OBJECTIVE: The goal was to identify the delays involved in diagnosing pediatric arterial ischemic stroke (AIS), a major cause of morbidity and death in children.
METHODS: Neonates (≤28 days of age) and children with a first presentation of radiologically confirmed AIS between June 1993 and January 2006 were identified retrospectively. The time to diagnosis of AIS (ie, time from clinical onset to radiologic confirmation) was calculated, and factors influencing stroke diagnosis were reviewed.
RESULTS: A total of 107 patients (19 neonates and 88 children) with a diagnosis of AIS were identified. The median time to AIS diagnosis was 87.9 hours for neonates, significantly longer than 24.8 hours for children (P = .0002). Sixty-nine percent of the children with AIS demonstrated a likely cardioembolic cause, and 51 (58%) of the 88 children were inpatients at the time of stroke. The inpatients were seen by a physician more quickly (P < .01) and received a diagnosis of AIS sooner (P < .01). Seventy-six (86%) of the 88 children had a focal neurologic deficit when first seen by a physician. Physicians documented a diagnosis/differential diagnosis for 44 (50%) of 88 children, and they documented a suspicion of AIS for only 23 (26%) of 88 children. The presence of seizures or focal signs was not associated with a quicker time to stroke confirmation.
CONCLUSIONS: The considerable delays in the diagnosis of pediatric AIS are most likely related to the lack of awareness of stroke among medical staff members, despite risk factors and focal signs at presentation.
Stroke is a major cause of morbidity and death in children. Long-term neurologic deficits occur in 50% to 85% of infants and children after arterial ischemic stroke (AIS).1 Limited awareness regarding pediatric stroke among physicians and in the community is a major issue. For adults, advertising campaigns (eg, “Time Is Brain Lost”) and education of primary care physicians have led to significant decreases in the time to presentation.2 One study recorded a mean time of 34.5 hours from clinical onset to presentation to any health care professional for children, and studies have demonstrated long delays to neuroimaging.3,4 For children, stroke symptoms are attributed frequently to stroke-mimickers such as migraine, encephalitis, tumors, and postictal Todd paralysis, which can account for up to one fifth of cases presenting with stroke-like symptoms.5,6 Recent pediatric studies suggested that factors such as delays in seeking medical attention and mode of onset were predictive of the delayed diagnosis of stroke and of the underlying cause.7,8
Various consensus guidelines have highlighted the importance of improving time to stroke diagnosis.9–11 Before assessment of acute thrombolysis in the pediatric population, the factors causing delays in stroke diagnosis must be identified and addressed. The aims of this study were to determine the time between clinical onset and diagnosis of AIS in a tertiary Australian pediatric hospital and to identify factors that influenced the time to diagnosis.
We identified retrospectively all children 0 to 18 years of age who were diagnosed as having a first episode of AIS during a 12.5-year period, from June 1, 1993, to January 30, 2006, at the Royal Children's Hospital (Melbourne, Australia). The Royal Children's Hospital is a tertiary pediatric referral center for the states of Victoria and Tasmania that treats 280000 children each year. This study was approved by the ethics committee.
Cases were ascertained by using International Classification of Diseases (ICD) revisions 9 and 10 codes applied to discharge diagnoses (Table 1), on the basis of standardized methods recommended by the International Pediatric Stroke Study. These cases were cross-referenced with the institution's hematology database and stroke registry (which commenced in 2002) to ensure that no patients were missed.
AIS was defined as (1) an acute neurologic deficit lasting ≥24 hours and (2) parenchymal infarction on neuroimaging scans. Inclusion criteria included a first presentation of AIS in patients from the newborn period (defined as <28 days of age) up to 18 years of age, with radiologic confirmation of AIS. Patients with the following conditions were excluded: sinovenous thrombosis and hemorrhagic stroke subtypes, systemic vasculitides, birth before 36 weeks of gestation, other non-AIS disorders associated with focal deficits, and recurrent strokes. Subjects who did not have documentation of the times of clinical onset and neuroimaging were excluded.
Medical records were reviewed retrospectively by Dr Srinivasan, and data on the pathways to stroke diagnosis were collected. “Prehospital” times recorded included the time of clinical onset and the time to the Royal Children's Hospital or another medical center. Whether the child presented to the tertiary center directly or indirectly (through a family physician or peripheral hospital) also was ascertained. “Posthospital” times recorded included the time of physician assessment, the time of initial imaging, and the time of neuroimaging confirming AIS. The time of neuroimaging confirming AIS was obtained from a time stamp recorded at the time of the procedure and might not have been equivalent to the time of initial neuroimaging, if a diagnosis of AIS was not made immediately after the first imaging study (for example, if the initial study was a computed tomographic [CT] study that did not reveal the AIS).
The time of clinical onset was determined in a number of ways. The preferred method was to use the exact time of clinical onset (such as 9:00 am), as documented in the medical chart by physicians or nurses. In the absence of such information, the clinical onset was extrapolated from other documented information. For example, if the physicians documented that “hemiparesis began 2 hours ago” at 11:00 am, then stroke onset was estimated at 9:00 am. If the symptoms were present at awakening, then the last time the patient was seen in normal condition was used. For the sedated population and for certain ICU patients, the time of clinical onset was determined from the detailed regular neurologic observations performed by ICU nursing staff members, which yielded a time of clinical onset within 15 minutes. In cases with discrepancies between the physicians and nurses, the order of preference was as follows: earliest precise time documented, neurologic observations charted by nurses, and information extrapolated from other documented data as described above. The following information also was collected: age at time of diagnosis, presence of localizing signs, seizures at presentation, risk factors, documented differential diagnoses (including suspicion of AIS), and neuroimaging modality.
Focal neurologic deficits included focal neurologic signs and focal seizures. A focal sign was defined as a motor (hemiplegia, monoplegia, or ataxia), sensory (paresthesia or dysesthesia), visual (visual field deficit or eye deviation), or speech (dysarthria or dysphasia) deficit. Stroke subtypes were classified into 8 categories by using the Pediatric Stroke Classification, a preliminary modification of the Trial of ORG 10172 in Acute Stroke Treatment system, that is, sickle cell disease, cardioembolic stroke, moyamoya syndrome, cervical arterial dissection, stenoocclusive arteriopathy, other determined cause, multiple probable/possible causes, and undetermined cause.12
The associations between variables were analyzed by using Stata 9.2 (Stata, College Station, TX), with the 2-sample rank-sum (Mann-Whitney) test and Kruskal-Wallis test where appropriate. P values of ≤.05 were considered statistically significant.
During this 12.5-year period, 375 patients had ≥1 of the listed ICD codes (Fig 1). A total of 107 patients, including 19 neonates and 88 children, met the inclusion criteria. Most patients (>70%) had a precise time of clinical onset documented.
The overall median age at the time of diagnosis was 20 months (interquartile range [IQR]: 62 days to 112 months), and the male/female ratio was 1.27:1. The median age at the time of diagnosis for children was 36 months (IQR: 6–118 months). The median age at the time of diagnosis for neonates was 6 days (IQR: 3–12 days). There were no significant differences between the excluded and included populations with respect to gender or age. Sixty percent of the study subjects were inpatients at the time of stroke onset. The remaining 40% were not inpatients at the time of AIS onset, with 13% presenting directly to the Royal Children's Hospital emergency department after clinical onset and 27% presenting through a peripheral center.
Children (Nonneonatal Population)
Times to AIS Confirmation
A median time of 24.8 hours (IQR: 10.2–67.0 hours) from clinical onset to radiologic confirmation of AIS was noted for the population of children (Tables 2 and 3). For inpatients, the median time was 21.3 hours (IQR: 5.8–54.1 hours); for outpatients, the median time was 27.4 hours (IQR: 18.9–74.6 hours; P < .001).
Six (6.8%) of 88 patients were diagnosed as having AIS within 3 hours (Table 2); all were inpatients with cardioembolic risk factors (Table 3). Five patients had structural heart disease. Five of the 6 children had focal signs. Five of the 6 children also had radiologic evidence of hemorrhagic conversion or involvement of more than one third of the middle cerebral artery territory on initial imaging scans. Fifty percent of all 88 children with AIS did not have their stroke radiologically confirmed until ≥24 hours after clinical onset.
Times to Seeing a Physician and Initial Imaging
The time of first physician assessment was documented for 37 (42%) of the 88 children, and the median time from clinical onset to the first evaluation by a physician in this group was 1 hour (IQR: 0–4.5 hours). Of all children, 75% were seen by a physician within 3 hours. For inpatients, the median time to physician assessment was 20 minutes (IQR: 0–180 minutes); for outpatients, the median time was 120 minutes (IQR: 57–408 minutes; P = .003) (Table 2). The time to initial imaging was 11.4 hours (IQR: 4.6–25.5 hours) for inpatients and 9.6 hours (IQR: 4.5–20.6 hours) for outpatients (P > .05).
Initial Suspicion of AIS
A suspected diagnosis or differential diagnosis was documented by the first physician to see the patient for 44 (50%) of the 88 patients. Physicians documented suspicion of AIS for only 23 (26%) of the 88 children. Other documented differential diagnoses included seizures, central nervous system infections, sepsis, tumors, increased intracranial pressure, and, less commonly, migraines, musculoskeletal conditions, and electrolyte disturbances. The small number of clinicians who documented a suspicion of AIS precluded analysis of whether suspicion of AIS influenced the time to neuroimaging.
Presence of Focal Signs or Focal Seizures
At the time of the first physician assessment, 76 (86%) of 88 patients had a focal sign (Table 4). The majority of focal signs were motor deficits except for the neonates, for whom focal seizures predominated. The presence of a focal sign was associated with increased suspicion of AIS (P = .02), with physicians suspecting AIS for 23 (30%) of the 76 children with focal deficits, compared with no suspicion of AIS for the 12 children without focal deficits. However, the presence of a focal sign was not associated with a shorter time to stroke confirmation.
There were no significant associations between seizures at presentation and suspicion of AIS or between seizures at presentation and time to confirmation. None of the cases reviewed in our retrospective study had documentation detailing cognitive deficits in children with acute stroke.
Risk Factors for AIS
By using the Pediatric Stroke Classification system, detailed stroke causes for the 88 nonneonates were categorized (Table 5). The 19 neonates were excluded from this analysis. Two of the 8 Pediatric Stroke Classification categories, namely, moyamoya syndrome and sickle cell disease, were not represented among the children.
Admission Status at Time of Stroke
Almost 58% of our children (n = 51) were inpatients at the time of AIS (Table 6). All inpatients with AIS had identifiable risk factors, compared with 84% of the outpatients. Almost 69% of inpatient strokes resulted from a likely cardioembolic mechanism secondary to previously recognized cardiac risk factors, including structural cardiac defects, previous cardiac surgery and/or cardiac catheterization, or treatment with an extracorporeal circulation device (eg, extracorporeal membrane oxygenation treatment).
The proportions of children who underwent brain MRI, brain CT imaging, or ultrasonography as initial imaging were 88.9%, 94.9%, and 30%, respectively. MRI and CT scanning were performed as initial imaging primarily in the older population, with more ultrasound studies in the neonatal population. Initial imaging yielded negative results despite AIS for 62 of 74 children who underwent CT imaging and 3 of 6 children who underwent ultrasonography. MRI, when used as the initial imaging modality, confirmed all strokes in 8 children.
The median time to confirmation of AIS for the neonates was 87.9 hours (IQR: 53.4–166.4 hours), which was significantly longer than the time for the children, for whom the median time to AIS confirmation was 24.8 hours (IQR: 10.2–67.0 hours; P = .0002). There was significantly less documentation of AIS suspicion in the medical records of neonates (P = .04), and seizures were more common at presentation for neonates, compared with children (P = .0001). Anterior circulation AIS predominated for both age groups (68% for neonates and 63% for children).
There is significant delay in the diagnosis of stroke in children, with a median time of almost 25 hours from clinical onset to radiologic confirmation of AIS. Gabis et al3 demonstrated a large prehospital delay (mean time: 28.5 hours) contributing to the overall delay in stroke diagnosis (mean time: 35.7 hours). In contrast, our study showed that the median time from clinical onset to assessment by a physician was 1 hour, with three fourths of the children being seen by a physician within 3 hours after clinical onset. These findings are similar to those of a recent study from the United Kingdom, which found that three fourths of children were seen by a physician within 6 hours.4 Although prehospital delay was demonstrated recently to be predictive of delayed AIS diagnosis,7 our findings suggest that the delay in diagnosis is more likely posthospital delay, largely related to the lack of awareness of stroke among medical staff members.
Eighty-six percent of our children had ≥1 focal sign at the time of presentation, but one third of our patients had nonmotor (ie, nonhemiplegic) presentations. One half of the physicians documented a differential diagnosis, and even fewer included stroke in the list of possibilities.
Our findings suggest that pediatric stroke is an underrecognized problem within the medical community, despite the significant associated long-term disability and the fact that more than one half of children have residual functional deficits.1 Furthermore, almost 58% of our older subjects were inpatients at the time of their AIS, and almost 69% had known cardiac risk factors for stroke. This is important because cardiac disease is the most common, previously recognized risk factor.13–15 Children with cardiac disease represent the one group that could potentially be targeted with primary prevention strategies.
Although they met the theoretical 3-hour criterion for intravenous thrombolysis, 5 of 6 children in our series who presented within 3 hours would not have been eligible for intravenous thrombolysis because of extensive infarction or the presence of hemorrhagic conversion.16,17 Cranial ultrasonography and CT scanning are not as sensitive as MRI of the brain when used as the initial imaging modality to diagnose AIS; our study showed that the stroke diagnosis was missed for 3 of 6 children who initially underwent ultrasonography and 62 of 74 children who initially underwent CT scanning. This is similar to the results of a United Kingdom study in which AIS was not detected on CT scans in 47% of cases.4
Stroke-mimickers, such as hemiplegic migraine or Todd paresis, can cause delay in AIS diagnosis.5 Emergency radiologic confirmation of acute infarction and arterial occlusion will undoubtedly be essential requirements in the first pilot safety trials of thrombolysis in childhood stroke.18
Hyperacute therapies have not been approved for children, and physicians may feel less urgency to perform rapid imaging. If such therapies were to be shown to be effective for children, then the time to diagnosis would likely decrease as clinicians and radiologists prioritize neuroimaging studies.
Pediatric stroke registry data suggest that 25% to 30% of all cases occur in neonates and almost 50% in children <1 year of age.13 Not surprisingly, diagnosis took longer for neonates with AIS than for the childhood population. This is because localizing signs often are absent, which makes stroke onset difficult to determine accurately for neonates. Focal seizures can predominate over motor deficits. Stroke should be suspected for any neonate presenting with seizures, lethargy, or apnea, the latter 2 features being seen in neonatal stroke in the setting of encephalopathy.19
Study Population and Limitations
Our study had a larger study population, compared with previous studies.3,4 We defined our time of stroke diagnosis as the time of neuroimaging confirming stroke. Our study was performed in a tertiary center where there is close collaboration between pediatric neurologists and radiologists, which makes use of the time of neuroimaging a more-reliable approach to assess the time of stroke diagnosis in our population. The results were not reviewed independently, and the timing of negative studies was not analyzed, because positive imaging results represented an inclusion criterion. All scans were evaluated by attending radiologists who were familiar with neuroimaging. When assessment is performed with a suspicion of AIS, the radiologist's interpretation of the neuroimaging scans at the Royal Children's Hospital is conveyed immediately to the referring doctor.
Other studies defined the time of AIS diagnosis to be the time the diagnosis was entered in the medical chart.4 We chose radiologic confirmation because a suspected diagnosis or differential diagnosis was documented by the first physician to see the patient for only 50% of our population and a suspicion of AIS was documented for only 26% of children. A limitation of our definition for determining the time to stroke diagnosis is that radiologic confirmation may not equal clinical confirmation; however, because there often are delays or omissions in medical documentation, we think that the time to radiologic diagnosis is most likely to represent the true time to AIS diagnosis in our population.
A limitation of this study is possible ascertainment bias, given the retrospective design and limitations in ICD codes for diagnosing stroke. Almost 50% of the cases in our initial population were miscoded with ICD codes and did not represent AIS. However, all of the eligible stroke cases seen in our tertiary center were included in this study.
Other limitations include our study population being composed of patients from a large tertiary pediatric hospital, which houses a major cardiac unit. Therefore, cardioembolic stroke is more frequent in our population than in other studies, and the results are not entirely applicable to the general population. However, we would have anticipated that such a high-risk population would have a shorter time to stroke diagnosis than that in a general, community-based center. The fact that only 17% of our patients were neonates reflects the fact that we are not colocated with an obstetric unit and most neonatal cases are treated at other centers. In addition, sickle cell disease is uncommon in the Australian pediatric population, and patients with moyamoya syndrome might have been excluded for reasons such as previous AIS.
Patient-related factors influencing the time to diagnosis include the possible need for a general anesthetic for imaging for young children, and comorbidities such as cardiac disease and critical illness may mean that children are too ill to be transported to the scanner. In addition, admission to an ICU, and thus the need for sedation, may mask focal signs. Many inpatients may have contraindications to hyperacute therapy, leading to delayed imaging because clinicians think that imaging results are unlikely to change management.
The delays associated with AIS are attributable to both decreased awareness in the medical community and problems with access to neuroimaging, including those associated with patient comorbidities precluding urgent scanning. The lack of available treatment options for the pediatric stroke population is the likely reason for delayed imaging for children, because similar findings have been observed for adult stroke patients.20 Current practice is the likely culprit in perpetuating delays, and this factor needs to be addressed. Although thrombolysis is an important factor, there are other measures for addressing delays in diagnosis that can change long-term outcomes. Measures that have been shown to improve outcomes in adults include control of fever and hypertension, maintenance of normal oxygenation, normalization of serum glucose levels, and early access to rehabilitation.11
It is hoped that progress will decrease the time lag to stroke diagnosis through better understanding of risk factors in the pediatric population.21 The American College of Chest Physicians has stated, “the infrequency of stroke in children results in delayed recognition and an inability to intervene early with medications that may reduce subsequent neurologic deficits.”9 Clearly, there are delays in childhood stroke diagnosis that must be addressed with increased awareness among medical professionals, particularly for high-risk groups such as children with cardiac disease. Ideally, development of hospital-based pediatric stroke services will improve early diagnosis of stroke, to allow children the benefits of acute medical interventions that have been shown to improve outcomes in the adult population.
We thank Drs James H. Lee and Manali Chitre for their valuable contributions in editing the manuscript.
- Accepted March 13, 2009.
- Address correspondence to Mark T. Mackay, MBBS, FRACP, Department of Paediatric Neurology, Children's Neuroscience Centre, Royal Children's Hospital, Melbourne, Australia 3052. E-mail:
Financial Disclosure: The authors have indicated they have no financial relationships relevant to this article to disclose.
What's Known on This Subject:
Pediatric AIS, like AIS in adults, is associated with high morbidity and mortality rates. Unlike that in adults, however, pediatric stroke often is unrecognized, for a variety of reasons.
What This Study Adds:
This study quantifies the time to diagnosis of pediatric stroke and aims to highlight the factors that perpetuate delay, namely, the lack of awareness of physicians regarding pediatric stroke. This is in contrast to many previous studies.
- ↵Saver JL. Time is brain: quantified. Stroke.2006;37 (1):263– 266
- ↵Gabis LV, Yangala R, Lenn NJ. Time lag to diagnosis of stroke in children. Pediatrics.2002;110 (5):924– 928
- ↵Shellhaas RA, Smith SE, O'Tool E, Licht DJ, Ichord RN. Mimics of childhood stroke: characteristics of a prospective cohort. Pediatrics.2006;118 (2):704– 709
- ↵Rafay MF, Pontigon AM, Chiang J, et al. Delay to diagnosis in acute pediatric arterial ischemic stroke. Stroke.2009;40 (1):58– 64
- ↵Braun KP, Rafay MF, Uiterwaal CS, Pontigon AM, DeVeber G. Mode of onset predicts etiological diagnosis of arterial ischemic stroke in children. Stroke.2007;38 (2):298– 302
- Royal College of Physicians, Paediatric Stroke Working Group. Stroke in Childhood: Clinical Guidelines for Diagnosis, Management and Rehabilitation. London, England: Royal College of Physicians; 2004. Available at: www.rcplondon.ac.uk/pubs/books/childstroke/childstroke_guidelines.pdf. Accessed June 25, 2009
- ↵Roach ES, Golomb MR, Adams R, et al. Management of stroke in infants and children: a scientific statement from a special writing group of the American Heart Association Stroke Council and the Council on Cardiovascular Disease in the Young. Stroke.2008;39 (9):2644– 2691
- ↵Chabrier S, Husson B, Lasjaunias P, Landrieu P, Tardieu M. Stroke in childhood: outcome and recurrence risk by mechanism in 59 patients. J Child Neurol.2000;15 (5):290– 294
- ↵Whelan HT, Cook JD, Amlie-Lefond CM, et al. Practical model-based dose finding in early-phase clinical trials: optimizing tissue plasminogen activator dose for treatment of ischemic stroke in children. Stroke.2008;39 (9):2627– 2636
- ↵Ramaswamy V, Miller SP, Barkovich AJ, Partridge JC, Ferriero DM. Perinatal stroke in term infants with neonatal encephalopathy. Neurology.2004;62 (11):2088– 2091
- ↵Rose KM, Rosamond WD, Huston SL, Murphy CV, Tegeler CH. Predictors of time from hospital arrival to initial brain-imaging among suspected stroke patients: the North Carolina Collaborative Stroke Registry. Stroke.2008;39 (12):3262– 3267
- Copyright © 2009 by the American Academy of Pediatrics