission was defined as admission to an inpatient psychiatric unit. Medical students using a template form performed chart reviews. Investigators entered data into a standardized database. Twenty percent of the study charts were reviewed by one of the study’s primary investigators.
A significant source of information about the study population came from notes written by LCSWs. To more thoroughly evaluate psychiatric patients presenting to the ED, LCSWs complete a crisis note detailing their assessment and recommendations for admission versus discharge following the initial evaluation by an EP. Reason for evaluation by a LCSW included suicidal ideation, suicide attempt, psychosis, substance abuse, or any other psychiatric complaint for which the attending EP requested an evaluation. Detailed crisis notes follow a template format and include patient age, gender, presentation, history of suicide attempts, psychiatric history, living situation, and current sources of stress in the patient’s life. All patient-disposition decisions (admission vs. discharge) are made by the LCSW in discussion with the attending EP. LCSWs in the ED follow up on discharged patients through hospital records and community psychiatric facility records.
The control group consisted of patients who presented to the ED with a complaint of a non-psychiatric nature in the time period concurrent with the study group. These complaints represented the full spectrum of potential ED visits including trauma, abdominal pain, chest pain, infection, etc. Patients were excluded from the control group if they had been evaluated by a LCSW for any reason during their stay, as the study group consisted of patients who had received an LCSW evaluation. These patients were found in the electronic database and selected randomly from the months of January through February 2007. We followed all patients included in the study for 30 days for return ED visit and hospitalization upon return visit. In the case of patients with multiple ED visits during the study period, the initial visit during this period was defined as the index visit, and additional visits were defined as repeat visits.
The admission rate, rate of return to ED within 30 days, and admission rate on return visits were compared for the two study groups. We performed statistical analysis using chi-square and Student’s t-test with p<0.05 considered statistically significant (SPSS v. 16.0).