Clinical Selection Strategies to Identify Ischemic Stroke

Patients With Large Anterior Vessel Occlusion

Results From SITS-ISTR (Safe Implementation of Thrombolysis in Stroke
International Stroke Thrombolysis Registry)
Jan F. Scheitz, MD*; Azmil H. Abdul-Rahim, MBChB, Msc(StrokeMed)*;
Rachael L. MacIsaac, PhD; Charith Cooray, MD; Heidi Sucharew, PhD; Dawn Kleindorfer, MD;
Pooja Khatri, MD, MSc; Joseph P. Broderick, MD; Heinrich J. Audebert, MD;
Niaz Ahmed, MD, PhD; Nils Wahlgren, MD, PhD; Matthias Endres, MD;
Christian H. Nolte, MD*; Kennedy R. Lees, MD, FRCP*; on behalf of SITS Scientific Committee

Background and PurposeThe National Institutes of Health Stroke Scale (NIHSS) correlates with presence of large
anterior vessel occlusion (LAVO). However, the application of the full NIHSS in the prehospital setting to select patients
Downloaded from http://stroke.ahajournals.org/ by guest on January 26, 2017

eligible for treatment with thrombectomy is limited. Therefore, we aimed to evaluate the prognostic value of simple
clinical selection strategies.
MethodsData from the Safe Implementation of Thrombolysis in Stroke International Stroke Thrombolysis Registry
(January 2012May 2014) were analyzed retrospectively. Patients with complete breakdown of NIHSS scores and
documented vessel status were included. We assessed the association of prehospital stroke scales and NIHSS symptom
profiles with LAVO (internal carotid artery, carotid-terminus or M1-segment of the middle cerebral artery).
ResultsAmong 3505 patients, 23.6% (n=827) had LAVO. Pathological finding on the NIHSS item best gaze was strongly
associated with LAVO (adjusted odds ratio 4.5, 95% confidence interval 3.85.3). All 3 facearmspeechtime test (FAST)
items identified LAVO with high sensitivity. Addition of the item best gaze to the original FAST score (G-FAST) or high
scores on other simplified stroke scales increased specificity. The NIHSS symptom profiles representing total anterior
syndromes showed a 10-fold increased likelihood for LAVO compared with a nonspecific clinical profile. If compared with
an NIHSS threshold of 6, the prehospital stroke scales performed similarly or even better without losing sensitivity.
ConclusionsSimple modification of the facearmspeechtime score or evaluating the NIHSS symptom profile may
help to stratify patients risk of LAVO and to identify individuals who deserve rapid transfer to comprehensive stroke
centers. Prospective validation in the prehospital setting is required. (Stroke. 2017;48:290-297. DOI: 10.1161/
STROKEAHA.116.014431.)
Key Words: FAST stroke management thrombectomy thrombolysis vessel occlusion

T he beneficial effects of endovascular treatment (EVT) in

addition to intravenous thrombolysis have been proven in
patients with acute ischemic stroke caused by large anterior
vessel occlusion (LAVO).15 Given the time-dependent effects
of revascularization,6,7 rapid recognition of potentially eligi-
ble patients for such treatment is critical, both in the prehos-
pital and in the early in-hospital triage stage. Because EVT
and complex diagnostic imaging resources often have limited
availability outside of comprehensive stroke centers (CSC)
or clinical trial settings, there is a pressing need to develop
strategies to identify patients who need bypass of a primary
stroke center and transfer to CSCs with EVT capability.
These strategies should balance well between sensitivity to
capture the majority of LAVO and appropriate specificity to

Received July 5, 2016; final revision received August 14, 2016; accepted September 14, 2016.
From the Center for Stroke Research Berlin (J.F.S., H.J.A., M.E., C.H.N.), Klinik fr Neurologie (J.F.S., H.J.A., M.E., C.H.N.), Excellence Cluster
NeuroCure (M.E.), German Center for Cardiovascular Research (DZHK) (M.E.), and German Center for Neurodegenerative Diseases (DZNE) (M.E.),
CharitUniversittsmedizin Berlin, Campus Benjamin Franklin, Germany; Stroke Research, Institute of Cardiovascular & Medical Sciences, University
of Glasgow, United Kingdom (A.H.A.-R., R.L.M., K.R.L.); Department of Clinical Neurosciences, Karolinska Institutet and Department of Neurology,
Karolinska University Hospital, Solna, Sweden (C.C., N.A., N.W.); Division of Biostatistics and Epidemiology, Cincinnati Childrens Hospital Medical
Center, OH (H.S.); and Department of Emergency Medicine, University of Cincinnati, College of Medicine, OH (D.K., P.K., J.P.B.).
Guest Editor for the article was Michael Brainin, MD, Dr (hon), FESO.
*Drs Scheitz, Abdul-Rahim, Nolte, and Lees contributed equally.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.
116.014431/-/DC1.
Correspondence to Jan F. Scheitz, MD, Department of Neurology, CharitUniversittsmedizin Berlin, Campus Benjamin Franklin, Hindenburgdamm
30, D-12200 Berlin, Germany. E-mail jan.scheitz@charite.de
2017 American Heart Association, Inc.
Stroke is available at http://stroke.ahajournals.org DOI: 10.1161/STROKEAHA.116.014431

290
 Scheitz et al Prediction of Large Anterior Vessel Occlusion 291
avoid overwhelming CSCs with patients who do not require
EVT.
A high National Institutes of Health Stroke Scale (NIHSS)
score is strongly associated with the presence of LAVO.810
Therefore, the NIHSS is frequently recommended to select
patients for EVT.11,12 Because of the complexity of a complete
NIHSS examination, simple stroke recognition scores like the
facearmspeechtime (FAST) test are commonly used by research studies.1315 The profiles grouped the 15 individual attributes
paramedics to evaluate patients with suspected stroke in the
field. Moreover, certain NIHSS items or symptom patterns
may be more informative of LAVO compared with simply
a score reflecting the overall severity of deficits. Recently, 6
profiles of NIHSS symptoms have been proposed and shown
to improve the clinical value of the overall NIHSS concerning
prediction of functional outcome and mortality.1315 To evalu-
ate different simple triage strategies beyond the total NIHSS
sum score, we aimed to analyze the value of the common pre-
hospital stroke scales and the NIHSS item profiles to predict
Downloaded from http://stroke.ahajournals.org/ by guest on January 26, 2017
LAVO in acute stroke patients.
Methods
Data Source, Design, Patients, and Outcomes
We conducted a retrospective analysis on individual patient data
obtained from the SITS-ISTR (Safe Implementation of Thrombolysis
in Stroke International Stroke Thrombolysis Registry) between
January 2012 and May 2014. SITS-ISTR is a multinational open
registry of acute ischemic stroke patients who received reperfusion
therapies.16,17 Patients from 132 participating centers with complete
breakdown of NIHSS scores and status of vessel occlusion were
included. Baseline characteristics included data on age, sex, stroke
severity according to the NIHSS, onset-to-treatment time, prestroke
modified Rankin Scale, and medical history (ie, previous stroke, cur-
rent smoking, hypertension, diabetes mellitus, atrial fibrillation, and
chronic heart failure).
Our outcome of interest was the presence of LAVO (ie, occlusion
within the internal carotid artery, carotid-T and M1 segment of the
middle cerebral artery). The definition of LAVO was in accordance
to the recent positive EVT trials in patients with anterior circulation
stroke.18,19 Vessel imaging was usually performed before treatment
with thrombolysis or shortly after application of bolus dose.
Two sensitivity analyses were conducted. First, we added basilar
artery occlusion (BAO) to the definition of large vessel occlusion.
Although currently EVT is not covered by Class I recommendation
in BAO, guidelines recommend clinical evaluation of patients with
BAO in CSCs.11 Thus, prehospital detection of patients with BAO is
important. Second, we confined the analysis to patients with mod-
erate stroke severity (NIHSS score =611). This group constitutes
a relevant subgroup because most false-positive or false-negative
identifications of LAVO occur, and the majority of patients evaluated
in the field have overall moderate stroke severity. The upper limit
of NIHSS score of 11 was chosen for consistency with optimal cut-
off for prediction of LAVO in the previous studies.9,2022 The lower
threshold of NIHSS score of 6 was chosen because this cutoff showed
at least 90% sensitivity for LAVO in the present cohort and previ-
ous reports.20,21 Stroke with NIHSS score <6 is often considered mild
stroke, with low probability of LAVO.23
Common Prehospital Stroke Scales
and NIHSS Item Profiles
We evaluated prehospital stroke recognition scales that could be
derived directly from the individual breakdown of NIHSS items
at baseline (ie, FAST test) and simplified NIHSS scores that have
been shown to be associated with LAVO (ie, 3-item Stroke Scale
[3I-SS], the Rapid Arterial Occlusion Evaluation Scale [RACE],
the Cincinnati Stroke Triage Assessment Tool [C-STAT], and the
Prehospital Acute Stroke Severity scale [PASS]).2427 Table I in the
online-only Data Supplement summarizes components of the prehos-
pital stroke scales that were analyzed. Because the NIHSS item best
gaze is missing in the typical FAST algorithm but strongly associated
with LAVO,21,2628 we tested the hypothesis that adding the item best
gaze to FAST (G-FAST) may improve its predictive value.
The NIHSS item profiles that were recently described and vali-
dated may prove useful for clinical stroke prognostication and
of NIHSS, using latent class analysis, into 6 clinical symptom pro-
files.1315 We applied the probabilities of profile membership gener-
ated by Sucharew et al13 to our cohort. Profile A represents a total
anterior circulation syndrome (TACS) of the dominant hemisphere;
Profile B, a TACS of the nondominant hemisphere; Profile C, a partial
anterior circulation syndrome (PACS) of the dominant hemisphere
with predominant language deficits; Profile D, a PACS of dominant
hemisphere without predominant language deficits; Profile E, a PACS
of the nondominant hemisphere; and Profile F, a mild clinical syn-
drome with low probability of abnormal findings on all NIHSS items
(Table II in the online-only Data Supplement).
Statistical Methods
Continuous variables were compared using the t test or Mann
Whitney U test and were presented as meanstandard deviation
(SD) or median (interquartile range), where appropriate. Categorical
variables were compared using the Pearson 2 test and presented as
percentages (n). Multivariable logistic regression analysis was per-
formed to assess the association of single NIHSS items, the prehospi-
tal stroke scales, and the NIHSS profiles with LAVO. Adjustment was
made for variables significantly associated with LAVO in the univari-
able comparison (sex, atrial fibrillation, onset-to-treatment time). We
also adjusted the analysis for age, prestroke modified Rankin Scale
score >2 (ie, being dependent from others in activities of daily liv-
ing), and history of previous stroke because prestroke disability and
residual neurological deficits from a previous stroke may affect the
NIHSS. For the prehospital stroke scales, the lowest score was used
as the reference. For the NIHSS item profiles, profile F (stroke with
low probabilities of abnormal findings on all 15 items) was used as
reference. Regarding single NIHSS items, we applied forward step-
wise regression analysis to identify the NIHSS item that improves the
model most.
We computed area under the receiver operating characteristics
curve (AUC) and the respective 95% confidence intervals (CIs) to
assess the global performance of the prehospital stroke scales and
NIHSS symptom profiles to predict LAVO. AUC values were com-
pared using the method of DeLong et al.29 The receiver operating
characteristics curvederived optimal cutoff for the scores was deter-
mined at the maximal Youden Index.30 Finally, we calculated sensitiv-
ity, specificity, positive predictive value (PPV), negative predictive
value (NPV), and overall accuracy for the prediction of LAVO at
high-sensitivity (>85%) and high-specificity (>75%) cutoffs of the
common prehospital stroke scales. For consistency with the sensitiv-
ity of the widely accepted FAST and Cincinnati Prehospital Stroke
Scale (CPSS) scores for recognition of stroke patients in the field,
a sensitivity >85% was considered the high-sensitivity threshold.31,32
The target specificity of >75% is even higher than the average speci-
ficity of common stroke recognition tools to discriminate strokes
from stroke mimics and suggests rate of futile transfers of <1 out
of 4.32,33 Data analysis was performed using SPSS (version 22, IBM,
New York) or MedCalc (Version 16.2, Ostend, Belgium).
Results
During the study period, 3505 patients with complete break-
down of NIHSS items and data on vessel occlusion site were
available for analysis. LAVO was present in 23.6% (n=827) of
patients. Baseline characteristics of patients with or without
LAVO are shown in Table1. Baseline NIHSS was strongly
 292StrokeFebruary 2017

associated with LAVO (adjusted odds ratio [OR] 1.15, 95% CI
1.131.16, per point). Sensitivity, specificity, PPV, and NPV at
different NIHSS cutoffs are shown in Table III in the online-
only Data Supplement.
As shown in Table2, there was a graded relationship
between prehospital stroke scales scores and NIHSS item pro-
files with presence of LAVO. Compared with a FAST score
of 0 or 1, patients with all 3 FAST items being positive had
an adjusted OR of 7.9 (95% CI 5.211.9) for LAVO (sensitiv-
ity 84%, specificity 44%, PPV 32%, and NPV 90%). Forward
stepwise multiple regression analysis suggested best gaze
to be the single NIHSS item with strongest association with
LAVO (adjusted OR 4.5, 95% CI 3.85.3). Addition of abnor-
mal gaze to FAST improved specificity (Table2) and resulted
in significant improvement of the AUC for LAVO compared
with FAST alone (P<0.001; Table IV in the online-only Data
Supplement).
The 6 NIHSS symptom profiles that represent different
Downloaded from http://stroke.ahajournals.org/ by guest on January 26, 2017
especially in patients presenting with moderate stroke sever-
ity. When compared with the NIHSS cutoff 6, which is rec-
ommended by current American Heart Association/American
Stroke Association guidelines to select patients for thrombec-
tomy (AUC 0.60, 95% CI 0.580.62), AUCs of the G-FAST3
(AUC 0.64, 95% CI 0.620.66) and C-STAT1 (AUC 0.63,
95% CI 0.610.65) were significantly higher (P<0.001), but
the cut-offs showed similar sensitivity (89%91%; Table3).
Sensitivity Analyses
Similar results were obtained after addition of 93 patients with
BAO to the large vessel occlusion definition (n=920; 26.2%),
although the overall strength of the association was slightly
weaker (Table V in the online-only Data Supplement). The
optimal NIHSS cutoff was also 12 (sensitivity 70%, specific-
ity 70%, PPV 45%, and NPV 87%), and the cutoff showing
at least 85% sensitivity was 7 (sensitivity 88%, specific-
ity 39%, PPV 34%, and NPV 90%). The optimal cutoffs for

clinical phenotypes were reproduced from previous analy-
ses.1315 Patients allocated to NIHSS symptom profiles rep-
resenting TACS (profile A and B) had a >6-fold increase of
LAVO compared with all other profiles combined (OR 6.2,
95% CI 5.17.5).
Table3 shows sensitivity, specificity, PPV, and NPV for
presence of LAVO of different cutoffs of the entire NIHSS, the
prehospital stroke scales, and NIHSS item profiles. High sen-
sitivity was observed for FAST2, G-FAST3, C-STAT1,
3I-SS1, PASS1, RACE3, and clinical signs of at least a
PACS (NIHSS symptom profiles A to E), while high speci-
ficity was observed for G-FAST=4, C-STAT3, 3I-SS=3,
PASS=3, RACE6, and clinical signs of TACS (NIHSS symp-
tom profile A and B).
The AUC of the prehospital stroke scales and NIHSS item
profiles to predict LAVO was similar and nearly as good as the
entire NIHSS (Table IV in the online-only Data Supplement),
detection of LVO including BAO were FAST=3, G-FAST=4,
C-STAT2, 3I-SS2, PASS2, and RACE5 (Table VI in the
online-only Data Supplement).
When we focused our analysis to patients with moderate
stroke severity (NIHSS 611, n=1257 patients), frequency of
LAVO was 12.6% (19.2% of all observed LAVO within the
cohort, 159 of 827). The common prehospital stroke scales
performances for prediction of LAVO did not differ from the
overall total NIHSS score (Table III in the online-only Data
Supplement), with the highest absolute AUC value for the
C-STAT. Similar to the entire cohort, increasing integer val-
ues of the common prehospital stroke scales showed disparate
associations with LAVO (Table VII in the online-only Data
Supplement). Profiles A and B (left and right TACS) were
associated with a nearly 3-fold increased risk of LAVO com-
pared with all other profiles (CF combined; adjusted OR 2.8,
95% CI 2.04.0).

Table 1. Baseline Characteristics of the Cohort According to Presence of LAVO

Entire Cohort, LAVO, No LAVO,
N=3505 N=827 N=2678 P Value
Age, y, mean (SD) 68.1 (13.6) 68.3 (13.5) 68.1 (13.6) 0.631
Sex, male, % (n) 56.1 (1967) 52.6 (435) 57.2 (1532) 0.020
Pre-mRS score >2, % (n), 164 missings 4.0 (134) 3.7 (29) 4.1 (105) 0.577
NIHSS sum, median (IQR) 9 (616) 16 (1120) 8 (513) <0.001
Endovascular treatment, % (n) 8.0 (282) 26.6 (220) 2.3 (62) <0.001
Atrial fibrillation, % (n), 16 missings 18.9 (661) 22.8 (188) 17.7 (473) 0.001
Chronic heart failure, % (n), 21 missings 7.7 (268) 8.6 (71) 7.4 (197) 0.255
Current smoker, % (n), 121 missings 17.5 (593) 18.2 (143) 17.3 (450) 0.560
Diabetes mellitus, % (n), 13 missings 18.5 (645) 16.9 (139) 19.0 (506) 0.187
Hyperlipidemia, % (n), 34 missings 29.6 (1028) 31.5 (257) 29.1 (771) 0.188
Hypertension, % (n), 12 missings 64.6 (2258) 64.8 (533) 64.6 (1725) 0.935
Previous stroke, % (n), 20 missings 12.0 (417) 11.1 (91) 12.2 (326) 0.358
Onset-to-needle, min, median (IQR) 150 (119195) 145 (115190) 150 (120195) 0.028
IQR indicates interquartile range; LAVO, large anterior vessel occlusion; mRS, modified Rankin Scale; and NIHSS, National
Institutes of Health Stroke Scale.
 Scheitz et al Prediction of Large Anterior Vessel Occlusion 293

Table 2. Association of NIHSS Categories, NIHSS Symptom Profile, and Simplified NIHSS Scores With LAVO
Number Needed to
Score Name Score Result LAVO n/N Screen* Adjusted OR for LAVO
NIHSS categories 05 72/828 11.5 1 (Reference)
610 128/1107 8.7 1.40 (1.021.92)
1115 191/679 3.6 4.29 (3.155.83)
>15 436/891 2.1 10.72 (8.0014.37)
NIHSS symptom profile A 318/793 2.5 10.24 (6.4016.38)
B 310/775 2.5 9.95 (6.2215.92)
C 38/352 9.3 1.85 (1.053.25)
D 70/644 9.2 1.62 (0.972.72)
E 68/591 8.7 1.87 (1.123.13)
F 23/350 15.2 1 (Reference)
FAST 0 3/63 20.8 1 (Reference)
Downloaded from http://stroke.ahajournals.org/ by guest on January 26, 2017

1 25/429 17.2 1.65 (0.387.18)

2 104/806 7.8 4.19 (1.0117.45)
3 695/2207 3.1 12.29 (2.9950.57)
G-FAST 0 3/56 18.6 1 (Reference)
1 22/398 18.2 1.37 (0.316.05)
2 68/686 10.0 2.85 (0.6811.99)
3 265/1334 5.0 5.96 (1.4424.68)
4 468/1029 2.2 20.52 (4.9784.99)
C-STAT 0 92/1091 11.9 1 (Reference)
1 145/953 6.6 1.85 (1.382.48)
2 108/410 3.8 4.17 (3.045.74)
3 266/606 2.3 8.81 (6.6511.66)
4 216/445 2.1 10.58 (7.8214.32)
PASS 0 23/296 12.8 1 (Reference)
1 153/1446 9.4 1.38 (0.862.24)
2 333/1074 3.2 5.63 (3.538.97)
3 318/689 2.2 11.16 (6.9218.00)
RACE 01 55/781 14.2 1 (Reference)
2 63/587 9.3 1.75 (1.182.60)
3 55/368 6.7 2.35 (1.543.57)
4 65/327 5.0 3.42 (2.275.14)
56 153/590 3.9 4.79 (3.386.79)
79 436/852 2.0 15.18 (10.9621.02)
3I-SS 0 68/803 11.8 1 (Reference)
1 221/1419 6.4 1.90 (1.412.56)
2 367/900 2.5 7.47 (5.5610.00)
3 171/383 2.2 8.44 (6.0311.81)
3I-SS indicates 3-item stroke scale; C-STAT, Cincinnati Stroke Triage Assessment Tool; FAST, facearmspeechtime test; G-FAST,
gazefacearmspeechtime score; LAVO, large anterior vessel occlusion; NIHSS, National Institutes of Health Stroke Scale; OR, odds ratio;
PASS, Prehospital Acute Stroke Severity scale; and RACE, Rapid Arterial Occlusion Evaluation Scale.
*Number needed to screen with the respective test result to identify a LAVO, per 100 patients.
Adjusted for age, sex, atrial fibrillation, pre-mRS, prior stroke, and onset-to-treatment time.
In G-FAST, the item best gaze was added to the FAST score (G-FAST).
In contrast to the original version, only one point was assigned per pathological item.
 294StrokeFebruary 2017

Table 3. Sensitivity, Specificity, PPV, and NPV for Presence of LAVO at Certain Cutoffs of the NIHSS, Simplified
NIHSS Scores, and NIHSS Symptom Profiles
n/N (%) Sensitivity Specificity PPV NPV Accuracy
Cutoffs of scores with >85% sensitivity and highest possible specificity
NIHSS8 2183/3505 (63.2) 85.6 44.9 28.2 91.3 54.5
NIHSS6* 2677/3505 (76.4) 91.3 28.2 28.2 91.3 43.1
FAST2 3013/3505 (86.0) 96.6 17.3 26.5 94.3 36.0
FAST=3 or abnormal item best
2410/3505 (68.8) 89.1 37.5 30.6 91.8 49.7
gaze
G-FAST3 2363/3505 (67.5) 88.7 39.1 31.0 91.8 50.8
C-STAT1 2414/3505 (68.9) 88.9 37.3 30.4 91.6 49.5
3I-SS1 2702/3505 (77.1) 91.8 27.5 28.1 91.5 42.7
PASS1 3209/3505 (91.6) 97.2 10.2 25.1 92.2 30.7
RACE3 2137/3505 (61.0) 85.7 46.7 33.2 91.4 55.9
NIHSS profile AE (at least PACS or
Downloaded from http://stroke.ahajournals.org/ by guest on January 26, 2017

3155/3505 (90.0) 97.2 12.2 25.5 93.4 32.3

worse) vs profile F
Cutoffs of scores with specificity >75% and highest possible sensitivity
NIHSS14 1133/3505 (32.3) 63.1 77.2 46.1 87.1 73.9
G-FAST=4 1029/3505 (29.4) 56.7 79.0 45.5 85.5 73.7
C-STAT3 1051/3505 (30.0) 58.3 78.8 45.9 85.9 73.9
3I-SS=3 383/3505 (10.9) 20.7 92.1 44.6 79.0 75.3
PASS=3 689/3505 (19.7) 38.5 86.2 46.2 81.9 74.9
RACE6 1154/3505 (32.9) 62.2 76.1 44.5 86.7 72.8
NIHSS profile A 793/3505 (22.6) 38.5 82.3 40.1 81.2 72.0
NIHSS profile B 775/3505 (22.1) 37.5 82.6 40.0 81.1 72.0
Statistically optimal cutoffs
NIHSS12 1420/3505 (40.5) 72.1 69.2 42.0 88.9 69.9
FAST=3 2207/3505 (63.0) 84.0 43.5 31.5 89.9 53.1
G-FAST3 2363/3505 (67.5) 88.7 39.1 31.0 91.8 50.8
C-STAT2 1461/3505 (41.7) 71.3 67.5 40.4 88.4 68.4
3I-SS2 1283/3505 (36.6) 65.0 72.2 41.9 87.0 70.5
PASS2 1763/3505 (50.3) 78.7 58.5 36.9 89.9 63.3
RACE5 1442/3505 (41.1) 71.2 68.2 40.8 88.5 68.9
NIHSS symptom profile A or B 1568/3505 (44.7) 75.9 64.9 40.1 89.7 67.5
3I-SS indicates 3-item Stroke Scale; AHA, American Heart Association; ASA, American Stroke Association; C-STAT, Cincinnati Stroke Triage
Assessment Tool; FAST, facearmspeechtime test; G-FAST, gazefacearmspeechtime score; LAVO, large anterior vessel occlusion;
NIHSS, National Institutes of Health Stroke Scale; NPV, negative predictive value; PACS, partial anterior circulation syndrome; PASS, Prehospital
Acute Stroke Severity scale; PPV, positive predictive value; and RACE, Rapid Arterial Occlusion Evaluation Scale.
*NIHSS6 is an inclusion criterion for endovascular treatment with stent retrievers according to current AHA/ASA Focused Update of the
2013 Guidelines for the Early Management of Patients With Acute Ischemic Stroke Regarding Endovascular Treatment.12
In contrast to the original version, only one point was assigned per pathological item.

Discussion
One of the major challenges of the current stroke care is to
translate the implications of the endovascular stroke trials into
clinical practice. Noninvasive vessel imaging and rapid trans-
fer of eligible patients to CSCs with EVT treatment option
need to be organized effectively. Because no triage strategy
performs perfectly, some patients with LAVO will be inevi-
tably missed, and many patients without LAVO will be trans-
ferred to CSCs.22 It is a political issue to decide what range
of false negatives and false positives are acceptable from the
perspective of society as a whole. Local circumstances should
also influence the choice of selection criteria.
In general, prehospital triage tools for detection of LAVO
should be as simple as possible and easily performed and mem-
orized by emergency medical services personnel. In addition,
the ideal scores are supposed to discriminate stroke patients
from stroke mimics. In our cohort, the simple prehospital
stroke scales performed nearly as well as the entire NIHSS in
 Scheitz et al Prediction of Large Anterior Vessel Occlusion 295
identifying LAVO and at least as well as the entire NIHSS in
patients with moderate severity. Importantly, highly sensitive
cutoffs of the prehospital scores performed as well as or even
better than NIHSS cutoff 6, which is recommended to select
patients for thrombectomy according to the current American
Heart Association/American Stroke Association guidelines.12
Thus, our findings emphasize the potential of simplified
NIHSS scores to detect LAVO in the prehospital setting. Our
sensitivity analysis demonstrated comparable findings when
BAO was included in the large vessel occlusion category.
Several stroke recognition tools have been validated for pre-
hospital evaluation of patients with suspected strokes.3133 The
FAST score is already widely used and shows the best sensi-
tivity for correct diagnosis of stroke together with the similar
CPSS.31,32 To identify stroke patients with underlying LAVO in
the field, it seems reasonable to use a 2-step screening process
starting with the FAST score. FAST has the advantage of using
the item facial palsy that has been shown to be the NIHSS item
Downloaded from http://stroke.ahajournals.org/ by guest on January 26, 2017
with best capability to discriminate between strokes and mim-
ics.34 In a second step, another tool is needed for the triage
regarding vessel imaging and facilitate transfer to EVT centers.
Ideally, this secondary score should be deduced from the initial
score but require addition of only a few more items with higher
sensitivity and specificity for presence of LAVO. In line with
previous studies,21,27,28 we found that gaze deviation was the
most sensitive clinical sign suggestive of LAVO. Thus, a simple
expansion of the typical FAST score by the NIHSS item best
gaze was developed in our study (G-FAST). G-FAST would
fulfill the criteria mentioned above and has the advantage of
mentioning all tested signs as an acronym. The C-STAT fol-
lows a similar concept and also seems promising at the second
stage as cortical signs (especially gaze) strengthen the score,
but still it maintains simplicity.18 By using the questions and
commands from the NIHSS instead of the language and speech
items that are complex for many emergency medical services,
the C-STAT makes the rating objective, rather than subjective.
Of note, G-FAST and C-STAT performed particularly well in
patients with moderate stroke severity, which represents the
majority of cases seen in the field.
Importantly, the optimal prehospital triage strategy depends
on various time variables. Besides time from symptom onset
until first evaluation by paramedics, transport time to next
CSC has to be considered. Our analysis adds relevant findings
in at least 3 different clinical scenarios.
First, we consider a patient with suspected stroke who is
evaluated by paramedics early after onset of symptoms or with
short transfer time to a CSC. In this case, a high sensitivity (ie,
low false-negative rate for LAVO) should be achieved, ideally
close to 90%. This was observed in different symptom com-
binations in our study with similar overall accuracy, namely,
all 3 FAST items positive or abnormal NIHSS item best gaze,
G-FAST3, C-STAT1, RACE3, and clinical signs of at
least a PACS (NIHSS symptom profiles A to E).
Second, we consider a patient with suspected stroke who is
evaluated by paramedics at the end of intravenous thromboly-
sis time window or with long transfer time to the nearest CSC.
In this case, a high specificity (ie, low false-positive rate for
LAVO) is warranted, such as >75%, resulting in <1 out of 4
futile transfers. This was observed for patients with abnormal
gaze and all 3 FAST items being positive (ie, G-FAST=4),
C-STAT3, 3I-SS=3, PASS=3, RACE6, and clinical signs of
TACS (NIHSS symptom profile A and B).
Third, we consider a patient who arrives at a primary
stroke center and is evaluated by trained stroke physicians.
In this case, full examination of the NIHSS is feasible. It
has been shown that no single variable beyond the NIHSS is
able to improve prediction of LAVO in a clinically meaning-
ful way.10 Current recommendations by the European Stroke
Organisation are based on the statistically optimal NIHSS cut
point observed in the large Bernese stroke registry (NIHSS9
within 3 hours, NIHSS7 within 6 hours).8,11 Our findings sug-
gest that lower NIHSS cutoffs could be used to improve sensi-
tivity (>90% with NIHSS6 and >95% with NIHSS5). Yet,
there are certain constellations in which application of highly
specific LAVO scores or the highly specific NIHSS symptom
profiles A or B could be helpful. Among others, these are late
arrival close to 6 hours, relative contraindications to computed
tomography-angiography like severely impaired kidney func-
tion or uncontrolled hyperthyroidism, and to avoid expenses
of screening failures in a randomized controlled trial.
Although all NIHSS items contribute equally to the sum
score, certain items and item constellations may reflect larger
ischemic lesions that carry a high attributable risk of an under-
lying LAVO. Not surprisingly, we observed a graded associa-
tion of the NIHSS item profiles with LAVO. The 2 symptom
profiles with the highest risks, profiles A and B, represent left
and right total hemispheric syndrome, respectively. Thus,
our findings suggest that patterns of deficit rather than sim-
ply scores reflecting severity of deficit will be more useful in
triage. Although the exact concept of the NIHSS symptom
profiles may be difficult to conduct by paramedics, our find-
ings support the notion that suspected stroke patients who
presented with NIHSS symptom profiles A or B, at any sever-
ity, should prompt an urgent neurovascular imaging and con-
sideration for transfer to a dedicated stroke center with EVT
capability. Given that right-hemispheric symptoms are under-
represented in the NIHSS, patients with right LAVO might be
missed in case of mild to moderate stroke severity based on
NIHSS scoring alone.
Our study has limitations. While the overall extent and accu-
racy of data collected within SITS-ISTR allow for statistically
robust analyses, the retrospective and observational design
inherits potential for bias. Our cohort consists of patients who
received revascularization treatments after a clinical diagnosis
of acute ischemic stroke was already established and hemor-
rhagic stroke was ruled out by brain imaging. Consequently,
sensitivity and specificity of the simplified NIHSS scores for
LAVO might differ in prehospital cohorts with suspected stroke
that include stroke mimics and hemorrhagic strokes. Majority
of data were derived from primary stroke centers with limited
availability of vessel imaging compared with CSCs. Moreover,
data on LAVO status were obtained from assessment by local
radiologists at the respective centers (not necessarily neuro-
radiologists). It is reassuring that 96% of patients within the
ESCAPE trial (Endovascular Treatment for Small Core and
Anterior Circulation Proximal Occlusion With Emphasis on
 296StrokeFebruary 2017
Minimizing CT to Recanalization Times) had the correct tar-
get-vessel occlusion status by using similar LAVO definition as
per our analysis, after review from central adjudication labora-
tory.2 In addition, it is possible that residual deficits related to
prior stroke or other reasons may affect baseline NIHSS score,
consequently leading to incorrect ratings. We have accounted
for this by adjusting the analysis for prestroke modified Rankin
Scale and previous stroke. Finally, we were not able to evalu-
ate other established stroke recognition tools (eg, Los Angeles
Motor Scale [LAMS], Los Angeles Prehospital Stroke Screen
[LAPSS], Melbourne Ambulance Stroke Screen [MASS],
Recognition of Stroke in the Emergency Room [ROSIER]),32,33
because grip strength was not part of the NIHSS recording,
and some scores require additional information other than the
NIHSS score (eg, history of seizures).
In summary, we found that the common simplified NIHSS
scores may be useful to stratify patients risk of LAVO in the
prehospital setting. Certain cutoffs seem sufficiently accurate to
Downloaded from http://stroke.ahajournals.org/ by guest on January 26, 2017
give guidance in clinical settings that require either high sensitiv-
ity (>85%) or high specificity (>75%). In general, patients with
abnormal findings on all 3 FAST items, and especially patients
with additional gaze deviation (G-FAST) may be considered for
urgent neurovascular imaging and transfer to CSC. This subset
of patients may be readily identifiable by paramedics during the
prehospital stage. Our findings deserve prospective validation,
ideally in the prehospital setting. The upcoming specialized
stroke ambulances seem to be one of the promising settings to
validate our findings and the feasibility of triage tools.
Sources of Funding
Dr Scheitz is a participant in the Charit Clinical Scientist Program
funded by the CharitUniversittsmedizin Berlin and the Berlin
Institute of Health. The research was done with support from European
Academy of Neurology via a Research Fellowship grant awarded to
Dr Scheitz. Dr Endres receives funding from the DFG (Excellence clus-
ter NeuroCure; SFB TR43, KFO 247, KFO 213), Bundesministerium
fr Bildung und Forschung (BMBF, Center for Stroke Research
Berlin), European Union (European Stroke Network, Wake-Up,
Counterstroke), and Volkswagen Foundation (Lichtenberg Program).
Disclosures
None.
References
1. Berkhemer OA, Fransen PS, Beumer D, van den Berg LA, Lingsma HF,
Yoo AJ, et al; MR CLEAN Investigators. A randomized trial of intraarte-
rial treatment for acute ischemic stroke. N Engl J Med. 2015;372:1120.
doi: 10.1056/NEJMoa1411587.
2. Goyal M, Demchuk AM, Menon BK, Eesa M, Rempel JL, Thornton J, et
al; ESCAPE Trial Investigators. Randomized assessment of rapid endo-
vascular treatment of ischemic stroke. N Engl J Med. 2015;372:1019
1030. doi: 10.1056/NEJMoa1414905.
3. Campbell BC, Mitchell PJ, Kleinig TJ, Dewey HM, Churilov L, Yassi
N, et al; EXTEND-IA Investigators. Endovascular therapy for ischemic
stroke with perfusion-imaging selection. N Engl J Med. 2015;372:1009
1018. doi: 10.1056/NEJMoa1414792.
4. Saver JL, Goyal M, Bonafe A, Diener HC, Levy EI, Pereira VM, et al;
SWIFT PRIME Investigators. Stent-retriever thrombectomy after intra-
venous t-PA vs. t-PA alone in stroke. N Engl J Med. 2015;372:2285
2295. doi: 10.1056/NEJMoa1415061.
5. Jovin TG, Chamorro A, Cobo E, de Miquel MA, Molina CA, Rovira
A, et al; REVASCAT Trial Investigators. Thrombectomy within 8 hours
after symptom onset in ischemic stroke. N Engl J Med. 2015;372:2296
2306. doi: 10.1056/NEJMoa1503780.
6. Emberson J, Lees KR, Lyden P, Blackwell L, Albers G, Bluhmki E, et al;
Stroke Thrombolysis Trialists Collaborative Group. Effect of treatment
delay, age, and stroke severity on the effects of intravenous thrombolysis
with alteplase for acute ischaemic stroke: a meta-analysis of individual
patient data from randomised trials. Lancet. 2014;384:19291935. doi:
10.1016/S0140-6736(14)60584-5.
7. Sheth SA, Jahan R, Gralla J, Pereira VM, Nogueira RG, Levy EI, et al;
SWIFT-STAR Trialists. Time to endovascular reperfusion and degree of
disability in acute stroke. Ann Neurol. 2015;78:584593. doi: 10.1002/
ana.24474.
8. Heldner MR, Zubler C, Mattle HP, Schroth G, Weck A, Mono ML, et al.
National Institutes of Health stroke scale score and vessel occlusion in
2152 patients with acute ischemic stroke. Stroke. 2013;44:11531157.
doi: 10.1161/STROKEAHA.111.000604.
9. Cooray C, Fekete K, Mikulik R, Lees KR, Wahlgren N, Ahmed N.
Threshold for NIH stroke scale in predicting vessel occlusion and func-
tional outcome after stroke thrombolysis. Int J Stroke. 2015;10:822829.
doi: 10.1111/ijs.12451.
10. Vanacker P, Heldner MR, Amiguet M, Faouzi M, Cras P, Ntaios G, et al.
Prediction of large vessel occlusions in acute stroke: National Institute of
Health Stroke Scale is hard to beat. Crit Care Med. 2016;44:e336e343.
doi: 10.1097/CCM.0000000000001630.
11. Wahlgren N, Moreira T, Michel P, Steiner T, Jansen O, Cognard C, et
al; ESO-KSU, ESO, ESMINT, ESNR and EAN. Mechanical thrombec-
tomy in acute ischemic stroke: consensus statement by ESO-Karolinska
Stroke Update 2014/2015, supported by ESO, ESMINT, ESNR and
EAN. Int J Stroke. 2016;11:134147. doi: 10.1177/1747493015609778.
12. Powers WJ, Derdeyn CP, Biller J, Coffey CS, Hoh BL, Jauch EC, et
al; American Heart Association Stroke Council. 2015 American Heart
Association/American Stroke Association Focused Update of the 2013
Guidelines for the Early Management of Patients With Acute Ischemic Stroke
Regarding Endovascular Treatment: A Guideline for Healthcare Professionals
From the American Heart Association/American Stroke Association. Stroke.
2015;46:30203035. doi: 10.1161/STR.0000000000000074.
13. Sucharew H, Khoury J, Moomaw CJ, Alwell K, Kissela BM, Belagaje S,
et al. Profiles of the National Institutes of Health Stroke Scale items as a
predictor of patient outcome. Stroke. 2013;44:21822187. doi: 10.1161/
STROKEAHA.113.001255.
14. Abdul-Rahim AH, Fulton RL, Sucharew H, Kleindorfer D, Khatri P,
Broderick JP, et al; VISTA Collaborators. National institutes of health
stroke scale item profiles as predictor of patient outcome: external vali-
dation on independent trial data. Stroke. 2015;46:395400. doi: 10.1161/
STROKEAHA.114.006837.
15. Abdul-Rahim AH, Fulton RL, Sucharew H, Kleindorfer D, Khatri P,
Broderick JP, et al; SITS-MOST Steering Committee. National Institutes
of Health Stroke Scale item profiles as predictor of patient outcome:
external validation on Safe Implementation of Thrombolysis in Stroke-
Monitoring Study data. Stroke. 2015;46:27792785. doi: 10.1161/
STROKEAHA.115.010380.
16. Mazya MV, Lees KR, Collas D, Rand VM, Mikulik R, Toni D, et al.
IV thrombolysis in very severe and severe ischemic stroke: results from
the SITS-ISTR Registry. Neurology. 2015;85:20982106. doi: 10.1212/
WNL.0000000000002199.
17. Mazya MV, Ahmed N, Ford GA, Hobohm C, Mikulik R, Nunes AP, et
al; Scientific Committee of SITS International. Remote or extraischemic
intracerebral hemorrhagean uncommon complication of stroke throm-
bolysis: results from the safe implementation of treatments in stroke-
international stroke thrombolysis register. Stroke. 2014;45:16571663.
doi: 10.1161/STROKEAHA.114.004923.
18. Demchuk AM, Goyal M, Menon BK, Eesa M, Ryckborst KJ, Kamal
N, et al; ESCAPE Trial Investigators. Endovascular Treatment for Small
Core and Anterior Circulation Proximal Occlusion With Emphasis on
Minimizing CT to Recanalization Times (ESCAPE) trial: methodology.
Int J Stroke. 2015;10:429438. doi: 10.1111/ijs.12424.
19. Saver JL, Goyal M, Bonafe A, Diener HC, Levy EI, Pereira VM, et
al; SWIFT PRIME Investigators. Solitaire With the Intention for
Thrombectomy as Primary Endovascular Treatment for Acute Ischemic
Stroke (SWIFT PRIME) trial: protocol for a randomized, controlled,
multicenter study comparing the Solitaire revascularization device
with IV tPA with IV tPA alone in acute ischemic stroke. Int J Stroke.
2015;10:439448. doi: 10.1111/ijs.12459.
20. Scheitz JF, Erdur H, Ttnc S, Fiebach JB, Audebert HJ, Endres M, et
al. National Institutes of Health Stroke Scale for prediction of proximal
vessel occlusion in anterior circulation stroke. Int J Stroke. 2015;10:E60.
doi: 10.1111/ijs.12556.
 Scheitz et al Prediction of Large Anterior Vessel Occlusion 297
21. Fischer U, Arnold M, Nedeltchev K, Brekenfeld C, Ballinari P,
Remonda L, et al. NIHSS score and arteriographic findings in acute
ischemic stroke. Stroke. 2005;36:21212125. doi: 10.1161/01.
STR.0000182099.04994.fc.
22. Turc G, Maer B, Naggara O, Seners P, Isabel C, Tisserand M, et al.
Clinical scales do not reliably identify acute ischemic stroke patients
with large-artery occlusion. Stroke. 2016;47:14661472. doi: 10.1161/
STROKEAHA.116.013144.
23. Heldner MR, Jung S, Zubler C, Mordasini P, Weck A, Mono ML, et al.
Outcome of patients with occlusions of the internal carotid artery or the
main stem of the middle cerebral artery with NIHSS score of less than
5: comparison between thrombolysed and non-thrombolysed patients.
J Neurol Neurosurg Psychiatry. 2015;86:755760. doi: 10.1136/
jnnp-2014-308401.
24. Singer OC, Dvorak F, du Mesnil de Rochemont R, Lanfermann H, Sitzer
M, Neumann-Haefelin T. A simple 3-item stroke scale: comparison with
the National Institutes of Health Stroke Scale and prediction of middle
cerebral artery occlusion. Stroke. 2005;36:773776. doi: 10.1161/01.
STR.0000157591.61322.df.
25. Perez de la Ossa N, Carrera D, Gorchs M, Querol M, Millan M, Gomis
M, et al. Design and validation of a prehospital stroke scale to predict
large arterial occlusion: the rapid arterial occlusion evaluation scale.
Stroke. 2014;45:8791.
Downloaded from http://stroke.ahajournals.org/ by guest on January 26, 2017
26. Katz BS, McMullan JT, Sucharew H, Adeoye O, Broderick JP. Design
and validation of a prehospital scale to predict stroke severity: Cincinnati
Prehospital Stroke Severity Scale. Stroke. 2015;46:15081512. doi:
10.1161/STROKEAHA.115.008804.
27. Hastrup S, Damgaard D, Johnsen SP, Andersen G. Prehospital acute
stroke severity scale to predict large artery occlusion: design and com-
parison with other scales. Stroke. 2016;47:17721776. doi: 10.1161/
STROKEAHA.115.012482.
28. Singer OC, Humpich MC, Laufs H, Lanfermann H, Steinmetz H,
Neumann-Haefelin T. Conjugate eye deviation in acute stroke: incidence,
hemispheric asymmetry, and lesion pattern. Stroke. 2006;37:27262732.
doi: 10.1161/01.STR.0000244809.67376.10.
29. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas
under two or more correlated receiver operating characteristic curves: a
nonparametric approach. Biometrics. 1988;44:837845.
30. Youden WJ. Index for rating diagnostic tests. Cancer. 1950;3:3235.
31. Kleindorfer DO, Miller R, Moomaw CJ, Alwell K, Broderick JP, Khoury
J, et al. Designing a message for public education regarding stroke: does
FAST capture enough stroke? Stroke. 2007;38:28642868. doi: 10.1161/
STROKEAHA.107.484329.
32. Purrucker JC, Hametner C, Engelbrecht A, Bruckner T, Popp E, Poli
S. Comparison of stroke recognition and stroke severity scores for
stroke detection in a single cohort. J Neurol Neurosurg Psychiatry.
2015;86:10211028. doi: 10.1136/jnnp-2014-309260.
33. Brandler ES, Sharma M, Sinert RH, Levine SR. Prehospital stroke scales
in urban environments: a systematic review. Neurology. 2014;82:2241
2249. doi: 10.1212/WNL.0000000000000523.
34. Ali SF, Viswanathan A, Singhal AB, Rost NS, Forducey PG, Davis LW, et
al; Partners Telestroke Network. The TeleStroke mimic -score: a predic-
tion rule for identifying stroke mimics evaluated in a Telestroke Network.
J Am Heart Assoc. 2014;3:e000838. doi: 10.1161/JAHA.114.000838.
 Clinical Selection Strategies to Identify Ischemic Stroke Patients With Large Anterior
Vessel Occlusion: Results From SITS-ISTR (Safe Implementation of Thrombolysis in
Stroke International Stroke Thrombolysis Registry)
Jan F. Scheitz, Azmil H. Abdul-Rahim, Rachael L. MacIsaac, Charith Cooray, Heidi Sucharew,
Dawn Kleindorfer, Pooja Khatri, Joseph P. Broderick, Heinrich J. Audebert, Niaz Ahmed, Nils
Downloaded from http://stroke.ahajournals.org/ by guest on January 26, 2017

Wahlgren, Matthias Endres, Christian H. Nolte and Kennedy R. Lees

on behalf of SITS Scientific Committee

Stroke. 2017;48:290-297; originally published online January 13, 2017;

doi: 10.1161/STROKEAHA.116.014431
Stroke is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
Copyright 2017 American Heart Association, Inc. All rights reserved.
Print ISSN: 0039-2499. Online ISSN: 1524-4628

The online version of this article, along with updated information and services, is located on the
World Wide Web at:
http://stroke.ahajournals.org/content/48/2/290

Data Supplement (unedited) at:

http://stroke.ahajournals.org/content/suppl/2017/01/13/STROKEAHA.116.014431.DC1.html

Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published
in Stroke can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office.
Once the online version of the published article for which permission is being requested is located, click
Request Permissions in the middle column of the Web page under Services. Further information about this
process is available in the Permissions and Rights Question and Answer document.

Reprints: Information about reprints can be found online at:

http://www.lww.com/reprints

Subscriptions: Information about subscribing to Stroke is online at:

http://stroke.ahajournals.org//subscriptions/
SUPPLEMENTAL MATERIAL

Clinical selection strategies to identify stroke patients with large anterior vessel occlusion-Results from SITS-ISTR

Jan F Scheitz1,2*,MD, Azmil H Abdul-Rahim3*, MBChB, MSc(StrokeMed), Rachael L MacIsaac3,PhD, Charith Cooray4,MD, Heidi Sucharew5, PhD, Dawn Kleindorfer6,
MD, Pooja Khatri6, MD, MSc, Joseph P Broderick6, MD, Heinrich J. Audebert1,2, MD, Niaz Ahmed4, MD, PhD, Nils Wahlgren4,MD, PhD, Matthias Endres1,2,7,8,9,MD,
Christian H Nolte1,2*,MD, Kennedy R Lees3*, MD, FRCP

On behalf of SITS Scientific Committee

1 Center for Stroke Research Berlin, Charit - Universittsmedizin Berlin, Berlin, Germany.
2 Klinik fr Neurologie, Charit- Universittsmedizin Berlin, Campus Benjamin Franklin, Berlin, Germany.
3 Stroke Research, Institute of Cardiovascular & Medical Sciences, University of Glasgow, Glasgow G12 8QQ, UK
4 Department of Clinical Neurosciences, Karolinska Institutet and Department of Neurology, Karolinska University Hospital, Solna, Sweden
5 Division of Biostatistics and Epidemiology, Cincinnati Childrens Hospital Medical Center, Cincinnati, OH, USA
6 Department of Emergency Medicine, University of Cincinnati, College of Medicine, 231 Albert Sabin Way, Cincinnati, OH 45267
7 Excellence Cluster NeuroCure, Charit Universittsmedizin Berlin, Berlin, Germany
8 German Center for Cardiovascular Research (DZHK), Charit - Universittsmedizin Berlin, Berlin, Germany.
9 German Center for Neurodegenerative Diseases (DZNE) Charit - Universittsmedizin Berlin, Berlin, Germany.
* authors contributed equally
Table I Components of the evaluated simple NIHSS scores
NIHSS Points FAST Points C-STAT Points PASS Points 3I-SS* Points RACE Points

LOC 0-3 - LOC questions AND LOC questions

commands (NIHSS>=1) Level of consciousness
LOC Questions 0-2 - - 1 1 1 - -
(NIHSS >=1 each) (NIHSS >=1)
LOC commands 0-2 -

Gaze 0-2 - - NIHSS >=1 2 NIHSS >=1 1 NIHSS >=1 1 Gaze 1

Visual field 0-3 - - - - - - - - -

Facial droop -
Facial palsy 0-3 1 - - - - Facial palsy 1-2
(NIHSS >=1)
Arm drift (NIHSS Arm weakness Arm weakness
Motor Arm 0-4 1 1 1 Hemiparesis 1 Arm motor function 1-2
>=1) (NIHSS >=2) (NIHSS >=1)
(Arm AND Leg motor
function on NIHSS >=1)
Motor Leg 0-4 - - - - - Leg motor function 1-2

Limb ataxia 0-2 - - - - - - - - -

Sensory 0-2 - - - - - - - - -

Speech problems Aphasia (if right

Best language 0-3 1 - - - - - 1-2
(dysarthria or hemiparesis)
language
Dysarthria 0-2 abnormality >=1) - - - - - - - -

Extinction and Agnosia (if left

0-2 - - - - - - - 1-2
Inattention hemiparesis)

Total 42 Total 3 Total 4 Total 3 Total 3 Total 9

Abbreviations: 3I-SS - 3-item Stroke Scale, C-STAT - Cincinnati Stroke Triage Assessment Tool, FAST face arm speech time test, LAVO large anterior vessel occlusion, LOC level of consciousness, NIHSS - National
Institutes of Health Stroke Scale, PASS prehospital acute stroke severity scale, RACE - Rapid Arterial Occlusion Evaluation Scale.
* In the original version 2 points were assignable to each of the three items depending on the overall severity
Table II - Clinical description of NIHSS symptom profiles

NIHSS symptom Profile Description

profiles

Profile A Severe Left TACS Severe stroke with decreased level of consciousness, facial palsy, abnormal
motor function on the right side, language deficit and dysarthria
Profile B Severe Right TACS Severe stroke with some decreased level of consciousness, facial palsy, abnormal
motor function on the left side and, dysarthria
Profile C Moderate Left PACS Stroke with language deficit and signs of dysarthria, mild abnormal normal motor
function on the right side possible
Profile D Moderate Left PACS Facial palsy, abnormal motor function on the right side, and
dysarthria
Profile E Moderate Right PACS Facial palsy and abnormal motor function on the left side, and dysarthria
Profile F Mild - Non-specific syndrome with low probabilities of
abnormal findings on all 15 items
Table III Sensitivity, specificity, PPV, NPV and accuracy for presence of LAVO according to different NIHSS cut-offs
n/N (%) Sensitivity Specificity PPV NPV Accuracy

NIHSS >=1 3494/3505 (99.7) 100.0 0.4 23.7 100.0 23.9

NIHSS >=2 3472/3505 (99.1) 99.6 1.1 23.7 90.9 24.3

NIHSS >=3 3389/3505 (96.7) 99.4 4.1 24.3 95.7 26.6

NIHSS >=4 3230/3505 (92.2) 97.8 9.6 25.0 93.5 30.4

NIHSS >=5 2972/3505 (84.8) 95.4 18.5 26.5 92.9 36.6

NIHSS >=6 2677/3505 (76.4) 91.3 28.2 28.2 91.3 43.1

NIHSS >=7 2395/3505 (68.3) 89.1 38.1 30.8 91.9 50.1

NIHSS >=8 2183/3505 (62.3) 85.6 44.9 32.4 91.0 54.5

NIHSS >=9 1928/3505 (55.0) 82.1 53.4 35.2 90.6 60.2

NIHSS >=10 1744/3505 (49.8) 79.4 59.4 37.7 90.3 64.1

NIHSS >=11 1570/3505 (44.8) 75.8 64.8 39.9 89.7 67.4

NIHSS >=12* 1420/3505 (40.5) 72.1 69.2 42.0 88.9 69.9

NIHSS >=13 1258/3505 (35.9) 67.4 73.2 44.3 88.0 71.8

NIHSS >=14 1133/3505 (32.3) 63.1 77.2 46.1 87.1 73.9

NIHSS >=15 1021/3505 (29.1) 58.0 80.1 47.4 86.1 74.9

NIHSS 16-20 891/3505 (25.4) 52.7 83.0 48.9 85.0 75.8

NIHSS >20 356/3505 (10.2) 23.0 93.8 53.4 79.8 77.1

Abbreviation: NIHSS - National Institutes of Health Stroke Scale

Table IV - AUC for the association between the entire NIHSS, the simplified pre-hospital scales, and NIHSS item profiles with LAVO

Score AUC (95% CI) Entire cohort* AUC (95% CI) patients with moderate stroke severity

N=3505 N= 1257

NIHSS 0.751 (0.736-0.765) 0.601 (0.573-0.628)

3I-SS 0.707 (0.692-0.722) 0.601 (0.573-0.628)

C-STAT 0.732 (0.717-0.747) 0.618 (0.590-0.645)

FAST 0.645 (0.629-0.661) 0.593 (0.566-0.621)

G-FAST 0.722 (0.707-0.737) 0.628 (0.600-0.655)

PASS 0.714 (0.699-0.729) 0.599 (0.572-0.627)

RACE 0.755 (0.740-0.769) 0.617 (0.589-0.644)

NIHSS profiles 0.709 (0.694-0.724) 0.619 (0.591-0.646)

*AUC of entire NIHSS=RACE (p=0.43). AUC of NIHSS and RACE >C-STAT (p<0.01). AUC of C-STAT=G-FAST (p=0.10). AUC of NIHSS, RACE, C-STAT, G-FAST > 3I-SS.
AUC of NIHSS, RACE, C-STAT, G-FAST > PASS (p<0.01). AUC of PASS=3I-SS (p=0.34). AUC of all scores > FAST (p<0.01).
moderate stroke severity was defined as NIHSS 6-11. AUC of G-FAST>FAST (p<0.01), all other scores not significantly different
Abbreviations: 3I-SS - 3-item Stroke Scale, AUC area under the receiver operating characteristics curve, CI confidence interval, C-STAT - Cincinnati Stroke Triage Assessment
Tool, FAST face, arm, speech and time test, G-FAST Go FAST score, LAVO large anterior vessel occlusion, NIHSS - National Institutes of Health Stroke Scale, PASS
prehospital acute stroke severity scale, RACE - Rapid Arterial Occlusion Evaluation Scale.
Table V - AUC for the association between entire NIHSS, simplified NIHSS scores and NIHSS item profiles with LVO (basilar artery or LAVO)

Score AUC (95% CI) Entire cohort* AUC (95% CI) patients with moderate stroke severity

N=3505 N= 1257

NIHSS 0.747 (0.733-0.762) 0.607 (0.564-0.650)

3I-SS 0.713 (0.698-0.728) 0.624 (0.580-0.669)

C-STAT 0.731 (0.716-0.746) 0.637 (0.594-0.681)

FAST 0.632 (0.615-0.648) 0.575 (0.533-0.617)

G-FAST 0.713 (0.698-0.728) 0.635 (0.592-0.678)

PASS 0.708 (0.693-0.723) 0.597 (0.569-0.624)

RACE 0.741 (0.726-0.755) 0.597 (0.566-0.621)

NIHSS profiles 0.698 (0.682-0.713) 0.578 (0.531-0.625)

*AUC of entire NIHSS=RACE (p=0.15). AUC of NIHSS>C-STAT (p<0.01). AUC of C-STAT=RACE (p=0.10). AUC of C-STAT, NIHSS, RACE>G-FAST (p<0.01), AUC of C-
STAT, NIHSS, RACE, G-FAST >3I-SS (p<0.01). AUC of C-STAT, NIHSS, RACE, G-FAST >PASS (p<0.01). AUC of all scores >FAST (p<0.01).

moderate stroke severity was defined as NIHSS 6-11. AUC of G-FAST, 3I-SS and C-STAT>FAST (p<0.05), AUC of C-STAT >RACE (p<0.05), all other scores not
significantly different
Abbreviations: 3I-SS - 3-item Stroke Scale, C-STAT - Cincinnati Stroke Triage Assessment Tool, FAST face, arm, speech and time test, G-FAST Go FAST score, LAVO large

anterior vessel occlusion, NIHSS - National Institutes of Health Stroke Scale, PASS prehospital acute stroke severity scale, RACE - Rapid Arterial Occlusion Evaluation Scale.
Table VI Sensitivity, specificity, PPV, and NPV for presence of large vessel occlusion (including BAO) at different cut-offs of the NIHSS, simplified

NIHSS scores and NIHSS symptom profiles.

Cut-offs of scores with >85% sensitivity and highest possible specificity.

n/N (%) Sensitivity Specificity PPV NPV Accuracy

NIHSS >=7 2395/3505 (66.3) 88.0 38.7 33.8 90.1 51.6

FAST >=2 3013/3505 (86.0) 96.0 17.6 29.3 92.5 38.1

G-FAST>=3 2363/3505 (67.5) 87.5 39.7 34.0 89.9 52.2

C-STAT>=1 2414/3505 (68.9) 88.0 40.0 33.6 89.9 55.5

3I-SS>=1 2702/3505 (77.1) 91.5 28.0 31.2 90.3 44.6

PASS >=1 3222/3505 (91.9) 96.4 10.2 27.6 88.9 32.8

RACE >=2 2724/3505 (77.7) 92.2 27.4 31.1 90.8 44.4

NIHSS profile A-E (at least PACS or 3155/3505 (90.0) 95.6 12.0 27.9 88.3 33.9
worse) vs profile F

Cut-offs of scores with specificity >75% and highest possible sensitivity

NIHSS >=14 1133/3505 (32.3) 60.7 77.8 49.2 84.7 73.3

G-FAST=4 1029/3505 (29.4) 54.3 79.5 48.5 83.0 72.9

C-STAT>=3 1051/3505 (30.0) 56.2 79.3 49.2 83.6 73.2

3I-SS =3 383/3505 (10.9) 21.4 92.8 51.4 76.8 74.1

PASS =3 565/3505 (16.1) 37.3 86.6 49.8 79.5 73.7

RACE >=6 1154/3505 (32.9) 59.9 76.7 47.7 84.3 72.3

NIHSS profile A 793/3505 (22.6) 38.3 82.9 44.3 79.1 71.2

NIHSS profile B 775/3505 (22.1) 34.9 82.4 41.4 78.1 70.0

Statistical optimal cut-offs

NIHSS >=12 1420/3505 (40.5) 69.5 69.8 45.0 86.5 69.7

FAST=3 2207/3505 (63.0) 81.2 43.5 33.8 86.7 53.4

G-FAST=4 1029/3505 (29.4) 54.3 79.5 48.5 83.0 72.9

C-STAT >=2 1461/3505 (41.7) 71.1 68.8 44.8 87.0 69.4

3I-SS >=2 1283/3505 (36.6) 64.6 73.4 46.3 85.3 71.1

PASS >=2 1802/3505 (51.4) 77.3 59.3 40.3 88.0 64.0

RACE >=5 1442/3505 (41.1) 68.9 68.8 44.0 86.1 68.9

NIHSS symptom profile A or B 1568/3505 (44.7) 73.0 65.3 44.9 87.2 67.3
Table VII Sensitivity, specificity, PPV, and NPV for presence of LAVO at different cut-offs of the NIHSS, simplified NIHSS scores and NIHSS symptom

profiles in patients with moderate stroke severity (NIHSS 6-11)

Cut-offs of scores with >85% sensitivity and highest possible specificity

n/N (%) Sensitivity Specificity PPV NPV Accuracy

NIHSS >=7 975/1257 (77.6) 88.7 24.0 14.5 93.6 32.1

FAST >=2 1115/1257 (88.7) 95.0 12.2 13.5 94.4 22.6

G-FAST>=2 1134/1257 (90.2) 96.2 10.7 13.5 95.1 21.5

3I-SS>=1 964/1257 (76.7) 85.5 24.6 14.1 92.2 32.3

PASS>=1 1228/1257 (97.7) 99.4 2.6 12.9 96.7 14.8

RACE >=2 1038/1257 (82.6) 91.8 18.8 14.1 94.1 28.0

NIHSS profile A-E (at least PACS or worse) vs profile F 1201/1257 (95.5) 98.7 4.9 13.1 96.4 16.7

Cut-offs of scores with specificity >75% and highest possible sensitivity

NIHSS >=10 324/1257 (25.8) 38.4 76.0 18.8 89.5 71.3

G-FAST=4 143/1257 (11.4) 24.5 90.0 27.4 89.2 81.7

C-STAT >=2 116/1257 (9.2) 32.7 82.1 20.9 89.4 75.9

3I-SS >=2 219/1257 (17.4) 30.8 84.5 22.4 89.4 77.7

PASS=3 52/1257 (4.1) 5.7 96.1 17.3 87.6 84.7

RACE >=5 73/1257 (5.8) 25.8 84.5 19.4 88.7 77.1

NIHSS profile A or B 316/1257 (25.1) 45.3 77.8 22.8 90.8 73.7