ORIGINAL CONTRIBUTION

EMS, trauma, triage, scoring

Prehospital Index: A Scoring System

for Field Triage of Trauma Victims

The Prehospital Index (PHI) is a triage-oriented trauma severity scoring Ws- John J Koehler, MD*
tern comprising four components: systolic blood pressure, pulse, respiratory Lawrence J Baer, PhD1-
status, and level of consciousness, each scored 0 to 5. The PHI was devel- Stuart A Malafa, MD*
oped after analysis of 313 cases to provide an objective prehospital scoring MS Meindertsma:~
Nancy R Navitskas*
system for distinguishing less seriously injured patients (minor trauma) James E Huizenga*
from those patients who are likely to di'e withi'n 72 hours after injury or Grand Rapids, Michigan
who require general or neurosurgical operative intervention within 24 hours Detroit, Michigan
(major trauma). A PHI of 0 to 3 indicated minor trauma, and a PHI of 4 to
20 signified major trauma. Retrospective analysis of an additional 465 con- From the Department of Emergency
secutive trauma cases revealed that patients with a PHI of 0 to 3 (minor Medicine, Butterworth Hospital,* and the
trauma) had a 0% mortality and a 2% rate of general or neurosurgical oper- Office of Research Consultation, Grand
ative intervention. Those with a PHI of 4 to 20 (major trauma) carried a Rapids Area Medical Education Center,
16.4% mortality and an emergency operative rate of 49.1%. The PHI was Grand Rapids, Michigan;t and the Wayne
applied prospectively to 388 consecutive trauma cases presenting to the But- State University Medical School, Detroit,
terworth Hospital Emergency Department from October through December Michigan/:
1984. Of the 351 patients scored as minor trauma in the field, there was a
0% mortality and only a 0.3% operative rate. Those scored as major trauma Received for publication June 3, 1985.
in the field had a mortality of 27% (PHI 4 to 7, 0%; PHI 8 to 20, 53%) and Revision received August 18, 1985.
an operative rate of 40.5% (PHI 4 to 7, 22%; PHI 8 to 20, 57.9%). These data Accepted for publication September 17,
1985.
demonstrate the ability of the PHI to predict mortality (P < .001) and the
need for emergency general or neurosurgical operative intervention (P <
.00I). The PHI compares favorably in both simplicity and statistical reliabil- Presented at the University Association for
Emergency Medicine Annual Meeting in
ity with previously published, triage-oriented trauma severity scoring sys- Kansas City, Missouri, May 1985.
tems such as the Trauma Index, Triage Index, Trauma Score, and CRAMS
Scale. Of these four, only the CRAMS Scale has been similarly field-tested
Address for reprints: John J Koehler, MD,
prospectively. The PHI has been found to be an easily implemented and Department of Emergency Medicine,
statistically significant prehospital indicator of trauma severity. [Koehler JJ, Butterworth Hospital, 100 Michigan, NE,
Baer LJ, Malafa SA, Meindertsma MS, Navitskas NR, Huizenga JE: Pre- Grand Rapids, Michigan 49503,
hospital index: A scoring system for field triage of trauma victims. Ann
Emerg Med February 1986;15:178-182.]

INTRODUCTION
The concept of prehospital trauma scoring, among n u m e r o u s recent ad-
vances in the care of trauma patients, has been developed out of the need to
use prehospital data more effectively for appropriate triage, prehospital man-
agement, and emergency department preparation.
In developing the Prehospital Index (PHI) we sought to create a simple and
reliable scoring system by which an accurate distinction could be made be-
tween major and m i n o r trauma at the scene of an accident. For our study,
major trauma was defined as any injury requiring emergency general surgery
or neurosurgery no more than 24 hours postinjury or that resulted in the
patient's death no more than 72 hours postinjury. All other injuries were
classified as m i n o r trauma.

METHODS
The PHI was developed and tested in the following four phases: compo-
nent selection, numerical weight assignments, retrospective validation, and
prospective field testing.

15:2 February 1986 Annals of Emergency Medicine 178/115

PREHOSPITAL INDEX
Koehler et al

Component Selection
Computerized analysis of 313 con- TABLE 1. Analysis of relative probability
secutive trauma patients was used to (Average relative probability = 1.0)
i d e n t i f y t h e variables t h a t are ob-
tained easily in the prehospital phase
and that correlate well w i t h subse- Surgery Mortality
quent mortality or the need for emer- Systolic Blood Pressure
gency general surgery or neurosurgery. < 70 3.78 18.25
In all, more than 30 variables were 71 to 85 2.58 .91
tested. 86 to 100 1.46 1.40
D e r a n g e m e n t s in s y s t o l i c b l o o d > 100 0.71 0.51
pressure, pulse, respiratory status, and
level of c o n s c i o u s n e s s were m o s t Pulse
closely correlated with major trauma < 50 2.84
(Table 1). For example, based On the 50 to 100 0.69 0.62
average relative probability of 1.0, pa- 101 to 120 t .70 0.29
tients who had a prehospital systolic 121 to 140 2.52 6.25
blood pressure less than 70 were 3.78 Respiratory Status
times more likely to require emergen-
Need for intubation 22.i3
cy general surgery or neurosurgery and
18.25 times more likely to die. Those Consciousness
patients with a pulse of more than 120 Alert 0.67 0.25
were 2.52 times more likely to under- Confused/Combative 1.44
go emergency surgery and 6.25 times Unresponsive 2.96 8.76
more likely to die. In addition, with
increasing deviation from normal for
each of the four components, a defi-
nite trend was established toward an TABLE 2. Prehospital Index*
increasing probability of surgery or
mortality (Table 1),
In light of this, systolic blood pres- Components Value Score
sure, pulse, respiratory status, and Blood Pressure > 100
level of consciousness were chosen to 86 to 100
make up the PHI (Table 2). This index 75 to 85
is calculated by adding the scores for 0 to 74
each of the four components.
Pulse I> 120
Numerical Weight A s s i g n m e n t s 51 to 119
The hierarchy of numerical weights < 50
assigned to each of the four compo-
nent subcategories was based on its Respirations Normal
relative degree of correlation with out- Labored/Shallow
come (Table 1), and the actual numer- < 10/Min/Needs intubation
ical c o m b i n a t i o n was based on the
computerized analysis of several dif-
ferent c o m b i n a t i o n s applied to the Consciousness Normal
ConfUsed/Combative 0
original 313 trauma cases. In this way
No intelligible words 3
objective criteria were developed to es-
5
timate the severity of a patient's con-
dition for each of these four catego- Total 0 - 20
ries. *0 to 3, minor trauma.
4 to 20, major trauma.
(Penetrating abdominal or chest injuries given four points in addition to the
Retrospective Validation calculated PHI.)
Retrospective analyses of an addi-
tional 465 consecutive trauma cases
then were used to validate our index.
These analyses revealed that patients Prospective Field Testing with emergency medical technicians
with a PHI of 0 to 3 had a 2% rate of The PHI then was applied prospec- (EMTs) was established were included
general surgical or neurosurgical oper- tively to 388 consecutive trauma cases in the study. The PHI was calculated
ative intervention and a 0% mortality. presenting to the Buttelworth Hospi- at that time in the ED solely on the
Those with a PHI of 4 to 20 carried a tal Emergency Department from Oc- basis of patient data obtained from the
49.1% operative rate and a mortality tober t h r o u g h D e c e m b e r 1984. All EMTS in the field. Final disposition
of 16.4%. trauma cases in which radio contact was determined by subsequent chart
116/179 Annals of Emergency Medicine 15:2 February 1986
 arrival at an intermediate care facility.
TABLE 3. Prospective field testing results This distinction is critical, because
widespread application of prehospital
trauma scoring methods to emergency
Minor Trauma PHI < 4 medical services (EMS) systems will
Surgery 1/351 = 0.3% require EMTs to gather the field data.
Mortality 0/351 = 0% Therefore, unless the scoring system
Major Trauma PHI i> 4 is tested in the context of its intended
use, accurate conclusions are difficult
Surgery 15/37 = 40.5% to make.
Mortality 10/37 = 27.0% As compared to our PHI, which has
four components, the CRAMS Scale 4
(Table 6) has five components and re-
review. For the 351 patients field-cate- other words, 45.9% of patients desig- quires abdominal and motor examina-
gorized as minor trauma, there was nated as major trauma by the PHI tions, both of which are sometimes
only a 0.3% operative rate and a 0% either died or required emergency sur- impossible to obtain at the scene be-
mortality. Those field-categorized as gery. cause of time constraints or the lack
major trauma had an operative rate of A further breakdown of the major of patient cooperation. In addition, ac-
40.5% and a 27% mortality (Table 3). trauma patients showed that although curate measurement of capillary refill
The ability of the PHI to predict the major trauma range of 4 to 20 is and correct identification of decere-
mortality (P < .001) and the need for large, the cutoff for major trauma at 4 brate posturing may exceedthe diag-
emergency surgery (P < .001) were was justified by the 22.2% operative nostic capabilities of basic EMTs, es-
demonstrated using chi square analy- rate in the PHI 4 to 7 group (Table 5). pecially when one considers the less-
sis. Furthermore, of the four patients who than-optimal conditions in which
required surgery in the 4 to 7 category, they operate. In contrast, the informa-
RESULTS two had PHIs of 4 and the other two tion necessary to calculate the PHI is
Of the 351 patients field-categorized had PHIs of 5. In addition, as the PHI obtained more easily by EMTs in that
as minor trauma, only one required increases from 4 to 20, the rates of aside from blood pressure and pulse, it
general surgery and was the only false s u r g e r y and m o r t a l i t y rise pro- can be completed by observation
negative. The false-negative rate, then, gressively. alone.
was 0.3%, and the negative predictive The PHI had a sensitivity of 99.4%,
value was 99.7%. Of the 37 patients DISCUSSION somewhat higher than that demon-
field-categorized as major trauma, 20 Other prehospital trauma severity strated by the CRAMS Scale 4 (Table
did not die or require surgery and thus scoring systems have been developed, 7]. In addition, only 11 of the 61 pa-
were false positives. Therefore, the including the Trauma Index, 1 Triage tients field-categorized as major trau-
false-positive rate was 54.1% and the Index, 2 T r a u m a Score,3 and the ma by the CRAMS Scale died or re-
positive predictive value was 45.9% CRAMS Scale. 4 Of these, only the q u i r e d e m e r g e n c y surgery. T h e
(Table 4). CRAMS Scale was field tested pro- resulting false-positive rate for the
The positive predictive value may spectively. In addition, an undeclared CRAMS Scale using our criteria was
seem low, but it is more easily under- number of patients in the CRAMS 50 of 61, and is represented by an
stood when one considers that the Scale report were not scored at the 18.0% positive predictive value.
outcome criteria for major trauma are time of the run from data gathered by Closer comparison of the PHI and
limited strictly to mortality, emergen- EMTs in the field, but were scored by CRAMS Scales revealed statistically
cy general surgery, or neurosurgery. In physicians or critical care nurses on significant differences (chi square, P <

TABLE 4. Prospective field testing results

General Surgery or
Field Neurosurgery Other
Categorization or Died Disposition Total
Minor PHI < 4 1 350 351
Major PHI i> 4 17 20 37
Total 18 370 388

False-negative rate 1 of 351 (0.3%)

Negative predictive value 99.7%
Sensitivity 94.4%
False-positive rate 20 of 37 (54.1%)
Positive predictive value 45,9%
Specificity 94.6%

15:2 February 1986 Annals of Emergency Medicine 180/117

PREHOSPITAL INDEX
Koehler et al

.01). Although the sensitivities and

specificities were comparable, the TABLE 5. Relationship of morbidity and mortality to the prehospital index
positive predictive values were not - -
45.9% for the PHI and 18.0% for the
CRAMS Scale {Table 8). Using these Major Trauma No. (%)
data, the PHI is 2.5 times more reli- PHI 4 to 7
able in predicting mortality or the Surgery 4 of 18 (22.2)
need for emergency surgery. These dif- Mortality 0 of 18 (0)
ferences led to a total misclassifica-
PHI 8 to 20
tion rate for the CRAMS Scale of
10.2%, compared to only 5.4% for the Surgery 11 of 19 (57.9)
PHI. Mortality 10 of 19 (52.6)
In the CRAMS Scale report 4 differ-
ent criteria were used to define the
false positives. Those patients field-
categorized as major trauma and sent TABLE 6. CRAMS Scale a
home from the ED were considered
false positives. Using this definition,
there were only six false positives out Components Score
of 61 patients, resulting in a positive Circulation
predictive value of 90%. If these same Normal capillary refill and blood pressure 2
criteria are applied to our data, we > 100
w o u l d have had o n l y f o u r false Delayed capillary refill or blood pressure > 1
positives and a positive predictive val- 85 < 100
ue of 89%. We believe that although No capillary refill or blood pressure < 85 0
the CRAMS method is reasonable, it Respir~ions
may tend to shift the focus away from Normal 2
determining what we feel the emer- Abnormal (labored or shallow) 1
gency physician really needs to know, Absent 0
that is, whether the trauma patient in
Abdomen
the field is going to die or need an
Abdomen and thorax nontender 2
emergency operation. Failure to pre-
Abdomen or thorax tender 1
dict this correctly should be reflected
Abdomen rigid or flail chest 0
in the false-positive rate and the
positive predictive value. In addition, Motor
basing the false positives on whether Normal 2
or not they were admitted may intro- Responds only to pain 1
duce bias based on the admitting prac- (other than decerebrate)
tices of the institution in which the No response (or decerebrate) 0
study is conducted. Speech
In addition, In the CRAMS Scale re- Normal 2
port only those patients who died in Confused 1
the ED or who went directly to the No intelligible words 0
operating room for surgery were con- Total 0-10
sidered major trauma. We have broad-
ened that definition to include pa- Score ~< 8, major trauma.
tients w h o are a d m i t t e d and die Score t> 9, minor trauma.
within 72 hours or who require gener-
al surgery or neurosurgery within 24
hours. Certainly patients included by trauma victims. Only 18 of 388 pa- tive emergency care delivery systems
these broadened criteria should be tients were classified as major trauma for trauma patients, we have created
considered major trauma and not false {prevalence, 4.6% I. Second, although the trauma center designation to iden-
positives. Of the 37 patients field-cate- statistical significance was demon- • tify those institutions equipped to
gorized by the PHI to be major trau- strated, the study included only 388 treat major trauma patients, but we
ma, two died and none went to sur- patients. A multicenter trial is now seem to have overlooked somewhat
gery following admission. Despite under way to provide the necessary the need to identify more accurately
c o u n t i n g these p a t i e n t s as false patient volume as well as demograph- patients who should be transported to
positives (as they would under the ic diversity. such institutions for their care. The
CRAMS definition) the PHI specificity PHI is an objective, accurate, and sta-
remains statistically superior to that CONCLUSION tistically superior method by which
for the CRAMS Scale (P < .02). There are two rationales for adopt- major and minor trauma can be differ-
There are two areas of weakness in ing the PHI in the prehospital deci- entiated at the scene and appropriate
our study. First, the population under sion-malting process. triage made. It would be especially ap-
study has a low percentage of major In the development of more effec- plicable in situations in which re-
118/181 Annals of Emergency Medicine 15:2 February1986
 TABLE 7. Results of prospective f i d d testing of C R A M S Scale 4

General Surgery or
Field Neurosurgery Other
Categorization or Died Disposition Total
Minor 1 438 439
Major 11 50 61
Total 12 488 500
False-negative rate 1 of 439 (0.2%)
Negative predictive value 99.8%
Sensitivity 91.7%
False-positive rate 50 of 61 (82.0%)
Positive predictive value 18.1%
Specificity 89.8%

ancillary help, blood products, the op-

TABLE 8. Results comparison: PHI vs C R A M S Scale erating room, and especially the trau-
ma surgeon.
On completion of the multicenter
CRAMS Scale trial, which is now under way, both
Prevalence of major trauma 12of500 2.4% surgical and mortality curves for the
False-negative rate 1 of 439 (0.2%) PHI will be made available. Based on
Negative predictive value 91.7% these data, each institution or EMS
Sensitivity 99.8% system will be able to select the PHI
level at which they activate their trau-
False-positive rate 50 of 61 (82.0%) ma care protocol.
Positive predictive value 18.0% The Prehospital Index has been
Specificity 89.8% shown to be an easily implemented
Total misclassification rate 51 of 500 (10.2%) and s t a t i s t i c a l l y m e a n i n g f u l pre-
hospital indicator of trauma severity
Prehospital Index
that can supply a more objective basis
Prevalence of major trauma 18of388 4.6% for decision making in the prehospital
False-negative rate 1 of 351 (0.3%) phase of trauma management.
Negative predictive value 99.7%
Sensitivity 94.4%
The authors extend special appreciation
False-positive rate 20 of 37 (54,1%) to Amanda K Stressman for her assistance
Positive predictive value 45.9% in the preparation of this manuscript.
Specificity 94.6%
Total misclassification rate 21 of 388 (5.4%) REFERENCES
1. Kirkpatrick JR, Youmans RL: Trauma
index. An aide in the evaluation of injury
sources to manage major trauma can blood bank, radiology department, res- victims, l Trauma 1971;11:711-714.
become exhausted (such as multicar piratory therapy, the blood drawing 2. Champion HR, Sacco WJ, Harmon DS,
motor vehicle accidents} or where re- team, and the trauma surgeon. Field- et al: Assessment of injury severity: The
sources are limited (as in some rural categorization of trauma patients by triage index. Crit Care Med 1980;8:
settings}. the PHI can supply the objective crite- 201,208.
Decisions critical to patient survival ria necessary for staff and departmen- 3. Champion HR, Sacco WI, Carnazzo, et
often are made prior to patient arrival. tal acceptance of trauma alert pro- al: The trauma score. Crit Care Med
These are made frequently on the tocols. Such protocols can be initiated 1981;9:672-676.
basis of the emergency physician's ini- long before the patient arrives and can 4. Gormican SP: CRAMS scale-field tri-
tial impression of trauma severity, and provide much-needed uniformity and age of trauma victims. Ann Eraerg Med
they can include notification of the effectiveness in the mobilization of 1982;11:132-135.

15:2 February 1986 Annals of Emergency Medicine 182/119