Full text of DOI:10.1080/10903120190939526

HOSPITAL FOLLOW-UP OF PATIENTS CATEGORIZED AS
NOT NEEDING AN AMBULANCE USING A SET OF
EMERGENCY MEDICAL TECHNICIAN PROTOCOLS
Terri A. Schmidt, MD, MS, Regina Atcheson, MD, Carol Federiuk, MD, PhD,
N. Clay Mann, PhD, MS, Tannert Pinney, MD, Dave Fuller, EMT-P, Kyran Colbry, BA
ABSTRACT
Emergency medical services (EMS) systems have
evolved to rapidly treat and transport seriously ill and
injured patients to the emergency department (ED).
Many current EMS systems transport all patients iden-
tified by 911 calls. Even in these systems, many people
for whom 911 was called are not transported to the
hospital via an ambulance. Some of these patients
refuse transport and others are directly, or indirectly,
encouraged by emergency medical technicians
(EMTs) not to use an ambulance. In 1996, only 17% of
EMS systems in large U.S. cities officially allowed
paramedic-initiated EMS refusal.1 There is recent evi-
dence that our EDs are becoming overburdened.2–4 As
EDs experience overcrowding, it may become neces-
sary to consider other resources and treatment options
for some patients.
There have been a number of proposed mechanisms
to allocate EMS resources by medical necessity. It may
be possible to design EMS systems to more closely
approximate patient needs by including programs
such as priority dispatch systems, non-911 access
numbers, alternatives to ambulance transport, and
medically appropriate treatment and release pro-
grams. It is important to ensure that all patients con-
sider themselves served by any system that is adopt-
ed.5 Further, before systems are redesigned, research
must show that these alternatives are safe. Albuquer-
que, New Mexico, attempted to implement a system of
paramedic triage but stopped the program because
protocols were not always followed and the para-
medics did not safely triage patients.6,7 Triage systems
allowing referral of patients with certain presenting
complaints to other resources have also been attempt-
ed in EDs.8–10
Objective. Using hospital outcomes, this study evaluated
emergency medical technicians’ (EMTs’) ability to safely
apply protocols to assign transport options. Methods.
Protocols were developed that categorized patients as: 1)
needs ambulance; 2) may go to emergency department (ED)
by alternative means; 3) contact primary care provider
(PCP); or 4) treat and release. After education on application
of the protocols, EMTs categorized patients at the scene
prior to transport but did not change current practice.
Hospital charts were reviewed to determine outcome of
patients whom EMTs categorized as not needing an ambu-
lance. Category 2 patients were assumed to need the ambu-
lance if they were admitted to a monitored bed or intensive
care unit. Category 3 and 4 patients were assumed to need
the ED if they were admitted. Results. The EMTs catego-
rized 1,300 study patients: 1,023 (79%) ambulance transport,
200 (15%) alternative means, 63 (5%) contact PCP, and 14
(1%) treat and release. Hospital data were obtained for 140
(51%) patients categorized as not needing ambulance trans-
port. Thirteen of 140 (9%) patients who transporting EMTs
determined did not need the ambulance were considered to
be undertriaged: five in category 2, six in category 3, and one
in category 4. Six of 13 (46%) undertriaged patients had
dementia or a psychiatric disorder as one of their presenting
complaints. Conclusion. These protocols led to a 9% under-
triage rate. Patients with psychiatric complaints and demen-
tia were at high risk for undertriage. Key words: EMS;
triage; guidelines; transport; emergency medical techni-
cians.
PREHOSPITAL EMERGENCY CARE 2001;5:366–370
Received October 5, 2000, from the Department of Emergency
Medicine, Oregon Health Sciences University (TAS, CF, NCM, KC),
Portland, Oregon; Kaiser Permanente (RA, TP), Portland, Oregon;
American Medical Response, Northwest (TAS, DF), Portland,
Oregon; and Intermountain Injury Control Research Center,
University of Utah, School of Medicine (NCM), Salt Lake City,
Utah. Revision received February 22, 2001; accepted for publication
February 22, 2001.
While a number of studies have evaluated the need
for transport in certain patients, no prospective guide-
lines for alternative transport or treatment have been
established.11–15 These criteria are needed in order to
design systems that efficiently use resources without
putting patients at risk.16
Presented at the Society for Academic Emergency Medicine
Western Regional Research Forum, Portland, Oregon, April 2000.
The goal of this study was to determine whether it is
feasible and safe to allow paramedics to offer alterna-
tives other than ambulance transport to an ED for
patients who access the 911 system. We previously
developed a set of triage protocols for EMTs and have
evaluated the safety of these protocols based on
Supported in part by Kaiser Permanente and American Medical
Response, Northwest.
Address correspondence and reprint requests to: Terri Schmidt,
MD, MS, Oregon Health Sciences University, Department of
Emergency Medicine, UHN-52, 3181 SW Sam Jackson Park Road,
Portland, OR 97201-3098. e-mail: <schmidtt@ohsu.edu>.
366

Schmidt et al. FOLLOW-UP OF PATIENTS NOT NEEDING AMBULANCE
367
adverse events in the ambulance.17 This study investi-
gated whether patients who were not classified in the
“needs ambulance” group by the EMTs had hospital
outcomes that would suggest that ambulance trans-
port was warranted.
METHODS
Protocol Development
Protocols were developed by an expert panel of para-
medics, EMTs, emergency physicians, and a county
EMS director. The protocols use presenting com-
plaints and paramedic assessments to recommend one
of four alternatives to patients: 1) ambulance transport
to an ED; 2) transport to an ED by alternative means;
3) referral to a primary care provider (PCP) within 24
hours; or 4) treatment at the scene only. After initial
development, the protocols were reviewed and
refined by members of the local EMS community
including a tricounty physician supervisors group and
a county system quality improvement group. The pro-
tocols are described in more detail in the previous
publication.17 Figure 1 shows a sample of the proto-
cols.
Study Design
This study was a prospective observational assess-
ment of scene categorization of patients to transporta-
tion alternatives and an associated retrospective hos-
pital chart review. During the study period, all
patients were treated according to the current proto-
cols, with no new alternatives being offered.
However, the EMTs were asked to categorize the
patient into the most appropriate defined triage classi-
fication and complete a data sheet.
Study Setting and Population
FIGURE 1. Sample protocol showing general guideline, followed by
an example of the protocol for a specific complaint.
The study was conducted in Clackamas County,
Oregon, with a population of 313,000 and a land mass
of more than 1,879 square miles. The county is one of
three counties that comprise the Portland metropoli-
tan area. It has one urban corridor and a large, rural
and wilderness area including Mt. Hood. First
response is provided by 12 fire districts and advanced
life support (ALS) transport is provided by American
Medical Response, a private ambulance company
under contract to the county. The fire department
units may be staffed with paramedics, EMT–interme-
diates, or EMT–basics. The ambulance is always
staffed with at least one paramedic. There is no tax
support for ambulance transport and patients are
billed for the service. For the fiscal year September
1998 through August 1999, there were 9,457 emer-
gency transports in the system.
arrived at the scene in Clackamas County from
August 1997 to mid-July 1998 were eligible for inclu-
sion. Interhospital transfers were excluded from the
study. Enrolled subjects represent a convenience sam-
ple, since not all of the fire departments in the county
chose to participate in the study and transporting
EMTs did not complete categorization forms for all
patients. However, we have little reason to believe
that any systematic exclusion of cases occurred based
on medical need or other characteristics. In the current
system, protocols do not allow EMTs to initiate a
refusal or offer alternatives to transport. This study
was approved by the Oregon Health Institutional
Review Board and the Tri-county EMS Scientific
Review Committee.
All patients for whom a 911-initiated ambulance

368
PREHOSPITAL EMERGENCY CARE OCTOBER/DECEMBER 2001 VOLUME 5 / NUMBER 4
TABLE 1. The Transporting Emergency Medical Technicians’ Triage Categorizations
Treatment and Transport-related Category
Total No. of Patients
Hospital Follow-up
Ambulance transport to an emergency department (ED)
Alternative transport to the ED
Contact primary care provider within 24 hours
Field treatment and release
1,023 (78.6%)
77
93
42
5
200 (15.4%)
63 (4.8%)
14 (1.2%)
of EMT categorization decisions. A “forced entry”
technique was used to enter contributory variables in
the logistic model. All analyses were conducted using
the SPSS for Windows (v 9.0) statistical program.18
Outcome Measures
Hospital outcome data were collected for patients in
categories II–IV, patients who were assessed as not
needing an ambulance, and category I patients who
were transported to two of the area hospitals. There
are 15 hospitals in the Portland area and patient infor-
mation was obtained from ten hospitals. One hospital
system, comprising several hospitals, was unwilling
to allow data collection. Patient outcome data were
not obtained from patients who were not transported
(n = 133). Hospital outcome data were entered into a
computerized Access database on a standardized
form by trained data abstractors. These data were
merged with data describing EMT categorization and
other prehospital information. Data included patient
vital signs, ED clinical impression, ED patient disposi-
tion (discharge, admission, monitored bed, and ICU),
surgeries and other procedures, receipt of blood prod-
ucts, discharge diagnoses for admitted patients, and
hospital death.
Using hospital outcome data, a criterion standard
was constructed indicating the need for ambulance
transport to the hospital. For patients categorized as
not needing an ambulance but requiring transport to
an ED (category II), it was assumed that any patient
who was admitted to a monitored bed or ICU,
required blood products in the ED, or required emer-
gent surgery was undertriaged. In addition, for any
patient categorized as not needing transport to an ED
(category III or IV), it was assumed that any patient
admitted to the hospital was undertriaged.
RESULTS
Demographic Information
From August 2, 1997, to July 19, 1998, there were 1,433
patients who were categorized by transporting EMTs
and for whom EMS transport information was avail-
able for review. A total of 133 (9.2%) patients refused
transport after EMT arrival and were eliminated from
further analysis. The mean age among the remaining
1,300 patients was 52.8 (±25.7) years, and 46% of the
patients were male. Based on the provided protocols,
the EMTs categorized 277 (21.3%) patients as did not
need ambulance transport and could seek treatment
either at an ED (n = 200) or from a PCP (n = 63), or
could be treated and released at the scene (n = 14)
(Table 1). Hospital outcome data were available for
140 (51%) patients considered not to need ambulance
transport. Hospital data were also collected for an
additional 77 patients considered by the transporting
EMTs to need ambulance transport. Thus, hospital
data were available for 217 patients, representing 15%
of the overall sample.
Comparison of EMT Triage
Categorizations and Hospital Outcomes
The EMTs’ triage categories were compared with hos-
pital outcome data using the prospectively developed
criterion standard. Thirteen of the 140 (9%) patients
who the transporting EMTs determined did not need
the ambulance were considered to be undertriaged
based on the criterion standard (Table 2). Of the
patients considered to need ED care but alternate
transport was advised (category 2 patients), four were
admitted to specialty care units (two admitted to the
ICU and two admitted to monitored beds). An addi-
tional patient required blood products in the ED and
emergent surgery. Among the patients considered by
the transporting EMTs not to require hospital care
(category 3 and 4 patients), three were admitted to
monitored beds and an additional five were admitted
to the hospital. No patient triaged as not needing an
ambulance died in the ED. Six of 13 (46%) under-
Statistical Analysis
Standard bivariate statistics [chi-square test, odds ratio
(OR), and unpaired t-test] were used to assess the per-
formance of EMT triage categorizations in relation to
actual hospital outcomes. Using hospital outcome as a
criterion standard, sensitivity and specificity rates
were calculated by collapsing triage categories into
“need ambulance” and “do not need ambulance.”
Subanalyses were also performed using multiple
logistic regression to determine the effect of other con-
tributing factors (i.e., patient age, gender, and present-
ing complaint) when regressing triage categorization
decisions on hospital outcomes. This analysis was con-
ducted to determine how patient demographics and
presenting signs and symptoms affected the accuracy

Schmidt et al. FOLLOW-UP OF PATIENTS NOT NEEDING AMBULANCE
369
triaged patients had dementia or a psychiatric disor-
der as one of their hospital diagnoses (Table 2).
explained by their age since age was not found to be a
factor in undertriage. Dementia and psychiatric disor-
ders may represent a group of patients who are more
difficult to assess due to their inability to provide a
clear history. The current protocols do not specifically
address dementia or mental illness. This may explain
why the logistic regression modeling did not find an
association between the specific protocols and under-
triage. It is possible that the undertriage rate could be
decreased by developing a protocol that addresses
behavioral complaints and dementia. The EMTs may
also need further education about assessment of this
high-risk group.
This study did not attempt to determine an accept-
able undertriage rate. Currently in most systems, a
number of patients are not transported who have
medical conditions that would warrant ambulance
transport. Prior to initiation of this study we evaluat-
ed patients who had refused ambulance transport
after 911 was called.19 In that study 6% were subse-
quently admitted to the hospital. In any system there
are some patients who should be transported but are
not. Emergency medical services systems and policy
makers will have to determine what is an acceptable
rate of undertriage.
Accuracy of the Transporting
EMTs’ Triage Categorizations
Nineteen of 77 patients whom the transporting EMTs
classified as requiring ambulance transport to the ED
(category 1 patients) had one of the criterion events,
for an overtriage rate of 400%. Thirteen of 140 patients
whom the EMTs classified as not requiring ambulance
transport had a criterion event (categories 2–4), repre-
senting an undertriage rate of 9%. When triage classi-
fications are collapsed to represent “need” and “not
need” ambulance transport, the participating EMTs
have a sensitivity of 0.25 (95% CI = 0.16, 0.36) and a
specificity of 0.90 (95% CI = 0.83, 0.94) based on hospi-
tal outcomes.
Multivariate logistic regression modeling was used
to determine whether selected patient characteristics
or protocol affected the precision of the EMTs’ trans-
port decisions. The model failed to identify any rela-
tionship between age, gender, or specific protocol and
an EMT’s ability to accurately classify patients based
on transport needs.
Based on the criterion standard, the developed pro-
tocols demonstrated a large overtriage rate. Many of
the patients who were classified as needing an ambu-
lance were not admitted to a monitored bed and did
not have any serious event in the ED. This is to be
expected since each protocol was designed to be con-
servative. It was assumed that it is safer to overtriage
than to undertriage patients. Other retrospective stud-
ies using various standards to define need for ambu-
lance transport have found that 11–61% of transports
may not be medically necessary.20–23
DISCUSSION
The EMTs used a set of protocols to assess patients
and determine whether those patients needed ambu-
lance transport to an ED. If the protocols had actually
been in place, the EMTs would have undertriaged 9%
of the patients. This undertriage occurred in two cate-
gories: patients who were determined not to need the
ambulance and then were admitted to a monitored
bed or an ICU and patients who were determined not
to need transport to an ED at all and then were admit-
ted to the hospital.
LIMITATIONS
We were unable to obtain hospital data from several
of the hospitals in the region, including some that
It is striking that nearly half of the patients who
were undertriaged had a psychiatric complaint or
dementia as part of their presentations. This cannot be
TABLE 2. Patients Requiring Ambulance Transport to the Hospital (Based on
Hospital Outcome Data) but Classified on Scene as Not Requiring Ambulance Transport
Emergency Medical Technician (EMT) Impression
Hospital Diagnosis
Critical Event
EMT Categorization
General medical
Hematemesis
Laceration, self-inflicted
Headache
Ethanol intoxication
Chest pain
Basic life support (BLS) trauma
Chest pain, rule out pneumonia
General medical
Fall
Hypercalcemia, anemia
Gastrointestinal bleed
Depression, laceration
Subarachnoid bleed
Suicidal ideation
Unstable angina
Fall, schizophrenia
Chest pain
Abdominal pain
Lumbar fracture
Dementia
Dementia, tibia fracture
Dementia, dehydration
Blood, surgery
intensive care unit (ICU), blood
Monitored bed
ICU
Monitored bed
Hospital admission
Hospital admission
Monitored bed
Monitored bed
Monitored bed
Category 2
Category 2
Category 2
Category 2
Category 2
Category 3
Category 3
Category 3
Category 3
Category 3
Category 3
Category 3
Category 4
Behavioral problem
Unkown medical
Dementia
Hospital admission
Hospital admission
Hospital admission

370
PREHOSPITAL EMERGENCY CARE OCTOBER/DECEMBER 2001 VOLUME 5 / NUMBER 4
Care. 1998;2:18-22.
received a large number of study patients. Thus, we
have hospital data for only about half of the patients
who were triaged as not needing an ambulance. We
do not know whether the results would differ if hos-
pital outcomes were available for the excluded
patients. One of the hospitals from which data were
obtained was a regional trauma center. This was one
of the hospitals with follow-up data for category 1
patients. Trauma system entry patients may be differ-
ent from other patients. However, the trauma system
patients would all have been categorized as needing
an ambulance, so this should not have affected
patients in the other three categories.
2. Derlet RW, Richards JR, Kravitz RL. Frequent overcrowding in
U.S. emergency departments. Acad Emerg Med. 2001;8:151-5.
3. Viccellio P. Emergency department overcrowding: an action
plan. Acad Emerg Med. 2001;8:185-7.
4. Henry MC. Overcrowding in America’s emergency depart-
ments: inpatient wards replace emergency care. Acad Emerg
Med. 2001;8:188-9.
5. Zachariah BS. The problem of ambulance misuse: whose prob-
lem is it, anyway? Acad Emerg Med. 1999;6:2-3.
6. Hauswald M, Brillman J, Raynovich W, Bralove D, Martinez A.
Training paramedics to determine who does not need ambu-
lance transport: validation of an educational program and pro-
tocol [abstract]. Acad Emerg Med. 1999;6:474.
7. Albuquerque halts EMS triage program. EMS Insider. 1999;Jul:
8.
Currently there is no accepted standard for who
needs an ambulance and who does not. The criteria
we used were based on the assumption that if a
patient was hospitalized, he or she should have gone
to the ED, and if a person needed monitoring, he or
she needed an ambulance. These assumptions are
open to debate. We previously reported an under-
triage rate of 3–10% using events in the ambulance as
the criterion standard for need for ambulance trans-
port. This study involved a subset of that same group
for whom hospital follow-up data were available.17
Thus, using either criterion, a group of patients who
the EMTs determined did not need an ambulance had
an event that suggested ambulance transport may be
warranted.
8. Washington DL, Stevens CD, Shekelle PG, et al. Safely directing
patients to appropriate levels of care: guideline driven triage in
the emergency service. Ann Emerg Med. 2000;36:15-22.
9. Shapiro MJ, O’Brien GM. Deferral out of the emergency depart-
ment: the wrong solution? [editorial]. Ann Emerg Med.
2000;36:71.
10. Derlet RW, Nishio D, Cole LM, Silva J. Triage of patients out of
the emergency department. Am J Emerg Med. 1992;10:195-9.
11. Cone DC, Kim DT, Davidson SJ. Patient-initiated refusals of
prehospital care: ambulance call report documentation, patient
outcome, and on-line medical control. Prehosp Disaster Med.
1995;10(1):3-9.
12. Zachariah BS, Bryan D, Pepe PE, Griffin M. Follow-up and out-
come of patients who decline or are denied transport by EMS.
Prehosp Disaster Med. 1992;7:359-71.
13. Sucov A, Verdile VP, Garettson D, Paris P. The outcome of
patients refusing prehospital transportation. Prehosp Disaster
Med. 1992;7:365-71.
14. Brown E, Sinclair J. The emergent problem of ambulance mis-
use. Ann Emerg Med. 1993;22:646-50.
15. Koenig KL. Unscheduled access to health care: re-engineering
the 911 system. Acad Emerg Med. 1996;3:989-91.
CONCLUSION
Although most people would agree that some patients
who are transported by ambulance do not need ambu-
lance transport, no criteria to prospectively determine
who can be triaged to other alternatives have been
established. We evaluated one set of protocols and
found that they produced an undertriage rate of near-
ly one in ten patients based on hospital outcome.
Patients with dementia and psychiatric disorders were
at particularly high risk to be undertriaged by this set
of protocols. Whether these protocols can be modified
to decrease that undertriage rate is yet to be deter-
mined. Emergency medical services systems also will
need to decide what is an acceptable undertriage rate.
16. Krohmer JR. Appropriate emergency medical transport. Acad
Emerg Med. 1999;6:5-7.
17. Schmidt TA, Atcheson R, Federiuk C, et al. Evaluation of proto-
cols allowing emergency medical technicians to determine need
for treatment and transport. Acad Emerg Med. 2000;7:663-9.
18. SPSS/PC + SPSS/PC v. 9.0: Statistical Package for the Social
Sciences. Chicago: SPSS Inc., 1998.
19. Schmidt TA, Mann NC, Federiuk CS, et al. Do patients refusing
transport remember descriptions of risks after initial ALS
assessment? Acad Emerg Med. 1998;5:796-801.
20. Billittier AJ IV, Moscati R, Janicke D, Lerner EB, Seymour J,
Olsson D. A multisite survey of factors contributing to medical-
ly unnecessary ambulance transports. Acad Emerg Med. 1996;3:
1046-52.
21. Morris DL, Cross AB. Is the emergency ambulance service
abused? Br Med J. 1980;281:121-3.
22. Richards JR, Ferral SJ. Inappropriate use of emergency medical
services transport: comparison of provider and patient perspec-
tives. Acad Emerg Med. 1999;6:14-20.
23. Camasso-Richardson K, Wilde JA, Petrack EM. Medically
unnecessary pediatric ambulance transports: a medical taxi
service? Acad Emerg Med. 1997;4:1137-41.
The authors are grateful to Denise Griffiths for database manage-
ment and analysis and Terri Davis and Jim Huie for data entry.
References
1. Jaslow D, Barbera JA, Johnson E, Moore W. EMS-initiated
refusal and alternative methods of transport. Prehosp Emerg