Objective: To derive and internally validate a simple prediction rule, using routinely collected prehospital patient data, that discriminates between hospital admission and emergency department (ED) discharge for adult patients who arrive by ambulance.
Methods: We performed a retrospective cohort study of consecutive adult nontrauma patients transported to two separate EDs over two months by a city-run emergency medical services (EMS) system. We tested whether specific prehospital variables could predict hospital admission using chi-square tests, logistic regression, and receiver-operating characteristic curves. We created a rule to predict the probabilities of hospital admission for individual patients.
Results: Of 401 patients, the mean age was 47 years; 60% were black and 32% were white; 51% were female; and 33% were admitted to an inpatient service after evaluation in the ED. Independent predictors of admission were dyspnea (adjusted odds ratio [OR] 6.8; awarded 3 points), chest pain (OR 5.2; 3 points), and dizziness, weakness, or syncope (OR 3.5; 2 points). Also predictive were age>or=60 years (OR 5.5; 3 points) and the prehospital identification of a history of diabetes (OR 1.9; 1 point) or cancer (OR 3.9; 2 points). Patients who had a score of 5 or higher had a greater than 69% chance of being admitted to an inpatient unit.
Conclusion: Routinely collected EMS patient information can help predict hospital admission for certain ED patients.