Cincinnati
Content
ORIGINAL CONTRIBUTION
Cincinnati Prehospital Stroke Scale:
Reproducibility and Validity
From the Department of Emergency
Medicine* and the Department of
Neurology,‡ University of Cincinnati
Medical Center, Cincinnati, OH.
Received for publication
July 27, 1998. Revision received
October 27, 1998. Accepted for
publication November 2, 1998.
Rashmi U Kothari, MD*
Arthur Pancioli, MD*
Tiepu Liu, MD, DPH*
Thomas Brott, MD‡
Joseph Broderick, MD‡
Supported by the Emergency Medicine
Foundation through a grant from Genentech.
Address for reprints: Rashmi
Kothari, MD, Department of
Emergency Medicine, University of
Cincinnati, Post Office Box 670769,
Cincinnati, OH 45267-0769;
513-558-5281, fax 513-558-5791;
E-mail [email protected].
Copyright © 1999 by the American
College of Emergency Physicians.
0196-0644/99/$8.00 + 0
47/1/96801
See editorial, p. 450.
Study objective: The Cincinnati Prehospital Stroke Scale
(CPSS) is a 3-item scale based on a simplification of the
National Institutes of Health (NIH) Stroke Scale. When performed by a physician, it has a high sensitivity and specificity in
identifying patients with stroke who are candidates for thrombolysis. The objective of this study was to validate and verify
the reproducibility of the CPSS when used by prehospital
providers.
Methods: The CPSS was performed and scored by a physician
certified in the use of the NIH Stroke Scale (gold standard).
Simultaneously, a group of 4 paramedics and EMTs scored the
same patient.
Results: A total of 860 scales were completed on a convenience sample of 171 patients from the emergency department
and neurology inpatient service. Of these patients, 49 had a
diagnosis of stroke or transient ischemic attack. High reproducibility was observed among prehospital providers for total
score (intraclass correlation coefficient [rI], .89; 95% confidence
interval [CI], .87 to .92) and for each scale item: arm weakness,
speech, and facial droop (.91, .84, and .75, respectively). There
was excellent intraclass correlation between the physician and
the prehospital providers for total score (rI, .92; 95% CI, .89 to
.93) and for the specific items of the scale (.91, .87, and .78,
respectively). Observation by the physician of an abnormality in
any 1 of the 3 stroke scale items had a sensitivity of 66% and
specificity of 87% in identifying a stroke patient. The sensitivity
was 88% for identification of patients with anterior circulation
strokes.
Conclusion: The CPSS has excellent reproducibility among
prehospital personnel and physicians. It has good validity in
APRIL 1999
33:4
ANNALS OF EMERGENCY MEDICINE
3 7 3
CINCINNATI PREHOSPITAL STROKE SCALE
Kothari et al
identifying patients with stroke who are candidates for thrombolytic therapy, especially those with anterior circulation stroke.
[Kothari RU, Pancioli A, Liu T, Brott T, Broderick J: Cincinnati
Prehospital Stroke Scale: Reproducibility and validity. Ann
Emerg Med April 1999;33:373-378.]
INTRODUCTION
Early recognition and prompt medical evaluation is critical for the use of thrombolytic therapy for patients with
acute ischemic stroke. Patients must be treated with tissue plasminogen activator (t-PA) within 180 minutes of
symptom onset for the treatment to be effective.1 To
accomplish this, clinical centers have emphasized “prehospital education” and “en-route notification by EMS
personnel.”2 Anecdotal experience at the University of
Cincinnati indicated that early notification by paramedics
via medical command may reduce time to treatment by
allowing early mobilization of appropriate personnel.2,3
Based on this experience, we developed a 3-item stroke
scale that could be used by prehospital providers to identify patients with stroke who are candidates for thrombolytic therapy.4
The Cincinnati Prehospital Stroke Scale (CPSS) was
derived from a simplification of the 15-item National
Figure.
The CPSS evaluates for facial palsy, arm weakness, and
speech abnormalities. Items are scored as either normal or
abnormal.
Facial Droop
(The patient shows teeth or smiles)
Normal: Both sides of face move equally.
Abnormal: One side of face does not move as well as the other.
Arm Drift
(The patient closes their eyes and extends both arms straight out for
10 seconds)
Normal: Both arms move the same, or both arms do not move at all.
Abnormal: One arm either does not move, or one arm drifts down
compared to the other.
Speech
(The patient repeats “The sky is blue in Cincinnati”)
Normal: The patient says correct words with no slurring of words.
Abnormal: The patient slurs words, says the wrong words, or is
unable to speak.
3 7 4
Institutes of Health (NIH) Stroke Scale.5,6 The CPSS evaluates the presence or absence of facial palsy, asymmetric
arm weakness, and speech abnormalities in potential
stroke patients (Figure). This prehospital scale has been
shown to have high sensitivity and specificity in identifying patients with stroke when performed by a physician
on an emergency department population.4
The goal of this study was to verify the reproducibility
of the CPSS scale when scored by prehospital care
providers and to validate its ability to identify patients
with stroke.
M AT E R I A L S A N D M E T H O D S
A total of 24 prehospital care providers (17 paramedics
and 7 EMTs) from University of Cincinnati Mobile Care
Unit were evaluated during 23 different sessions. Groups
of 4 to 11 patients with or without a final discharge diagnosis of stroke were identified from the ED and the inpatient neurology service for each of these 23 different sessions. The testing physician identified a convenience
sample of patients from the ED. An attempt was made to
identify patients with chief complaints that were suggestive of stroke or of other diseases that could be mistaken
for stroke. To further increase the proportion of stroke
patients and patients with stroke-mimicking conditions,
patients were recruited from the inpatient neurology service as well as the ED. The inpatient physician on the neurology service identified patients with stroke, transient
ischemic attack (TIA), a stroke-mimicking condition, or a
combination of these conditions, as well as patients with
other neurologic disorders, and gave a list of these patients
to the testing physicians. The testing physicians were aware
of the patients’ chief complaints but not of their clinical
findings or final diagnoses. Paramedics and EMTs were
blinded to all patient information, including the patients’
medical histories, clinical findings, and diagnoses.
Table 1.
Patient demographics.
Variable
No. patients
Mean age (y)
No. (%) male
No. (%) black
No. (%) white
All Patients
Stroke/TIA
171
57.8
72 (42)
102 (60)
69 (40)
49
62.5
17 (35)
29 (59)
20 (41)
ANNALS OF EMERGENCY MEDICINE
Nonstroke
122
55.8
55 (45)
73 (60)
49 (40)
33:4 APRIL 1999
CINCINNATI PREHOSPITAL STROKE SCALE
Kothari et al
Sensitivity, specificity, and positive and negative predictive values were determined by using the physicians’
and prehospital care providers’ scores and comparing them
with the final hospital discharge stroke diagnoses. Patients
with a diagnosis of TIA were excluded from this analysis
since it was unknown when the neurologic deficits resolved.
All statistical analyses were performed with the use of
programs from SAS Institute Inc (Cary, NC). An SAS
macro program (INTRACC) was used to calculate intraclass correlation coefficients.
The CPSS was performed and scored by 1 of 2 physicians (gold standard) certified in the use of the NIH
Stroke Scale during each session. Simultaneously, a group
of 4 paramedics and EMTs scored the same patient. To
avoid fatiguing the patient, the scale was performed only
by the physician while the prehospital personnel scored
the patient’s response. Before each evaluation session, the
physician conducted a 10-minute review on how to perform and score the CPSS with the 4 paramedics and EMTs.
Only verbal instructions were given. This protocol was
approved by the University of Cincinnati Institutional
Review Board, and verbal consent was obtained from all
patients.
The intraclass correlation coefficient (rI) was calculated as the variance component among patients divided
by all variance components (patients, raters, and residuals). The 95% confidence intervals (CIs) were calculated
from F distribution.7 For reproducibility, rI was calculated for prehospital providers. For validity, raters were
grouped as physicians versus prehospital personnel
(EMTs and paramedics). Agreement between these 2
groups of raters was evaluated by r I for the total score
and for each of the 3 stroke scale item scores. The rI
has the same interpretation as κ statistics for measuring
agreement and reliability.8 Because the total score was
treated as a quantitative rating, there are multiple raters
(EMTs and paramedics) for each patient, and the raters
were different for each group of patients; therefore, we
used rI instead of κ statistics. According to Fleiss’8 criteria, values greater than .75 represent excellent agreement,
those between .4 and .75 represent fair to good agreement, and those lower than .4 represent poor agreement.
R E S U LT S
A total of 860 scales were completed on 171 patients. Of
these patients, 38 had a final diagnosis of stroke and 11 a
final diagnosis of TIA. There was no difference in terms of
race or sex between stroke/TIA and nonstroke patients;
however, nonstroke patients were significantly younger
(mean difference, 6.7 years, 95% CI, 11.7 to 1.7 years), as
shown in Table 1.
Of the 38 patients with stroke, 14 (37%) had deficits
involving the posterior circulation. Thirty-two patients
(18.7%) had nonstroke neurologic disorders or altered
mental status; 7 (21%) of these 32 patients had at least 1
abnormality on the CPSS (Table 2).
Excellent reproducibility was observed among prehospital care providers for total score (rI , .89; 95% CI, .87 to
.92) and for each scale item: arm weakness (rI, .91; 95%
CI, .88 to .93); speech (rI, .84; 95% CI, .80 to .87); and
facial droop (rI, .75; 95% CI, .69 to .80). In addition,
there was excellent correlation for total score between the
physician (gold standard) and the prehospital providers
Table 2.
Percentage of stroke and nonstroke patients with abnormalities detected on the CPSS (n = number of CPSS ratings performed).
Abnormality
Type
Face
Arm
Speech
No.
None
1
2
3
APRIL 1999
All Patients
MD
EMS
(n=171)
(n=689)
Stroke
TIA
MD
(n=38)
EMS
(n=154)
MD
(n=11)
EMS
(n=45)
Neurologic Nonstroke
MD
EMS
(n=32)
(n=129)
Nonneurologic
MD
EMS
(n=90)
(n=361)
7
15
15
9
14
13
18
47
37
25
37
38
0
0
9
0
0
0
9
16
13
11
18
12
2
3
6
3
4
4
73
15
6
3
78
13
5
4
34
39
16
11
41
32
14
13
91
9
0
0
100
0
0
0
78
9
9
3
77
13
5
5
91
7
2
0
92
6
1
1
33:4
ANNALS OF EMERGENCY MEDICINE
3 7 5
CINCINNATI PREHOSPITAL STROKE SCALE
Kothari et al
(rI, .92; 95% CI, .89 to .93), with no difference related to
level of training (rI for paramedics, .88, with 95% CI, .85
to .91; rI for EMTs, .85, with 95% CI, .81 to .89). Agreement
on scoring of specific items between physicians and prehospital personnel was excellent for all 3 items—arm
weakness (rI, .91), speech (rI, .87), and facial droop (rI,
.78). Again, there was no difference in terms of level of
training (Table 3).
Presence of a single abnormality on the CPSS had a
sensitivity of 66% and a specificity of 87% in identifying a
patient with stroke when scored by a physician, and 59%
and 89%, respectively, when scored by prehospital
providers (Table 4). Of the 13 patients with stroke who
were not identified by an abnormality on the prehospital
stroke scale, 10 had posterior circulation stroke (Table 5).
The CPSS correctly identified 21 of the 24 patients
with anterior circulation stroke (sensitivity, 87.5%; 95%
CI, 67% to 97%). The 3 patients with anterior circulation
strokes who were missed had minimal or atypical symptoms and would not have been candidates for thrombolysis. One of these patient presented with altered mental
status and no focal deficits and was noted to have a subacute infarct in her left caudate. Another patient had a
small subcortical infarct with only mild leg weakness. The
third patient had a small lacunar infarct with only minimal symptoms.
DISCUSSION
The rapid identification of potential stroke patients and
early ED notification are important components of the
prehospital management of stroke patients.9 The CPSS is
a 3-item neurologic examination that was developed to
assist paramedics and EMTs in identifying patients with
stroke who are candidates for thrombolysis. This scale
has been shown to be effective in identifying such stroke
patients when it is performed by a trained physician.4 It
can be taught in approximately 10 minutes and per-
formed in less than 1 minute.4 In this study, we found that
the CPSS was easily taught, was reproducible, and was a
valid tool when performed by paramedics and EMTs in
identifying stroke patients who may be candidates for
thrombolysis.
Interobserver reproducibility measures the agreement
among different persons using the same assessment tool.
Poor reproducibility (variation among separate observers)
increases error. Reproducibility among prehospital care
providers using the CPSS was excellent both for total
score (rI, .89) and for the individual stroke scale items (rI
ranging from .75 to .91). Brott et al5 reported similar
results using the 15-item NIH Stroke Scale, from which
our stroke scale was derived. The NIH Stroke Scale uses a
graded scoring system (eg, 0 to 2 or 0 to 4 points) for each
item rather than the binary (normal/abnormal) score used
by our CPSS. Brott et al5 found good interrater reliability
for all items of the NIH Stroke Scale (mean κ, .69) among
neurologists, emergency physicians, residents, and
nurses. Interrater reliability ranged from excellent for
motor arm (κ, .85), to moderate for dysarthria and best
language (κ, .64 and .55, respectively), to poor for facial
palsy (κ, .39).5 Similar results have been noted by other
investigators.6,10,11 Reproducibility for other prehospital
stroke scales has not been reported.
A measurement tool is valid if it correctly describes the
underlying phenomenon (eg, a specific neurologic
deficit) or disease (eg, stroke).12 The CPSS accurately
identifies specific neurologic abnormalities (ie, facial
palsy, arm weakness, and speech abnormality) when performed by prehospital care providers. Correlation for the
total score between prehospital care providers and the
physician was excellent (rI, .92). Similarly, agreement
between the physician and the prehospital providers on
individual scale items was good (rI ranging from .78 to
.91), with arm weakness being the most consistent. The
Table 4.
Sensitivity and specificity of the CPSS in identifying patients with
stroke.
Table 3.
Agreement (rI and 95% CI) between physicians and prehospital
care providers in scoring of individual CPSS items.
Item
Paramedics
Facial palsy
Arm
Speech
Total
.72 (.67–.78)
.87 (.84–.90)
.82 (.78–.86)
.88 (.85–.91)
3 7 6
EMTs
.64 (.55–.71)
.86 (.81–.89)
.77 (.71–.82)
.85 (.81–.89)
Combined
.78 (.74–.83)
.91 (.89–.93)
.87 (.34–.90)
.92 (.89–.93)
Physicians
No. of
Sensitivity
Specificity
Abnormalities (95% CI)
(95% CI)
Prehospital
Care Providers
Sensitivity
Specificity
(95% CI)
(95% CI)
1
2
3
59 (51–67)
27 (21–35)
13 (8–20)
66 (49–80)
26 (14–43)
11 (3–26)
87 (80–92)
95 (90–98)
99 (95–100)
ANNALS OF EMERGENCY MEDICINE
88 (86–91)
96 (94–97)
98 (96–99)
33:4 APRIL 1999
CINCINNATI PREHOSPITAL STROKE SCALE
Kothari et al
CPSS is also valid in identifying patients with stroke (sensitivity, 66%; specificity, 87%), especially anterior circulation stroke (sensitivity, 88%). Of the 13 stroke patients
who were missed by the CPSS, 10 had posterior circulation strokes and the other 3 remaining had minimal or
atypical symptoms of their anterior circulation strokes
and would not have been candidates for thrombolysis
based on our current protocols.
Presence of a single abnormality on the CPSS identified
all patients with anterior circulation stroke who would
have been candidates for thrombolytic therapy. The addition of a test for ataxia would have identified 6 of the 10
missed patients with posterior circulation stroke.
However, posterior circulation strokes are difficult to
diagnosis even by physicians13 and in our experience
have been infrequently treated with intravenous thrombolytic agents. Furthermore, ataxia is one of the most
poorly reproducible items on the NIH Stroke Scale.5,6,10
We previously reported that, when performed by a
physician on an ED population, a single abnormality on
the CPSS had a sensitivity of 100% and a specificity of
almost 90% in identifying stroke patients who are candidates for thrombolysis. The difference between those
results and the findings of the current study may relate to
differences in the populations being studied or differences in the severity of the strokes being evaluated. In the
previous study, patients were recruited only from the ED.
In contrast, the patients in this study were recruited from
both the ED and the inpatient neurology service (to
increase the number of patients with strokes or strokemimicking conditions). This led to a greater proportion of
patients (14 of 38) with posterior circulation stroke and a
Table 5.
Stroke patients not identified by the CPSS.
Patient Sex Age (y)
1
2
3
4
5
6
7
8
9
10
11
12
13
M
F
M
F
F
F
F
F
F
F
M
F
M
APRIL 1999
42
70
72
73
75
72
72
54
70
59
50
79
40
33:4
Symptoms
Ataxia and numbness
Vertigo/dizziness
Vertigo/dizziness
Diploplia and vertigo
Ataxia
Mental status changes
Vertigo/dizziness
Visual disturbances
Minimal unilateral weakness
Minimal unilateral weakness
Ataxia
Mild unilateral leg weakness
Minimal unilateral weakness
Final Diagnosis
Cerebellar infarct
Cerebellar infarct
Pontine infarct
Pontine infarct
Cerebellar infarct
Subacute caudate infarct
Cerebellar infarct
Occipital infarct
Occipital infarct
Brainstem infarct
Cerebellar infarct
Subcortical infarct
Lacuanar infarct
ANNALS OF EMERGENCY MEDICINE
greater proportion of nonstroke patients with neurologic
disorders and deficits, compared with the previous study
or what would be expected in a prehospital population. In
addition, in the previous study, the “final diagnosis” was
that of a patient with stroke who was a candidate for
thrombolysis (all patients scored 4 or higher on the NIH
Stroke Scale, and all were treated with t-PA). In the current study, the “final diagnosis” combined all patients
with a final hospital discharge diagnosis of stroke
regardless of the severity of their deficit. The CPSS in
this study could have missed a stroke patient with minimal symptoms, and in fact, the 3 patients with anterior
circulation stroke who were missed all had minimal or
atypical symptoms and would not have been candidates
for thrombolysis by our current protocol.
The only other prehospital stroke scale described in
the medical literature is the Los Angeles Prehospital
Stroke Scale. It evaluates arm strength, hand grip
strength, and facial droop. Kidwell et al 14 reported that
prehospital providers could effectively learn this scale
and increase their stroke knowledge base after watching a brief training video. However, the validity and
reproducibility of this scale have not been reported.
There are a number of limitations to our study. First,
the patient mix was not representative of the prehospital patient population. Almost a half of the study population had a stroke or another neurologic diagnosis. We
purposefully tried to increase the number of patients
with stroke or stroke-mimicking conditions to test this
tool. The proportion of patients with posterior circulation stroke was also much higher than would be
expected in the general population. The sensitivity of
this tool may be higher when it is used in a population
with a higher proportion of patients with anterior circulation stroke. The mode of testing was also somewhat
artificial in that prehospital personnel were instructed
on how to use the tool immediately before the exercise
and were asked only to score the findings and not to try
to elicit them. Ultimately this tool needs to be studied
in the prehospital setting with paramedics and EMTs
performing the examination under routine work conditions.
The CPSS has excellent reproducibility among prehospital care providers and can be taught in less than 10
minutes. It can accurately identify patients with stroke,
especially those with anterior circulation stroke, who
are candidates for thrombolysis.
We thank the men and women of the University of Cincinnati Mobile Care Unit for their
assistance in this project.
3 7 7
CINCINNATI PREHOSPITAL STROKE SCALE
Kothari et al
REFERENCES
1. National Institute of Neurological Disorders and Stroke rt-PA Stroke Study Group: Tissue
plasminogen activator for acute ischemic stroke. N Engl J Med 1995;333:1581-1587.
2. Barsan WG, Brott TG, Broderick JP, et al: Time of hospital presentation in patients with
acute stroke. Arch Intern Med 1993;153:2558-2561.
3. Brott TG, Haley EC, Levy DE, et al: Urgent therapy for stroke: Part I. Pilot study of tissue
plasminogen activator administered within 90 minutes. Stroke 1992;23:632-640.
4. Kothari RU, Hall K, Brott T, et al: Early stroke recognition: Developing an out-of-Hospital NIH
Stroke Scale. Acad Emerg Med 1997;4:986-990.
5. Brott T, Adams HP Jr, Olinger CP, et al: Measurements of acute cerebral infarction: A clinical examination scale. Stroke 1989;20:864-870.
6. Lyden P, Brott T, Tilley B, et al: Improved reliability of the NIH Stroke Scale using video
training. Stroke 1994;25:2220-2226.
7. Snedecor GW, Cochran WG: Statistical Methods, ed 7. Ames, IA: The Iowa State University
Press, 1980:243-246.
8. Fleiss JL: Statistical Methods for Rates and Proportions, ed 2. New York: John Wiley &
Sons, 1981:212-236.
9. Emergency Stroke Care Task Force: Acute stroke, in Cummins RO (ed): Advanced Cardiac
Life Support. Dallas, TX: American Heart Association, 1997:10-1,10-28.
10. Schmulling S, Grond M, Rudolf J: Training as a prerequisite for reliable use of NIH Stroke
Scale [letter]. Stroke 1998;29:1258-1259.
11. Goldstein LB, Bartels C, Davis JN: Interrater reliability of the NIH Stroke Scale. Arch Neurol
1989;46:660-662.
12. Lyden PD, Lau GT: A critical appraisal of stroke evaluation and rating scales. Stroke
1991;22:1345-1352.
13. Ferro JM, Pinto AN, Falcao I, et al: Diagnosis of stroke by the non-neurologist: A validation
study. Stroke 1998;29:1106-1109.
14. Kidwell CS, Saver JL, Eckstein M, et al: High accuracy of emergency medical technician
identification of stroke using the Los Angeles Prehospital Stroke Screen (LAPSS). Stroke
1998;29:313.
3 7 8
ANNALS OF EMERGENCY MEDICINE
33:4 APRIL 1999
Cincinnati Prehospital Stroke Scale:
Reproducibility and Validity
From the Department of Emergency
Medicine* and the Department of
Neurology,‡ University of Cincinnati
Medical Center, Cincinnati, OH.
Received for publication
July 27, 1998. Revision received
October 27, 1998. Accepted for
publication November 2, 1998.
Rashmi U Kothari, MD*
Arthur Pancioli, MD*
Tiepu Liu, MD, DPH*
Thomas Brott, MD‡
Joseph Broderick, MD‡
Supported by the Emergency Medicine
Foundation through a grant from Genentech.
Address for reprints: Rashmi
Kothari, MD, Department of
Emergency Medicine, University of
Cincinnati, Post Office Box 670769,
Cincinnati, OH 45267-0769;
513-558-5281, fax 513-558-5791;
E-mail [email protected].
Copyright © 1999 by the American
College of Emergency Physicians.
0196-0644/99/$8.00 + 0
47/1/96801
See editorial, p. 450.
Study objective: The Cincinnati Prehospital Stroke Scale
(CPSS) is a 3-item scale based on a simplification of the
National Institutes of Health (NIH) Stroke Scale. When performed by a physician, it has a high sensitivity and specificity in
identifying patients with stroke who are candidates for thrombolysis. The objective of this study was to validate and verify
the reproducibility of the CPSS when used by prehospital
providers.
Methods: The CPSS was performed and scored by a physician
certified in the use of the NIH Stroke Scale (gold standard).
Simultaneously, a group of 4 paramedics and EMTs scored the
same patient.
Results: A total of 860 scales were completed on a convenience sample of 171 patients from the emergency department
and neurology inpatient service. Of these patients, 49 had a
diagnosis of stroke or transient ischemic attack. High reproducibility was observed among prehospital providers for total
score (intraclass correlation coefficient [rI], .89; 95% confidence
interval [CI], .87 to .92) and for each scale item: arm weakness,
speech, and facial droop (.91, .84, and .75, respectively). There
was excellent intraclass correlation between the physician and
the prehospital providers for total score (rI, .92; 95% CI, .89 to
.93) and for the specific items of the scale (.91, .87, and .78,
respectively). Observation by the physician of an abnormality in
any 1 of the 3 stroke scale items had a sensitivity of 66% and
specificity of 87% in identifying a stroke patient. The sensitivity
was 88% for identification of patients with anterior circulation
strokes.
Conclusion: The CPSS has excellent reproducibility among
prehospital personnel and physicians. It has good validity in
APRIL 1999
33:4
ANNALS OF EMERGENCY MEDICINE
3 7 3
CINCINNATI PREHOSPITAL STROKE SCALE
Kothari et al
identifying patients with stroke who are candidates for thrombolytic therapy, especially those with anterior circulation stroke.
[Kothari RU, Pancioli A, Liu T, Brott T, Broderick J: Cincinnati
Prehospital Stroke Scale: Reproducibility and validity. Ann
Emerg Med April 1999;33:373-378.]
INTRODUCTION
Early recognition and prompt medical evaluation is critical for the use of thrombolytic therapy for patients with
acute ischemic stroke. Patients must be treated with tissue plasminogen activator (t-PA) within 180 minutes of
symptom onset for the treatment to be effective.1 To
accomplish this, clinical centers have emphasized “prehospital education” and “en-route notification by EMS
personnel.”2 Anecdotal experience at the University of
Cincinnati indicated that early notification by paramedics
via medical command may reduce time to treatment by
allowing early mobilization of appropriate personnel.2,3
Based on this experience, we developed a 3-item stroke
scale that could be used by prehospital providers to identify patients with stroke who are candidates for thrombolytic therapy.4
The Cincinnati Prehospital Stroke Scale (CPSS) was
derived from a simplification of the 15-item National
Figure.
The CPSS evaluates for facial palsy, arm weakness, and
speech abnormalities. Items are scored as either normal or
abnormal.
Facial Droop
(The patient shows teeth or smiles)
Normal: Both sides of face move equally.
Abnormal: One side of face does not move as well as the other.
Arm Drift
(The patient closes their eyes and extends both arms straight out for
10 seconds)
Normal: Both arms move the same, or both arms do not move at all.
Abnormal: One arm either does not move, or one arm drifts down
compared to the other.
Speech
(The patient repeats “The sky is blue in Cincinnati”)
Normal: The patient says correct words with no slurring of words.
Abnormal: The patient slurs words, says the wrong words, or is
unable to speak.
3 7 4
Institutes of Health (NIH) Stroke Scale.5,6 The CPSS evaluates the presence or absence of facial palsy, asymmetric
arm weakness, and speech abnormalities in potential
stroke patients (Figure). This prehospital scale has been
shown to have high sensitivity and specificity in identifying patients with stroke when performed by a physician
on an emergency department population.4
The goal of this study was to verify the reproducibility
of the CPSS scale when scored by prehospital care
providers and to validate its ability to identify patients
with stroke.
M AT E R I A L S A N D M E T H O D S
A total of 24 prehospital care providers (17 paramedics
and 7 EMTs) from University of Cincinnati Mobile Care
Unit were evaluated during 23 different sessions. Groups
of 4 to 11 patients with or without a final discharge diagnosis of stroke were identified from the ED and the inpatient neurology service for each of these 23 different sessions. The testing physician identified a convenience
sample of patients from the ED. An attempt was made to
identify patients with chief complaints that were suggestive of stroke or of other diseases that could be mistaken
for stroke. To further increase the proportion of stroke
patients and patients with stroke-mimicking conditions,
patients were recruited from the inpatient neurology service as well as the ED. The inpatient physician on the neurology service identified patients with stroke, transient
ischemic attack (TIA), a stroke-mimicking condition, or a
combination of these conditions, as well as patients with
other neurologic disorders, and gave a list of these patients
to the testing physicians. The testing physicians were aware
of the patients’ chief complaints but not of their clinical
findings or final diagnoses. Paramedics and EMTs were
blinded to all patient information, including the patients’
medical histories, clinical findings, and diagnoses.
Table 1.
Patient demographics.
Variable
No. patients
Mean age (y)
No. (%) male
No. (%) black
No. (%) white
All Patients
Stroke/TIA
171
57.8
72 (42)
102 (60)
69 (40)
49
62.5
17 (35)
29 (59)
20 (41)
ANNALS OF EMERGENCY MEDICINE
Nonstroke
122
55.8
55 (45)
73 (60)
49 (40)
33:4 APRIL 1999
CINCINNATI PREHOSPITAL STROKE SCALE
Kothari et al
Sensitivity, specificity, and positive and negative predictive values were determined by using the physicians’
and prehospital care providers’ scores and comparing them
with the final hospital discharge stroke diagnoses. Patients
with a diagnosis of TIA were excluded from this analysis
since it was unknown when the neurologic deficits resolved.
All statistical analyses were performed with the use of
programs from SAS Institute Inc (Cary, NC). An SAS
macro program (INTRACC) was used to calculate intraclass correlation coefficients.
The CPSS was performed and scored by 1 of 2 physicians (gold standard) certified in the use of the NIH
Stroke Scale during each session. Simultaneously, a group
of 4 paramedics and EMTs scored the same patient. To
avoid fatiguing the patient, the scale was performed only
by the physician while the prehospital personnel scored
the patient’s response. Before each evaluation session, the
physician conducted a 10-minute review on how to perform and score the CPSS with the 4 paramedics and EMTs.
Only verbal instructions were given. This protocol was
approved by the University of Cincinnati Institutional
Review Board, and verbal consent was obtained from all
patients.
The intraclass correlation coefficient (rI) was calculated as the variance component among patients divided
by all variance components (patients, raters, and residuals). The 95% confidence intervals (CIs) were calculated
from F distribution.7 For reproducibility, rI was calculated for prehospital providers. For validity, raters were
grouped as physicians versus prehospital personnel
(EMTs and paramedics). Agreement between these 2
groups of raters was evaluated by r I for the total score
and for each of the 3 stroke scale item scores. The rI
has the same interpretation as κ statistics for measuring
agreement and reliability.8 Because the total score was
treated as a quantitative rating, there are multiple raters
(EMTs and paramedics) for each patient, and the raters
were different for each group of patients; therefore, we
used rI instead of κ statistics. According to Fleiss’8 criteria, values greater than .75 represent excellent agreement,
those between .4 and .75 represent fair to good agreement, and those lower than .4 represent poor agreement.
R E S U LT S
A total of 860 scales were completed on 171 patients. Of
these patients, 38 had a final diagnosis of stroke and 11 a
final diagnosis of TIA. There was no difference in terms of
race or sex between stroke/TIA and nonstroke patients;
however, nonstroke patients were significantly younger
(mean difference, 6.7 years, 95% CI, 11.7 to 1.7 years), as
shown in Table 1.
Of the 38 patients with stroke, 14 (37%) had deficits
involving the posterior circulation. Thirty-two patients
(18.7%) had nonstroke neurologic disorders or altered
mental status; 7 (21%) of these 32 patients had at least 1
abnormality on the CPSS (Table 2).
Excellent reproducibility was observed among prehospital care providers for total score (rI , .89; 95% CI, .87 to
.92) and for each scale item: arm weakness (rI, .91; 95%
CI, .88 to .93); speech (rI, .84; 95% CI, .80 to .87); and
facial droop (rI, .75; 95% CI, .69 to .80). In addition,
there was excellent correlation for total score between the
physician (gold standard) and the prehospital providers
Table 2.
Percentage of stroke and nonstroke patients with abnormalities detected on the CPSS (n = number of CPSS ratings performed).
Abnormality
Type
Face
Arm
Speech
No.
None
1
2
3
APRIL 1999
All Patients
MD
EMS
(n=171)
(n=689)
Stroke
TIA
MD
(n=38)
EMS
(n=154)
MD
(n=11)
EMS
(n=45)
Neurologic Nonstroke
MD
EMS
(n=32)
(n=129)
Nonneurologic
MD
EMS
(n=90)
(n=361)
7
15
15
9
14
13
18
47
37
25
37
38
0
0
9
0
0
0
9
16
13
11
18
12
2
3
6
3
4
4
73
15
6
3
78
13
5
4
34
39
16
11
41
32
14
13
91
9
0
0
100
0
0
0
78
9
9
3
77
13
5
5
91
7
2
0
92
6
1
1
33:4
ANNALS OF EMERGENCY MEDICINE
3 7 5
CINCINNATI PREHOSPITAL STROKE SCALE
Kothari et al
(rI, .92; 95% CI, .89 to .93), with no difference related to
level of training (rI for paramedics, .88, with 95% CI, .85
to .91; rI for EMTs, .85, with 95% CI, .81 to .89). Agreement
on scoring of specific items between physicians and prehospital personnel was excellent for all 3 items—arm
weakness (rI, .91), speech (rI, .87), and facial droop (rI,
.78). Again, there was no difference in terms of level of
training (Table 3).
Presence of a single abnormality on the CPSS had a
sensitivity of 66% and a specificity of 87% in identifying a
patient with stroke when scored by a physician, and 59%
and 89%, respectively, when scored by prehospital
providers (Table 4). Of the 13 patients with stroke who
were not identified by an abnormality on the prehospital
stroke scale, 10 had posterior circulation stroke (Table 5).
The CPSS correctly identified 21 of the 24 patients
with anterior circulation stroke (sensitivity, 87.5%; 95%
CI, 67% to 97%). The 3 patients with anterior circulation
strokes who were missed had minimal or atypical symptoms and would not have been candidates for thrombolysis. One of these patient presented with altered mental
status and no focal deficits and was noted to have a subacute infarct in her left caudate. Another patient had a
small subcortical infarct with only mild leg weakness. The
third patient had a small lacunar infarct with only minimal symptoms.
DISCUSSION
The rapid identification of potential stroke patients and
early ED notification are important components of the
prehospital management of stroke patients.9 The CPSS is
a 3-item neurologic examination that was developed to
assist paramedics and EMTs in identifying patients with
stroke who are candidates for thrombolysis. This scale
has been shown to be effective in identifying such stroke
patients when it is performed by a trained physician.4 It
can be taught in approximately 10 minutes and per-
formed in less than 1 minute.4 In this study, we found that
the CPSS was easily taught, was reproducible, and was a
valid tool when performed by paramedics and EMTs in
identifying stroke patients who may be candidates for
thrombolysis.
Interobserver reproducibility measures the agreement
among different persons using the same assessment tool.
Poor reproducibility (variation among separate observers)
increases error. Reproducibility among prehospital care
providers using the CPSS was excellent both for total
score (rI, .89) and for the individual stroke scale items (rI
ranging from .75 to .91). Brott et al5 reported similar
results using the 15-item NIH Stroke Scale, from which
our stroke scale was derived. The NIH Stroke Scale uses a
graded scoring system (eg, 0 to 2 or 0 to 4 points) for each
item rather than the binary (normal/abnormal) score used
by our CPSS. Brott et al5 found good interrater reliability
for all items of the NIH Stroke Scale (mean κ, .69) among
neurologists, emergency physicians, residents, and
nurses. Interrater reliability ranged from excellent for
motor arm (κ, .85), to moderate for dysarthria and best
language (κ, .64 and .55, respectively), to poor for facial
palsy (κ, .39).5 Similar results have been noted by other
investigators.6,10,11 Reproducibility for other prehospital
stroke scales has not been reported.
A measurement tool is valid if it correctly describes the
underlying phenomenon (eg, a specific neurologic
deficit) or disease (eg, stroke).12 The CPSS accurately
identifies specific neurologic abnormalities (ie, facial
palsy, arm weakness, and speech abnormality) when performed by prehospital care providers. Correlation for the
total score between prehospital care providers and the
physician was excellent (rI, .92). Similarly, agreement
between the physician and the prehospital providers on
individual scale items was good (rI ranging from .78 to
.91), with arm weakness being the most consistent. The
Table 4.
Sensitivity and specificity of the CPSS in identifying patients with
stroke.
Table 3.
Agreement (rI and 95% CI) between physicians and prehospital
care providers in scoring of individual CPSS items.
Item
Paramedics
Facial palsy
Arm
Speech
Total
.72 (.67–.78)
.87 (.84–.90)
.82 (.78–.86)
.88 (.85–.91)
3 7 6
EMTs
.64 (.55–.71)
.86 (.81–.89)
.77 (.71–.82)
.85 (.81–.89)
Combined
.78 (.74–.83)
.91 (.89–.93)
.87 (.34–.90)
.92 (.89–.93)
Physicians
No. of
Sensitivity
Specificity
Abnormalities (95% CI)
(95% CI)
Prehospital
Care Providers
Sensitivity
Specificity
(95% CI)
(95% CI)
1
2
3
59 (51–67)
27 (21–35)
13 (8–20)
66 (49–80)
26 (14–43)
11 (3–26)
87 (80–92)
95 (90–98)
99 (95–100)
ANNALS OF EMERGENCY MEDICINE
88 (86–91)
96 (94–97)
98 (96–99)
33:4 APRIL 1999
CINCINNATI PREHOSPITAL STROKE SCALE
Kothari et al
CPSS is also valid in identifying patients with stroke (sensitivity, 66%; specificity, 87%), especially anterior circulation stroke (sensitivity, 88%). Of the 13 stroke patients
who were missed by the CPSS, 10 had posterior circulation strokes and the other 3 remaining had minimal or
atypical symptoms of their anterior circulation strokes
and would not have been candidates for thrombolysis
based on our current protocols.
Presence of a single abnormality on the CPSS identified
all patients with anterior circulation stroke who would
have been candidates for thrombolytic therapy. The addition of a test for ataxia would have identified 6 of the 10
missed patients with posterior circulation stroke.
However, posterior circulation strokes are difficult to
diagnosis even by physicians13 and in our experience
have been infrequently treated with intravenous thrombolytic agents. Furthermore, ataxia is one of the most
poorly reproducible items on the NIH Stroke Scale.5,6,10
We previously reported that, when performed by a
physician on an ED population, a single abnormality on
the CPSS had a sensitivity of 100% and a specificity of
almost 90% in identifying stroke patients who are candidates for thrombolysis. The difference between those
results and the findings of the current study may relate to
differences in the populations being studied or differences in the severity of the strokes being evaluated. In the
previous study, patients were recruited only from the ED.
In contrast, the patients in this study were recruited from
both the ED and the inpatient neurology service (to
increase the number of patients with strokes or strokemimicking conditions). This led to a greater proportion of
patients (14 of 38) with posterior circulation stroke and a
Table 5.
Stroke patients not identified by the CPSS.
Patient Sex Age (y)
1
2
3
4
5
6
7
8
9
10
11
12
13
M
F
M
F
F
F
F
F
F
F
M
F
M
APRIL 1999
42
70
72
73
75
72
72
54
70
59
50
79
40
33:4
Symptoms
Ataxia and numbness
Vertigo/dizziness
Vertigo/dizziness
Diploplia and vertigo
Ataxia
Mental status changes
Vertigo/dizziness
Visual disturbances
Minimal unilateral weakness
Minimal unilateral weakness
Ataxia
Mild unilateral leg weakness
Minimal unilateral weakness
Final Diagnosis
Cerebellar infarct
Cerebellar infarct
Pontine infarct
Pontine infarct
Cerebellar infarct
Subacute caudate infarct
Cerebellar infarct
Occipital infarct
Occipital infarct
Brainstem infarct
Cerebellar infarct
Subcortical infarct
Lacuanar infarct
ANNALS OF EMERGENCY MEDICINE
greater proportion of nonstroke patients with neurologic
disorders and deficits, compared with the previous study
or what would be expected in a prehospital population. In
addition, in the previous study, the “final diagnosis” was
that of a patient with stroke who was a candidate for
thrombolysis (all patients scored 4 or higher on the NIH
Stroke Scale, and all were treated with t-PA). In the current study, the “final diagnosis” combined all patients
with a final hospital discharge diagnosis of stroke
regardless of the severity of their deficit. The CPSS in
this study could have missed a stroke patient with minimal symptoms, and in fact, the 3 patients with anterior
circulation stroke who were missed all had minimal or
atypical symptoms and would not have been candidates
for thrombolysis by our current protocol.
The only other prehospital stroke scale described in
the medical literature is the Los Angeles Prehospital
Stroke Scale. It evaluates arm strength, hand grip
strength, and facial droop. Kidwell et al 14 reported that
prehospital providers could effectively learn this scale
and increase their stroke knowledge base after watching a brief training video. However, the validity and
reproducibility of this scale have not been reported.
There are a number of limitations to our study. First,
the patient mix was not representative of the prehospital patient population. Almost a half of the study population had a stroke or another neurologic diagnosis. We
purposefully tried to increase the number of patients
with stroke or stroke-mimicking conditions to test this
tool. The proportion of patients with posterior circulation stroke was also much higher than would be
expected in the general population. The sensitivity of
this tool may be higher when it is used in a population
with a higher proportion of patients with anterior circulation stroke. The mode of testing was also somewhat
artificial in that prehospital personnel were instructed
on how to use the tool immediately before the exercise
and were asked only to score the findings and not to try
to elicit them. Ultimately this tool needs to be studied
in the prehospital setting with paramedics and EMTs
performing the examination under routine work conditions.
The CPSS has excellent reproducibility among prehospital care providers and can be taught in less than 10
minutes. It can accurately identify patients with stroke,
especially those with anterior circulation stroke, who
are candidates for thrombolysis.
We thank the men and women of the University of Cincinnati Mobile Care Unit for their
assistance in this project.
3 7 7
CINCINNATI PREHOSPITAL STROKE SCALE
Kothari et al
REFERENCES
1. National Institute of Neurological Disorders and Stroke rt-PA Stroke Study Group: Tissue
plasminogen activator for acute ischemic stroke. N Engl J Med 1995;333:1581-1587.
2. Barsan WG, Brott TG, Broderick JP, et al: Time of hospital presentation in patients with
acute stroke. Arch Intern Med 1993;153:2558-2561.
3. Brott TG, Haley EC, Levy DE, et al: Urgent therapy for stroke: Part I. Pilot study of tissue
plasminogen activator administered within 90 minutes. Stroke 1992;23:632-640.
4. Kothari RU, Hall K, Brott T, et al: Early stroke recognition: Developing an out-of-Hospital NIH
Stroke Scale. Acad Emerg Med 1997;4:986-990.
5. Brott T, Adams HP Jr, Olinger CP, et al: Measurements of acute cerebral infarction: A clinical examination scale. Stroke 1989;20:864-870.
6. Lyden P, Brott T, Tilley B, et al: Improved reliability of the NIH Stroke Scale using video
training. Stroke 1994;25:2220-2226.
7. Snedecor GW, Cochran WG: Statistical Methods, ed 7. Ames, IA: The Iowa State University
Press, 1980:243-246.
8. Fleiss JL: Statistical Methods for Rates and Proportions, ed 2. New York: John Wiley &
Sons, 1981:212-236.
9. Emergency Stroke Care Task Force: Acute stroke, in Cummins RO (ed): Advanced Cardiac
Life Support. Dallas, TX: American Heart Association, 1997:10-1,10-28.
10. Schmulling S, Grond M, Rudolf J: Training as a prerequisite for reliable use of NIH Stroke
Scale [letter]. Stroke 1998;29:1258-1259.
11. Goldstein LB, Bartels C, Davis JN: Interrater reliability of the NIH Stroke Scale. Arch Neurol
1989;46:660-662.
12. Lyden PD, Lau GT: A critical appraisal of stroke evaluation and rating scales. Stroke
1991;22:1345-1352.
13. Ferro JM, Pinto AN, Falcao I, et al: Diagnosis of stroke by the non-neurologist: A validation
study. Stroke 1998;29:1106-1109.
14. Kidwell CS, Saver JL, Eckstein M, et al: High accuracy of emergency medical technician
identification of stroke using the Los Angeles Prehospital Stroke Screen (LAPSS). Stroke
1998;29:313.
3 7 8
ANNALS OF EMERGENCY MEDICINE
33:4 APRIL 1999
