Validation of the DRAGON Score in 12 Stroke Centers in Anterior and Posterior Circulation

Validation of the DRAGON Score in 12 Stroke Centers
                in Anterior and Posterior Circulation
                   Daniel Strbian, MD, PhD; David J. Seiffge, MD; Lorenz Breuer, MD;
             Heikki Numminen, MD, PhD; Patrik Michel, MD; Atte Meretoja, MD, PhD, MSc;
           Skye Coote, CCRN; Régis Bordet, MD, PhD; Victor Obach, MD, Bruno Weder, MD;
         Simon Jung, MD; Valeria Caso, MD, PhD; Sami Curtze, MD, PhD; Jyrki Ollikainen, MD;
               Philippe A. Lyrer, MD; Ashraf Eskandari, RN; Heinrich P. Mattle, MD, PhD;
            Angel Chamorro, MD, PhD; Didier Leys, MD, PhD; Christopher Bladin, MD, PhD;
         Stephen M. Davis, MD, PhD; Martin Köhrmann, MD, PhD; Stefan T. Engelter, MD, PhD;
                                      Turgut Tatlisumak, MD, PhD

Background and Purpose—The DRAGON score predicts functional outcome in the hyperacute phase of intravenous
  thrombolysis treatment of ischemic stroke patients. We aimed to validate the score in a large multicenter cohort in anterior
  and posterior circulation.
Methods—Prospectively collected data of consecutive ischemic stroke patients who received intravenous thrombolysis in
  12 stroke centers were merged (n=5471). We excluded patients lacking data necessary to calculate the score and patients
  with missing 3-month modified Rankin scale scores. The final cohort comprised 4519 eligible patients. We assessed the
  performance of the DRAGON score with area under the receiver operating characteristic curve in the whole cohort for
  both good (modified Rankin scale score, 0–2) and miserable (modified Rankin scale score, 5–6) outcomes.
Results—Area under the receiver operating characteristic curve was 0.84 (0.82–0.85) for miserable outcome and 0.82 (0.80–
  0.83) for good outcome. Proportions of patients with good outcome were 96%, 93%, 78%, and 0% for 0 to 1, 2, 3, and 8
  to 10 score points, respectively. Proportions of patients with miserable outcome were 0%, 2%, 4%, 89%, and 97% for 0
  to 1, 2, 3, 8, and 9 to 10 points, respectively. When tested separately for anterior and posterior circulation, there was no
  difference in performance (P=0.55); areas under the receiver operating characteristic curve were 0.84 (0.83–0.86) and
  0.82 (0.78–0.87), respectively. No sex-related difference in performance was observed (P=0.25).
Conclusions—The DRAGON score showed very good performance in the large merged cohort in both anterior and
  posterior circulation strokes. The DRAGON score provides rapid estimation of patient prognosis and supports clinical
  decision-making in the hyperacute phase of stroke care (eg, when invasive add-on strategies are considered).   (Stroke.
                              Key Words: ischemic stroke ◼ outcome ◼ prognosis ◼ thrombolysis

I  ntravenous thrombolysis (IVT) with alteplase is the only
   approved clot-busting medical therapy in acute ischemic
stroke, yet approximately half of the patients treated achieve
                                                                               [mRS] score, 0–2).2,3 For patients who do not benefit from
                                                                               thrombolysis, alternative treatment strategies may be con-
                                                                               sidered. In such cases, it is of utmost importance to predict
recanalization in acute setting.1 Regarding functional benefit,                the benefit of IVT as soon as possible after patient admis-
a similar percentage of patients achieve good outcome (inde-                   sion, because its efficacy decreases substantially with increas-
pendence in activities of daily life; modified Rankin scale                    ing onset-to-treatment time.3 We recently described such a

  Received May 3, 2013; final revision received June 20, 2013; accepted July 8, 2013.
  From the Department of Neurology and Stroke Unit, Helsinki University Central Hospital, Helsinki, Finland (D.S., A.M., S.C., T.T.); Neurology and
Stroke Center, University Hospital Basel, Basel, Switzerland (D.J.S., P.A.L., S.T.E.); Department of Neurology, Universitätsklinikum Erlangen, Erlangen,
Germany (L.B., M.K.); Department of Neurology, Tampere University Hospital, Tampere, Finland (H.N., J.O.); Department of Neurology, Centre
Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne, Switzerland (P.M., A.E.); Department of Neurology, The Royal Melbourne
Hospital, Parkville, Australia (A.M., C.B., S.M.D.); Department of Neurology, Box Hill Hospital, Box Hill, Australia (S.C., C.B.); Department of
Neurology, University Lille Nord de France, Lille, France (R.B., D.L.); Department of Neurology, Hospital Clínic Institut d’Investigacions Biomediques
August Pi i Sunyer, Barcelona, Spain (V.O., A.C.); Department of Neurology, Kantonsspital St. Gallen, St. Gallen, Switzerland (B.W.); Department of
Neurology, University of Bern, Bern, Switzerland (S.J., H.P.M.); and Department of Neurology, University of Perugia, Perugia, Italy (V.C.).
  Guest Editor for this article was Emmanuel Touzé, PhD.
  The online-only Data Supplement is available with this article at
  Correspondence to Daniel Strbian, MD, PhD, Department of Neurology, Helsinki University Central Hospital, Haartmaninkatu 4, 00290 Helsinki,
Finland. E-mail
  © 2013 American Heart Association, Inc.
  Stroke is available at                                                       DOI: 10.1161/STROKEAHA.113.002033

                                  Downloaded from by guest on July 30, 2015
Strbian et al   Validation of the DRAGON Score   2719

Table 1.     The DRAGON Score                                                       centers were compiled in the coordinating center (Helsinki), where the
                                                                                    analyses of the pooled data (n=5471) were performed. We only con-
Parameter                                           Category           Points       sidered patients with information available for all necessary baseline
(hyper)Dense cerebral artery sign or early            None               0          parameters, some of which were missing for 820 patients (footnote
infarct signs on admission CT head scan                                             of Table 2), leaving 4519 eligible patients. None of the patients from
                                                 Either of them          1          the original derivation cohort4 was included in the current analysis.
                                                      Both               2          The merged validation cohort comprises patients with both anterior
Modified Rankin scale score >1, prestroke              No                0          and posterior circulation strokes. CT scans were evaluated by the local
                                                                                    investigators, who were blinded to the results of CT angiography (if
                                                      Yes                1          performed). Early infarct signs refer to an observation of any of the
Age                                                  8 mmol/L (144 mg/dL)       1          original publication.4 Discrimination statistics (AUC-ROC) tested the
Onset-to-treatment time                             ≤90 min              0
                                                                                    performance of the score for good (mRS score, 0–2) and miserable
                                                                                    (mRS score, 5–6) outcomes. Two prespecified subgroup analyses were
                                                    >90 min              1          performed: AUC-ROC in anterior versus posterior circulation strokes
NIHSS on admission                                    0–4                0          and in women versus men. Positive and negative likelihood ratios, pos-
                                                                                    itive and negative predictive values, and their 95% confidence intervals
                                                      5–9                1
                                                                                    were calculated for different cut-off values of the score. Calibration of
                                                     10–15               2          the model was tested with the Hosmer–Lemeshow test. Analyses were
                                                      >15                3          performed on IBM SPSS 18 (IBM Corp, Armonk, NY).
   CT indicates computed tomography; and NIHSS, National Institutes of Health
Stroke Scale.                                                                                                         Results
                                                                                    The demographics and baseline characteristics of patients
prognostic tool, the DRAGON score, that was derived in the                          included in the merged cohort (n=4519) are presented in
Helsinki cohort (n=1319) and validated in the Basel cohort
(n=333).4 It consists of parameters that are available soon                         Table 2. Demographics, Baseline Characteristics, and
                                                                                    3-Month Outcome
after patient admission—before IVT administration (Table 1).
The score was developed for ultraearly estimation of patient                        Parameter                                                     N=4519
prognosis. It reliably identifies patients who will very likely                     Age, median (IQR), y                                        72 (61–80)
benefit from IVT but also those with high likelihood of mis-                        Women                                                     1951 (43.2%)
erable outcome, despite IVT. In the latter group, an invasive
                                                                                    Onset-to-treatment time, median (IQR), min                 141 (105–180)
endovascular procedure5,6 can be an alternative treatment of
                                                                                    Baseline NIHSS score, median (IQR), points                  10 (6–16)
choice, although currently that is not evidence-based. Such
an approach is an alternative in some centers in situations in                      Hyperdense cerebral artery sign                           1075 (23.8%)
which severe symptoms persist, despite thrombolytic treat-                          Early infarct signs                                       1187 (26.3%)
ment, as a rescue strategy.7,8                                                      Prestroke mRS score >1                                     278 (6.2%)
   The DRAGON score showed very good performance in the                             BP and glucose level
original derivation and validation cohorts: area under the receiver                  Systolic BP before, median (IQR), mm Hg                  156 (140–172)
operating characteristic curve (AUC-ROC) 0.84 (0.80–0.87)                            Diastolic BP before, median (IQR), mm Hg                  85 (75–95)
and 0.80 (0.74–0.86), respectively.4 Furthermore, the DRAGON
                                                                                     Glucose, median (IQR), mmol/L                             6.6 (5.8–8.0)
score was recently adapted for patients undergoing MRI instead
                                                                                    Medical history, n (%)
of CT as the first-line imaging modality,9 reporting performance
similar to that in the original publication. In this study, we                       Hypertension                                            2775 (61.4)
validated the DRAGON score in a large multicenter cohort to                          Diabetes mellitus                                        914 (20.2)
address its generalizability. We also tested the score separately in                 Previous stroke                                          565 (12.5)
anterior and posterior circulation and in women and men.                             Hyperlipidemia                                          1487 (32.9)
                                                                                     Atrial fibrillation                                     1094 (24.2)
                     Patients and Methods                                           Proportions of 3-month outcome, mRS, n (%)
Study Setting                                                                        mRS, 0–2                                                2543 (56.3)
The study includes data from 12 European and Australian stroke cen-                  mRS, 3–4                                                1094 (24.2)
ters and was approved by the relevant authorities in each participating              mRS, 5–6                                                 882 (19.5)
center, if required. This study was approved in the coordinating center
(Helsinki) as a quality registry and did not require ethical board re-                BP indicates blood pressure; IQR, interquartile range; mRS, modified Rankin
view. All patients were prospectively included in the database. Data                scale; and NIHSS, National Institutes of Health Stroke Scale. Data are presented
from individual consecutive patients receiving IVT for acute ischemic               as median (IQR) or percentage. Age and baseline NIHSS scores of the 952
stroke (without subsequent endovascular procedure) were collected                   excluded patients were 70 (19) and 9 (9), respectively. Reasons for exclusion
with a standardized form with predefined variables. Local study inves-              were ≥1 of the following missing data: hyperdense cerebral artery sign (n=475),
tigators completed the forms systematically using prospectively ascer-              early infarct signs (n=464), prestroke mRS (n=201), baseline glucose (n=231),
tained in-hospital thrombolysis registries. Completed forms from all                onset-to-treatment time (n=106), and baseline NIHSS (n=64).

                                      Downloaded from by guest on July 30, 2015
2720  Stroke  October 2013

Table 2. Numbers of included patients per center are as fol-              score 0 to 2, positive likelihood ratio was 8.3 (6.6–10.5), neg-
lows: Barcelona (n=226), Basel (n=694), Bern (n=79), Box                  ative likelihood ratio was 0.7 (0.7–0.7), positive predictive
Hill (n=455), Erlangen (n=688), Helsinki (n=467), Lausanne                value was 91.5% (89.4–93.3), negative predictive value was
(n=463), Lille (n=277), Melbourne (n=336), Perugia (n=58),                51.8% (50.2–53.4), specificity was 96.4 (95.4–97.1), and sen-
St. Gallen (n=153), and Tampere (n=623). Proportions of                   sitivity was 30.3 (28.5–32.1).
patients per each cut-off of each parameter of the score are out-
lined in Table I in the online-only Data Supplement together                                       Discussion
with adjusted odds ratios for miserable and good outcomes per             In a large multicenter cohort of 4519 patients, we successfully
each parameter of the score.                                              validated the DRAGON score, which accurately predicts out-
   Distributions of 3-month outcome per increasing points of              come of IVT-treated ischemic stroke patients. Performance of
the DRAGON score are outlined in Figure 1. The Hosmer–                    the score was similarly good (AUC-ROC=0.85) in the origi-
Lemeshow test showed good calibration of the model in                     nal derivation cohort4 and in the multicenter-merged cohort
the merged cohort (χ2=1.2; degrees of freedom=5; P=0.94).                 (Figure 2). Baseline characteristics were similar between
Predicted risk of miserable and good outcomes per categories              original derivation4 cohort and the current merged cohort
of the DRAGON score and observed percentages are depicted                 (Table 2). Advantages of the DRAGON score are manifold:
in Figures I and II in the online-only Data Supplement.                   (1) it is simple, (2) does not require any computer calcula-
Performance of the DRAGON score in the entire cohort as                   tions, (3) consists of parameters that are available immediately
judged with AUC-ROC was 0.84 (0.82–0.85; Figure 2) for                    after patient admission, (4) is very fast to calculate, and (5) is
miserable outcome and 0.82 (0.80–0.83) for good outcome,                  cost-free. In this multicenter study, we improved the gener-
without statistical difference between them (P=0.10). The                 alizability of the DRAGON score by validating it in patients
majority of patients had stroke in anterior circulation (n=3944;          with both anterior and posterior circulation ischemic strokes
87.3%). There was no difference in AUC-ROC (P=0.55)                       and in women and men separately.
between anterior and posterior circulation strokes: 0.84 (0.83–              The score estimates which patients have a very high likeli-
0.86) and 0.82 (0.78–0.87), respectively. The same held true              hood of miserable outcome (bedridden incontinent patient in
for sex-related analysis: AUC-ROC was 0.85 (0.82–0.87) in                 institutional care or death; mRS score, 5 and 6, respectively),
women and 0.82 (0.81–0.84) in men (P=0.25).                               despite administration of the only approved recanalization
   Proportions of patients with miserable outcomes (mRS score,            treatment for acute ischemic stroke, IVT with alteplase. Such
5–6) were 0%, 2%, 4%, 89%, and 97% for 0 to 1, 2, 3, 8, and 9             patients are, in some centers, offered rescue add-on therapy
to 10 points, respectively. For mRS score 5 to 6 and DRAGON               (eg, endovascular procedure). These procedures are, however,
score 9 to 10, positive likelihood ratio was 45.4 (13.9–147.6),           not currently evidence-based. In case the use of such add-on
negative likelihood ratio was 1.0 (0.9–1.0), positive predictive          rescue strategies is in line with local institutional protocols,
value was 91.7% (77.5–98.2), negative predictive value was                the clinicians must act promptly to not lose time.10 In such
81.1% (79.9–82.2), specificity was 99.9% (99.8–100), and sen-             situations, rapid arrangements of the endovascular procedures
sitivity was 3.7% (2.6–5.2). In case of DRAGON score 8 to                 can be based on the DRAGON score, which is calculated even
10, the corresponding numbers were as follows: positive likeli-           before IVT administration.
hood ratio was 24.9 (16.3–38.1), negative likelihood ratio was               Based on the performance of the DRAGON score in the der-
0.8 (0.8–0.9), positive predictive value was 85.8% (79.6–90.7),           ivation and validation cohorts, patients with scores 0 to 3 have
negative predictive value was 83.1% (81.9–84.2), specificity              very high likelihood for good recovery after IVT. For patients
was 99.3% (99.0–99.6), and sensitivity was 16.4 (14.1–19.1).              with scores 7 to 10, the chances for good recovery are very
   Proportions of patients with good outcome (mRS score,
0–2) were 96%, 93%, 78%, and 0% for 0 to 1, 2, 3, and 8 to 10
score points, respectively. For mRS score 0 to 2 and DRAGON

Figure 1. Three-month outcome per increasing points in the
DRAGON score in the merged cohort (n=4519). Good 3-month out-
come (gray bars), modified Rankin scale (mRS) score 0 to 2, indicates
independence in activities of daily living. Miserable outcome (black
bars) corresponds to being bedridden, incontinent, and requiring con-     Figure 2. Performance (AUC-ROC curve) of the score in the
stant nursing care and attention (mRS score, 5) or dead (mRS score,       whole cohort. AUC-ROC indicates area under receiver operating
6). Note that for better clarity, mRS score 3 to 4 is not in the chart.   characteristic curve; and CI, confidence interval.

                                  Downloaded from by guest on July 30, 2015
Strbian et al   Validation of the DRAGON Score   2721

low. In institutions where rescue endovascular procedures are              (Dr Seiffge), the Swiss National Science Foundation No. 33CM30-
not performed because of a lack of randomized trial data, these            124119 (Drs Engelter and Lyrer), and Australian National Health
                                                                           and Medical Research Council Center for Research Excellence Grant
patients may be recruited into randomized controlled trials                1001216 (Drs Meretoja and Davis). The other authors have no con-
testing endovascular procedures or other add-on experimen-                 flicts to report.
tal treatments. Naturally, not all patients with high DRAGON
scores would be suitable candidates for endovascular proce-                                             References
dures (eg, a patient >80 years with extensive early infarction,              1. Rha JH, Saver JL. The impact of recanalization on ischemic stroke out-
high glucose, previous disability, and late presentation). Hence,               come: a meta-analysis. Stroke. 2007;38:967–973.
institutional inclusion criteria for endovascular approach have              2. Hacke W, Kaste M, Bluhmki E, Brozman M, Dávalos A, Guidetti D,
                                                                                et al; ECASS Investigators. Thrombolysis with alteplase 3 to 4.5 hours
to be considered in individual cases after obtaining informed                   after acute ischemic stroke. N Engl J Med. 2008;359:1317–1329.
consent from patients or their caregivers. Finally, because                  3. Lees KR, Bluhmki E, von Kummer R, Brott TG, Toni D, Grotta JC, et
thrombolysis treatment has a risk of intracranial hemorrhage                    al; ECASS, ATLANTIS, NINDS, and EPITHET rt-PA Study Group.
that can influence patient outcome,11 the final decision can be                 Time to treatment with intravenous alteplase and outcome in stroke: an
                                                                                updated pooled analysis of ECASS, ATLANTIS, NINDS, and EPITHET
based both on the DRAGON score (predicting functional out-                      trials. Lancet. 2010;375:1695–1703.
come) and a score predicting risk of symptomatic intracranial                4. Strbian D, Meretoja A, Ahlhelm FJ, Pitkäniemi J, Lyrer P, Kaste M, et al.
hemorrhage after IVT (eg, the SEDAN score).12                                   Predicting outcome of IV thrombolysis-treated ischemic stroke patients:
                                                                                the DRAGON score. Neurology. 2012;78:427–432.
   There are some limitations of this study. The datasets include
                                                                             5. Broderick JP. Endovascular therapy for acute ischemic stroke. Stroke.
mostly white patients. Also, assessments of early ischemic                      2009;40(suppl 3):S103–S106.
changes on pretreatment CT (eg, posterior circulation ASPECTS                6. Hemmen TM, Lyden PD. Induced hypothermia for acute stroke. Stroke.
score)13 and the role of hyperdense signs in the basilar artery or              2007;38(suppl 2):794–799.
                                                                             7. Rubiera M, Ribo M, Pagola J, Coscojuela P, Rodriguez-Luna D,
posterior cerebral artery are warranted. The strength of the study              Maisterra O, et al. Bridging intravenous-intra-arterial rescue strategy
is the large number of included patients. Furthermore, baseline                 increases recanalization and the likelihood of a good outcome in non-
characteristics and the outcome distribution in our dataset were                responder intravenous tissue plasminogen activator-treated patients: a
                                                                                case-control study. Stroke. 2011;42:993–997.
similar to those of the Safe Implementation of Thrombolysis
                                                                             8. Strbian D, Mustanoja S, Pekkola J, Putaala J, Haapaniemi E, Paananen
registry,14 with somewhat different baseline National Institutes                T, et al. Intravenous alteplase versus rescue endovascular procedure in
of Health Stroke Scale scores (median, 10 in our database and                   patients with proximal middle cerebral artery occlusion. Int J Stroke.
12 in the SITS database).                                                       September 27, 2012. doi: 10.1111/j.1747-4949.2012.00918.x.
                                                                             9. Turc G, Apoil M, Naggara O, Calvet D, Lamy C, Tataru AM, et al.
   In conclusion, the DRAGON score provides an accurate                         Magnetic Resonance Imaging-DRAGON score: 3-month outcome pre-
and immediate estimation of functional outcome in IVT-                          diction after intravenous thrombolysis for anterior circulation stroke.
treated ischemic stroke patients. Validation in a large mul-                    Stroke. 2013;44:1323–1328.
                                                                            10. Saver JL. Time is brain–quantified. Stroke. 2006;37:263–266.
ticenter cohort and in prespecified subgroups confirms its
                                                                            11. Strbian D, Sairanen T, Meretoja A, Pitkäniemi J, Putaala J, Salonen O,
generalizability.                                                               et al; Helsinki Stroke Thrombolysis Registry Group. Patient outcomes
                                                                                from symptomatic intracerebral hemorrhage after stroke thrombolysis.
                                                                                Neurology. 2011;77:341–348.
                    Sources of Funding                                      12. Strbian D, Engelter S, Michel P, Meretoja A, Sekoranja L, Ahlhelm FJ, et
This study was supported by the Helsinki University Central Hospital            al. Symptomatic intracranial hemorrhage after stroke thrombolysis: the
(HUCH) governmental subsidiary (EVO) funds for clinical research.               SEDAN score. Ann Neurol. 2012;71:634–641.
                                                                            13. Puetz V, Sylaja PN, Coutts SB, Hill MD, Dzialowski I, Mueller P, et al.
                                                                                Extent of hypoattenuation on CT angiography source images predicts
                         Disclosures                                            functional outcome in patients with basilar artery occlusion. Stroke.
The authors received funding from the HUCH (Drs Strbian, Meretoja,              2008;39:2485–2490.
and Tatlisumak), the Biomedicum Helsinki Foundation (Dr Meretoja),          14. Mazya MV, Lees KR, Markus R, Roine RO, Seet RC, Wahlgren N, et al.
the Sigrid Juselius Foundation (Drs Meretoja and Tatlisumak), the               Safety of IV thrombolysis for ischemic stroke in patients treated with
Finnish Medical Foundation (Dr Meretoja), the Basel Stroke Fund                 warfarin. Ann Neurol. June 6, 2013. doi: 10.1002/ana.23924.

                                Downloaded from by guest on July 30, 2015
Online Supplement

      Validation of the DRAGON Score in 12 stroke centers in anterior and posterior circulation

  Daniel Strbian, MD, PhD; 2David J Seiffge, MD; 3Lorenz Breuer, MD; 4Heikki Numminen, MD, PhD;
  Patrik Michel, MD; 1,6Atte Meretoja, MD, PhD, MSc (Stroke Med); 7Skye Coote, CCRN; 8Régis Bordet,
MD, PhD; 9Victor Obach, MD, 10,11Bruno Weder, MD; 11Simon Jung, MD; 12Valeria Caso, MD, PhD; 1Sami
Curtze, MD, PhD; 4Jyrki Ollikainen, MD; 2Philippe A Lyrer, MD; 5Ashraf Eskandari, RN; 11Heinrich P Mat-
tle, MD, PhD; 9Angel Chamorro, MD, PhD; 8Didier Leys, MD, PhD; 7Christopher Bladin, MD, PhD;
  Stephen M Davis, MD, PhD; 3Martin Köhrmann, MD, PhD; 2Stefan T Engelter, MD, PhD; 1Turgut Tatlisu-
mak, MD, PhD

Supplementary Table I. Analysis of selected parameters per NIHSS and OTT category

                                        no (%)       adjusted OR (95% CI)     adjusted OR (95% CI)
                                                      miserable outcome          good outcome
hyperDense cerebral artery sign                        1.64 (1.34-2.00)         0.69 (0.58-0.83)
                               yes   1075 (23.8%)
                                no   3444 (76.2%)
early infarct signs                                     1.30 (1.07-1.58)         0.64 (0.54-0.75)
                               yes   1187 (26.3%)
                                no   3332 (73.7%)
pre-stroke mRS>1                                        1.94 (1.44-2.62)         0.14 (0.09-0.20)
                               yes    278 (6.2%)
                                no   4241 (93.8%)
Age                                                     1.06 (1.05-1.07)       0.965 (0.960-0.970)
                         80 years    1166 (25.8%)
                          65-79 y    1954 (43.2%)
                            < 65 y   1399 (31.0%)
Glucose on admission                                    1.06 (1.03-1.08)         0.95 (0.93-0.98)
          > 8 mmol/L (144 mg/dL)     1106 (24.5%)
             8 mmol/L (144mg/dL)     3413 (75.5%)
Onset-to-treatment time                               1.001 (1.000-1.002)      0.999 (0.998-1.000)
                         > 90 min    3766 (83.3%)
                           90 min     753 (16.7%)
NIHSS on admission                                      1.16 (1.14-1.17)         0.86 (0.85-0.87)
                              > 15   1286 (28.5%)
                            10-15    1113 (24.6%)
                               5-9   1431 (31.7%)
                               0-4    689 (15.2%)
Supplementary Figure I

Legend: Predicted risk of miserable outcome (mRS 5-6) per categories of the DRAGON score and ob-
served percentages.

Supplementary Figure II

Legend: Predicted risk of good outcome (mRS 0-2) per categories of the DRAGON score and observed
Validation of the DRAGON Score in 12 Stroke Centers in Anterior and Posterior
  Daniel Strbian, David J. Seiffge, Lorenz Breuer, Heikki Numminen, Patrik Michel, Atte
Meretoja, Skye Coote, Régis Bordet, Victor Obach, Bruno Weder, Simon Jung, Valeria Caso,
Sami Curtze, Jyrki Ollikainen, Philippe A. Lyrer, Ashraf Eskandari, Heinrich P. Mattle, Angel
 Chamorro, Didier Leys, Christopher Bladin, Stephen M. Davis, Martin Köhrmann, Stefan T.
                               Engelter and Turgut Tatlisumak

             Stroke. 2013;44:2718-2721; originally published online August 8, 2013;
                            doi: 10.1161/STROKEAHA.113.002033
      Stroke is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
                    Copyright © 2013 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:

                                      Data Supplement (unedited) at:

 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:

 Subscriptions: Information about subscribing to Stroke is online at:

                       Downloaded from by guest on July 30, 2015
You can also read
NEXT SLIDES ... Cancel