For Peer Review Only

BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

FAST Warning Signs in Younger Stroke Patients: Implications for Stroke Education Campaigns

ForJournal: peerBMJ Open review only Manuscript ID: bmjopen2014005276

Article Type: Research

Date Submitted by the Author: 17Mar2014

Complete List of Authors: Kaps, M; JustusLiebig Universit�t Giessen Grittner, Ulrike; Department of Biostatistics and Clinical Epidemiology, Charité University Medicine Berlin, Germany, Tanislav, Christian; Justus Liebig University , Neurology Jungehuelsing, Gerhard; Charit� Berlin, Department of Neurology Tatlisumak, Turgut; Helsinki University Central Hospital, Department of Neurology Kessler, Christof; University Greifswald, Neurology Schmidt, Reinhold; Department of Neurology, University of Graz, Austria, Putaala, Jukka; University of Helsinki, Department of Neurological Sciences; Helsinki University Central Hospital, Department of Neurology Norrving, Bo; Lund University Hospital Rolfs, Arndt; University of Rostock, Department of Neurology

Primary Subject http://bmjopen.bmj.com/ Health services research Heading:

Cardiovascular medicine, Diagnostics, Epidemiology, Health services Secondary Subject Heading: research, Neurology

Stroke < NEUROLOGY, HEALTH SERVICES ADMINISTRATION & Keywords: MANAGEMENT, EPIDEMIOLOGY, MEDICAL EDUCATION & TRAINING, NEUROLOGY

on September 25, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 FAST Warning Signs in Younger Stroke Patients: Implications for Stroke Education 4 5 Campaigns 6 7 8 9 Manfred Kaps professor1*, Ulrike Grittner PhD2*, Christian Tanislav MD1*, Gerhard 10 3 4 5 11 Jungehülsing, MD , Turgut Tatlisumak, PhD , Christoph Kessler professor , Reinhold 8 4 7 8 12 Schmidt professor , Putaala Jukka MD , Bo Norrving professor , Arndt Rolfs professor , on TM 13 behalf of the sifap1 Investigators 14 15 *equal contributionFor peer review only 16 17 1Department of Neurology, Justus Liebig University, Giessen, Germany 18 2Department of Biostatistics and Clinical Epidemiology, Charité- University Medicine Berlin, 19 Germany 20 3 21 Department of Neurology, Charité, Berlin, Germany 4 22 Department of Neurology, University, Helsinki, Finland 5 23 Department of Neurology, Moritz Arndt University Greifswald, Germany 24 6Department of Neurology, University of Graz, Austria 25 7Department of Neurology, Lund, Sweden 26 8Albrecht-Kossel Institute for Neuroregeneration, University of Rostock, Rostock, Germany 27 28

29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 Corresponding author: 38 Christian Tanislav, MD 39 40 Department of Neurology 41 Justus Liebig University, Giessen, Germany

42 Klinikstr. 33 on September 25, 2021 by guest. Protected copyright. 43 35385 Giessen 44 Germany 45 Tel.: +49/641 98545301; Fax: +49/641 98545409 46 E-mail: [email protected]; 47 48

49 50 51 Key words: young stroke, awareness signs, FAST, clinical presentation. 52 Word count (manuscript): 2882 53 References: 14 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 25 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 Abstract 4 5 6 7 Background Clinical recognition instruments are important in a complex chain towards 8 timely treatment management in acute stroke. The present study aimed to evaluate the 9 10 frequency of individual warning signs reflected by the ‘FAST’ public stroke recognition 11 12 instrument (face, arm, speech, act timely) in younger stroke patients. 13 14 15 Methods InFor a multinational peer prospective review cross sectional studyonly (Stroke in Young Fabry 16 17 Patients; sifap1) including stroke patients aged below 56 the frequency of presenting 18 symptoms was analysed. The performance of the FAST scheme was investigated. 19 20 21 22 Results 5023 patients were enrolled in the study between 2007 and 2010. Information of 4535 23 patients with TIA (n=1071), ischemic stroke (n=3396) or other (n=68) were selected for this 24 25 analysis. FAST symptoms could be traced in 76.5% of all cases. 35% of those with at least 26 27 one FAST symptom had all three symptoms. The FAST instrument performed better in older 28 age groups: At least one FAST symptom could be recognised in 69.1% of 18-24 years old 29 30 patients, in 74.0% of those aged 25-34 years, in 75.4% of those aged 35-44 years, and77.8% 31 32 in 45-55 years old patients. Patients severely disabled (NIHSS<5: 69.0%, NIHSS 6-15: 33 98.9%, NIHSS>15: 100.0%) FAST signs were less sensitive for indicating cerebrovascular 34 http://bmjopen.bmj.com/ 35 events as well as in the posterior circulation (65.2%) and in patients with transient ischemic 36 37 attacks (62.3%). 38 39 40 Conclusions Nearly one-quarter of young stroke patients is not targeted by the FAST scheme; 41 the performance is higher in patients with worse stroke severity addressing almost all these 42 on September 25, 2021 by guest. Protected copyright. 43 individuals. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 Article Summary 4 5 6 Warning signs of stroke are probably the most important step in the complex chain towards 7 timely treatment management. For this purpose different stroke recognition instruments with 8 9 different symptoms and different wording have been developed, offering criteria to identify 10 11 stroke for public education. Currently many distinguished awareness campaigns to propagate 12 the signs of stroke use the simplified FAST message, which adopts weakness of face, arm as 13 14 well as speech and essentially time. In the present study we aimed to ascertain the 15 For peer review only 16 performance of FAST criteria in the sifap1 cohort, a unique prospectively collected 17 population of young (18-55 years) stroke victims. 18 19 The main finding indicates nearly one quarter of young strokes is not targeted by the FAST 20 21 scheme. However FAST fits very well to capture patients with acute stroke eligible for 22 thrombolysis due to the increasing performance in worse disability. It performs far less 23 24 sensitive for the large number of cases not eligible for thrombolysis but for secondary 25 26 prevention, for young stroke and for the posterior circulation. However, the main purpose of 27 sifap1 was not to validate a stroke recognition instrument. This might by the main limitation 28 29 of the present investigation. The “FAST wording” was primarily not covered with a specific 30 31 slot in the case report form. FAST items were derived from different sources after the 32 cerebrovascular event (e.g. after employing the NIH Stroke Scale immediately after hospital 33

34 admission (median delay: one day)). http://bmjopen.bmj.com/ 35 36 In conclusion we confirm FAST as a useful tool in the recognition of strokes in this specific 37 group of young strokes. Because risk factors and aetiology profiles in the sifap1 cohort 38 39 resembled those found in elderly stroke patients, conclusions from our study may be also 40 41 valid in older age groups.

42 on September 25, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 25 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 Introduction 4 5 6 Getting stroke patients to the acute care hospital on time is the major requirement for effective 7 8 stroke therapy. Knowing warning signs of stroke is the first and most important step in a 9 10 complex chain towards timely treatment management. Effective public education follows 11 simple rules: the message must be simple to remember, effective and consistent. For this 12 13 purpose different stroke recognition instruments with different symptoms and different wording 14 15 have been developedFor [1-6], peer offering criteria review to identify stroke only for public education [1-3;7;8]. 16 Currently many distinguished awareness campaigns to propagate the signs of stroke use the 17 18 simplified FAST message [9], which adopts weakness of face, arm as well as speech and 19 20 essentially time (National Health Service: “When stroke strikes act FAST”). Although the 21 mnemonic FAST is most frequently adopted, there are so far no prospective multicentre studies 22 23 from large cohorts, which evaluate distinct stroke signs for the use in public campaigns. 24 25 26 The aim of the present study is to ascertain the performance of FAST signs in a large cohort of 27 28 younger stroke patients the Stroke in Young Fabry Patients study (sifap1) [10] and to analyse 29 30 factors which influence the affectivity in identifying CVEs. This is a secondary analysis of the 31 sifap1 data which was originally supposed to investigating the relation of juvenile stroke and a 32 33 genetic disorder known as Fabry disease. Fabry disease was diagnosed in 0.9% of all patients

34 http://bmjopen.bmj.com/ 35 (n=45) [11]. 36 37 38 Methods 39 40 41 The protocol of the sifap1 study was published recently [10]. Described briefly, the sifap1

42 on September 25, 2021 by guest. Protected copyright. 43 study was designed as a multicenter multinational prospective observational study of young 44 45 stroke patients across Europe (Table 1). A total of 5023 stroke patients below 56 years were 46 enrolled in the sifap1 study, 271 patients with primary haemorrhages and 217 patients without 47 48 classification of the cerebrovascular event (CVE) were excluded. The remaining cohort of 49 50 4535 patients was extensively analysed including detailed medical history, 51 sociodemographics, clinical characteristics, stroke severity, laboratory values, genetics, 52 53 cardiac work up as well as presenting symptoms on hospital admission. Neurological deficits 54 55 were measured at the time of maximum impairment according to previous hospital-based 56 stroke registers [12]. A transient ischemic attack (TIA) was defined as a CVE with clinical 57 58 symptoms lasting <24 hours. Cerebral magnetic resonance imaging with standardized MRI 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 sequences was a mandatory procedure. Images were assessed centrally at the Department of 4 5 Neurology, Medical University of Graz, Austria, blinded to clinical and demographic data. 6 Items according to FAST were constructed as follows: 7 8 Face: facial palsy (minor asymmetry, partial or complete) according to NIHSS item. 9 10 Arm/Paresis: left or right arm some effort against gravity, no effort against gravity or no 11 movement according to NIHSS item or Paresis in arm or leg according to presenting 12 13 symptoms. 14 15 Speech: severeFor aphasia orpeer mute according review to NIHSS, or dysarthria only (mid-moderate slurring, or 16 severe, nearly intelligible or worse) according to NIHSS or dysphasia or aphasia or dysarthria 17 18 according to presenting symptoms. 19 20 The effectiveness of the FAST message in relation to different vascular territories was based 21 on MRI data. Anterior circulation included the anterior and middle cerebral artery, the 22 23 posterior circulation the vertebrobasilar territory and posterior cerebral artery. 24 25 According to the case report form (CRF) 210 out of 1537 patients underwent intravenous 26 thrombolysis. Information about thrombolysis was missing for 2999 patients because this item 27 28 was included later and not asked for all patients. The indication for thrombolysis was left to 29 30 the participating centres, according to their local structured operating procedures. 31 Frequencies of presenting symptoms were analysed in two modifications. At first the number 32 33 of cases exhibiting FAST symptoms i.e. face palsy or problems with arm/paresis or speech

34 http://bmjopen.bmj.com/ 35 were simply calculated as percentage from the whole cohort. Additionally a Venn diagram 36 was plotted for those with at least one FAST symptom to check frequencies of combinations 37 38 of symptoms. 39 40 Alternatively, the frequency of each symptom was calculated in a sequential approach. After 41 sizing the most frequent symptom, the second frequent symptom was extracted from the

42 on September 25, 2021 by guest. Protected copyright. 43 remaining cases and so on.. This approach allows establishing a “hierarchy” of stroke signs 44 45 with regard to those symptoms that most often occur aiming to identify as many as possible 46 stroke patients employing as few symptoms as possible to create a comprehensive public 47 48 message. 49 50 To test differences in subgroups according to age, sex or severity of stroke we used multilevel 51 logistic models to account for the heterogeneity between centres. All analyses were calculated 52 53 using commercially available software: PASW Statistics 18, Release Version 18.0.2 (© SPSS, 54 55 Inc., 2009, Chicago, IL, USA) and SAS software, Version 9.2 of the SAS System for 56 Windows. (© 2008 SAS Institute Inc., Cary, NC, USA). 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 25 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 Results 4 5 6 Frequency of presenting symptoms 7 8 5024 patients were enrolled in the study between 2007 and 2010. Information of 4535 patients 9 10 with TIA (n=1071), ischemic stroke (n=3396) or other (n=68) were selected for this analysis. 11 In general 76.5 % of the young stroke cohort could be captured by FAST criteria. Table 2 12 13 specifies the frequency of FAST symptoms according to gender and age. Face was more 14 15 frequently affectedFor in men.peer FAST performed review better in older only age groups, this difference 16 becomes even more obvious in younger patients below 25 yrs where the capture rate of FAST 17 18 signs is below 70%. 19 20 Figure 1 shows the frequency of combinations of FAST symptoms. 34.7% of those with at 21 least one FAST symptom have all three symptoms. 17.3% have only arm problems or paresis. 22 23 16.6% have only speech problems. 15.7 % have both arm problems/paresis and speech 24 25 problems. The isolated symptom “Face” is very rare (2.5%). 26 Symptoms which were registered in the sifap1 cohort but were not considered for FAST are 27 28 listed in Table 3. These “non-traditional” stroke symptoms also disclosed age and gender 29 30 differences: headache and nausea were more common in women even after controlling for 31 migraine and age (results not shown), whereas hemianopia, headache, nausea, and 32 33 somatosensory deficits were more common in younger age groups even after controlling for

34 http://bmjopen.bmj.com/ 35 migraine and sex (results not presented). 36 Capturing cerebrovascular events by FAST signs depended on the severity (Table 4). FAST 37 38 signs were less frequent in TIA than in strokes (62.3% vs. 81.5%). Severe strokes were nearly 39 40 completely covered by FAST signs when a NIH score of at least 6 was present. FAST signs 41 highlighted 96.7% of patients who received thrombolysis.

42 on September 25, 2021 by guest. Protected copyright. 43 44 45 Association between presenting symptoms and MRI lesions 46 1419 patients had no infarct in MRI, more than half of those patients (56.5%) had a TIA. 47 48 From those 1419 patients only 65.1% were detected by the FAST items. For 252 patients no 49 50 information about lesions in MRI was available. For patients with definite infarct lesions in 51 MRI (n=2865) we tested the differences in performance of FAST regarding vascular 52 53 territories: the FAST instrument performed worse in the posterior circulation, where only 54 55 65.2% of all cases with an apparent MRI lesion could be depicted, whereas 92.4% of anterior 56 circulation strokes matched FAST criteria. There was no difference between left and right 57 58 hemispheric stroke (82.9% vs 82.1%). 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 Hierarchy of presenting symptoms 6 The item “arm/paresis” was by far the most frequent sign in all stroke patients aged below 56 7 8 years, and “speech” the second (Fig.2). Both items together were covering almost 75 % of all 9 10 recruited stroke and TIA patients. Under hierarchic perspective the frequency of the “face” 11 item (0.2%) becomes irrelevant if the four most common signs of stroke “arm/paresis” 12 13 (57.7%), “speech” (16.9%), “vertigo” (10.6%) and “somatosensory deficit” (6.8%) were prior 14 15 sequentially For used as selectionpeer criteria review [13;14]. Hemianopia only occurred in general with a 16 frequency of 14.4% but only 3.2% of the cases experienced this symptom independent from 17 18 arm/paresis, speech, vertigo or somatosensory deficits. Common stroke signs in young 19 20 patients such as headache, leg paresis and diplopia were seldom or not independently 21 differentiated from other stroke signs. 22 23 24 25 Discussion 26 27 28 Principal findings 29 30 Frequency of presenting symptoms 31 In our cohort the FAST symptoms could be traced in 76.5% of all cases. This is notably less 32 33 than in a previously reported study in which study nurses screened retrospectively the medical

34 http://bmjopen.bmj.com/ 35 records of 3498 stroke cases who presented in an acute care hospital [4]. The “capture rate” in 36 this study was 88.9% for all strokes including haemorrhage and even higher for ischemic 37 38 stroke and TIA (91.1% and 91.8% resp.). Interestingly Kleindorfer and co-workers found that 39 40 cases missed by FAST tended to be significantly younger than cases with FAST symptoms 41 (mean 68.9 vrs 71.5 yrs) [4]. In our selected cohort with younger patients (median age 46

42 on September 25, 2021 by guest. Protected copyright. 43 years), we observed a comparable trend with significantly fewer patients presenting with 44 45 symptoms covered by FAST in younger age groups. We therefore conclude that FAST might 46 perform less robustly in younger stroke patients. 47 48 With respect to FAST signs there was no gender difference in stroke presentation. Other 49 50 presenting symptoms such as headache, nausea/vomiting and somatosensory deficits occurred 51 more frequently in women. This specific feature could not be explained by larger number of 52 53 vertebrobasilar strokes in women. Gender differences in these symptoms were also significant 54 55 when TIA patients were excluded. In a previous study on gender differences in acute stroke 56 symptoms a higher prevalence of symptoms termed “non-traditional” (pain, mental status 57 58 change, light-headedness, headache, non-neurological symptoms) could be registered in 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 25 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 women [7]. However, there were no differences observed regarding traditional symptoms 4 5 such as hemiparesis, aphasia, facial weakness, hemibody numbness, which is in line with our 6 findings [7]. 7 8 FAST performed better with increasing stroke severity and disclosed nearly all cases who 9 10 received thrombolysis irrespective of gender. This is of major interest, because a recent 11 metaanalysis indicated that women with stroke had 30% lower odds of receiving tissue 12 13 plasminogen activator thought possibly to result from different symptom presentation in 14 15 women. For peer review only 16 17 18 Hierarchy of presenting symptoms 19 20 “Arm/paresis” and “Speech” are by far the most relevant ischemic stroke signs. “Face” can 21 be quantitatively neglected, because it occurred as an isolated symptom very rarely. 22 23 Nevertheless, 34.7% of patients with at least one FAST symptom report all three FAST 24 25 symptoms, meaning that “Face” is rare as an isolated symptom but often accompanied by 26 “Arm/Paresis” or “Speech”. Similarly, headache is not relevant to identify ischemic stroke but 27 28 may characterize cases with subarachnoid haemorrhage. Remarkably, one single stroke sign 29 30 (“Arm/Paresis”) was effective enough to select 86.7% of patients in the subgroup of patients 31 who underwent thrombolytic therapy. In other words the FAST message is very capable of 32 33 capturing (or detecting) nearly all stroke patients eligible for thrombolysis. However FAST

34 http://bmjopen.bmj.com/ 35 did not apply for 23.5% of stroke patients in our cohort. Moreover 37.7% of patients with TIA 36 were not addressed by the symptoms in the FAST message. Although this is not relevant for 37 38 thrombolytic treatment, it may exclude many young patients at risk for stroke from mandatory 39 40 secondary prevention. Performance was also worse in young patients with vertebrobasilar 41 infarctions (64.5%). This is relevant, because “non-FAST” signs are frequently seen as first

42 on September 25, 2021 by guest. Protected copyright. 43 symptoms of progressing stroke especially in basilar thrombosis and patients missed by FAST 44 45 do not necessarily have a benign outcome [4]. 46 47 48 Other recognition instruments 49 50 Diverse stroke recognition instruments have been developed in order to identify strokes in 51 public [5;9], for triage by ambulance paramedics [2;3;6], to guide paramedics and emergency 52 53 physicians to direct patients with acute neurological signs to appropriate care for emergency 54 55 room physicians [1;2] or as a screening instrument in pre-hospital stroke research [6;14]. 56 Selection and number of alarming symptoms as well as wording differ; three to six warning 57 58 signs are usually promoted. The ASA Stroke Warning Signs [5] emphasize the term 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 “sudden” as prefix, and add a broad spectrum of signs (weakness of face, arm or leg, 4 5 confusion, trouble speaking or understanding, trouble seeing in one or both eyes, trouble 6 walking, dizziness, loss of balance or coordination, severe headache). This extensive 7 8 summary of neurological symptoms is likely to cover every brain injury. This kind of 9 10 messaging takes advantage of the fact that acute onset (“sudden”) is the most discriminating 11 factor between stroke and non-stroke and targets 96% of all strokes and TIA but can include 12 13 47% of differential non- stroke diagnoses [1]. The Cincinnati Prehospital Stroke Scale 14 15 (CPSS) [3] isFor a 3-item scalepeer derived from review a simplification of theonly National Institutes of Health 16 (NIH) Stroke Scale. It evaluates facial droop and arm drift, speech is tested by asking the 17 18 patient to repeat sentences. Thereby the CPSS design is very similar to FAST [9] and its 19 20 reproducibility has proven excellent among pre-hospital care providers [3]. In agreement with 21 our findings, the CPSS has also proven high validity to identify candidates for thrombolysis 22 23 and strokes preferably in the anterior circulation. The Los Angeles Prehospital Stroke 24 25 Screen (LAPPS) [2] considers unilateral motor weakness (“facial smile/grimace”, hand grip 26 and arm strength/drift) as well as 4 screening criteria (Age > 45, history of seizure disorder 27 28 absent, symptoms duration less than 24 hours, not wheel chair bound or bedridden prior to the 29 30 event) and a glucose measurement. It was designed to allow pre-hospital personnel to rapidly 31 identify most common stroke patients and exclude those unlikely to qualify for or benefit 32 33 from acute interventions. Thereby stroke patients below 45 years or with longer symptom

34 http://bmjopen.bmj.com/ 35 duration were purposefully not targeted. The Melbourne Ambulance Stroke Screen 36 (MASS) [6] combines clinical signs of the CPSS with the LAPPS screening criteria. The 37 38 ROSSIER scale [1] lists 5 key signs scoring positive (asymmetric facial/arm/leg weakness, 39 40 speech disturbances and visual field defect) and 2 items scoring negative (loss of 41 consciousness and syncope, seizure activity) adding to a total score ranging between -2 and

42 on September 25, 2021 by guest. Protected copyright. 43 +5. The purpose of the ROSSIER design was to develop a simple and practical instrument for 44 45 emergency room physicians in order to reduce the number of non-stroke referrals from 46 emergency room to stroke unit. 47 48 Stroke recognition instruments must be differentiated with regard to the addressee. For public 49 50 education and campaigns there is no choice but to promote a selected number of clinical 51 signs. Additional assessments (i.e. glucose measurements) to rule out non-strokes or stroke 52 53 subgroups are reserved for paramedic use or for triage purposes in hospital. 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 25 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 Strengths and weaknesses of the study 4 5 6 The sifap1 study is by far the largest prospective multicenter study analysing presenting 7 8 stroke symptoms in the young with broad clinical work up including MRI. Age limit up to 55 9 10 years allows exploring subgroups with increasing age including patients with risk factor and 11 aetiology profiles resembling those found in elderly stroke patients [10]. Therefore the 12 13 findings may be extrapolated to some extend to the stroke population in general. Nevertheless 14 15 there are someFor limitations peer regarding thisreview study: The main purposeonly of sifap1 was not to 16 validate a stroke recognition instrument. Therefore “FAST wording” was primarily not 17 18 covered with a specific item in the CRF. Instead we asked for paresis directly after CVE and 19 20 employed the NIH Stroke Scale immediately after hospital admission (median delay: one 21 day). Assessments ruling out stroke mimics (i.e. blood glucose level) influence specificity but 22 23 cannot be used for public education [6]. Because we excluded stroke mimics by definition, 24 25 our study is not designed to calculate for positive and negative predictive values of distinct 26 stroke signs. Moreover, there is a problem in general to calculate for false-negative diagnoses, 27 28 i.e. undiagnosed strokes in a population under survey. It has to be noted also, that we did not 29 30 consider haemorrhage, subarachnoid haemorrhage or venous thrombosis in our calculations. 31 Addressing all these strokes types may further add to complexity and dilute the awareness 32 33 message.

34 http://bmjopen.bmj.com/ 35 Extensive MRI documentation allowed us to validate the clinical stroke diagnosis and 36 constituted a robust additional aspect of FAST performance. One-third of vertebrobasilar 37 38 strokes and transient attacks could not be visualized on MRI. In these cases the appraisal of an 39 40 experienced neurologist was decisive. Notably there was no relevant difference comparing the 41 FAST performance in patients with and without proven MRI lesions.

42 on September 25, 2021 by guest. Protected copyright. 43 44 45 Implications for public campaigns 46 47 48 Instruments that help the lay public to identify stroke in pre-hospital setting are elementary to 49 50 trigger early treatment. Our study in stroke patients below 56 proves that FAST fits very well 51 to capture patients with acute stroke eligible for thrombolysis. It is much less sensitive for the 52 53 large number of cases not eligible for thrombolysis but for secondary prevention, for young 54 55 stroke and for the posterior circulation. A tailored message to reach especially women would 56 not increase performance of FAST, because “non-traditional symptoms” occurred 57 58 infrequently in isolation from common stroke symptoms. Because risk factors and aetiology 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 profiles in the sifap1 cohort resembled those found in elderly stroke patients [10], conclusions 4 5 from our study may be also valid in older age groups. 6 Design and usefulness of stroke recognition instruments depend strongly on the context and 7 8 the target group to be addressed. The proper diagnosis of stroke requires profound clinical 9 10 knowledge, which cannot be expected from lay persons or ambulance paramedics. On one 11 hand simplified but specific signs are needed for public campaigns, on the other hand stroke 12 13 recognition in pre-hospital settings cannot be simplified without losing a substantial number 14 15 of patients. Obviously,For signspeer such as “Vertigo”review and “Somatosensory only Deficits” target much 16 more stroke patients but coincidently capture more stroke mimics and thereby dilute the 17 18 message and burden the health care system. 19 20 It is important to recognize that sudden neurological symptoms need expeditious referral and 21 neurological expertise, because sudden neurological diseases other than stroke may represent 22 23 emergencies as well. Messaging in public awareness programs is therefore subject to more 24 25 complex considerations. Besides catchy wording, frequency of symptoms, capture rates, and 26 positive and negative predictive values, the capacity and readiness of regional health care 27 28 systems to cope with stroke mimics will enhance public campaigns. 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41

42 on September 25, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 25 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 Acknowledgements 4 5 We thank especially Kristin Brüderlein, Ines Federow, Doreen Niemann, Gesine Makowei, 6 Frances König, Jan Burmeister, and Carla Biedermann for organizational aspects. In 7 8 particular, Sabine Rösner and Susanne Zielke spent a lot of effort in realizing the entire 9 10 workflow. Elmar Beck (Anfomed GmbH, Möhrendorf, Germany) programmed the web-based 11 data bank as well as hosted the data. 12 13 14 15 Conflict of interestFor disclosures peer review only 16 All authors have completed the Unified Competing Interest form at 17 18 http://www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) 19 20 and declare that they have no non-financial interests that may be relevant to the submitted 21 work. 22 23 24 25 Ethical approval 26 The design of the study was approved by the ethics committee of the Medical Association 27 28 Mecklenburg-Vorpommern (board 2), Medical Faculty, and University of Rostock. Approval 29 30 of the local ethics committees was obtained in all participating centers. The study was 31 registered in www.clinicaltrials.gov and (No. NCT00414583). 32 33

34 http://bmjopen.bmj.com/ 35 Funding 36 The sifap1 study (Stroke in Young Fabry Patients, www.sifap.eu; www.clinicaltrials.gov: No. 37 38 NCT00414583) has been supported partially by an unrestricted scientific grant from Shire 39 40 Human Genetic Therapies. 41

42 on September 25, 2021 by guest. Protected copyright. 43 Sponsors’ role in design and conduction of the study, data analysis and interpretation 44 45 The sponsors of the study had no role in the study design, data collection, data analysis, 46 interpretation, writing of the manuscript, or the decision to submit the manuscript for 47 48 publication. 49 50 The corresponding author had full access to all of the data in the study and takes full 51 responsibility for the integrity of the data and the accuracy of the data analysis, and had the 52 53 final responsibility for the decision to submit for publication. 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 Data sharing statement 4 5 No additional data available. 6 7 8 Authors’ contributions 9 10 The concept of the present analysis was developed by MC. MC and UG analysed the data. All 11 authors were involved in the analysis and interpretation of the data and in drafting the 12 13 manuscript. All authors read and approved the final manuscript. 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41

42 on September 25, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 25 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 References 4 5 1. Nor AM, Davis J, Sen B, Shipsey D, Louw SJ, Dyker AG, Davis M, Ford GA. The 6 Recognition of Stroke in the Emergency Room (ROSIER) scale: development and 7 validation of a stroke recognition instrument. Lancet Neurol 2005;4:727-34. 8 9 2. Kidwell CS, Starkman S, Eckstein M, Weems K, Saver JL. Identifying stroke in the 10 11 field. Prospective validation of the Los Angeles prehospital stroke screen (LAPSS). 12 Stroke 2000;31:71-6. 13 14 3. Kothari RU, Pancioli A, Liu T, Brott T, Broderick J. Cincinnati Prehospital Stroke 15 Scale:For reproducibility peer and validity. review Ann Emerg Med 1999; only33:373-8. 16 17 4. Kleindorfer DO, Miller R, Moomaw CJ, Alwell K, Broderick JP, Khoury J, Woo D, 18 Flaherty ML, Zakaria T, Kissela BM. Designing a message for public education 19 regarding stroke: does FAST capture enough stroke? Stroke 2007;38:2864-8. 20 21 5. American Stroke Association. Learn to Recognize a Stroke. 22 http://www.strokeassociation.org/STROKEORG/WarningSigns/Warning- 23 Signs_UCM_308528_SubHomePage.jsp. 2011. 24 25 6. Bray JE, Coughlan K, Barger B, Bladin C. Paramedic diagnosis of stroke: examining 26 long-term use of the Melbourne Ambulance Stroke Screen (MASS) in the field. Stroke 27 2010;41:1363-6. 28 29 7. Lisabeth LD, Brown DL, Hughes R, Majersik JJ, Morgenstern LB. Acute stroke 30 symptoms: comparing women and men. Stroke 2009;40:2031-6. 31 32 8. Hodgson CS. To FAST or not to FAST? Stroke 2007;38:2631-2. 33

34 9. Harbison J, Hossain O, Jenkinson D, Davis J, Louw SJ, Ford GA. Diagnostic accuracy http://bmjopen.bmj.com/ 35 of stroke referrals from primary care, emergency room physicians, and ambulance 36 staff using the face arm speech test. Stroke 2003;34:71-6. 37 38 10. Rolfs A, Martus P, Heuschmann PU, Grittner U, Holzhausen M, Tatlisumak T, 39 40 Bottcher T, Fazekas F, Enzinger C, Ropele S, Schmidt R, Riess O, Norrving B. 41 Protocol and methodology of the Stroke in Young Fabry Patients (sifap1) study: a prospective multicenter European study of 5,024 young stroke patients aged 18-55 42 on September 25, 2021 by guest. Protected copyright. 43 years. Cerebrovasc Dis 2011;31:253-62. 44 45 11. Rolfs A, Fazekas F, Grittner U, Dichgans M, Martus P, Holzhausen M, Bottcher T, 46 Heuschmann PU, Tatlisumak T, Tanislav C, Jungehulsing GJ, Giese AK, Putaala J, 47 Huber R, Bodechtel U, Lichy C, Enzinger C, Schmidt R, Hennerici MG, Kaps M, 48 Kessler C, Lackner K, Paschke E, Meyer W, Mascher H, Riess O, Kolodny E, 49 Norrving B. Acute cerebrovascular disease in the young: the Stroke in Young Fabry 50 Patients study. Stroke 2013;44:340-9. 51 52 53 12. Bhalla A, Grieve R, Rudd AG, Wolfe CD. Stroke in the young: access to care and 54 outcome; a Western versus eastern European perspective. J Stroke Cerebrovasc Dis 55 2008;17:360-5. 56 57 13. Fussman C, Rafferty AP, Lyon-Callo S, Morgenstern LB, Reeves MJ. Lack of 58 association between stroke symptom knowledge and intent to call 911: a population- 59 based survey. Stroke 2010;41:1501-7. 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 14. Tirschwell DL, Longstreth WT, Jr., Becker KJ, Gammans RE, Sr., Sabounjian LA, 4 Hamilton S, Morgenstern LB. Shortening the NIH Stroke scale for use in the 5 prehospital setting. Stroke 2002;33:2801-6. 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41

42 on September 25, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 25 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 Table 1. Inclusion criteria of patients in sifap1 [10] 4 5 age 18-55 years 6 Diagnosis Acute CVE of any etiology (ischemic stroke, 7 8 TIA, intracranial hemorrhage) 9 Time since event Less than 3 months before inclusion in the study 10 11 Verification of diagnosis Verification of brain infarction or hemorrhage 12 by MRI scan 13 14 In case of negative MRI diagnosis confirmed by 15 For peerstroke-experienced review neurologist (more only than 2 16 17 years of experience in stroke and at least 6 years 18 of experience in general neurology) 19 20 Diagnostic information MRI documentation available 21 Diagnostic procedures according to EUSI/ESO 22 23 recommendations 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41

0.014 <0.001

5 % 56.5 40. 45-55 45-55 N=2706 p

% 54.8 36.8 35-44 N=1250 N=1250

% 53.3 34.1 25-34 25-34 N=443

% 47.1 29.4 18-24 18-24 N=136

0.062 0.002

BMJ Open

% 52.6 35.8 http://bmjopen.bmj.com/ N=1860 p Women

4 % 57.4 40. Men N=2675

5 % on September 25, 2021 by guest. Protected copyright. All 76.5 77.2 75.4 0.072 69.1 74.0 75.4 77.8 0.002 57.7 58.9 56.0 0.062 50.0 56.2 54.8 59.7 0.002 55.4 38. n=4535 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml For peer review only Frequency of presenting symptoms according to FAST stratified by gender and age. and age. gender by to stratified according FAST symptoms presenting of Frequency

stroke signs. more indicated of the one or by (%) identified subsumes patients *Total Speech *Total Face Arm/Paresis Table 2. Page 17 of 25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open Page 18 of 25 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 Table 3. Frequency of presenting symptoms not accounted for FAST. 4

5 6 All Men Women p 18-44 44-55 p 7 yrs yrs 8 9 % % % % % 10 Vertigo 29.4 29.5 29.4 0.836 30.5 28.7 0.329 11 (n=4496) 12 13 Hemianopia 14.4 14.2 14.7 0.615 16.3 13.1 0.006 14 (n=4535) 15 For peer review only 16 Diplopia 10.3 10.8 9.6 0.201 10.2 10.4 0.840 17 (n=4502) 18 Headache 31.9 27.1 38.7 <0.0001 38.8 27.3 <0.001 19 20 (n=4499) 21 Nausea/Vomiting 19.8 18.0 22.3 <0.001 22.3 18.1 0.001 22 23 (n=4516) 24 Amaurosis fugax 2.3 2.0 2.8 0.113 2.4 2.3 0.760 25 (n=4505) 26 27 Somatosens.deficit 55.9 54.4 58.1 0.032 59.4 53.6 <0.001 28 (n=4147) 29 Gaze deviation 8.4 8.7 7.9 0.320 8.5 8.3 0.884 30 31 (n=4534) 32 Leg paresis 35.0 35.3 34.5 0.543 33.2 36.1 0.095 33

34 (n=4535) http://bmjopen.bmj.com/ 35 36

37 38 39 40 41

p p <0.0001 <0.0001

out 31.0 52.6 Lysis n=1327 With

5 6 67. 86. Lysis n=209

p With <0.0001 <0.0001

BMJ Open 92.6 99.5 > 15> NIH http://bmjopen.bmj.com/ n=204

81.8 92.0 6-15 NIH n=927

< 5< 23.5 45.9 n=3403

on September 25, 2021 by guest. Protected copyright. p p NIH <0.0001 <0.0001 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

46.4 63.3

For peer review only n=3396 Stroke Ischemic

15.4 41.9 n=1071

5 68 patients were classified as “other” and are not included here included here not and are “other” as classified were patients 68 1 % % % % % % % % % All All TIA 38. 57.7

Performance of FAST depending on stroke severity, TIA or Stroke, and eligibility for thrombolysis. thrombolysis. for eligibility and or Stroke, TIA severity, on stroke of FAST depending Performance

total 76.5 62.3 81.5 <0.0001 69.0 98.9 100.0 <0.0001 96.7 72.6 <0.0001 Face Arm/Paresis Speech 55.4 38.5 61.3 <0.0001 44.5 86.1 98.5 <0.0001 82.3 50.5 <0.0001

Table Table 4.

Page 19 of 25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open Page 20 of 25 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 Figure 1: Venn diagram of FAST symptoms in those with at least one symptom (n=3469) 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41

42 on September 25, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 21 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 Figure 2: Initial presenting symptoms (shown as a percentage of all cases) 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41

42 on September 25, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 22 of 25 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 131x120mm (150 x 150 DPI) 39 40 41

42 on September 25, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 23 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 166x133mm (96 x 96 DPI) 36 37 38 39 40 41

42 on September 25, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 24 of 25 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 STROBE Statement—checklist of items that should be included in reports of observational 3 studies 4 5 6 Manuscript jnnp-2013-307526 entitled "FAST Warning Signs in Younger Stroke Patients: Implications for 7 Stroke Education Campaigns" 8 9 10 Item No Recommendation 11 Title and abstract 1 (a) the recommendation was considered 12 (b) the recommendation was considered 13 14 Introduction 15 Background/rationaleFor peer 2 the recommendation review was considered only 16 Objectives 3 the recommendation was considered 17 18 Methods 19 Study design 4 the recommendation was considered 20 Setting 5 the recommendation was considered 21 22 Participants 6 (a) Cohort study— the recommendation was considered 23 Case-control study— not applicable 24 Cross-sectional study— the recommendation was considered 25 (b) Cohort study— not applicable 26 Case-control study— not applicable 27 28 Variables 7 the recommendation was considered 29 Data sources/ measurement 8* the recommendation was considered 30 Bias 9 the recommendation was considered 31 Study size 10 the recommendation was considered 32 33 Quantitative variables 11 the recommendation was considered

34 Statistical methods 12 (a) the recommendation was considered http://bmjopen.bmj.com/ 35 (b) the recommendation was considered 36 (c) the recommendation was considered 37 38 (d) Cohort study— not applicable 39 Case-control study— not applicable 40 Cross-sectional study— the recommendation was considered 41 (e) the recommendation was considered

42 on September 25, 2021 by guest. Protected copyright. 43 44 Results 45 Participants 13* (a) the recommendation was considered 46 47 (b) the recommendation was considered 48 (c) the recommendation was considered 49 Descriptive data 14* (a) the recommendation was considered 50 (b) the recommendation was considered 51 52 (c) Cohort study— not applicable 53 Outcome data 15* Cohort study— not applicable 54 Case-control study— the recommendation was considered 55 Cross-sectional study— not applicable 56 57 Main results 16 (a) not applicable 58 (b) the recommendation was considered 59 (c) not applicable 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml1 Page 25 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 Other analyses 17 the recommendation was considered 3 Discussion 4 5 Key results 18 the recommendation was considered 6 Limitations 19 the recommendation was considered 7 Interpretation 20 the recommendation was considered 8 Generalisability 21 the recommendation was considered 9 10 Other information 11 Funding 22 the recommendation was considered 12 13

14 15 For peer review only 16 17 18 C. Tanislav 19

20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41

42 on September 25, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml2 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

Clinical Signs in Young Stroke Patients related to FAST – Results of the sifap1 study –

ForJournal: peerBMJ Open review only Manuscript ID: bmjopen2014005276.R1

Article Type: Research

Date Submitted by the Author: 25Sep2014

Complete List of Authors: Kaps, M; JustusLiebig Universit�t Giessen Grittner, Ulrike; Department of Biostatistics and Clinical Epidemiology, Charité University Medicine Berlin, Germany, Jungehuelsing, Gerhard; Charit� Berlin, Department of Neurology Tatlisumak, Turgut; Helsinki University Central Hospital, Department of Neurology Kessler, Christof; University Greifswald, Neurology Schmidt, Reinhold; Department of Neurology, University of Graz, Austria, Putaala, Jukka; University of Helsinki, Department of Neurological Sciences; Helsinki University Central Hospital, Department of Neurology Norrving, Bo; Lund University Hospital Rolfs, Arndt; University of Rostock, Department of Neurology Tanislav, Christian; Justus Liebig University , Neurology

Primary Subject http://bmjopen.bmj.com/ Health services research Heading:

Cardiovascular medicine, Diagnostics, Epidemiology, Health services Secondary Subject Heading: research, Neurology

Stroke < NEUROLOGY, HEALTH SERVICES ADMINISTRATION & Keywords: MANAGEMENT, EPIDEMIOLOGY, MEDICAL EDUCATION & TRAINING, NEUROLOGY

on September 25, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 48 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 Clinical Signs in Young Stroke Patients related to FAST 4 – Results of the sifap1 study – 5 6 7 8

9 10 Manfred Kaps professor1*, Ulrike Grittner PhD2*, Gerhard Jungehülsing, MD3, Turgut 11 Tatlisumak, PhD4, Christoph Kessler professor5 , Reinhold Schmidt professor8, Putaala Jukka 12 MD4, Bo Norrving professor7, Arndt Rolfs professor8, Christian Tanislav MD1*, on behalf of 13 the sifap1TM Investigators 14 15 For peer review only 16 *equal contribution

17 1 18 Department of Neurology, Justus Liebig University, Giessen, Germany 19 2Department of Biostatistics and Clinical Epidemiology, Charité- University Medicine Berlin, 20 Germany 21 3Department of Neurology, Charité, Berlin, Germany 22 4Department of Neurology, University, Helsinki, Finland 23 5Department of Neurology, Moritz Arndt University Greifswald, Germany 24 6Department of Neurology, University of Graz, Austria 25 7 26 Department of Neurology, Lund, Sweden 8 27 Albrecht-Kossel Institute for Neuroregeneration, University of Rostock, Rostock, Germany 28 29 30 31 32 33 34 http://bmjopen.bmj.com/ 35 36 37 Corresponding author: 38 39 Christian Tanislav, MD 40 Department of Neurology 41 Justus Liebig University, Giessen, Germany 42 Klinikstr. 33 on September 25, 2021 by guest. Protected copyright. 43 35385 Giessen 44 45 Germany 46 Tel.: +49/641 98545301; Fax: +49/641 98545409 47 E-mail: [email protected]; 48 49 50 51 Key words: young stroke, awareness signs, FAST, clinical presentation. 52 Word count (manuscript): 3350 53 References: 14 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 48 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 Abstract 4 5 6 7 Objectives The present study aimed to evaluate the frequency of warning signs in younger 8 stroke patients with a special regard to the ‘FAST’ scheme, a public stroke recognition 9 10 instrument (face, arm, speech, act timely). 11 12 13 Setting Primary stroke care in participating centres of a multinational European prospective 14 15 cross sectionalFor study (Stroke peer in Young Fabryreview Patients; sifap1). only 47 centres from 15 European 16 17 countries participate in sifap1. 18 19 20 Participants 5023 acute stroke patients (aged 18-55 years) patients (96.5% Caucasians) were 21 22 enrolled in the study between April 2007 and January 2010. 23 24 25 Primary and secondary outcome measures: sifap1 was originally design to investigate the 26 27 relation of juvenile stroke and Fabry disease. A secondary aim of sifap1 was to investigate 28 stroke patterns in this specific group of patients. The present investigation is a secondary 29 30 analysis addressing stroke presenting symptoms with a special regard to signs included in the 31 32 FAST scheme. 33 34 http://bmjopen.bmj.com/ 35 Results 4535 patients with TIA (n=1071), ischemic stroke (n=3396) or other (n=68) were 36 37 considered in the presented analysis. FAST symptoms could be traced in 76.5% of all cases. 38 35% of those with at least one FAST symptom had all three symptoms. At least one FAST 39 40 symptom could be recognised in 69.1% of 18-24 years old patients, in 74.0% of those aged 41 25-34 years, in 75.4% of those aged 35-44 years, and 77.8% in 45-55 years old patients. With 42 on September 25, 2021 by guest. Protected copyright. 43 increasing stroke severity signs included in the FAST scheme were more prevalent 44 45 (NIHSS<5: 69.0%, NIHSS 6-15: 98.9%, NIHSS>15: 100.0%). Clustering clinical signs 46 47 according to FAST lower percentages of strokes in the posterior circulation (65.2%) and in 48 patients with transient ischemic attacks (62.3%) were identified. 49 50 51 52 Conclusions FAST may be applied as a useful and rapid tool to identify stroke symptoms in 53 young individuals aged 18-55 years. Especially in patients eligible for thrombolysis FAST 54 55 might address the majority of individuals. 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 48 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 Strengths and limitations of this study 4 5 • The present analysis investigated clinical symptoms in young stroke patients, 6 7 participants in a unique prospectively collected population of young (18-55 years) 8 9 stroke victims (Stroke in Young Fabry Patients study; sifap1). A special focus was 10 the analysis of symptoms included in the FAST scheme. The FAST message is a tool 11 12 used in awareness campaigns to propagate clinical sign of stroke; it includes 13 14 weakness of face, arm as well as speech and essentially time. 15 • The mainFor finding peerindicates that nearlyreview one quarter of young only strokes is not targeted by 16 17 the FAST scheme. However, FAST signs are more frequent in patients with acute 18 19 stroke eligible for thrombolysis. This might be due to increasing stroke severity. 20 • The main purpose of sifap1 was not to validate a stroke recognition instrument. The 21 22 “FAST wording” was primarily not covered with a specific slot in the case report 23 24 form. FAST items were derived from different sources after the cerebrovascular event 25 (e.g. after employing the NIH Stroke Scale immediately after hospital admission). 26 27 This limitation need to be taken into account when interpreting our results. 28 29 • However, clustering clinical signs according to FAST the majority of patients 30 included in the sifap1 study are targeted. Therefore, FAST may be applied as a useful 31 32 and rapid tool to identify stroke symptoms in young individuals aged 18-55 years. 33

34 • Because risk factors and aetiology profiles in the sifap1 cohort resembled those found http://bmjopen.bmj.com/ 35 in elderly stroke patients, conclusions from our study may be also valid in older age 36 37 groups. 38 39 40 41

42 on September 25, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 48 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 Introduction 4 5 6 Getting stroke patients to the acute care hospital on time is the major requirement for effective 7 8 stroke therapy. Knowing warning signs of stroke is the first and most important step in a 9 10 complex chain towards timely treatment management. Effective public education follows 11 simple rules: the message must be simple to remember, effective and consistent. For this 12 13 purpose different stroke recognition instruments with different symptoms and different wording 14 15 have been developedFor [1-6], peer offering criteria review to identify stroke foronly public education [1-3;7;8, 9]. 16 They are useful for identify strokes in public [5;9], for triage by ambulance paramedics [2;3;6], 17 18 to guide paramedics and emergency physicians to direct patients with acute neurological signs 19 20 to appropriate care for emergency room physicians [1;2] or as a screening instrument in pre- 21 hospital stroke research [6]. Selection and number of alarming symptoms as well as wording 22 23 differ; three to six warning signs are usually promoted. Currently many distinguished 24 25 awareness campaigns to propagate the signs of stroke use the simplified FAST message [9], 26 which adopts weakness of face, arm as well as speech and essentially time (National Health 27 28 Service: “When stroke strikes act FAST”). Although the mnemonic FAST is most frequently 29 30 adopted, there are so far no prospective multicentre studies from large cohorts, which evaluate 31 distinct stroke signs for the use in public campaigns. 32 33 The aim of the present study is to investigate symptoms in acute young stroke patients with a

34 http://bmjopen.bmj.com/ 35 special focus on signs included in the FAST scheme in a large cohort of younger stroke patients 36 the Stroke in Young Fabry Patients study (sifap1) and to analyse factors which influence the 37 38 effectiveness in identifying CVEs [10]. This is a secondary analysis of the sifap1 data which 39 40 was originally supposed to investigate the relation of juvenile stroke and a genetic disorder 41 known as Fabry disease. Fabry disease is an x-linked storage disorder which might affect the

42 on September 25, 2021 by guest. Protected copyright. 43 entire body. In sifap1 in 0.9% of all patients Fabry disease was identified [11]. 44 45 46 Methods 47 48 49 50 The protocol of the sifap1 study was published recently [10]. Described briefly, the sifap1 51 study was designed as a multicenter multinational prospective observational study of young 52 53 stroke patients across Europe (Table 1). A total of 5023 stroke patients (aged 18-55 years) 54 55 were enrolled in the sifap1 study, 271 patients with primary haemorrhages and 217 patients 56 without classification of the cerebrovascular event (CVE) were excluded. All patients or legal 57 58 representatives gave written consent to inclusion. The remaining cohort of 4535 patients was 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 48 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 extensively analysed including detailed medical history, socio-demographics, clinical 4 5 characteristics, stroke severity, laboratory values, genetics, cardiac work up as well as 6 presenting symptoms on hospital admission. Neurological deficits were measured at the time 7 8 of maximum impairment according to previous hospital-based stroke registers [12]. A 9 10 transient ischemic attack (TIA) was defined as a CVE with clinical symptoms lasting <24 11 hours. Cerebral magnetic resonance imaging with standardized MRI sequences was a 12 13 mandatory procedure. Images were assessed centrally at the Department of Neurology, 14 15 Medical UniversityFor of peer Graz, Austria, review blinded to clinical andonly demographic data. Items 16 according to FAST were constructed as follows: 17 18 Face: facial palsy (minor asymmetry, partial or complete) according to NIHSS item (the scale 19 20 was assessed within 48h after admission). 21 Arm/Paresis: left or right arm some effort against gravity, no effort against gravity or no 22 23 movement according to NIHSS item (the scale was assessed within 48h after admission) or 24 25 Paresis in arm or leg according to presenting symptoms (documented on inclusion in the 26 study). 27 28 Speech: severe aphasia or mute according to NIHSS (documented on inclusion in the study), 29 30 or dysarthria (mid-moderate slurring, or severe, nearly intelligible or worse) according to 31 NIHSS (documented on inclusion in the study) or dysphasia or aphasia or dysarthria 32 33 according to presenting symptoms (documented on inclusion in the study).

34 http://bmjopen.bmj.com/ 35 Based on MRI data clinical signs addressed by FAST were analysed in relation to the 36 different vascular territories. Anterior circulation included the anterior and middle cerebral 37 38 artery, the posterior circulation the vertebra-basilar territory and posterior cerebral artery. 39 40 Apart from NIHSS, the following presenting symptoms were recorded additionally in the case 41 report form on inclusion: headache, nausea-vomiting, hemianopia, diplopia and vertigo.

42 on September 25, 2021 by guest. Protected copyright. 43 According to the case report form (CRF) 210 out of 1537 patients underwent intravenous 44 45 thrombolysis. Information about thrombolysis was missing for 2999 patients because this item 46 was included later and not asked for all patients. The indication for thrombolysis was left to 47 48 the participating centres, according to their local structured operating procedures. 49 50 Frequencies of presenting symptoms were analysed in two modifications. At first the number 51 of cases exhibiting FAST symptoms i.e. face palsy or problems with arm/paresis or speech 52 53 were simply calculated as percentage from the whole cohort. Additionally a Venn diagram 54 55 was plotted for those with at least one FAST symptom to check frequencies of combinations 56 of symptoms. 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 48 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 Alternatively, the frequency of each symptom was calculated in a sequential approach. After 4 5 sizing the most frequent symptom, the second frequent symptom was extracted from the 6 remaining cases and so on. This approach allows establishing a “hierarchy” of stroke signs 7 8 with regard to those symptoms that most often occur aiming to identify as many as possible 9 10 stroke patients employing as few symptoms as possible to create a comprehensive public 11 message. 12 13 To test differences in subgroups according to age, sex or severity of stroke we used multilevel 14 15 logistic modelsFor to account peer for the heterogeneity review between centres. only All analyses were calculated 16 using commercially available software: PASW Statistics 18, Release Version 18.0.2 (© SPSS, 17 18 Inc., 2009, Chicago, IL, USA) and SAS software, Version 9.2 of the SAS System for 19 20 Windows. (© 2008 SAS Institute Inc., Cary, NC, USA). 21 22 23 Results 24 25 26 Frequency of presenting symptoms 27 28 5024 patients were enrolled in the study between 2007 and 2010. 4535 patients with TIA 29 30 (n=1071), ischemic stroke (n=3396) or other (n=68, refers to cerebral vein thrombosis or no 31 documented entity) were included in the analysis. In general 76.5 % of the young strokes 32 33 included in sifap1 had clinical signs covered by FAST criteria. Table 2 specifies the

34 http://bmjopen.bmj.com/ 35 frequency of FAST symptoms according to gender and age. Face was more frequently 36 affected in men. FAST signs were more prevalent in older age groups; this difference 37 38 becomes even more obvious in younger patients below 25 years where the capture rate of 39 40 FAST signs is below 70%. 41 Figure 1 shows the frequency of combinations of FAST symptoms. 34.7% of those with at

42 on September 25, 2021 by guest. Protected copyright. 43 least one FAST symptom have all three symptoms. 17.3% have only arm problems or paresis. 44 45 16.6% have only speech problems. 15.7 % have both arm problems/paresis and speech 46 problems. The isolated symptom “Face” is very rare (2.5%). 47 48 Symptoms which were registered in the sifap1 cohort but were not considered for FAST are 49 50 listed in Table 3. These stroke symptoms (not included in FAST) also disclosed age and 51 gender differences: headache and nausea were more common in women even after controlling 52 53 for migraine and age (results not shown), whereas hemianopia, headache, nausea, and 54 55 somatosensory deficits were more common in younger age groups even after controlling for 56 migraine and sex (results not presented). 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 48 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 With increasing stroke severity signs included in the FAST scheme were more prevalent 4 5 (Table 4). FAST signs were less frequent in TIA than in strokes (62.3% vs. 81.5%). Severe 6 strokes were nearly completely covered by FAST signs when a NIH score of at least 6 was 7 8 present. FAST signs occurred in 96.7% of patients who received thrombolysis. 9 10 11 Association between presenting symptoms and MRI lesions 12 13 1419 patients had no infarct in MRI, more than half of those patients (56.5%) had a TIA. 14 15 From those 1419For patients peer only 65.1% were review detected by the FAST only items. For 252 patients no 16 information about lesions in MRI was available. For patients with definite infarct lesions in 17 18 MRI (n=2865) we tested the differences in occurrence of symptoms included in the FAST 19 20 scheme regarding vascular territories: FAST symptoms were less frequent in patients with 21 strokes in the posterior circulation, where only 65.2% of all cases with an apparent MRI 22 23 lesion could be identified, whereas 92.4% of anterior circulation strokes matched symptoms 24 25 included in the FAST criteria. There was no difference between left and right hemispheric 26 stroke (82.9% vs 82.1%). 27 28 29 30 Hierarchy of presenting symptoms 31 The item “arm/paresis” was by far the most frequent sign in all stroke patients (aged 18-55 32 33 years), and “speech” the second (Fig.2). Both items together were covering almost 75 % of all

34 http://bmjopen.bmj.com/ 35 recruited stroke and TIA patients. Under hierarchic perspective the frequency of the “face” 36 item (0.2%) becomes irrelevant if the four most common signs of stroke “arm/paresis” 37 38 (57.7%), “speech” (16.9%), “vertigo” (10.6%) and “somatosensory deficit” (6.8%) were prior 39 40 sequentially used as selection criteria [13;14]. Hemianopia occurred in general with a 41 frequency of 14.4% but only 3.2% of the cases experienced this symptom independent from

42 on September 25, 2021 by guest. Protected copyright. 43 arm/paresis, speech, vertigo or somatosensory deficits. Common stroke signs in young 44 45 patients such as headache, leg paresis and diplopia were seldom or not independently 46 differentiated from other stroke signs. 47 48 49 50 Discussion 51 52 53 Principal findings 54 55 Frequency of presenting symptoms 56 In our cohort the FAST symptoms could be traced in 76.5% of all cases. This is notably less 57 58 than in a previously reported study in which study nurses screened retrospectively the medical 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 48 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 records of 3498 stroke cases who presented in an acute care hospital [4]. The “capture rate” in 4 5 this study was 88.9% for all strokes including haemorrhage and even higher for ischemic 6 stroke and TIA (91.1% and 91.8% resp.). Interestingly Kleindorfer and co-workers found that 7 8 cases missed by FAST tended to be significantly younger than cases with FAST symptoms 9 10 (mean 68.9 vrs 71.5 yrs) [4]. Even though Kleindorfer and co-workers investigates a general 11 stroke population, in our selected cohort with younger patients (median age 46 years), we 12 13 observed a comparable trend with significantly fewer patients presenting with symptoms 14 15 covered by FASTFor in younger peer age groups. review Therefore clinical signs only included in FAST might be 16 less prevalent in younger stroke patients. 17 18 With respect to FAST signs there was no gender difference in stroke presentation. Other 19 20 presenting symptoms such as headache, nausea/vomiting and somatosensory deficits occurred 21 more frequently in women. This specific feature could not be explained by larger number of 22 23 vertebra-basilar strokes in women. Gender differences in these symptoms were also 24 25 significant when TIA patients were excluded. In a previous study on gender differences in 26 acute stroke symptoms a higher prevalence of symptoms termed “non-traditional” (pain, 27 28 mental status change, light-headedness, headache, non-neurological symptoms) could be 29 30 registered in women [7]. However, there were no differences observed regarding traditional 31 symptoms such as hemiparesis, aphasia, facial weakness, hemibody numbness, which is in 32 33 line with our findings [7].

34 http://bmjopen.bmj.com/ 35 Clinical signs clustered according to the FAST scheme disclosed more stroke patients with 36 increasing stroke severity. Nearly all cases who received thrombolysis irrespective of gender 37 38 were indicated by symptoms included in FAST. This is of major interest, because a recent 39 40 metaanalysis indicated that women with stroke had 30% lower odds of receiving tissue 41 plasminogen activator thought possibly to result from different symptom presentation in

42 on September 25, 2021 by guest. Protected copyright. 43 women. 44 45 46 Hierarchy of presenting symptoms 47 48 49 50 “Arm/paresis” and “Speech” are by far the most relevant ischemic stroke signs. “Face” can 51 be quantitatively neglected, because it occurred as an isolated symptom very rarely. 52 53 Nevertheless, 34.7% of patients with at least one FAST symptom report all three FAST 54 55 symptoms, meaning that “Face” is rare as an isolated symptom but often accompanied by 56 “Arm/Paresis” or “Speech”. Similarly, headache is not relevant to identify ischemic stroke but 57 58 may characterize cases with subarachnoid haemorrhage. Remarkably, one single stroke sign 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 48 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 (“Arm/Paresis”) was effective enough to select 86.7% of patients in the subgroup of patients 4 5 who underwent thrombolytic therapy. However, clinical signs included in the FAST scheme 6 were absent in 23.5% of stroke patients included in sifap1. Moreover 37.7% of patients with 7 8 TIA had symptoms other than those mentioned by FAST. Sign included in the FAST scheme 9 10 were also less frequent in young patients with vertebro-basilar infarctions (64.5%). This is 11 relevant, because “non-FAST” signs are frequently seen as first symptoms of progressing 12 13 stroke, especially in basilar thrombosis and patients missed by FAST do not necessarily have 14 15 a benign outcomeFor [4]. peer review only 16 17 18 Other recognition instruments 19 20 21 The ASA Stroke Warning Signs [5] emphasize the term “sudden” as prefix, and add a broad 22 23 spectrum of signs (weakness of face, arm or leg, confusion, trouble speaking or 24 25 understanding, trouble seeing in one or both eyes, trouble walking, dizziness, loss of balance 26 or coordination, severe headache). This extensive summary of neurological symptoms is 27 28 likely to cover every brain injury. This kind of messaging takes advantage of the fact that 29 30 acute onset (“sudden”) is the most discriminating factor between stroke and non-stroke and 31 targets 96% of all strokes and TIA but can include 47% of differential non- stroke diagnoses 32 33 [1]. The Cincinnati Prehospital Stroke Scale (CPSS) [3] is a 3-item scale derived from a

34 http://bmjopen.bmj.com/ 35 simplification of the National Institutes of Health (NIH) Stroke Scale. It evaluates facial 36 droop and arm drift, speech is tested by asking the patient to repeat sentences. Thereby the 37 38 CPSS design is very similar to FAST [9] and its reproducibility has proven excellent among 39 40 pre-hospital care providers [3]. In agreement with our findings, the CPSS has also proven 41 high validity to identify candidates for thrombolysis and strokes preferably in the anterior

42 on September 25, 2021 by guest. Protected copyright. 43 circulation. The Los Angeles Prehospital Stroke Screen (LAPPS) [2] considers unilateral 44 45 motor weakness (“facial smile/grimace”, hand grip and arm strength/drift) as well as 4 46 screening criteria (Age > 45, history of seizure disorder absent, symptoms duration less than 47 48 24 hours, not wheel chair bound or bedridden prior to the event) and a glucose measurement. 49 50 It was designed to allow pre-hospital personnel to rapidly identify most common stroke 51 patients and exclude those unlikely to qualify for or benefit from acute interventions. Thereby 52 53 stroke patients below 45 years or with longer symptom duration were purposefully not 54 55 targeted. The Melbourne Ambulance Stroke Screen (MASS) [6] combines clinical signs of 56 the CPSS with the LAPPS screening criteria. The ROSSIER scale [1] lists 5 key signs 57 58 scoring positive (asymmetric facial/arm/leg weakness, speech disturbances and visual field 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 48 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 defect) and 2 items scoring negative (loss of consciousness and syncope, seizure activity) 4 5 adding to a total score ranging between -2 and +5. The purpose of the ROSSIER design was 6 to develop a simple and practical instrument for emergency room physicians in order to 7 8 reduce the number of non-stroke referrals from emergency room to stroke unit. 9 10 Stroke recognition instruments must be differentiated with regard to the addressee. For public 11 education and campaigns there is no choice but to promote a selected number of clinical 12 13 signs. Additional assessments (i.e. glucose measurements) to rule out non-strokes or stroke 14 15 subgroups areFor reserved forpeer paramedic use review or for triage purposes only in hospital. 16 17 18 Strengths and weaknesses of the study 19 20 21 The sifap1 study is by far the largest prospective multicenter study analysing presenting 22 23 stroke symptoms in the young with broad clinical work up including MRI. Age limit up to 55 24 25 years allows exploring subgroups with increasing age including patients with risk factor and 26 aetiology profiles resembling those found in elderly stroke patients [10]. Therefore the 27 28 findings may be extrapolated to some extend to the stroke population in general. Nevertheless 29 30 there are some limitations regarding this study: The main purpose of sifap1 was not to 31 validate a stroke recognition instrument. Therefore “FAST wording” was primarily not 32 33 covered with a specific item in the CRF. Instead we asked for paresis directly after CVE and

34 http://bmjopen.bmj.com/ 35 employed the NIH Stroke Scale immediately after hospital admission (median delay: one 36 day). Assessments ruling out stroke mimics (i.e. blood glucose level) influence specificity but 37 38 cannot be used for public education [6]. Because we excluded stroke mimics by definition, 39 40 our study is not designed to calculate for positive and negative predictive values of distinct 41 stroke signs. Moreover, there is a problem in general to calculate for false-negative diagnoses,

42 on September 25, 2021 by guest. Protected copyright. 43 i.e. undiagnosed strokes in a population under survey. It has to be noted also, that we did not 44 45 consider haemorrhage, subarachnoid haemorrhage or venous thrombosis in our calculations. 46 Addressing all these strokes types may further add to complexity and dilute the awareness 47 48 message. The FAST scheme was develop to screen for potential stroke victims in a preclinical 49 50 setting [9]. In contrast our patients were included after admission to a neurological 51 department. This needs to be taken into account when interpreting our results. 52 53 Extensive MRI documentation allowed us to validate the clinical stroke diagnosis and 54 55 constituted a robust additional aspect in identifying presenting symptoms in acute young 56 stroke patients. One-third of vertebra-basilar strokes and transient attacks could not be 57 58 visualized on MRI. In these cases the appraisal of an experienced neurologist was decisive. 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 48 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 Notably, there was no relevant difference regarding signs included in the FAST scheme 4 5 comparing patients with and without proven MRI lesions. 6 7 8 9 10 Implications for public campaigns 11 12 13 Instruments that help the lay public to identify stroke in pre-hospital setting are elementary to 14 15 trigger early Fortreatment. Ourpeer study in stroke review patients (aged 18-55 only years) proves that symptoms 16 considered in the FAST scheme may be useful for identifying young stroke patients. 17 18 Especially young stroke patients eligible for thrombolysis might be targeted by a FAST 19 20 evaluation. In contrast, clustering only clinical symptoms according to FAST, it might be less 21 effective in young stroke patients with TIA and infarcts in the posterior circulation. Because 22 23 risk factors and aetiology profiles in the sifap1 cohort resembled those found in elderly stroke 24 25 patients [10], conclusions from our study may be also valid in older age groups. 26 Design and usefulness of stroke recognition instruments depend strongly on the context and 27 28 the target group to be addressed. The proper diagnosis of stroke requires profound clinical 29 30 knowledge, which cannot be expected from lay persons or ambulance paramedics. On one 31 hand simplified but specific signs are needed for public campaigns, on the other hand stroke 32 33 recognition in pre-hospital settings cannot be simplified without losing a substantial number

34 http://bmjopen.bmj.com/ 35 of patients. Obviously, signs such as “Vertigo” and “Somatosensory Deficits” target much 36 more stroke patients but coincidently capture more stroke mimics and thereby dilute the 37 38 message and burden the health care system. 39 40 It is important to recognize that sudden neurological symptoms need expeditious referral and 41 neurological expertise, because sudden neurological diseases other than stroke may represent

42 on September 25, 2021 by guest. Protected copyright. 43 emergencies as well. Messaging in public awareness programs is therefore subject to more 44 45 complex considerations. Besides catchy wording, frequency of symptoms, capture rates, and 46 positive and negative predictive values, the capacity and readiness of regional health care 47 48 systems to cope with stroke mimics will enhance public campaigns. 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 48 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

34 http://bmjopen.bmj.com/ 35 Study Registration 36 The study was registered in the http://www.clinicaltrials.gov registry with the identifier 37 38 number NCT004414583. 39 40 41 Funding

42 on September 25, 2021 by guest. Protected copyright. 43 The sifap1 study (Stroke in Young Fabry Patients, www.sifap.eu; www.clinicaltrials.gov: No. 44 45 NCT00414583) has been supported partially by an unrestricted scientific grant from Shire 46 Human Genetic Therapies. 47 48 49 50 Sponsors’ role in design and conduction of the study, data analysis and interpretation 51 The sponsors of the study had no role in the study design, data collection, data analysis, 52 53 interpretation, writing of the manuscript, or the decision to submit the manuscript for 54 55 publication. The corresponding author had full access to all of the data in the study and takes 56 full responsibility for the integrity of the data and the accuracy of the data analysis, and had 57 58 the final responsibility for the decision to submit for publication. 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 48 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 Data sharing statement 4 5 No additional data available. 6 7 8 Authors’ contributions 9 10 The concept of the present analysis was developed by MC. MC and UG analysed the data. 11 UG conducted the statistical analysis. MC, UG and CT were responsible for the development 12 13 of the manuscript. AR initiated the sifap1 study. GJ, TT, JP, CT participated in the data 14 15 acquisition. For MC, UG, GJ,peer TT, CK, RS,review PJ, BN, AR and only CT were responsible for the 16 intellectual discussion, review, analysis and interpretation of the results. MC, UG, GJ, TT, 17 18 CK, RS, PJ, BN, AR and CT contributed with substantial and important intellectual content in 19 20 drafting and reviewing the manuscript. 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41

42 on September 25, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 48 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 References 4 5 1. Nor AM, Davis J, Sen B, Shipsey D, Louw SJ, Dyker AG, Davis M, Ford GA. The 6 Recognition of Stroke in the Emergency Room (ROSIER) scale: development and 7 validation of a stroke recognition instrument. Lancet Neurol 2005;4:727-34. 8 9 2. Kidwell CS, Starkman S, Eckstein M, Weems K, Saver JL. Identifying stroke in the 10 11 field. Prospective validation of the Los Angeles prehospital stroke screen (LAPSS). 12 Stroke 2000;31:71-6. 13 14 3. Kothari RU, Pancioli A, Liu T, Brott T, Broderick J. Cincinnati Prehospital Stroke 15 Scale:For reproducibility peer and validity. review Ann Emerg Med 1999; only33:373-8. 16 17 4. Kleindorfer DO, Miller R, Moomaw CJ, Alwell K, Broderick JP, Khoury J, Woo D, 18 Flaherty ML, Zakaria T, Kissela BM. Designing a message for public education 19 regarding stroke: does FAST capture enough stroke? Stroke 2007;38:2864-8. 20 21 5. American Stroke Association. Learn to Recognize a Stroke. 2011. Homepage: 22 http://www.strokeassociation.org/STROKEORG/WarningSigns/Warning- 23 Signs_UCM_308528_SubHomePage.jsp. 24 25 26 6. Bray JE, Coughlan K, Barger B, Bladin C. Paramedic diagnosis of stroke: examining 27 long-term use of the Melbourne Ambulance Stroke Screen (MASS) in the field. Stroke 28 2010;41:1363-6. 29 30 7. Lisabeth LD, Brown DL, Hughes R, Majersik JJ, Morgenstern LB. Acute stroke 31 symptoms: comparing women and men. Stroke 2009;40:2031-6. 32 33 8. Hodgson CS. To FAST or not to FAST? Stroke 2007;38:2631-2.

34 http://bmjopen.bmj.com/ 35 9. Harbison J, Hossain O, Jenkinson D, Davis J, Louw SJ, Ford GA. Diagnostic accuracy 36 of stroke referrals from primary care, emergency room physicians, and ambulance 37 staff using the face arm speech test. Stroke 2003;34:71-6. 38 39 40 10. Rolfs A, Martus P, Heuschmann PU, Grittner U, Holzhausen M, Tatlisumak T, 41 Bottcher T, Fazekas F, Enzinger C, Ropele S, Schmidt R, Riess O, Norrving B. Protocol and methodology of the Stroke in Young Fabry Patients (sifap1) study: a 42 on September 25, 2021 by guest. Protected copyright. 43 prospective multicenter European study of 5,024 young stroke patients aged 18-55 44 years. Cerebrovasc Dis 2011;31:253-62. 45 46 11. Rolfs A, Fazekas F, Grittner U, Dichgans M, Martus P, Holzhausen M, Bottcher T, 47 Heuschmann PU, Tatlisumak T, Tanislav C, Jungehulsing GJ, Giese AK, Putaala J, 48 Huber R, Bodechtel U, Lichy C, Enzinger C, Schmidt R, Hennerici MG, Kaps M, 49 Kessler C, Lackner K, Paschke E, Meyer W, Mascher H, Riess O, Kolodny E, 50 Norrving B. Acute cerebrovascular disease in the young: the Stroke in Young Fabry 51 52 Patients study. Stroke 2013;44:340-9. 53 54 12. Bhalla A, Grieve R, Rudd AG, Wolfe CD. Stroke in the young: access to care and 55 outcome; a Western versus eastern European perspective. J Stroke Cerebrovasc Dis 56 2008;17:360-5. 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 48 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 13. Fussman C, Rafferty AP, Lyon-Callo S, Morgenstern LB, Reeves MJ. Lack of 4 association between stroke symptom knowledge and intent to call 911: a population- 5 based survey. Stroke 2010;41:1501-7. 6 7 14. Tirschwell DL, Longstreth WT, Jr., Becker KJ, Gammans RE, Sr., Sabounjian LA, 8 Hamilton S, Morgenstern LB. Shortening the NIH Stroke scale for use in the 9 prehospital setting. Stroke 2002;33:2801-6. 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41

42 on September 25, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 48 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 Table 1. Inclusion criteria of patients in sifap1 [10] 4 5 age 18-55 years 6 Diagnosis Acute CVE of any aetiology (ischemic stroke, 7 8 TIA, intracranial haemorrhage) 9 Time since event Less than 3 months before inclusion in the study 10 11 Verification of diagnosis Verification of brain infarction or haemorrhage 12 by MRI scan 13 14 In case of negative MRI diagnosis confirmed by 15 For peerstroke-experienced review neurologist (more only than 2 16 17 years of experience in stroke and at least 6 years 18 of experience in general neurology) 19 20 Diagnostic information MRI documentation available 21 Diagnostic procedures according to EUSI/ESO 22 23 recommendations 24 25 CVE refers to cerebro-vascular event. 26 TIA refers to transient ischaemic attack 27 MRI refers to magnet-resonance imaging 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41

0.002 <0.001

40.5 59.7 45-55 45-55 n=2706 p

36.8 54.8 35-44 n=1250

34.1 56.2 25-34 25-34 n=443

29.4 50.0 18-24 18-24 n=136

0.002 0.062

BMJ Open 35.8 56.0 http://bmjopen.bmj.com/ n=1860 p Women

40.4 58.9 Men n=2675

% % % % % % % % on September 25, 2021 by guest. Protected copyright. All 76.5 77.2 75.4 0.072 69.1 74.0 75.4 77.8 0.002 38.5 57.7 55.4 57.4 52.6 0.062 47.1 53.3 54.8 56.5 0.014 n=4535 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml For peer review only

Frequency of presenting symptoms according to FAST stratified by gender and age. and age. gender by to stratified according FAST symptoms presenting of Frequency

*Total subsumes patients (%) identified by one or more of the indicated stroke stroke signs. more indicated of the one or by (%) identified subsumes patients *Total Total *Total Face Arm/Paresis Speech

Table 2. Page 17 of 48 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open Page 18 of 48 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 Table 3. Frequency of presenting symptoms not accounted for FAST. 4

5 6 All Men Women p 18-44 44-55 p 7 years years 8 9 % % % % % 10 Vertigo 29.4 29.5 29.4 0.836 30.5 28.7 0.329 11 (n=4496) 12 13 Hemianopia 14.4 14.2 14.7 0.615 16.3 13.1 0.006 14 (n=4535) 15 For peer review only 16 Diplopia 10.3 10.8 9.6 0.201 10.2 10.4 0.840 17 (n=4502) 18 Headache 31.9 27.1 38.7 <0.0001 38.8 27.3 <0.001 19 20 (n=4499) 21 Nausea/Vomiting 19.8 18.0 22.3 <0.001 22.3 18.1 0.001 22 23 (n=4516) 24 Amaurosis fugax 2.3 2.0 2.8 0.113 2.4 2.3 0.760 25 (n=4505) 26 27 Somatosens.deficit 55.9 54.4 58.1 0.032 59.4 53.6 <0.001 28 (n=4147) 29 Gaze deviation 8.4 8.7 7.9 0.320 8.5 8.3 0.884 30 31 (n=4534) 32 Leg paresis 35.0 35.3 34.5 0.543 33.2 36.1 0.095 33

34 (n=4535) http://bmjopen.bmj.com/ 35 36

37 38 39 40 41

p p <0.0001 <0.0001

out 31.0 52.6 Lysis n=1327 With

67.5 86.6 Lysis n=209

p With <0.0001 <0.0001

BMJ Open 92.6 99.5 > 15> http://bmjopen.bmj.com/ n=204 NIHSS

81.8 92.0 6-15 n=927 NIHSS

< 5< 23.5 45.9 n=3403

on September 25, 2021 by guest. Protected copyright. p p NIHSS <0.0001 <0.0001 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

46.4 63.3

For peer review only n=3396 Stroke Ischemic

15.4 41.9 n=1071

68 patients were classified as “other” and are not included here included here not and are “other” as classified were patients 68 1 TIA refers to transient ischaemic attack ischaemic TIA refers to transient NIHSS refers to National Institute of Health stroke scale Health stroke to National of Institute NIHSS refers % % % % % % % % % All All TIA 38.5 57.7

Sings included in the FAST scheme depending on stroke severity, TIA or Stroke, and eligibility for thrombolysis. for and thrombolysis. eligibility or TIA Stroke, on severity, stroke scheme FAST depending the in Sings included

total 76.5 62.3 81.5 <0.0001 69.0 98.9 100.0 <0.0001 96.7 72.6 <0.0001 Face Arm/Paresis Speech 55.4 38.5 61.3 <0.0001 44.5 86.1 98.5 <0.0001 82.3 50.5 <0.0001

Table Table 4.

Page 19 of 48 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open Page 20 of 48 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 Figure 1: Venn diagram of FAST symptoms in those with at least one symptom (n=3469) 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41

42 on September 25, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 21 of 48 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 Figure 2: Initial presenting symptoms (shown as a percentage of all cases) 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41

42 on September 25, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 22 of 48 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 6 7 Clinical Signs in Young Stroke Patients related to FAST 8 – Results of the sifap1 study – 9 FAST Warning Signs in Younger Stroke Patients: Implications for Stroke Education 10 Campaigns 11

12 13 14 Manfred Kaps professor1*, Ulrike Grittner PhD2*, Gerhard Jungehülsing, MD3, Turgut 15 Tatlisumak, PhD4, ChristophFor Kessler peerprofessor5 , Reinhold review Schmidt professor8, Putaalaonly Jukka 16 MD4, Bo Norrving professor7, Arndt Rolfs professor8, Christian Tanislav MD1* on behalf of 17 the sifap1TM Investigators 18 19 *equal contribution 20 21 1Department of Neurology, Justus Liebig University, Giessen, Germany 22 2Department of Biostatistics and Clinical Epidemiology, Charité- University Medicine Berlin, 23 Germany 24 3Department of Neurology, Charité, Berlin, Germany 25 4Department of Neurology, University, Helsinki, Finland 26 5Department of Neurology, Moritz Arndt University Greifswald, Germany 27 6Department of Neurology, University of Graz, Austria 7 28 Department of Neurology, Lund, Sweden 8 29 Albrecht-Kossel Institute for Neuroregeneration, University of Rostock, Rostock, Germany 30 31 32 33

34 http://bmjopen.bmj.com/ 35

37 38 Corresponding author: 39 Christian Tanislav, MD 40 Department of Neurology 41 Justus Liebig University, Giessen, Germany

42 Klinikstr. 33 on September 25, 2021 by guest. Protected copyright. 43 35385 Giessen 44 Germany 45 Tel.: +49/641 98545301; Fax: +49/641 98545409 46 E-mail: [email protected]; Formatted: German (Germany) 47 Formatted: German (Germany) 48 Formatted: German (Germany) 49 50 Key words: young stroke, awareness signs, FAST, clinical presentation. 51 Word count (manuscript): 28823350 52 References: 14 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 23 of 48 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 6 7 Abstract 8

9 10 Objectives The present study aimed to evaluate the frequency of warning signs in younger 11 stroke patients with a special regard to the ‘FAST’ scheme, a public stroke recognition 12 13 instrument (face, arm, speech, act timely). 14 15 For peer review only 16 Setting Primary stroke care in participating centres of a multinational European prospective 17 cross sectional study (Stroke in Young Fabry Patients; sifap1). 47 centres from 15 European 18 19 countries participate in sifap1. 20 21 Participants 5023 acute stroke patients (aged 18-55 years) patients (96.5% Caucasians) were 22 23 enrolled in the study between April 2007 and January 2010. 24 25 26 Primary and secondary outcome measures: sifap1 was originally design to investigate the 27 relation of juvenile stroke and Fabry disease. A secondary aim of sifap1 was to investigate 28 29 stroke patterns in this specific group of patients. The present investigation is a secondary 30 analysis addressing stroke presenting symptoms with a special regard to signs included in the 31 32 FAST scheme. 33

34 http://bmjopen.bmj.com/ Results 4535 patients with TIA (n=1071), ischemic stroke (n=3396) or other (n=68) were 35 36 considered in the presented analysis. FAST symptoms could be traced in 76.5% of all cases. 37 35% of those with at least one FAST symptom had all three symptoms. At least one FAST 38 39 symptom could be recognised in 69.1% of 18-24 years old patients, in 74.0% of those aged 40 25-34 years, in 75.4% of those aged 35-44 years, and 77.8% in 45-55 years old patients. With 41 increasing stroke severity signs included in the FAST scheme were more prevalent 42 on September 25, 2021 by guest. Protected copyright. 43 (NIHSS<5: 69.0%, NIHSS 6-15: 98.9%, NIHSS>15: 100.0%). Clustering clinical signs 44 45 according to FAST lower percentages of strokes in the posterior circulation (65.2%) and in 46 patients with transient ischemic attacks (62.3%) were identified. 47

48 49 Conclusions FAST may be applied as a useful and rapid tool to identify stroke symptoms in 50 young individuals aged 18-55 years. Especially in patients eligible for thrombolysis FAST 51 52 might address the majority of individuals. 53 Background Clinical recognition instruments are important in a complex chain towards 54 55 timely treatment management in acute stroke. The present study aimed to evaluate the 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 24 of 48 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 6 7 frequency of individual warning signs reflected by the ‘FAST’ public stroke recognition 8 instrument (face, arm, speech, act timely) in younger stroke patients. 9

10 11 Methods In a multinational prospective cross sectional study (Stroke in Young Fabry 12 Patients; sifap1) including stroke patients aged below 56 the frequency of presenting 13 14 symptoms was analysed. The performance of the FAST scheme was investigated. 15 For peer review only 16 17 Results 5023 patients were enrolled in the study between 2007 and 2010. Information of 4535 18 patients with TIA (n=1071), ischemic stroke (n=3396) or other (n=68) were selected for this 19 20 analysis. FAST symptoms could be traced in 76.5% of all cases. 35% of those with at least 21 one FAST symptom had all three symptoms. The FAST instrument performed better in older 22 age groups: At least one FAST symptom could be recognised in 69.1% of 18-24 years old 23 24 patients, in 74.0% of those aged 25-34 years, in 75.4% of those aged 35-44 years, and77.8% 25 in 45-55 years old patients. Patients severely disabled (NIHSS<5: 69.0%, NIHSS 6-15: 26 27 98.9%, NIHSS>15: 100.0%) FAST signs were less sensitive for indicating cerebrovascular 28 events as well as in the posterior circulation (65.2%) and in patients with transient ischemic 29 30 attacks (62.3%). 31 32 33 Conclusions Nearly one-quarter of young stroke patients is not targeted by the FAST scheme;

34 the performance is higher in patients with worse stroke severity addressing almost all these http://bmjopen.bmj.com/ 35 individuals. 36 37 38 39 40 41

42 on September 25, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 25 of 48 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 6 7 Article Summary 8 9 Warning signs of stroke are probably the most important step in the complex chain towards 10 timely treatment management. For this purpose different stroke recognition instruments with 11 12 different symptoms and different wording have been developed, offering criteria to identify 13 stroke for public education. Currently many distinguished awareness campaigns to propagate 14 15 the signs of stroke use theFor simplified peerFAST message, whichreview adopts weakness of fonlyace, arm as 16 well as speech and essentially time. In the present study we aimed to ascertain the 17 18 performance of FAST criteria in the sifap1 cohort, a unique prospectively collected 19 population of young (18-55 years) stroke victims. 20 21 The main finding indicates nearly one quarter of young strokes is not targeted by the FAST 22 scheme. However FAST fits very well to capture patients with acute stroke eligible for 23 thrombolysis due to the increasing performance in worse disability. It performs far less 24 25 sensitive for the large number of cases not eligible for thrombolysis but for secondary 26 prevention, for young stroke and for the posterior circulation. However, the main purpose of 27 28 sifap1 was not to validate a stroke recognition instrument. This might by the main limitation 29 of the present investigation. The “FAST wording” was primarily not covered with a specific 30 31 slot in the case report form. FAST items were derived from different sources after the 32 cerebrovascular event (e.g. after employing the NIH Stroke Scale immediately after hospital 33

34 admission (median delay: one day)). http://bmjopen.bmj.com/ 35 In conclusion we confirm FAST as a useful tool in the recognition of strokes in this specific 36 group of young strokes 37 38 Because risk factors and aetiology profiles in the sifap1 cohort resembled those found in 39 elderly stroke patients, conclusions from our study may be also valid in older age groups. 40 41 Strengths and limitations of this study 42 on September 25, 2021 by guest. Protected copyright. 43 44 • The present analysis investigated clinical symptoms in young stroke patients, 45 participants in a unique prospectively collected population of young (18-55 years) 46 47 stroke victims (Stroke in Young Fabry Patients study; sifap1). A special focus was 48 the analysis of symptoms included in the FAST scheme. The FAST message is a tool 49 50 used in awareness campaigns to propagate clinical sign of stroke; it includes 51 weakness of face, arm as well as speech and essentially time. 52 • The main finding indicates that nearly one quarter of young strokes is not targeted by 53 54 the FAST scheme. However, FAST signs are more frequent in patients with acute 55 stroke eligible for thrombolysis. This might be due to increasing stroke severity. 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 26 of 48 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 6 7 • The main purpose of sifap1 was not to validate a stroke recognition instrument. The 8 “FAST wording” was primarily not covered with a specific slot in the case report 9 10 form. FAST items were derived from different sources after the cerebrovascular event 11 (e.g. after employing the NIH Stroke Scale immediately after hospital admission). 12 This limitation need to be taken into account when interpreting our results. 13 14 • However, clustering clinical signs according to FAST the majority of patients 15 included in the sifap1For study arepeer targeted. Therefore, review FAST may be applied only as a useful 16 17 and rapid tool to identify stroke symptoms in young individuals aged 18-55 years. 18 • Because risk factors and aetiology profiles in the sifap1 cohort resembled those found 19 20 in elderly stroke patients, conclusions from our study may be also valid in older age 21 groups. 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41

42 on September 25, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 27 of 48 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 6 7 Introduction 8 9 Getting stroke patients to the acute care hospital on time is the major requirement for effective 10 11 stroke therapy. Knowing warning signs of stroke is the first and most important step in a 12 complex chain towards timely treatment management. Effective public education follows 13 14 simple rules: the message must be simple to remember, effective and consistent. For this 15 purpose different strokeFor recognition instrumentspeer with differentreview symptoms and different only wording 16 17 have been developed [1-6], offering criteria to identify stroke for public education [1-3;7;8, 9]. 18 They are useful for identify strokes in public [5;9], for triage by ambulance paramedics [2;3;6], 19 20 to guide paramedics and emergency physicians to direct patients with acute neurological signs 21 to appropriate care for emergency room physicians [1;2] or as a screening instrument in pre- 22 hospital stroke research [6]. Selection and number of alarming symptoms as well as wording 23 24 differ; three to six warning signs are usually promoted. Currently many distinguished 25 awareness campaigns to propagate the signs of stroke use the simplified FAST message [9], 26 27 which adopts weakness of face, arm as well as speech and essentially time (National Health 28 Service: “When stroke strikes act FAST”). Although the mnemonic FAST is most frequently 29 30 adopted, there are so far no prospective multicentre studies from large cohorts, which evaluate 31 distinct stroke signs for the use in public campaigns. 32 33 The aim of the present study is to investigate symptoms in acute young stroke patients with a

34 special focus on signs included in the FAST scheme in a large cohort of younger stroke patients http://bmjopen.bmj.com/ 35 the Stroke in Young Fabry Patients study (sifap1) and to analyse factors which influence the 36 37 effectiveness in identifying CVEs [10]. This is a secondary analysis of the sifap1 data which 38 was originally supposed to investigate the relation of juvenile stroke and a genetic disorder Field Code Changed 39 Field Code Changed 40 known as Fabry disease. Fabry disease is an x-linked storage disorder which might affect the Field Code Changed 41 entire body. In sifap1 in 0.9% of all patients Fabry disease was identified [11].

42 on September 25, 2021 by guest. Protected copyright. 43 Getting stroke patients to the acute care hospital on time is the major requirement for effective 44 stroke therapy. Knowing warning signs of stroke is the first and most important step in a 45 46 complex chain towards timely treatment management. Effective public education follows 47 simple rules: the message must be simple to remember, effective and consistent. For this 48 purpose different stroke recognition instruments with different symptoms and different wording 49 50 have been developed [1-6], offering criteria to identify stroke for public education [1-3;7;8]. 51 Currently many distinguished awareness campaigns to propagate the signs of stroke use the 52 53 simplified FAST message [9], which adopts weakness of face, arm as well as speech and 54 essentially time (National Health Service: “When stroke strikes act FAST”). Although the 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 28 of 48 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 6 7 mnemonic FAST is most frequently adopted, there are so far no prospective multicentre studies 8 from large cohorts, which evaluate distinct stroke signs for the use in public campaigns. 9

10 11 The aim of the present study is to ascertain the performance of FAST signs in a large cohort of 12 younger stroke patients the Stroke in Young Fabry Patients study (sifap1) [10] and to analyse 13 14 factors which influence the affectivity in identifying CVEs. This is a secondary analysis of the 15 sifap1 data which was originallyFor supposed peer to investigating review the relation of juvenile only stroke and a 16 17 genetic disorder known as Fabry disease. Fabry disease was diagnosed in 0.9% of all patients 18 (n=45) [11]. 19 20 21 Methods 22

23 24 The protocol of the sifap1 study was published recently [10]. Described briefly, the sifap1 25 study was designed as a multicenter multinational prospective observational study of young 26 27 stroke patients across Europe (Table 1). A total of 5023 stroke patients (aged 18-55 years) 28 below 56 years were enrolled in the sifap1 study, 271 patients with primary haemorrhages and 29 30 217 patients without classification of the cerebrovascular event (CVE) were excluded. All 31 patients or legal representatives gave written consent to inclusion. The remaining cohort of 32 33 4535 patients was extensively analysed including detailed medical history,

34 sociodemographics, clinical characteristics, stroke severity, laboratory values, genetics, http://bmjopen.bmj.com/ 35 cardiac work up as well as presenting symptoms on hospital admission. Neurological deficits 36 37 were measured at the time of maximum impairment according to previous hospital-based 38 stroke registers [12]. A transient ischemic attack (TIA) was defined as a CVE with clinical 39 40 symptoms lasting <24 hours. Cerebral magnetic resonance imaging with standardized MRI 41 sequences was a mandatory procedure. Images were assessed centrally at the Department of

42 on September 25, 2021 by guest. Protected copyright. 43 Neurology, Medical University of Graz, Austria, blinded to clinical and demographic data. 44 Items according to FAST were constructed as follows: 45 46 Face: facial palsy (minor asymmetry, partial or complete) according to NIHSS item (the scale 47 was assessed within 48h after admission).. 48 Arm/Paresis: left or right arm some effort against gravity, no effort against gravity or no 49 50 movement according to NIHSS item (the scale was assessed within 48h after admission) or 51 Paresis in arm or leg according to presenting symptoms (documented on inclusion in the 52 53 study). 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 29 of 48 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 6 7 Speech: severe aphasia or mute according to NIHSS (documented on inclusion in the study), 8 or dysarthria (mid-moderate slurring, or severe, nearly intelligible or worse) according to 9 NIHSS (documented on inclusion in the study) or dysphasia or aphasia or dysarthria 10 11 according to presenting symptoms (documented on inclusion in the study). 12 The effectiveness of Based on MRI data theclinical signs addressed by FAST FAST message 13 14 were analysed in relation to the different vascular territories territorieswas based on MRI 15 data. Anterior circulationFor included peer the anterior and review middle cerebral artery, theonly posterior 16 17 circulation the vertebrobasilar territory and posterior cerebral artery. 18 Apart from NIHSS, the following presenting symptoms were recorded additionally in the case 19 20 report form on inclusion: headache, nausea-vomiting, hemianopia, diplopia and vertigo. 21 According to the case report form (CRF) 210 out of 1537 patients underwent intravenous 22 thrombolysis. Information about thrombolysis was missing for 2999 patients because this item 23 24 was included later and not asked for all patients. The indication for thrombolysis was left to 25 the participating centres, according to their local structured operating procedures. 26 27 Frequencies of presenting symptoms were analysed in two modifications. At first the number 28 of cases exhibiting FAST symptoms i.e. face palsy or problems with arm/paresis or speech 29 30 were simply calculated as percentage from the whole cohort. Additionally a Venn diagram 31 was plotted for those with at least one FAST symptom to check frequencies of combinations 32 33 of symptoms.

34 Alternatively, the frequency of each symptom was calculated in a sequential approach. After http://bmjopen.bmj.com/ 35 sizing the most frequent symptom, the second frequent symptom was extracted from the 36 37 remaining cases and so on.. This approach allows establishing a “hierarchy” of stroke signs 38 with regard to those symptoms that most often occur aiming to identify as many as possible 39 40 stroke patients employing as few symptoms as possible to create a comprehensive public 41 message.

42 on September 25, 2021 by guest. Protected copyright. 43 To test differences in subgroups according to age, sex or severity of stroke we used multilevel 44 logistic models to account for the heterogeneity between centres. All analyses were calculated 45 46 using commercially available software: PASW Statistics 18, Release Version 18.0.2 (© SPSS, 47 Inc., 2009, Chicago, IL, USA) and SAS software, Version 9.2 of the SAS System for 48 Windows. (© 2008 SAS Institute Inc., Cary, NC, USA). 49 50 51 Results 52 53 54 Frequency of presenting symptoms 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 30 of 48 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 6 7 5024 patients were enrolled in the study between 2007 and 2010. 4535 patients with TIA 8 (n=1071), ischemic stroke (n=3396) or other (n=68, refers to cerebral vein thrombosis or no 9 documented entity) were included in the analysis. In general 76.5 % of the young strokes 10 11 included in sifap1 had clinical signs covered by FAST criteria. Information of 4535 patients 12 with TIA (n=1071), ischemic stroke (n=3396) or other (n=68) were selected for this analysis. 13 14 In general 76.5 % of the young stroke cohort could be captured by FAST criteria. Table 2 15 specifies the frequencyFor of FAST symptomspeer according review to gender and age. Face only was more 16 17 frequently affected in men. FAST signs were more prevalent in older age groups; . FAST 18 performed better in older age groups, this difference becomes even more obvious in younger 19 20 patients below 25 years where the capture rate of FAST signs is below 70%. 21 Figure 1 shows the frequency of combinations of FAST symptoms. 34.7% of those with at 22 least one FAST symptom have all three symptoms. 17.3% have only arm problems or paresis. 23 24 16.6% have only speech problems. 15.7 % have both arm problems/paresis and speech 25 problems. The isolated symptom “Face” is very rare (2.5%). 26 27 Symptoms which were registered in the sifap1 cohort but were not considered for FAST are 28 listed in Table 3. These “non-traditional” stroke symptoms (not included in FAST) also 29 30 disclosed age and gender differences: headache and nausea were more common in women 31 even after controlling for migraine and age (results not shown), whereas hemianopia, 32 33 headache, nausea, and somatosensory deficits were more common in younger age groups

34 even after controlling for migraine and sex (results not presented). http://bmjopen.bmj.com/ 35 Capturing cerebrovascular events by FAST signs depended on the severityWith increasing Formatted: Not Highlight 36 37 stroke severity signs included in the FAST scheme were more prevalent (Table 4). FAST 38 signs were less frequent in TIA than in strokes (62.3% vs. 81.5%). Severe strokes were nearly 39 40 completely covered by FAST signs when a NIH score of at least 6 was present. FAST signs Formatted: Not Highlight 41 occurred in 96.7% of patients who received thrombolysis.FAST signs highlighted 96.7% of

42 on September 25, 2021 by guest. Protected copyright. 43 patients who received thrombolysis. 44 45 46 Association between presenting symptoms and MRI lesions 47 1419 patients had no infarct in MRI, more than half of those patients (56.5%) had a TIA. 48 From those 1419 patients only 65.1% were detected by the FAST items. For 252 patients no 49 50 information about lesions in MRI was available. For patients with definite infarct lesions in 51 MRI (n=2865) we tested the differences in occurrence of symptoms included in the FAST 52 53 scheme regarding vascular territories: FAST symptoms were less frequent in patients with 54 strokes in the posterior circulation, where only 65.2% of all cases with an apparent MRI 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 31 of 48 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 6 7 lesion could be identified, whereas 92.4% of anterior circulation strokes matched symptoms Formatted: Not Highlight 8 included in the FAST criteria. There was no difference between left and right hemispheric 9 stroke (82.9% vs 82.1%).1419 patients had no infarct in MRI, more than half of those patients 10 11 (56.5%) had a TIA. From those 1419 patients only 65.1% were detected by the FAST items. 12 For 252 patients no information about lesions in MRI was available. For patients with definite 13 14 infarct lesions in MRI (n=2865) we tested the differences in performance of FAST regarding 15 vascular territories: theFor FAST instrument peer performed worse review in the posterior circulation, only where 16 17 only 65.2% of all cases with an apparent MRI lesion could be depicted, whereas 92.4% of 18 anterior circulation strokes matched FAST criteria. There was no difference between left and 19 20 right hemispheric stroke (82.9% vs 82.1%). 21 22 Hierarchy of presenting symptoms 23 24 The item “arm/paresis” was by far the most frequent sign in all stroke patients (aged 18-55 Formatted: Not Highlight 25 years)aged below 56 years, and “speech” the second (Fig.2). Both items together were 26 27 covering almost 75 % of all recruited stroke and TIA patients. Under hierarchic perspective 28 the frequency of the “face” item (0.2%) becomes irrelevant if the four most common signs of 29 30 stroke “arm/paresis” (57.7%), “speech” (16.9%), “vertigo” (10.6%) and “somatosensory 31 deficit” (6.8%) were prior sequentially used as selection criteria [13;14]. Hemianopia 32 33 occurred in general with a frequency of 14.4% but only 3.2% of the cases experienced this

34 symptom independent from arm/paresis, speech, vertigo or somatosensory deficits. Common http://bmjopen.bmj.com/ 35 stroke signs in young patients such as headache, leg paresis and diplopia were seldom or not 36 37 independently differentiated from other stroke signs. 38

39 40 Discussion 41

42 on September 25, 2021 by guest. Protected copyright. 43 Principal findings 44 Frequency of presenting symptoms 45 46 In our cohort the FAST symptoms could be traced in 76.5% of all cases. This is notably less 47 than in a previously reported study in which study nurses screened retrospectively the medical 48 records of 3498 stroke cases who presented in an acute care hospital [4]. The “capture rate” in 49 50 this study was 88.9% for all strokes including haemorrhage and even higher for ischemic 51 stroke and TIA (91.1% and 91.8% resp.). Interestingly Kleindorfer and co-workers found that 52 53 cases missed by FAST tended to be significantly younger than cases with FAST symptoms 54 (mean 68.9 vrs 71.5 yrs) [4]. Even though Kleindorfer and co-workers investigates a general Formatted: Not Highlight 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 32 of 48 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 6 7 stroke population, in our selected cohort with younger patients (median age 46 years), we 8 observed a comparable trend with significantly fewer patients presenting with symptoms 9 covered by FAST in younger age groups. Therefore clinical signs included in FAST might be Formatted: Not Highlight 10 11 less prevalent in younger stroke patients.Interestingly Kleindorfer and co-workers found that 12 cases missed by FAST tended to be significantly younger than cases with FAST symptoms 13 14 (mean 68.9 vrs 71.5 yrs) [4]. In our selected cohort with younger patients (median age 46 15 years), we observed a For comparable trendpeer with significantly review fewer patients presenting only with 16 17 symptoms covered by FAST in younger age groups. We therefore conclude that FAST might 18 perform less robustly in younger stroke patients. 19 20 With respect to FAST signs there was no gender difference in stroke presentation. Other 21 presenting symptoms such as headache, nausea/vomiting and somatosensory deficits occurred 22 more frequently in women. This specific feature could not be explained by larger number of 23 24 vertebro-basilar strokes in women. Gender differences in these symptoms were also 25 significant when TIA patients were excluded. In a previous study on gender differences in 26 27 acute stroke symptoms a higher prevalence of symptoms termed “non-traditional” (pain, 28 mental status change, light-headedness, headache, non-neurological symptoms) could be 29 30 registered in women [7]. However, there were no differences observed regarding traditional 31 symptoms such as hemiparesis, aphasia, facial weakness, hemibody numbness, which is in 32 33 line with our findings [7].

34 Clinical signs clustered according to the FAST scheme disclosed more stroke patients with Formatted: Not Highlight http://bmjopen.bmj.com/ 35 increasing stroke severity. Nearly all cases who received thrombolysis irrespective of gender 36 37 were indicated by symptoms included in FAST.FAST performed better with increasing stroke 38 severity and disclosed nearly all cases who received thrombolysis irrespective of gender. This 39 40 is of major interest, because a recent metaanalysis indicated that women with stroke had 30% 41 lower odds of receiving tissue plasminogen activator thought possibly to result from different

42 on September 25, 2021 by guest. Protected copyright. 43 symptom presentation in women. 44 Formatted: Tab stops: 3.69", Left + Not at 45 0.22" 46 Hierarchy of presenting symptoms 47 48 “Arm/paresis” and “Speech” are by far the most relevant ischemic stroke signs. “Face” can 49 50 be quantitatively neglected, because it occurred as an isolated symptom very rarely. 51 Nevertheless, 34.7% of patients with at least one FAST symptom report all three FAST 52 53 symptoms, meaning that “Face” is rare as an isolated symptom but often accompanied by 54 “Arm/Paresis” or “Speech”. Similarly, headache is not relevant to identify ischemic stroke but 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 33 of 48 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 6 7 may characterize cases with subarachnoid haemorrhage. Remarkably, one single stroke sign 8 (“Arm/Paresis”) was effective enough to select 86.7% of patients in the subgroup of patients 9 who underwent thrombolytic therapy. However, clinical signs included in the FAST scheme Formatted: Not Highlight 10 11 were absent in 23.5% of stroke patients included in sifap1. Moreover 37.7% of patients with 12 TIA had symptoms other than those mentioned by FAST. Signs included in the FAST scheme Formatted: Not Highlight 13 14 were also less frequent in young patients with vertebro-basilar infarctions (64.5%). This is 15 relevant, because “non-FAST”For signs peer are frequently seenreview as first symptoms of only progressing 16 17 stroke, especially in basilar thrombosis and patients missed by FAST do not necessarily have 18 a benign outcome [4]. 19 20 “Arm/paresis” and “Speech” are by far the most relevant ischemic stroke signs. “Face” can 21 be quantitatively neglected, because it occurred as an isolated symptom very rarely. 22 Nevertheless, 34.7% of patients with at least one FAST symptom report all three FAST 23 24 symptoms, meaning that “Face” is rare as an isolated symptom but often accompanied by 25 “Arm/Paresis” or “Speech”. Similarly, headache is not relevant to identify ischemic stroke but 26 27 may characterize cases with subarachnoid haemorrhage. Remarkably, one single stroke sign 28 (“Arm/Paresis”) was effective enough to select 86.7% of patients in the subgroup of patients 29 30 who underwent thrombolytic therapy. In other words the FAST message is very capable of 31 capturing (or detecting) nearly all stroke patients eligible for thrombolysis. However FAST 32 33 did not apply for 23.5% of stroke patients in our cohort. Moreover 37.7% of patients with TIA

34 were not addressed by the symptoms in the FAST message. Although this is not relevant for http://bmjopen.bmj.com/ 35 thrombolytic treatment, it may exclude many young patients at risk for stroke from mandatory 36 37 secondary prevention. Performance was also worse in young patients with vertebrobasilar 38 infarctions (64.5%). This is relevant, because “non-FAST” signs are frequently seen as first 39 40 symptoms of progressing stroke especially in basilar thrombosis and patients missed by FAST 41 do not necessarily have a benign outcome [4].

42 on September 25, 2021 by guest. Protected copyright. 43 44 Other recognition instruments 45 46 Diverse stroke recognition instruments have been developed in order to identify strokes in 47 public [5;9], for triage by ambulance paramedics [2;3;6], to guide paramedics and emergency 48 physicians to direct patients with acute neurological signs to appropriate care for emergency 49 50 room physicians [1;2] or as a screening instrument in pre-hospital stroke research [6;14]. 51 Selection and number of alarming symptoms as well as wording differ; three to six warning 52 53 signs are usually promoted. 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 34 of 48 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 6 7 The ASA Stroke Warning Signs [5] emphasize the term “sudden” as prefix, and add a broad 8 spectrum of signs (weakness of face, arm or leg, confusion, trouble speaking or 9 understanding, trouble seeing in one or both eyes, trouble walking, dizziness, loss of balance 10 11 or coordination, severe headache). This extensive summary of neurological symptoms is 12 likely to cover every brain injury. This kind of messaging takes advantage of the fact that 13 14 acute onset (“sudden”) is the most discriminating factor between stroke and non-stroke and 15 targets 96% of all strokesFor and TIA butpeer can include 47% review of differential non- stroke only diagnoses 16 17 [1]. The Cincinnati Prehospital Stroke Scale (CPSS) [3] is a 3-item scale derived from a 18 simplification of the National Institutes of Health (NIH) Stroke Scale. It evaluates facial 19 20 droop and arm drift, speech is tested by asking the patient to repeat sentences. Thereby the 21 CPSS design is very similar to FAST [9] and its reproducibility has proven excellent among 22 pre-hospital care providers [3]. In agreement with our findings, the CPSS has also proven 23 24 high validity to identify candidates for thrombolysis and strokes preferably in the anterior 25 circulation. The Los Angeles Prehospital Stroke Screen (LAPPS) [2] considers unilateral 26 27 motor weakness (“facial smile/grimace”, hand grip and arm strength/drift) as well as 4 28 screening criteria (Age > 45, history of seizure disorder absent, symptoms duration less than 29 30 24 hours, not wheel chair bound or bedridden prior to the event) and a glucose measurement. 31 It was designed to allow pre-hospital personnel to rapidly identify most common stroke 32 33 patients and exclude those unlikely to qualify for or benefit from acute interventions. Thereby

34 stroke patients below 45 years or with longer symptom duration were purposefully not http://bmjopen.bmj.com/ 35 targeted. The Melbourne Ambulance Stroke Screen (MASS) [6] combines clinical signs of 36 37 the CPSS with the LAPPS screening criteria. The ROSSIER scale [1] lists 5 key signs 38 scoring positive (asymmetric facial/arm/leg weakness, speech disturbances and visual field 39 40 defect) and 2 items scoring negative (loss of consciousness and syncope, seizure activity) 41 adding to a total score ranging between -2 and +5. The purpose of the ROSSIER design was

42 on September 25, 2021 by guest. Protected copyright. 43 to develop a simple and practical instrument for emergency room physicians in order to 44 reduce the number of non-stroke referrals from emergency room to stroke unit. 45 46 Stroke recognition instruments must be differentiated with regard to the addressee. For public 47 education and campaigns there is no choice but to promote a selected number of clinical 48 signs. Additional assessments (i.e. glucose measurements) to rule out non-strokes or stroke 49 50 subgroups are reserved for paramedic use or for triage purposes in hospital. 51

52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 35 of 48 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 6 7 Strengths and weaknesses of the study 8 9 The sifap1 study is by far the largest prospective multicenter study analysing presenting 10 11 stroke symptoms in the young with broad clinical work up including MRI. Age limit up to 55 12 years allows exploring subgroups with increasing age including patients with risk factor and 13 14 aetiology profiles resembling those found in elderly stroke patients [10]. Therefore the 15 findings may be extrapolatedFor to some peer extend to the stroke review population in general. Neverthelessonly 16 17 there are some limitations regarding this study: The main purpose of sifap1 was not to 18 validate a stroke recognition instrument. Therefore “FAST wording” was primarily not 19 20 covered with a specific item in the CRF. Instead we asked for paresis directly after CVE and 21 employed the NIH Stroke Scale immediately after hospital admission (median delay: one 22 day). Assessments ruling out stroke mimics (i.e. blood glucose level) influence specificity but 23 24 cannot be used for public education [6]. Because we excluded stroke mimics by definition, 25 our study is not designed to calculate for positive and negative predictive values of distinct 26 27 stroke signs. Moreover, there is a problem in general to calculate for false-negative diagnoses, 28 i.e. undiagnosed strokes in a population under survey. It has to be noted also, that we did not 29 30 consider haemorrhage, subarachnoid haemorrhage or venous thrombosis in our calculations. 31 Addressing all these strokes types may further add to complexity and dilute the awareness 32 33 message. The FAST scheme was develop to screen for potential stroke victims in a preclinical

34 setting [9]. In contrast our patients were included after admission to a neurological http://bmjopen.bmj.com/ 35 department. This needs to be taken into account when interpreting our results. 36 37 Extensive MRI documentation allowed us to validate the clinical stroke diagnosis and 38 constituted a robust additional aspect in identifying presenting symptoms in acute young 39 40 stroke patients. One-third of vertebrobasilar strokes and transient attacks could not be 41 visualized on MRI. In these cases the appraisal of an experienced neurologist was decisive.

42 on September 25, 2021 by guest. Protected copyright. 43 Notably, there was no relevant difference regarding signs included in the FAST scheme 44 comparing patients with and without proven MRI lesions. 45 46 Notably there was no relevant difference comparing the FAST performance in patients with 47 and without proven MRI lesions. 48 Extensive MRI documentation allowed us to validate the clinical stroke diagnosis and Formatted: Tab stops: Not at 0.22" 49 50 constituted a robust additional aspect of FAST performance. One-third of vertebrobasilar 51 strokes and transient attacks could not be visualized on MRI. In these cases the appraisal of an 52 53 experienced neurologist was decisive. Notably there was no relevant difference comparing the 54 FAST performance in patients with and without proven MRI lesions. 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 36 of 48 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 6 7 8 Implications for public campaigns 9

10 11 Instruments that help the lay public to identify stroke in pre-hospital setting are elementary to 12 trigger early treatment. Our study in stroke patients (aged 18-55 years) proves that symptoms Formatted: Not Highlight 13 14 considered in the FAST scheme may be useful for identifying young stroke patients. 15 Especially young strokeFor patients eligiblepeer for thrombolysis review might be targeted only by a FAST 16 17 evaluation. In contrast, clustering only clinical symptoms according to FAST, it might be less 18 effective in young stroke patients with TIA and infarcts in the posterior circulation. Our study 19 20 in stroke patients below 56 proves that FAST fits very well to capture patients with acute 21 stroke eligible for thrombolysis. It is much less sensitive for the large number of cases not 22 eligible for thrombolysis but for secondary prevention, for young stroke and for the posterior 23 24 circulation. A tailored message to reach especially women would not increase performance of 25 FAST, because “non-traditional symptoms” occurred infrequently in isolation from common 26 27 stroke symptoms. Because risk factors and aetiology profiles in the sifap1 cohort resembled 28 those found in elderly stroke patients [10], conclusions from our study may be also valid in 29 30 older age groups. 31 Design and usefulness of stroke recognition instruments depend strongly on the context and 32 33 the target group to be addressed. The proper diagnosis of stroke requires profound clinical

34 knowledge, which cannot be expected from lay persons or ambulance paramedics. On one http://bmjopen.bmj.com/ 35 hand simplified but specific signs are needed for public campaigns, on the other hand stroke 36 37 recognition in pre-hospital settings cannot be simplified without losing a substantial number 38 of patients. Obviously, signs such as “Vertigo” and “Somatosensory Deficits” target much 39 40 more stroke patients but coincidently capture more stroke mimics and thereby dilute the 41 message and burden the health care system.

42 on September 25, 2021 by guest. Protected copyright. 43 It is important to recognize that sudden neurological symptoms need expeditious referral and 44 neurological expertise, because sudden neurological diseases other than stroke may represent 45 46 emergencies as well. Messaging in public awareness programs is therefore subject to more 47 complex considerations. Besides catchy wording, frequency of symptoms, capture rates, and 48 positive and negative predictive values, the capacity and readiness of regional health care 49 50 systems to cope with stroke mimics will enhance public campaigns. 51

52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 37 of 48 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 6 7 Acknowledgements 8 We thank especially Kristin Brüderlein, Ines Federow, Doreen Niemann, Gesine Makowei, 9 Frances König, Jan Burmeister, and Carla Biedermann for organizational aspects. In 10 11 particular, Sabine Rösner and Susanne Zielke spent a lot of effort in realizing the entire 12 workflow. Elmar Beck (Anfomed GmbH, Möhrendorf, Germany) programmed the web-based 13 14 data bank as well as hosted the data. 15 For peer review only 16 17 Conflict of interest disclosures 18 All authors have completed the Unified Competing Interest form at 19 20 http://www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) 21 and declare that they have no non-financial interests that may be relevant to the submitted 22 work. 23 24 25 Ethical approval 26 27 The design of the study was approved by the ethics committee of the Medical Association 28 Mecklenburg-Vorpommern (board 2), Medical Faculty, and University of Rostock. Approval 29 30 of the local ethics committees was obtained in all participating centers. The study was 31 registered in www.clinicaltrials.gov and (No. NCT00414583). 32 33

34 Study Registration http://bmjopen.bmj.com/ 35 The study was registered in the http://www.clinicaltrials.gov registry with the identifier 36 37 number NCT004414583. 38

39 40 Funding 41 The sifap1 study (Stroke in Young Fabry Patients, www.sifap.eu; www.clinicaltrials.gov: No. Field Code Changed

42 on September 25, 2021 by guest. Protected copyright. 43 NCT00414583) has been supported partially by an unrestricted scientific grant from Shire 44 Human Genetic Therapies. 45 46 47 Sponsors’ role in design and conduction of the study, data analysis and interpretation 48 The sponsors of the study had no role in the study design, data collection, data analysis, 49 50 interpretation, writing of the manuscript, or the decision to submit the manuscript for 51 publication. 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 38 of 48 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 6 7 The corresponding author had full access to all of the data in the study and takes full 8 responsibility for the integrity of the data and the accuracy of the data analysis, and had the 9 final responsibility for the decision to submit for publication. 10 11 12

13 14 15 Data sharing statementFor peer review only 16 17 No additional data available. 18 19 20 Authors’ contributions 21 Authors’ contributions 22 The concept of the present analysis was developed by MC. MC and UG analysed the data. 23 24 UG conducted the statistical analysis. MC, UG and CT were responsible for the development 25 of the manuscript. AR initiated the sifap1 study. GJ, TT, JP, CT participated in the data 26 27 acquisition. MC, UG, GJ, TT, CK, RS, PJ, BN, AR and CT were responsible for the 28 intellectual discussion, review, analysis and interpretation of the results. MC, UG, GJ, TT, 29 30 CK, RS, PJ, BN, AR and CT contributed with substantial and important intellectual content in 31 drafting and reviewing the manuscript. 32 33 The concept of the present analysis was developed by MC. MC and UG analysed the data. All

34 authors were involved in the analysis and interpretation of the data and in drafting the http://bmjopen.bmj.com/ 35 manuscript. All authors read and approved the final manuscript. 36 37 38

39 40 41

42 on September 25, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 39 of 48 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 6 7 References 8 1. Nor AM, Davis J, Sen B, Shipsey D, Louw SJ, Dyker AG, Davis M, Ford GA. The 9 Recognition of Stroke in the Emergency Room (ROSIER) scale: development and 10 validation of a stroke recognition instrument. Lancet Neurol 2005;4:727-34. 11 12 2. Kidwell CS, Starkman S, Eckstein M, Weems K, Saver JL. Identifying stroke in the 13 field. Prospective validation of the Los Angeles prehospital stroke screen (LAPSS). 14 Stroke 2000;31:71-6. 15 For peer review only 16 3. Kothari RU, Pancioli A, Liu T, Brott T, Broderick J. Cincinnati Prehospital Stroke 17 Scale: reproducibility and validity. Ann Emerg Med 1999;33:373-8. 18 19 4. Kleindorfer DO, Miller R, Moomaw CJ, Alwell K, Broderick JP, Khoury J, Woo D, 20 Flaherty ML, Zakaria T, Kissela BM. Designing a message for public education 21 regarding stroke: does FAST capture enough stroke? Stroke 2007;38:2864-8. 22 5. American Stroke Association. Learn to Recognize a Stroke. 2011. Homepage: 23 http://www.strokeassociation.org/STROKEORG/WarningSigns/Warning- 24 Signs_UCM_308528_SubHomePage.jsp. 25 5. American Stroke Association. Learn to Recognize a Stroke. 26 http://www.strokeassociation.org/STROKEORG/WarningSigns/Warning- 27 Signs_UCM_308528_SubHomePage.jsp. 2011. 28 29 6. Bray JE, Coughlan K, Barger B, Bladin C. Paramedic diagnosis of stroke: examining 30 long-term use of the Melbourne Ambulance Stroke Screen (MASS) in the field. Stroke 31 2010;41:1363-6. 32 33 7. Lisabeth LD, Brown DL, Hughes R, Majersik JJ, Morgenstern LB. Acute stroke

34 symptoms: comparing women and men. Stroke 2009;40:2031-6. http://bmjopen.bmj.com/ 35 36 8. Hodgson CS. To FAST or not to FAST? Stroke 2007;38:2631-2. 37 9. Harbison J, Hossain O, Jenkinson D, Davis J, Louw SJ, Ford GA. Diagnostic accuracy 38 of stroke referrals from primary care, emergency room physicians, and ambulance 39 staff using the face arm speech test. Stroke 2003;34:71-6. 40 41 10. Rolfs A, Martus P, Heuschmann PU, Grittner U, Holzhausen M, Tatlisumak T,

42 Bottcher T, Fazekas F, Enzinger C, Ropele S, Schmidt R, Riess O, Norrving B. on September 25, 2021 by guest. Protected copyright. 43 Protocol and methodology of the Stroke in Young Fabry Patients (sifap1) study: a 44 prospective multicenter European study of 5,024 young stroke patients aged 18-55 45 years. Cerebrovasc Dis 2011;31:253-62. 46 47 11. Rolfs A, Fazekas F, Grittner U, Dichgans M, Martus P, Holzhausen M, Bottcher T, 48 Heuschmann PU, Tatlisumak T, Tanislav C, Jungehulsing GJ, Giese AK, Putaala J, 49 Huber R, Bodechtel U, Lichy C, Enzinger C, Schmidt R, Hennerici MG, Kaps M, 50 Kessler C, Lackner K, Paschke E, Meyer W, Mascher H, Riess O, Kolodny E, 51 Norrving B. Acute cerebrovascular disease in the young: the Stroke in Young Fabry 52 Patients study. Stroke 2013;44:340-9. 53 12. Bhalla A, Grieve R, Rudd AG, Wolfe CD. Stroke in the young: access to care and 54 outcome; a Western versus eastern European perspective. J Stroke Cerebrovasc Dis 55 2008;17:360-5. 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 40 of 48 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 6 7 13. Fussman C, Rafferty AP, Lyon-Callo S, Morgenstern LB, Reeves MJ. Lack of 8 association between stroke symptom knowledge and intent to call 911: a population- based survey. Stroke 2010;41:1501-7. 9 10 14. Tirschwell DL, Longstreth WT, Jr., Becker KJ, Gammans RE, Sr., Sabounjian LA, 11 Hamilton S, Morgenstern LB. Shortening the NIH Stroke scale for use in the 12 prehospital setting. Stroke 2002;33:2801-6. 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41

42 on September 25, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 41 of 48 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 6 7 Table 1. Inclusion criteria of patients in sifap1 [10] 8 age 18-55 years 9 Diagnosis Acute CVE of any etiology (ischemic stroke, 10 11 TIA, intracranial hemorrhage) 12 Time since event Less than 3 months before inclusion in the study 13 Verification of diagnosis Verification of brain infarction or hemorrhage 14 15 Forby MRI scanpeer review only 16 In case of negative MRI diagnosis confirmed by 17 stroke-experienced neurologist (more than 2 18 19 years of experience in stroke and at least 6 years 20 of experience in general neurology) 21 Diagnostic information MRI documentation available 22 23 Diagnostic procedures according to EUSI/ESO 24 recommendations 25 CVE refers to cerebro-vascular event. Formatted: Not Highlight 26 TIA refers to transient ischaemic attack 27 MRI refers to magnet-resonance imaging 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41

42 on September 25, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from Page 42 of 48

<0.001

40.5 45-55 N=2706 p

36.8 35-44 N=1250 N=1250

34.1 25-34 N=443

29.4 18-24 N=136 BMJ Open

http://bmjopen.bmj.com/ 0.002

35.8 N=1860 p Women

40.4 Men Men on September 25, 2021 by guest. Protected copyright. N=2675

% % % % % % % For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml All 76.5 77.2 75.4 0.072 69.1 74.0 75.4 77.8 0.002 57.7 58.9 56.0 0.062 50.0 56.2 54.8 59.7 0.002 55.4 57.4 52.6 0.062 47.1 53.3 54.8 56.5 0.014 38.5 Forn=4535 peer review only Frequency of presenting symptoms according to FAST stratified by gender and age. stratified by to gender according symptoms FAST presenting of Frequency

signs. stroke indicatedof the one or identified more by (%) patients subsumes *Total Total * Speech Face Arm/Paresis

2.Table 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 43 of 48 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 6 7 Table 3. Frequency of presenting symptoms not accounted for FAST. 8 9 All Men Women p 18-44 44-55 p 10 yrs yrs 11 12 % % % % % 13 Vertigo 29.4 29.5 29.4 0.836 30.5 28.7 0.329 14 (n=4496) 15 Hemianopia 14.4 For 14.2 14.7peer 0.615 16.3review 13.1 0.006 only 16 (n=4535) 17 18 Diplopia 10.3 10.8 9.6 0.201 10.2 10.4 0.840 19 (n=4502) 20 Headache 31.9 27.1 38.7 <0.0001 38.8 27.3 <0.001 21 (n=4499) 22 Nausea/Vomiting 19.8 18.0 22.3 <0.001 22.3 18.1 0.001 23 24 (n=4516) 25 Amaurosis fugax 2.3 2.0 2.8 0.113 2.4 2.3 0.760 26 (n=4505) 27 Somatosens.deficit 55.9 54.4 58.1 0.032 59.4 53.6 <0.001 28 (n=4147) 29 30 Gaze deviation 8.4 8.7 7.9 0.320 8.5 8.3 0.884 31 (n=4534) 32 Leg paresis 35.0 35.3 34.5 0.543 33.2 36.1 0.095 33 (n=4535) 34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41

42 on September 25, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from Page 44 of 48 Justified, Indent: Left: 0.98", Justified, Left: Indent: Formatted: lines After:1.5 spacing: Line 0 pt, Space

p p <0.0001

out out

% 50.5 Lysis n=1327 With

% BMJ Open 82.3 Lysis n=209

http://bmjopen.bmj.com/

p p With <0.0001

% 98.5 > 15 NIH n=204 n=204

% < 5 44.5 n=3403

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

p NIH

For peer<0.0001 review only

% 61.3 n=3396 Stroke Ischemic Ischemic

% 38.5 n=1071

68 patients were classified as “other” and are not included here here not included and are “other” as classified were patients 68

TIA refers to transient ischaemic toattack transient refersischaemic TIA 1 NIHSS refers to National Institute of Health stroke scaleof stroke Health refersInstitute to National NIHSS % All All TIA 55.4 Sings Sings included in the FAST scheme depending on stroke severity, TIA or Stroke, and eligibility for thrombolysis.Performance of FAST

Speech total 76.5 62.3 81.5 <0.0001 69.0 98.9 100.0 <0.0001 96.7 72.6 <0.0001 Arm/Paresis 57.7 41.9 63.3 <0.0001 45.9 92.0 99.5 <0.0001 86.6 52.6 <0.0001 Face 38.5 15.4 46.4 <0.0001 23.5 81.8 92.6 <0.0001 67.5 31.0 <0.0001

depending on stroke severity, TIA or Stroke, and eligibility for thrombolysis. for thrombolysis. and Stroke, or eligibility severity, stroke onTIA depending Table Table 4. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 45 of 48 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 6 7 Figure 1: Venn diagram of FAST symptoms in those with at least one symptom (n=3469) 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41

42 on September 25, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 46 of 48 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 6 7 Figure 2: Initial presenting symptoms (shown as a percentage of all cases) 8

9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41

42 on September 25, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 47 of 48 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41

42 on September 25, 2021 by guest. Protected copyright. 43 123x130mm (300 x 300 DPI) 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 48 of 48 BMJ Open: first published as 10.1136/bmjopen-2014-005276 on 7 November 2014. Downloaded from

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 175x145mm (300 x 300 DPI) 37 38 39 40 41

42 on September 25, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml