2023 14th Annual NUBE Conference

The Crisis Response Center (CRC) of Temple University sees the highest number of patients of any Psychiatric Emergency Service (PES) in the city of Philadelphia, PA. Due in part to the high rates of comorbid substance use and other clinical unknowns, some patients are placed into a 23-hour observation status before a final clinical decision is made. Our study evaluated the charts of observation patients over a 3-month period in 2019 (pre-COVID) to determine if there are any clinical factors associated with psychiatric admission at the conclusion of the observation period. Being able to better predict which patients will ultimately need psychiatric admission, we can eliminate the need for observation and get them to the appropriate level of care more promptly. Our research team developed a list of 41 clinical variables, from literature and chart reviews, that might have influence psychiatric admission rates. These variables include, but were not limited to, demographics, symptoms at presentation, commitment status, urine drug screen (UDS) results, and history from previous CRC presentations. After obtaining IRB approval from the Temple IRB, a total of 320 patient records were evaluated. Patient records were first anonymized. Each observation encounter was reviewed by study physicians looking at all 41 variables. Data was then entered into a REDcap database. Descriptive, univariate, and, multivariate analyses were used to determine if any variables were significantly associated with subsequent acute psychiatric hospitalization. Our analysis demonstrated that several variables appear to have a significant association with psychiatric admission versus referral to a lower level of care (inpatient drug and alcohol, outpatient, etc). Clinical features strongly linked with subsequent psychiatric admission after observation include the presence of delusions (OR-2.7), symptoms of mania (OR-5.04), an inability to perform a breathalyzer analysis (OR-3.11), and an involuntary commitment status (OR-2.31). Conversely, some variables were found to be associated with lower rates of admission. These include acute intoxication (OR-0.39), reported recent substance use (OR-0.43), and multiple previous discharges after CRC observation periods (OR-0.27). Interestingly, neither a positive breathalyzer nor any particular UDS result were found to have a significant effect on disposition. Reported synthetic cannabinoid use however, does appear to be linked with a lower likelihood of admission (OR-0.43).  This research is the first step to possibly developing a predictive clinical tool that may aid in preventing unnecessary observation in those patients that will require acute psychiatric admission.  This could help ensure that delay to appropriate treatment for these patients is reduced. Our future directions with this project are two-fold. The first is to evaluate the data to ascertain if there are any combinations of variables that are linked with a certain clinical outcome. Additionally, we plan to further investigate ways to add specificity to the variables (ex: the quality of delusions) that might be improve the accuracy of the data. A comparison of data from more recent months would also help us determine what, if any, effect COVID has had on post-observation dispositions.

Learning Objectives:

1. Discuss the use of psychiatric observation in Psychiatric Emergency Services prior to final disposition determination.
2. Determine if there are any variables in a patient's initial presentation that might help predict the likelihood of psychiatric hospitalization after the completion of an observation period.
3. Evaluate what role a predictive tool based on these variables may have in clinical practice.