Objectives: To assess both the variability of interhospital trauma transfer practices and nonclinical factors associated with the transfer of injured patients from emergency departments (EDs) of non-tertiary care hospitals.
Methods: The authors analyzed a retrospective cohort of trauma patients initially presenting to one of 42 non-tertiary care hospitals (Level 3 or 4 hospitals) and requiring admission or transfer from January 1998 to December 2003. Twenty-one clinical, demographic, and hospital-level variables were included in multivariable logistic regression models (outcome = ED transfer to a tertiary care hospital), with hospital and year included as fixed effects to adjust for clustering. Classification and regression tree analysis was used to determine the importance of different covariates in predicting whether or not a patient was transferred from the ED.
Results: Included in the analysis were 10,176 persons, of whom 3,785 (37%) were transferred to a tertiary care hospital from the ED. The hospital of initial presentation was the factor of greatest importance in predicting transfer, and there was substantial variability in transfer practices between hospitals. Several additional nonclinical variables were independently associated with transfer, including type and level of hospital, patient age, increasing distance from the nearest higher-level hospital (a measure of geographic isolation), and the patient's insurance status (particularly among Level 3 hospitals).
Conclusions: The non-tertiary care hospital of initial presentation is the strongest predictor for whether an injured patient is transferred to a tertiary center from the ED. There is substantial variability in transfer practices between hospitals after accounting for important clinical factors, and several nonclinical variables are independently associated with transfer.