EMS/ORIGINAL CONTRIBUTION
Patient-Specific Predictors of Ambulance Use
From the Division of Emergency
Medicine, Department of Medicine,
Beth Israel Deaconess Medical
Center*; Harvard University School
of Public Health¢; the Division of
General Medicine, Department of
Medicine, Brigham and Women's
HospitaP; and Partners HealthCare
System", Boston, Massachusetts.
Receivedfor publicationApril 22, 1996.
Revision received October31, 1996.
Acceptedfor publication
November i2, 1996.
Abstract presented at the 5ociety for
Academic EmergencyMedicineAnnual
Meeting, San Antonio, May 1995.
Partially sponsored by the Risk
Management Foundation, Cambridge,
Massachusetts.
Copyright © by the American College
of Emergency Physicians.
Donald W Rucker, MD, MBA*
Roger A Edwards, ScB*"
Helen R Borstin, MD, MPH §*
Anne C O'Neil, MPH*
Troyen A Brennan, MD, JD, MPH*§
Study objective: To determine patient-specific socioeconomic
and health status characteristics for patients arriving by ambulance
at an emergency department.
Methods: Ambulance use among adult ED patients presenting
with abdominal pain, chest pain, head trauma, or shortness of
breath was studied at five urban teaching hospitals in the north-
eastern United States. Cross-sectional analysis within a prospec-
tive cohort study of 4,979 consecutive patients was performed
using an interval sequence subset of 2,315 patients (84% of those
eligible) to whom questionnaires were administered. Ambulance
use (21% of surveyed patients; 26% of all patients) was analyzed
with logistic regression.
Results: Predictors of ambulance use included age greater than
65 years (odds ratio [OR], 1.95; 95% confidence interval lOll, 1.34
to 2.82); clinical severity (OR,3.11; 95% CI, 2.27 to 4.25); poverty
(OR, 1.40; 95% CI, 1.08 to 1.83); physical function (OR, 1.05; 95%
CI, 1.02 to 1.09 for each point of worsening function on a 12-point
physical function scale); and various types of health insurance
coverage. Race, sex, education, Medicaid coverage, frequency of
E0 use, living arrangements, and primary physician availability
were not predictive in multivariate analysis of surveyed patients.
Conclusion: Ambulance use varies by age, clinical severity,
income, patient-specific characteristics of physical function, and
type of health insurance. Medicaid coverage and frequent ED use
are not predictive of increased ambulance use.
[Rucker DW, Edwards RA, Burstin HR, O'Neil AC, Brennan TA:
Patient-specific predictors of ambulance use. Ann Emerg Med
April 1997;29:484-491.]
4 8 4 ANNALS OF EMERGENCY MEDICINE 29:4 APRIL 1997