PARAMEDIC ACCURACY USING SALT TRIAGE AFTER A BRIEF INITIAL TRAINING
Matthew R. Deluhery, MD, E. Brooke Lerner, PhD, Ronald G. Pirrallo, MD, MHSA,
Richard B. Schwartz, MD
ABSTRACT
Objective. To determine paramedics’ understanding of and
accuracy using SALT (sort–assess–lifesaving interventions–
treatment/transport) triage, a proposed national guideline
for primary triage during mass-casualty incidents, immedi-
ately and four months after training. Methods. A 20-minute
lecture on SALT triage was provided to all paramedics
(n = 320) from a single county during mandatory contin-
uing education. Triage concepts were reemphasized during
a 10-minute small-group lecture throughout the study pe-
riod as part of standard refresher training. After the initial
training, all paramedics were asked to complete a posttest
consisting of three general knowledge questions about SALT
triage and 10 patient scenarios in which they had to assign a
triage category. The same test was administered four months
after the original educational session. Demographic and job
experience information was also obtained. Responses were
scored and matched for each paramedic and compared using
paired t-test. Results. A total of 290 (91%) paramedics com-
pleted the initial posttest. They correctly answered an aver-
age (± standard deviation) of 10.7 ± 2.3 of the 13 questions
(82%). For the 10 patient scenarios, they correctly triaged
an average of 8.1 ± 2.0 patients. A total of 159 paramedics
completed both tests. Sixty-seven percent had more than 10
years of emergency medical services (EMS) experience; 72%
had prior mass-casualty drill experience; 51% had prior ac-
tual mass-casualty experience; and 23% had heard of SALT
triage prior to the training. There were no statistically signif-
icant differences in initial test scores for any of these demo-
graphic groups. For those subjects who completed both tests,
the mean overall score for the initial test was 10.9 ± 1.9 (84%)
and for the later test was 11.0 ± 1.9 (85%) (p < 0.770; 95% con-
fidence interval [CI] –0.3 to 0.3). For the 10 patient scenarios,
Received January 12, 2011, from the Department of Emergency
Medicine (MRD, EBL, RGP), Medical College of Wisconsin, Milwau-
kee, Wisconsin; and the Department of Emergency Medicine (RBS),
Medical College of Georgia, Augusta, Georgia. Revision received
February 23, 2011; accepted for publication February 24, 2011.
Presented at the Society for Academic Emergency Medicine annual
meeting, Phoenix, Arizona, June 2010.
The authors thank the Medical Directors and all the Paramedics of
Milwaukee County EMS System for participating in this project.
The authors report no conflicts of interest. The authors alone are re-
sponsible for the content and writing of the paper.
Address correspondence and reprint requests to: E. Brooke Lerner,
PhD, Department of Emergency Medicine, Medical College of Wis-
consin, 9200 West Wisconsin Avenue, Milwaukee, WI 53226. e-mail:
eblerner@mcw.edu
doi: 10.3109/10903127.2011.569852
526
the paramedics correctly triaged an average of 8.3 ± 1.7 pa-
tients on the initial test and 8.3 ±1.4 patients on the later test
(p < 0.565; 95% CI –0.4 to 0.2). Conclusion. Following a short
didactic course, paramedics were able to accurately perform
SALT triage during a written scenario. Four months after the
training, they had retained their understanding of and accu-
racy using SALT triage. It appears that a brief educational
tool was effective for training EMS providers in SALT triage.
Key words: disaster; triage; SALT triage; emergency medi-
cal services
PREHOSPITAL EMERGENCY CARE 2011;15:526–532
INTRODUCTION
During a mass-casualty incident (MCI), emergency
medical services (EMS) providers must rapidly assess
and prioritize patients for treatment and transport.
Mass-casualty triage systems are intended to assist
EMS providers with this process. Many different
mass-casualty triage systems have been created.1,2 In
the United States no single national standard mass-
casualty triage system has been established. This may
limit interoperability when an incident requires that
multiple EMS agencies respond.
In 2006, the Centers for Disease Control and
Prevention–funded Terrorism Injuries Information,
Dissemination and Exchange Project sponsored a
working group to review the scientific literature on
mass-casualty triage systems and recommend a mass-
casualty triage system that could be adopted as the
U.S. national standard. This working group found
that sufficient literature does not exist to support
any of the currently available triage systems as a
national standard.2 To address this gap, the work-
ing group selected the best practices from existing
systems and created a proposed national standard
called SALT (sort–assess–lifesaving interventions–
treatment/transport) triage (Fig. 1). Limited research
has been conducted to identify the most effective
method to learn SALT triage and assess an EMS
provider’s ability to use SALT triage.3,4
A systemwide protocol change provided a unique
opportunity to assess the introduction of SALT triage
to a large, urban EMS agency. The three specific goals
of this study were to measure EMS providers’ un-
derstanding of SALT triage after an initial 20-minute
lecture, evaluate their ability to accurately triage hy-
pothetical victims of an MCI, and then reassess their
knowledge of SALT triage and triage accuracy four
months later.
Deluhery et al. SALT TRIAGE ACCURACY AFTER TRAINING
FIGURE 1. SALT triage algorithm. LSI = lifesaving interventions.
METHODS
This prospective observational study was conducted
during two regularly scheduled Milwaukee County
EMS paramedic continuing education classes. Mil-
waukee County EMS employs dual-trained firefighter
paramedics in an urban fire-based system serving a
population of 960,000. At the time of the study, the Mil-
waukee County EMS system included 320 paramedics.
This study was considered exempt from review by the
institutional review board at the Medical College of
Wisconsin.
In 2009, Milwaukee County EMS adopted SALT
triage as its MCI triage method, replacing START (sim-
ple triage and rapid treatment) triage. As part of the
protocol rollout, training in SALT triage was provided
as two consecutive mandatory continuing education
classes in February 2009. A 20-minute, 34-slide Pow-
erPoint (Microsoft Corp., Redmond, WA) lecture was
given by the same presenter (EBL). The lecture in-
cluded historical information on triage, the origins of
SALT triage, a description of the SALT triage protocol,
and a review of the process using a fictitious shooting
at a local sporting event.
After the educational class ended, when convenient,
and on their own time, participants were asked to com-
plete a knowledge test of SALT triage and survey at the
online survey site www.surveymonkey.com. The sur-
vey gathered respondents’ demographic information
and their prior real-life experience with patient triage.
The test was scenario based, involving assigning cor-
527
rect triage categories to hypothetical victims of a mass-
casualty event (Fig. 2). Three questions on basic SALT
triage concepts and 10 patient scenarios were included
on the test. Only a single answer was allowed per ques-
tion. No time restrictions were placed on the comple-
tion of the examination. The survey Website was closed
four weeks after the training.
After the initial training, paramedics also attended
additional continuing education (“refresher training”)
sessions held between March and May 2009. These
sessions were small-group interactions and covered a
wide range of state-required topics. During these re-
fresher sessions, a 10-minute review of SALT triage (20
slides with written narrative) was incorporated. A sec-
ond online test using questions and methodology iden-
tical to those of the first test was administered to the
participants in June 2009, approximately four months
after the original lecture.
The correct triage designation was determined by
comparing paramedic responses to the test creators’
intended triage category. One of the 10 questions
(patient scenario 5) was later determined to be am-
biguous, so two different triage categories were
accepted as the correct answer. The participants’
first and second tests were matched using a unique
paramedic identifying number. Several instances
occurred in which the same paramedic number was
used on more than one set of examinations. If there
were multiple tests started under the same paramedic
number, the test that had the most questions com-
pleted was used for the analysis. If multiple tests were
528 PREHOSPITAL EMERGENCY CARE OCTOBER/DECEMBER 2011 VOLUME 15 / NUMBER 4

equally completed, the one completed at the latest
date was used. Paramedics who completed only the
second examination were excluded.
Data were extracted from the survey site into Ex-
cel (Microsoft Corp.) and analyzed using SPSS (SPSS
Inc., Chicago, IL). The initial posttest and four-month
posttest results were compared for each paramedic.
Descriptive analysis and paired t-test were used to an-
alyze data.
RESULTS
A total of 290 of the 320 eligible paramedics (91%) com-
pleted the initial survey and 193 (60%) completed the
second survey (Fig. 3). Of the participants who com-
pleted the initial test, the average score was 82%, a
mean (± standard deviation) of 10.7 ± 2.3 correct an-
swers out of 13 total questions. For the 10 patient sce-
narios, the paramedics correctly triaged an average of
8.1 ± 2.0 patients. For those participants who com-
pleted both tests (n = 159), there was not a statisti-
cally significant difference in scores immediately after
the training compared with four months after the train-
ing (Table 1). The participants who completed the first
test but not the second test had similar initial scores to
those who completed both tests (10.8 ± 2.0 compared
with 10.9 ± 1.9).
In general, the participants had extensive experience
in EMS. Approximately half of the participants had
greater than 15 years of EMS experience, while 17%
had seven years or less. Experience in EMS did not ap-
pear to affect test scores (Table 2).
Seventy-two percent of the paramedics completing
both surveys had previously participated in one or
more mass-casualty drills. The mean number of drills
was four. Just over half of the respondents (51%) had
previously responded to an actual MCI. The mean
number of events was two. There was no statistical
difference in mean scores on the initial test based on

Question 1: You are one of the first medical responders to arrive at a large-scale mass casualty
event and are asked by the incident commander to initially triage patients using SALT triage, what
is the first step in triage?

286
(99%) Announce loudly for all patients who can to walk to a set point
1 (1%) Start at the farthest away patient and work your way to a set point
3 (1%) Asses all patients that appear to have serious injuries

Question 2: You are one of the first medical responders to arrive at a large-scale mass casualty
event and are asked by the incident commander to perform an individual assessment, what
lifesaving intervention could be done on a young adult who is unresponsive, with agonal breathing,
obvious jugular venous distention (JVD) and a trachea that appears to deviate towards the left
(assume you have the proper training and equipment on hand)?

3 (1%) CPR, intubation, and bag-valve mask until assistance arrives

210
(72%) Needle decompression
77 (24%) When using SALT triage you do not stop to perform lifesaving interventions

Question 3: Which of the following patients should be the first to be individually assessed during
as mass casualty incident?

An elderly man who is able to walk but has an obvious deformity to the right upper
1 (1%) extremity
A child who is unable to walk but is screaming for his mother. There are no
4 (1%) immediately obvious injuries, but the child is screaming loudly and sounds injured
285 A young man who is breathing but unresponsive and has a large, swollen left
(98%) lower extremity

Dead Expectant Immediate Delayed Minimal

Patient 1: Doesn’t wave or move
when instructed. 55 year old male,
95% total body surface area burns
with inhalation injury. There is no
267 (92%) 20 (7%) 3 (1%) __ __
radial pulse present, and no
respirations.

FIGURE 2. Initial survey results. Boldfaced answers are the authors’ intended correct responses. CPR = cardiopulmonary resuscitation; SALT =
sort–assess–lifesaving interventions–treatment/transport. ∗ The scenario for patient 5 was found to be ambiguous, so two answers were accepted
as correct.
Deluhery et al. SALT TRIAGE ACCURACY AFTER TRAINING 529

Patient 2: Doesn’t wave or move

when instructed. 36 year old female
with a penetrating shrapnel wound to
the head that is through and through.
The patient is unresponsive, has 69 (24%) 218 (75%) 2 (1%) __ 1 (1%)
shallow respirations, approximately
2/minute, and no palpable radial
pulse.

Patient 3: Doesn’t wave or move

when instructed. 82 year old male,
very large pool of blood surrounding
the patient. Left lower extremity
amputation above the knee actively
hemorrhaging. Unresponsive, weak 3 (1%) 238 (82%) 49 (17%) __ __
carotid pulse, occasional shallow
respirations. A tourniquet has been
applied but it does not appear to be
controlling the bleeding.

Patient 4: Doesn’t wave or move

when instructed. 20 year old male
with massive chest and abdomen
wounds. Unresponsive, has no 286 (98%) 3 (1%) 1 (1%) __ __
respirations, and no palpable radial
pulse.

Patient 5: Doesn’t move when

instructed, is able to wave. 18 year
old male with left leg injury. Injury
is spurting blood (arterial bleeding).
__ 2 (1%) 140 (48%) 145 (50%) 3 (1%)
Pulse is 110, respirations 20. A
tourniquet has been placed and the
bleeding is controlled.*

Patient 6: Able to wave and move

when instructed. 24 year old male.
Penetrating injuries with avulsion to
upper arm and uncontrolled arterial 1 (1%) 6 (2%) 269 (93%) 14 (5%) __
bleeding that cannot be controlled.
Oriented to person and place but
does not know what day it is. The

FIGURE 2. Continued.

drill or actual mass-casualty experience (drill p < 0.79;
actual p < 0.86).
Thirty-seven (23%) of the participants reported hav-
ing previously heard of SALT triage. However, test
scores were not significantly different when a respon-
dent had prior knowledge of SALT triage (p < 0.57).
The number of correct answers for each question was
relatively constant between the first and second tests,
with the exception of patient scenario 3. In this ques-
tion, 82% correctly triaged the patient as “expectant”
on the first test and 45% on the second test, with the
majority changing their answer to “immediate.”
DISCUSSION
The EMS providers in our study accurately applied
SALT triage on an online written test, averaging over
80% on both the initial and follow-up tests. Although
this is a high rate of correct triage, there is no standard
for the expected rate of accurate triage. Prior studies
of SALT triage have found accuracy rates of 78%3 to
81%.4 Studies of other triage systems have found accu-
racy rates of 48% to 85%.2,5,6 Of note, our study popu-
lation was composed of experienced EMS personnel,
with the majority having more than 15 years of ex-
perience. This may have contributed to the high test
scores.
The four-month knowledge retention was high in
our study group. The test scores were essentially
equivalent on the first and second examinations. This
demonstrates that a 20-minute training and 10-minute
refresher may be sufficient to teach SALT triage to
a highly experienced group of EMS providers. This
is similar to the findings of previous studies using
530 PREHOSPITAL EMERGENCY CARE OCTOBER/DECEMBER 2011 VOLUME 15 / NUMBER 4

airway is clear. Respirations are

rapid and labored. The pulse is
weak. He states he is thirsty and
needs some water.

Patient 7: Able to wave and move

when instructed. 36 year old female.
Complains of chest pain with
palpitations and difficulty breathing.
No obvious trauma and states she
was not near the crash site but has a __ __ 71 (24%) 4 (1%)
215 (74%)
weak heart. Alert and oriented to 3
questions. Clearly in respiratory
distress and has a weak radial pulses
that are too fast to count.

Patient 8: Doesn’t move when

instructed, is able to wave. 24 year
old male with amputated fingers and
a small laceration to the forehead.
Bleeding is controlled by the patient.
Alert and oriented to 3 questions but __ __ 9 (3%) 55 (19%)
226 (78%)
unable to walk because he is too
dizzy. Respirations are normal but
there is an odor of alcohol. Pulses
are normal.

Patient 9: Able to walk when

instructed. 22 year old female.
Complains of ringing in her ears and
an inability to hear clearly. She has a __ 1 (1%) __ 19 (7%)
270 (93%)
strong regular pulse and normal
respirations.

Patient 10: Able to walk when

instructed. 30 year old female.
Complains of a headache and pain in
her left forearm, but no deformity.
She is very concerned about her
__ __ __ 15 (5%)
friend who is unresponsive. She is 275 (95%)
crying and yelling loudly that you
should help her friend. She has a
strong regular pulse and normal
respirations.

FIGURE 2. Continued.

other mass-casualty triage systems, which have shown
improvement in triage rates after a one-hour didactic
session,7 after a two-hour educational session,8 after
an approximately 30-minute Internet training session,9
and after being given decision-making materials.10
Of note, a previous study showed that providers’
comfort with mass-casualty triage increases when
they participate in a drill after a didactic training
session.4 While drills likely improve performance,
they are expensive and difficult to coordinate for large
groups. In our population, 72% of the respondents
had previously participated in a mass-casualty drill.
It is unclear from our study what effect prior MCI
drill participation may have had on provider comfort

TABLE 1. Overall Study Results for Those Participants (n = 159) Who Completed Both Tests
First Test Second Test
Number of Correct Standard Number of Standard Difference in Confidence
Answers Deviation Correct Answers Deviation Score p-Value Interval

Overall questions out of 13 10.9 1.9 11.0 1.9 0.1 <0.770 −0.3 to 0.3
Patient scenarios out of 10 8.3 1.7 8.3 1.4 0.0 <0.565 −0.4 to 0.2
Deluhery et al. SALT TRIAGE ACCURACY AFTER TRAINING 531

as immediate. We have no way to know what caused

320 eligible participants
this change, as the rest of the answers were basically
consistent between the two examinations. The change
could have been related to our inability to dictate the
290 completed first test 193 completed second test available scene resources that would further assist
with their patient care decision making.
Many communities are concerned about the costs
associated with training EMS providers to use a new
167 completed both tests triage system. For this study a simple 20-minute Pow-
erPoint lecture was chosen to train paramedics, for sev-
eral reasons: first, to minimize cost, as repetitive didac-
8 left at least one test
completely blank
tic sessions can become costly; and second, to ensure
reproducibility, as a large EMS system requires mul-
tiple days of training to reach all of the providers. A
159 used for analysis narrated PowerPoint didactic training session is easily
repeated and reproducible on DVD for an individual
FIGURE 3. The number of eligible participants who participated in who was unable to attend the face-to-face sessions.
each stage of the study.

LIMITATIONS AND FUTURE RESEARCH

and confidence. However, based on our findings, it
did not have a statistically significant effect on test
score.
Originally, the study population had been trained
to use START triage. One of the differences between
the two triage systems is the use of an “expectant”
category. This category is assigned to patients who
have injuries that are unlikely to be survivable given
available resources. While for some clinical situations
a casualty is likely to be expectant regardless of the
resources (e.g., an 85-year-old woman with burns
over 95% of her body surface area), in other scenarios
expectant would be considered with certain resource
constraints (e.g., a 20-year-old man with blunt trau-
matic brain injury and agonal breathing). In a written
scenario it is difficult to represent the dynamic nature
of these conditions. Additionally, the survey scenarios
allowed the respondent to self-interpret the available
scene resources that may influence patient care deci-
sions. The third patient scenario involved an elderly
man with lower-extremity amputation and arterial
bleeding. The authors selected expectant as the correct
answer, and on the first survey 82% of the paramedics
got this question correct. However, on the second
survey, fewer than half of the paramedics selected
expectant, with many respondents triaging this patient
TABLE 2. Mean Number of Correct Answers Based on
Years Involved in Emergency Medical Services
First Test Second Test
Correct Standard Correct
n Answers Deviation Answers
<1–7 years 27 10.9 1.57
8–10 years 26 11.3 1.67
11–15 years 30 10.8 2.18
16–20 years 31 10.6 1.62
>20 years 45 10.9 2.04
This study has several limitations. Our study was con-
ducted on paramedics from one county EMS system
and may not be representative of all paramedics. Fur-
ther, while our initial response rate was high, we had
a high rate of loss to follow-up for completion of the
second test. Potential selection and response bias may
have occurred, but it is not possible to know the effect
size. Reassuringly, the initial test scores were similar in
the two groups.
Our study included only paramedic-level EMS
providers from an urban fire-based EMS system. We
did not study other levels of EMS professionals or
other geographic locations. Additionally, our partici-
pants had a very high level of EMS experience that
may limit the ability to generalize these findings to
other EMS systems. All testing was completed online
without a proctor. Some respondents completed the
survey during work hours and there was no way to
determine whether group test taking or sharing of an-
swers occurred. This online testing method is used
commonly in our EMS system, and providers were
told that their individual scores would not be provided
to the medical director or other members of the fire de-
partment. We believe that there was no incentive to use
any outside resources to complete the test.
Regrettably, after the testing was started, we noted a
typo in question 2. The answer listed “needle compres-
sion” instead of “needle decompression.” We chose not
to correct the error since we felt that any change would
confound our results. Interestingly, when the results
Standard were analyzed, it appears that the typo did not con-
Deviation fuse respondents, as the majority of test takers still an-
11.1 1.29
swered the question correctly.
11.2 1.30 The effect of the 10-minute refresher training that
10.9 1.66 occurred between the two tests is unknown and may
11.1 1.29 have been responsible for our results. This ham-
10.8 1.8
pered our ability to gauge long-term retention of the
532 PREHOSPITAL EMERGENCY CARE
information. More research is needed on how fre-
quently mass-casualty triage refresher training should
be offered.
Clearly the ability to correctly triage patients dur-
ing an online written scenario may not represent one’s
ability to triage patients in a true MCI. Retrospec-
tive studies have examined triage accuracy during an
MCI,5,11–14 but no study has compared triage perfor-
mance during a written scenario test with triage per-
formance during a real MCI. It is unknown how the
paramedics who completed the study training sessions
will perform during an MCI.
CONCLUSION
Following a short didactic course, paramedics were
able to accurately perform SALT triage during a writ-
ten scenario. Four months after the training, they had
retained their understanding of and accuracy using
SALT triage. It appears that a brief educational tool
was effective for training EMS providers in SALT
triage.
References
1. Jenkins JL, McCarthy ML, Sauer LM, et al. Mass-casualty triage:
time for an evidence-based approach. Prehosp Disaster Med.
2008;23:3–8.
2. Lerner EM, Schwartz RB, Coule PL, et al. Mass casualty triage:
an evaluation of the data and development of a proposed na-
tional guideline. Disaster Med Public Health Prep. 2008;2(suppl
1):S25–S34.
OCTOBER/DECEMBER 2011 VOLUME 15 / NUMBER 4
3. Cone DC, Serra J, Burns K, MacMillan DS, Kurland L, Van
Gelder C. Pilot test of the SALT mass casualty triage system.
Prehosp Emerg Care. 2009;13:536–40.
4. Lerner EB, Schwartz RB, Coule PL, Pirrallo RG. Use of SALT
triage in a simulated mass-casualty incident. Prehosp Emerg
Care. 2010;14:21–5.
5. Kahn CA, Schultz CH, Miller KT, Anderson CL. Does START
triage work? An outcomes assessment after a disaster. Ann
Emerg Med. 2009;54:424–430.e1.
6. Garner A, Lee A, Harrison K, Schultz CH. Comparative analy-
sis of multiple-casualty incident triage algorithms. Ann Emerg
Med. 2001;38:541–8.
7. Sanddal TL, Loyacono T, Sanddal ND. Effect of JumpSTART
training on immediate and short-term pediatric triage perfor-
mance. Pediatr Emerg Care. 2004;20:749–53.
8. Risavi BL, Salen PN, Heller MB, Arcona S. A two-hour interven-
tion using START improves prehospital triage of mass casualty
incidents. Prehosp Emerg Care. 2001;5:197–9.
9. Baez AA, Sztajnkrycer MD, Smester P, Giraldez E, Vargas LE.
Effectiveness of a simple Internet-based disaster triage edu-
cational tool directed toward Latin-American EMS providers.
Prehosp Emerg Care. 2005;9:227–30.
10. Kilner T, Hall FJ. Triage decisions of United Kingdom police
firearms officers using a multiple-casualty scenario paper exer-
cise. Prehosp Disaster Med. 2005;20:40–6.
11. Beyersdorf SR, Nania JN, Luna GK. Community medical re-
sponse to the Fairchild mass casualty event. Am J Surg.
1996;171:467–70.
12. Kahn CA, Schultz CH, Miller KT, Anderson CL. Does START
triage work? An outcomes-level assessment of use at a mass
casualty event. Acad Emerg Med. 2007;14:S12a–S13a
13. Asaeda G. The day that the START triage system came to a
stop: observations from the World Trade Center disaster. Acad
Emerg Med. 2002;9:255–6.
14. Malik ZU, Pervez M, Safdar A, Massod T, Tariq M. Triage and
management of mass casualties in a train accident. J Coll Phys
Surg Pak. 2004;14:108–11.
Copyright of Prehospital Emergency Care is the property of Taylor & Francis Ltd and its content may not be
copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written
permission. However, users may print, download, or email articles for individual use.