[Multimorbidity as a predictor for inpatient admission in clinical emergency and acute medicine : Single-center cluster analysis].

Abstract

Background: Parallel to demographic trends, an increase of multimorbid patients in emergency and acute medicine is prominent. To define easily applicable criteria for the necessity of inpatient admission, a hierarchical cluster analysis was performed.

Methods: In a retrospective, single-center study data of n = 35,249 emergency cases (01/2016-05/2018) were statistically analyzed. Multimorbidity (MM) was defined by at least five ICD-10-GM diagnoses resulting from treatment. A hierarchical cluster analysis was performed for those diagnoses initially summarized into 112 diagnosis subclusters to determine specific clusters of in- and outpatient cases.

Results: Hospital admission was determined in 81.2% of all ED patients (n = 28,633); 54.7% of inpatients (n = 15,652) and 0.97% of outpatient cases (n = 64) met the criteria for multimorbidity and the age difference between them was highly significant (68.7/60.8 years; p < 0.001). Using a hierarchical cluster analysis, 13 clusters with different diagnoses were identified for inpatient multimorbid patients (MP) and 7 clusters with primarily hematological malignancies for outpatient MP. The length of stay in the ED of inpatient MP was more than twice as long (max. 8.3 h) as for outpatient MP (max. 3.2 h.).

Conclusions: The combination of diagnoses typical for MM were characterized as clusters in this study. In contrast to single or combined single diagnoses, the statistically determined characterization of clusters allows for a significantly more accurate prediction of ED patients' disposition as well as for economic process allocation.