Factors associated with psychiatric admission from observation beds in the psychiatric emergency room

Psychiatric emergency services frequently utilize observation units to support disposition decisions when diagnostic uncertainty exists, allowing for extended psychiatric supervision to inform medical decision-making. Yet there is limited data describing predictors of inpatient admission following observation, and understanding which factors are associated with this outcome may improve clinical decision-making and resource utilization.

We conducted a retrospective chart review of adult patients placed in an observation unit within a high-volume urban psychiatric emergency service between April 1 and June 30, 2019. Demographic, psychiatric, and substance-related variables were extracted from the electronic medical record. The primary outcome was disposition to acute inpatient psychiatric admission versus no psychiatric admission. Univariate analyses were performed, followed by multivariable logistic regression to identify factors independently associated with admission.

Among 269 patients placed in observation, 81 (30.1%) were admitted to an acute inpatient psychiatric unit and 188 (69.9%) were not admitted to an acute psychiatric inpatient unit. In multivariable analysis, factors associated with increased odds of psychiatric admission included delusions (aOR 2.94, 95% CI 1.46-5.94), depression (aOR 2.27, 95% CI 1.17-4.43), involuntary commitment status at presentation by either a civilian (aOR 3.65, 95% CI 1.43-9.37) or physician/law enforcement (aOR 2.07, 95% CI 1.10-3.90), and absence of a breathalyzer test (aOR 3.58, 95% CI 1.56-8.19). A decreased likelihood of psychiatric admission was associated with patients with two or more prior observation discharges (aOR 0.26, 95% CI 0.08-0.84), and self-reported substance use within the prior 72 hours (aOR 0.46, 95% CI 0.25-0.84).

In this psychiatric emergency cohort, delusions, depressive symptoms, involuntary commitment status at presentation, and absence of a completed breathalyzer were associated with increased odds of acute psychiatric admission, while self-reported substance use within the prior 72 hours and two or more prior observation discharges were associated with decreased odds. These findings should be interpreted as hypothesis-generating given the observational design, limited study window, and single-site sample. Validation in larger, multi-site, multi-season cohorts is needed before these associations can inform clinical decision-making or be incorporated into disposition prediction tools.