Factors Associated with Hospital Admission Decisions in Hematologic Emergency Department Patients: A Hierarchical Modeling Study.

Background and

Objectives

To identify factors associated with hospital admission decisions among hematologic patients presenting to the emergency department (ED) and to evaluate the incremental contribution of clinical, laboratory, and process-related variables using a hierarchical modeling framework. Materials and

Methods

Single-center, retrospective observational cohort study conducted between January 2016 and December 2018. Adult patients for whom a hematology consultation was requested at ED presentation were included. Hierarchical block modeling was used as a sequential multivariable logistic regression approach in which clinical, laboratory, and process-related variables were entered in separate blocks. Process-related variables were defined as ED workflow-related factors, including mode of arrival, timing of presentation, and transfusion requirement. Model performance was assessed using the area under the receiver operating characteristic curve (AUC), Hosmer-Lemeshow goodness-of-fit test, calibration plot, calibration slope/intercept, and nested bootstrap internal validation.

Results

Of 428 patients included, 297 (69.4%) were hospitalized. In the updated final model, impaired mental status (aOR 2.682; 95% CI 1.393-5.161), ambulance arrival (aOR 2.607; 95% CI 1.532-4.434), body temperature (aOR 1.399 per °C; 95% CI 1.049-1.866), respiratory rate (aOR 1.105 per breath/min; 95% CI 1.019-1.197), and heart rate (aOR 1.023 per bpm; 95% CI 1.007-1.040) were independently associated with hospital admission. Transfusion requirement showed an inverse association with admission (aOR 0.397; 95% CI 0.244-0.644), whereas platelet count showed the same direction of association but did not reach statistical significance in the updated final model. The final model demonstrated moderate apparent discrimination (AUC 0.771; 95% CI 0.725-0.818), with no statistically significant lack of fit on the Hosmer-Lemeshow test (p = 0.188). After nested bootstrap internal validation, the optimism-corrected AUC was 0.729.

Conclusions

In hematologic patients presenting to the ED, admission decisions were associated mainly with acute physiological instability-particularly impaired mental status and abnormal vital signs-rather than by age or sex. These findings should be interpreted as exploratory associations with real-world disposition decisions, not as a validated tool for predicting objective outcomes such as mortality, ICU transfer, or 30-day adverse events.