Predicting admission of patients by their presentation to the emergency department

Area of Science:

• Emergency Medicine
• Clinical Decision Making
• Health Services Research

Background:

• Predicting hospital admission from the emergency department (ED) is crucial for resource allocation and patient flow.
• Triage nurses play a key role in initial patient assessment and disposition decisions.

Purpose of the Study:

• To identify key factors predicting hospital admission for emergency department (ED) patients.
• To compare the predictive accuracy of patient characteristics and triage nurse assessments with logistic regression models.

Main Methods:

• Retrospective observational cohort study of 100,123 patients presenting to an urban tertiary referral hospital ED.
• Logistic regression models were developed using patient characteristics and blood test results to predict disposition.
• Model predictions were compared against triage nurses' predictions.

Main Results:

• Patient age, presenting symptoms, triage category, mode of arrival, prior referral, and triage timing significantly predicted disposition (P < 0.001).
• Ordering and abnormality of blood tests slightly increased the likelihood of admission.
• Triage nurses' admission predictions showed similar accuracy to models based on presentation characteristics alone.
• Including blood tests in the model improved accuracy by only 3%.

Conclusions:

• Patient presentation characteristics are significant predictors of hospital admission.
• Triage nurse prediction accuracy is comparable to models using patient characteristics.
• Blood test results offer only a marginal improvement in predicting patient disposition, advising against delaying admission decisions for these results.