The aim of this study was to identify predictors for an increased probability of admission of inpatient psychiatric emergency patients after presentation to an ED. To date, there has been no further development of clear SOPs although there have been few international studies on this issue since the 1980s, two of which were from Germany.

One reason may be the lack of data and inconsistent results of previous studies because of the lack of development and implementation of SOPs for psychiatric emergency patients.

The patients we studied largely matched those of other studies [1, 2, 8, 12, 15, 16, 19, 25,26,27] in terms of age, diagnostic spectrum, and admission rate.

Consistencies in predictors of inpatient psychiatric admission include prior psychiatric treatment [7, 11, 28], presence of legally initiated admissions against the patient's will [7, 21], and physician referral [7, 12, 21, 22]. The psychiatric emergency syndromes from the German guideline Emergency Psychiatry [9] are also predictive, supporting the validity of the syndromes.

In contrast to other studies, none of the ICD-10 diagnoses were associated with an increased probability of admission. Some of these studies found a diagnosis of schizophrenia, schizotypal and delusional disorders (F2), or mood [affective] disorders (F3) as predictors [11,12,13, 16]. Although neurotic, stress-related, and somatoform disorders (F4) account for a significant proportion of psychiatric patients in the ED, they appear to be associated with a decreased likelihood of inpatient admission. This has also been found by other studies [7, 12, 28]. In our study, suicidality was not a predictor of inpatient admission, although this has been found in most other studies. This may be due to the comparatively low documented frequency of suicide attempts or existing acute suicidality (3%) among patients in our study. The reason for this is probably the incomplete documentation of suicidality in the emergency department and, because of the study design with anonymization, the impossibility of combining different patient records from the emergency physician service, emergency department, and psychiatric hospital. Differences between our study and most other studies were also found with regard to the factors age and gender. In other studies, older age was a predictor [11, 14, 15, 19, 22]. Regarding gender, the results are inconsistent. In an Italian study, a higher probability was found for men [12], while in an American study, the probability was higher for women [19]. Factors that have been shown to be predictive in other studies, such as the presence of aggressiveness, apathy, psychotic perception, thought disorder [7, 11, 16, 28], or homelessness [19], were not examined.

A limitation of our study was the retrospective design. Therefore, maybe in some cases it was a challenge to assess the severity of clinical syndromes, which may have led to a lack of clarity in the distinction between the presence or non-presence of a psychiatric emergency. Due to anonymization of data collection, it was not possible to merge different data sources from prehospital emergency medicine, EDs, and psychiatric hospitals. Because of that, prehospital treatment in another clinic could not be excluded with certainty. Consequently, in some cases possible conclusions about patients with a first psychiatric illness could have been over- or underestimated. The results are data from a single ED and are therefore not representative of Germany.