Pediatric Neurology 58 (2016) 53e56
Contents lists available at ScienceDirect
Pediatric Neurology journal homepage: www.elsevier.com/locate/pnu
Original Article
Sensitivity and Specificity of an Adult Stroke Screening Tool in Childhood Ischemic Stroke Kerri Neville MD a, Warren Lo MD a, b, * a b
Department of Pediatrics, The Ohio State University and Nationwide Children’s Hospital, Columbus, Ohio Department of Neurology, The Ohio State University and Nationwide Children’s Hospital, Columbus, Ohio
abstract BACKGROUND: There are frequent delays in the diagnosis of acute pediatric ischemic stroke. A screening tool that could increase the suspicion of acute ischemic stroke could aid early recognition and might improve initial care. An earlier study reported that children with acute ischemic stroke have signs that can be recognized with two adult stroke scales. We tested the hypothesis that an adult stroke scale could distinguish children with acute ischemic stroke from children with acute focal neurological deficits not due to stroke. METHODS: We retrospectively applied an adult stroke scale to the recorded examinations of 53 children with acute symptomatic acute ischemic stroke and 53 age-matched control subjects who presented with focal neurological deficits. We examined the sensitivity and specificity of the stroke scale and the occurrence of acute seizures as predictors of stroke status. RESULTS: The total stroke scale did not differentiate children with acute ischemic stroke from those who had acute deficits from nonstroke causes; however, the presence of arm weakness was significantly associated with stroke cases. Acute seizures were significantly associated with stroke cases. CONCLUSIONS: An adult stroke scale is not sensitive or specific to distinguish children with acute ischemic stroke from those with nonstroke focal neurological deficits. The development of a pediatric acute ischemic stroke screening tool should include arm weakness and perhaps acute seizures as core elements. Such a scale must account for the limitations of language in young or intellectually disabled children. Keywords: pediatric, ischemic stroke, stroke recognition, stroke detection, stroke scale
Pediatr Neurol 2016; 58: 53-56 Ó 2016 Elsevier Inc. All rights reserved.
Introduction
Delayed recognition of acute ischemic stroke in children is common1-5 and delays evaluation and initial treatment. A tool that might lead to increased recognition of possible stroke would be a major contribution to the field, and could Financial disclosures: None. Funding source: No specific support. Conflict of interest: None. Contributor’s Statement Page: Dr. Neville identified the subjects, performed the chart reviews, collected the data, performed the initial data analysis, drafted the initial manuscript, and approved the final manuscript as submitted. Dr. Lo designed the study, performed the final data analysis, revised the manuscript, and approved the final manuscript as submitted.
Article History: Received November 2, 2015; Accepted in final form February 2, 2016 * Communications should be addressed to: Dr. Lo; EDU 582; 700 Children’s Drive; Columbus, OH 43205. E-mail address:
[email protected] 0887-8994/$ e see front matter Ó 2016 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.pediatrneurol.2016.02.001
potentially lead to changes in the acute treatment of thromboembolic stroke.6 Multiple tools have been developed to identify potential stroke in adults before or upon arrival to an emergency room.7-11 These tools have a range of sensitivities and specificities for acute stroke,12 and some have been tested in the identification of stroke before hospital arrival.13-15 Adult stroke scales were not developed with children in mind, but one study tested whether these scales could be used to identify acute stroke in children. In 47 children who had radiologically confirmed arterial ischemic stroke, the Face Arm Speech Test (FAST)11 and the Recognition of Stroke in the Emergency Room (ROSIER)10 were applied retrospectively to identify whether the children’s strokes could have been detected with these scales.16 Most subjects had a positive ROSIER (þ1) and at least one positive variable on the FAST, and the authors called for a prospective cohort study of children with stroke and nonstroke brain attacks.
54
K. Neville, W. Lo / Pediatric Neurology 58 (2016) 53e56
The finding suggested the possibility that an adult stroke tool might be used to identify children with acute ischemic stroke, although the study was limited by a lack of a comparison group. We hypothesized that an adult stroke tool could distinguish children who had acute symptomatic stroke from children who had acute neurological deficits from other causes. To address this question we tested the sensitivity and specificity of an adult stroke tool in a retrospective case-control study.
Methods We searched for cases of acute ischemic stroke from the discharge records of the Nationwide Children’s Hospital (NCH), Columbus, Ohio, and from prospectively collected records of a hospital-based stroke clinic. We identified cases that had the International Classification of Disease 9th Edition codes of 433.x, 434.x, 436, 437.4, 437.5 in the first or second diagnosis position, and who were discharged from January 1, 2006, through June 30, 2013. We included all children younger than 19 years who had documented acute clinical symptoms and ischemic stroke verified by abnormal findings on diffusion-weighted magnetic resonance imaging. We excluded patients who had strokes in the newborn period or who were less than 37 weeks’ gestation at the time of stroke, stroke due to trauma, infarcts due to diffuse hypoxia-ischemia, venous infarcts, nonacute stroke, subarachnoid hemorrhage, primary intraparenchymal hemorrhage, and intraventricular hemorrhage. Subjects who were unexaminable at initial diagnosis because of pharmacologic paralysis were excluded. For controls we selected patients who presented to the NCH emergency department with a focal motor deficit. Patients who had International Classification of Disease 9th Edition codes of 342.90 and 781.94, acute facial droop, or hemiparesis were matched to the stroke cases by age (3 years). Radiological imaging and medical records were reviewed to verify that the control subjects did not have a pre-existing stroke. Other exclusions for patients with stroke and control subjects were pre-existing neurological injury (cerebral palsy, prior stroke, or severe traumatic brain injury) to ensure that the acute presentations represented new neurological findings only. We scored the subjects with an adult stroke tool (termed the Central Ohio Trauma System [COTS] scale) that was derived from the Cincinnati prehospital stroke scale (CPSS) and Los Angeles prehospital stroke scale.17 The COTS scale records four common signs of stroke in adults: decreased consciousness, slurred speech, facial droop, and unilateral arm drift or paralysis as present or absent. Each item is scored as 0 (no sign) or 1 (positive sign) with scores ranging from 0 for no deficits to 4 in a patient with deficits in all four measures. The COTS scale was calculated based upon the neurological examination in the first neurologist’s notes, and for control subjects from the first emergency physician’s notes. Most control subjects were not examined by a neurologist. Of the control subjects, 70% had neuroimaging with computed tomography, magnetic resonance imaging, and in some cases magnetic resonance angiography. The control subjects who did not undergo imaging commonly had Bell’s palsy or postictal paralysis. Some of the neurologists’ examinations were performed after the radiological diagnosis of a stroke was already known. Age at presentation, presence of concurrent seizures, and associated diagnoses were recorded. Descriptive statistics were used to describe the basic demographics of the groups. X2 analysis was used to test whether the total COTS score, the individual component scores of the COTS, or the presence of acute seizures distinguished patients with stroke from control subjects. Receiver operating characteristic (ROC) curves were used to examine the sensitivity and specificity of the COTS scale to detect stroke. Analyses were performed with SPSS 21 (IBM, Armonk, NY). This study was reviewed and approved by the NCH Institutional Review Board.
Results
There were 53 patients with acute arterial ischemic stroke and 53 control subjects, with 33 male patients with stroke and 34 control subjects, and 20 female patients with stroke and 19 control subjects. Median age for the patients was 7.0 years (interquartile range 3.0-14.5 years), and for the control subjects was 10.0 years (interquartile range 5.014.0 years). There were 38 Caucasian patients and 42 control subjects; 11 African-American patients and six control subjects; two Latino control subjects; ethnicity was listed as “other” in three patients and two control subjects; and ethnicity was listed as unknown in one patient and one control subject. Seizures were witnessed in 11 patients and in four control subjects. The associated diagnoses in the stroke cases were cerebral vasculopathy (N ¼ 12), systemic illness (11), idiopathic (11), and cardiovascular disease (9), recent surgical procedures (6), central nervous system (CNS) infection (2), and CNS tumor (2). Among the control subjects the associated diagnoses were Bell palsy (10), CNS infection (7), postictal paralysis (7), idiopathic (6), CNS tumor (5), migraine (4), behavioral or psychogenic (4), miscellaneous neurological conditions (demyelinating disease and syringomyelia; 3), and other miscellaneous conditions ranging from systemic illness to otitis media (7). The median score on the total COTS scale was 1.0 for both cases (interquartile range 1-2 for patients and control subjects). Eight of the control subjects had intermittent or transient symptoms that cleared before the physical examination was documented, which resulted in a score of 0. In patients with stroke, the most common abnormal finding was arm drift or paralysis in 35 patients (71%), facial droop in 21 (43%), altered consciousness in 11 (21%), and slurred speech in nine (27%). Speech could not be assessed because of age, developmental delay, or alteration of consciousness in 20 patients with stroke. The most common presenting symptom in control subjects was unilateral facial droop in 29 (55%), followed by unilateral arm droop or paralysis in 23 (44%). Altered consciousness was present in eight (15%) and slurred speech in seven (17%). Speech could not be assessed in 11 control subjects, because of age, underlying disease (autism or inborn error of metabolism), or developmental delay. There were nine patients with stroke who had a total COTS scale score of 3-4, and 43 who had a score of 0 to 2 (Table). One patient was unresponsive and could not be examined further, although not paralyzed. The children with stroke who had scores of 0 presented with seizures, ataxia, headache, or visual change. There were four control subjects with a COTS scale score of 3, and the rest had scores of 0 to 2. There was a weak trend for patients with stroke to have a higher total COTS score compared with control subjects, but this did not achieve significance (Pearson c2 8.02, TABLE. Total Central Ohio Trauma System Scores for Patients With Stroke and Control Subjects
Patients with stroke Control subjects Total
0
1
2
3
4
Total
11 6 17
17 30 47
15 13 28
7 4 11
2 0 2
52 53
K. Neville, W. Lo / Pediatric Neurology 58 (2016) 53e56
P ¼ 0.091). When we examined the individual components of the COTS score, unilateral arm weakness was significantly associated with patients with stroke (c2 7.633, P ¼ 0.006). Alteration of consciousness, slurred speech, and facial droop were not significantly associated with either group. The presence of acute seizures was significantly associated with stroke cases (c2 3.81, P ¼ 0.051), whereas age at presentation was not. ROC analysis showed that the COTS total score had poor sensitivity and specificity for stroke status, with an area under the curve of 0.541 (Figure). Because six of the control subjects had clinical signs that resolved by the time they were examined (stroke score ¼ 0) we repeated the ROC analysis after eliminating these control subjects and the corresponding patients with stroke. There was no difference in the ROC curve or area under the curve (data not shown).
Discussion
In this study a screening scale for adult stroke did not distinguish children with stroke from children who presented with focal neurological signs from other conditions. The total score was not sensitive or specific for stroke cases, although one component of the scale, unilateral arm weakness, was significantly associated with stroke cases. A sensitive and specific scale for identifying childhood stroke would certainly be a major contribution to the field. There are a growing number of centers that are prepared to treat children with acute stroke,6 but they can only be effective if a stroke or the possibility of a stroke is recognized. Multiple studies have shown that there are delays in identifying children with acute ischemic stroke, even though parents bring their children to medical attention promptly.1-3,18 The problem is that stroke mimics are
FIGURE. Receiver operating characteristics curve for the total Central Ohio Trauma System scale detection of stroke cases. (The color version of this figure is available in the online edition.)
55
common,19 but childhood ischemic stroke is uncommon.20 Many primary physicians do not consider the possibility that a child might have a stroke. Additional delays occur as a provider sorts through the many possible alternative diagnoses, and then has to work to obtain brain imaging to identify the presence of cerebral ischemia. As noted earlier, a recent study of 47 children with proven arterial ischemic stroke16 tested whether the children’s clinical signs would be listed as positive on the ROSIER and FAST scales. The investigators found that 81% of subjects had a positive ROSIER (þ1) and 76% had at least one positive variable on the FAST. The most common signs were arm (61%), face (62%), or leg weakness (57%), or dysarthria (34%). The authors concluded that ROSIER and FAST had good to fair sensitivity, respectively, for detecting stroke symptoms in children. The authors noted that these scales were less sensitive in children than in adults. That study was limited by the lack of a control population for comparison. The investigators subsequently performed a prospective study of 287 children who presented to an emergency room with a brain attack (an apparently abrupt onset focal brain dysfunction).21 They found that stroke was the fourth most common disorder and occurred in 7% of the subjects, so the need for a tool that can raise awareness of stroke is clinically important. Our results suggest that a scale to screen children for acute ischemic stroke should build upon the association with unilateral arm weakness. Speech had limited usefulness, because it could not be reliably assessed in very young children or children who had moderate or severe intellectual disability. Acute seizures might be one component of a scale because they were significantly associated with stroke cases. However, because seizures can account for 15% of “brain attacks” presenting to an emergency room,21 additional features are needed to use acute seizures to distinguish patients with stroke from those without. Perhaps seizures together with age might comprise one element, because young children may present with seizures as the initial sign of stroke,22,23 but a robust sample of infantile stroke will be needed to test the validity of such an element. All the currently accepted adult stroke scales have limitations in sensitivity and specificity. A recent review of stroke scales found that the sensitivity of the CPSS ranged from 79% to 95%, whereas specificity ranged from 24% to 79%.12 The Los Angeles pre-hospital stroke scale sensitivity ranged from 78% to 91% and specificity ranged from 85% to 99%. The ROSIER and FAST scales both had sensitivities of 97%, but specificity ranged from 13% to 18%. These limitations together with our study results suggest that adult stroke scales will have substantial limitations for direct screening of children for acute stroke. How might one develop a stroke scale for use in children? One approach might follow the development of the CPSS for adults.7 Those investigators assessed the National Institutes of Health stroke scale in 74 patients with acute ischemic stroke and 225 patients without stroke presenting to an emergency room. The authors then selected three stroke scale elements with the strongest association with stroke for a screening tool. A pediatric version of the National Institutes of Health stroke scale has been developed,24 so a similar approach is possible. In children, however, a stroke scale will need
56
K. Neville, W. Lo / Pediatric Neurology 58 (2016) 53e56
additional elements that are more strongly associated with stroke than alertness, speech, or facial droop. One strength of our study was the onclusion of agematched control subjects who presented with focal motor deficits. Medical records and imaging review confirmed the diagnosis of acute stroke or eliminated it in the control subjects. The major limitation was the retrospective nature of our study. The stroke scale was applied to clinical examinations that were performed at variable times after the patient’s presentation. Because the first neurologist’s examination may have occurred after the diagnosis of stroke was established, some neurologists’ examinations may have been skewed toward a positive finding. In addition, this study was conducted at a large tertiary care children’s hospital in North America, so the results might not generalize to other settings. Conclusions
The simple application of an adult stroke scale is unlikely to distinguish children with acute ischemic stroke from children with stroke mimics. Elements such as unilateral arm weakness may be used to develop a pediatric ischemic stroke scale, but other elements are needed. A recent review of pediatric stroke noted that early recognition is a key educational target for health care professionals.25 The recent study by Mackay et al.,21 cited above, called for the modification of adult stroke recognition tools to help frontline providers identify affected children. This study supports the need for a modified stroke scale for children and provides directions for the development of such a scale. The authors thank Ms. Marie Robinette, Executive Director for the Central Ohio Trauma System, for sharing the COTS tool for this study.
References 1. Rafay MF, Pontigon AM, Chiang J, et al. Delay to diagnosis in acute pediatric arterial ischemic stroke. Stroke. 2009;40:58-64. 2. McGlennan C, Ganesan V. Delays in investigation and management of acute arterial ischaemic stroke in children. Dev Med Child Neurol. 2008;50:537-540. 3. Martin C, von EE, El-Koussy M, Boltshauser E, Steinlin M. Delayed diagnosis of acute ischemic stroke in children - a registry-based study in Switzerland. Swiss Med Wkly. 2011;141:w13281. 4. Srinivasan J, Miller SP, Phan TG, Mackay MT. Delayed recognition of initial stroke in children: need for increased awareness. Pediatrics. 2009;124:e227-e234. 5. Mallick AA, Ganesan V, Kirkham FJ, et al. Diagnostic delays in paediatric stroke. J Neurol Neurosurg Psychiatry. 2015;86:917-921.
6. Bernard TJ, Rivkin MJ, Scholz K, et al. Emergence of the primary pediatric stroke center: impact of the thrombolysis in pediatric stroke trial. Stroke. 2014;45:2018-2023. 7. Kothari R, Hall K, Brott T, Broderick J. Early stroke recognition: developing an out-of-hospital NIH Stroke Scale. Acad Emerg Med. 1997;4:986-990. 8. Llanes JN, Kidwell CS, Starkman S, Leary MC, Eckstein M, Saver JL. The Los Angeles Motor Scale (LAMS): a new measure to characterize stroke severity in the field. Prehosp Emerg Care. 2004;8: 46-50. 9. Bray JE, Martin J, Cooper G, Barger B, Bernard S, Bladin C. Paramedic identification of stroke: community validation of the Melbourne ambulance stroke screen. Cerebrovasc Dis. 2005;20:28-33. 10. Nor AM, Davis J, Sen B, et al. The Recognition of Stroke in the Emergency Room (ROSIER) scale: development and validation of a stroke recognition instrument. Lancet Neurol. 2005;4:727-734. 11. Harbison J, Hossain O, Jenkinson D, Davis J, Louw SJ, Ford GA. Diagnostic accuracy of stroke referrals from primary care, emergency room physicians, and ambulance staff using the face arm speech test. Stroke. 2003;34:71-76. 12. Brandler ES, Sharma M, Sinert RH, Levine SR. Prehospital stroke scales in urban environments: a systematic review. Neurology. 2014;82:2241-2249. 13. Asimos AW, Ward S, Brice JH, Rosamond WD, Goldstein LB, Studnek J. Out-of-hospital stroke screen accuracy in a state with an emergency medical services protocol for routing patients to acute stroke centers. Ann Emerg Med. 2014;64:509-515. 14. Berglund A, Svensson L, Wahlgren N, von Euler M. Face Arm Speech Time Test use in the prehospital setting, better in the ambulance than in the emergency medical communication center. Cerebrovasc Dis. 2014;37:212-216. 15. Bray JE, Coughlan K, Barger B, Bladin C. Paramedic diagnosis of stroke: examining long-term use of the Melbourne Ambulance Stroke Screen (MASS) in the field. Stroke. 2010;41:1363-1366. 16. Yock-Corrales A, Babl FE, Mosley IT, Mackay MT. Can the FAST and ROSIER adult stroke recognition tools be applied to confirmed childhood arterial ischemic stroke? BMC Pediatr. 2011;11:93. 17. COTS Stroke Alert Tool. Available at: http://www.goodhealth columbus.org/attachments/COTSStrokeAlertPostItPadsRevII124350. pdf. Accessed August 7, 2015. 18. Mallick AA, Ganesan V, Kirkham FJ, et al. Childhood arterial ischaemic stroke incidence, presenting features, and risk factors: a prospective population-based study. Lancet Neurol. 2014;13:35-43. 19. Shellhaas RA, Smith SE, O’Tool E, Licht DJ, Ichord RN. Mimics of childhood stroke: characteristics of a prospective cohort. Pediatrics. 2006;118:704-709. 20. Agrawal N, Johnston SC, Wu YW, Sidney S, Fullerton HJ. Imaging data reveal a higher pediatric stroke incidence than prior US estimates. Stroke. 2009;40:3415-3421. 21. Mackay MT, Chua ZK, Lee M, et al. Stroke and nonstroke brain attacks in children. Neurology. 2014;82:1434-1440. 22. Singh RK, Zecavati N, Singh J, et al. Seizures in acute childhood stroke. J Pediatr. 2012;160:291-296. 23. Abend NS, Beslow LA, Smith SE, et al. Seizures as a presenting symptom of acute arterial ischemic stroke in childhood. J Pediatr. 2011;159:479-483. 24. Ichord RN, Bastian R, Abraham L, et al. Interrater reliability of the Pediatric National Institutes of Health Stroke Scale (PedNIHSS) in a multicenter study. Stroke. 2011;42:613-617. 25. Kirton A, deVeber G. Paediatric stroke: pressing issues and promising directions. Lancet Neurol. 2015;14:92-102.