Age at stroke onset influences the clinical outcome and health-related quality of life in pediatric ischemic stroke survivors
Abstract
Aim
Stroke in children occurs across different phases of brain development. Age at onset may affect outcome and health-related quality of life (HRQL). We evaluated the influence of age at stroke onset on the long-term neurological outcomes and HRQL of pediatric stroke survivors.
Method
Children with ischemic stroke were recruited into three groups according to their age at onset of stroke (presumed perinatal, neonatal, and childhood). Neurological outcomes were assessed using the Pediatric Stroke Recovery and Recurrence Questionnaire. HRQL was evaluated using proxy report versions (2–18y) of the Pediatric Quality of Life Inventory (PedsQL 4.0). A χ2/Fisher's exact test and multivariable logistic regression analysis was performed for the neurological outcomes. HRQL scores from the different age groups were compared using linear regression.
Results
Ninety participants (presumed perinatal stroke, n=31; neonatal stroke, n=36; childhood stroke, n=23) were enrolled. Median age at the onset of stroke was 0.5 days and 3.7 years in neonatal and childhood participants respectively. Of the three groups, participants with presumed perinatal stroke demonstrated the worst global (p<0.002) and motor (p<0.001) outcomes and the lowest level of independence in daily activities (p<0.001). Parents reported the best global outcome and overall HRQL (p=0.007) after neonatal stroke.
Interpretation
The age at stroke onset has important implications regarding long-term clinical outcomes and HRQL for survivors. Individuals with presumed perinatal stroke should be considered at high-risk for poor outcomes.
What this paper adds
- Age at stroke onset influences the health-related quality of life (HRQL) of survivors in a bimodal fashion.
- Individuals with presumed perinatal and childhood stroke appear to have the lowest morbidity and HRQL.
- Acute ischemic stroke in the newborn at term exhibits the best long-term HRQL.
- Age at stroke onset is an important prognostic marker for long-term clinical outcomes.
Abbreviations
-
- HRQL
-
- Health-related quality of life
-
- RRQ
-
- Pediatric Stroke Recurrence and Recovery Questionnaire
-
- PedsQL
-
- Pediatric Quality of Life Inventory
Once considered rare, pediatric stroke is being increasingly recognized, with a frequency of more than 3 in 100 000, which is equivalent to that of childhood cancers.1 Stroke not only occurs in infants and children, but can also occur in the developing brain of the fetus and newborn. The perinatal period is one of most likely times in life to suffer a stroke, with an incidence ranging from 1 in 2300 to 1 in 5000 live births, and is the leading cause of hemiplegic cerebral palsy.2
Pediatric stroke is classified as perinatal or childhood stroke. Perinatal stroke is a vascular event that causes the focal interruption of blood supply, and can occur between 20 weeks gestational age and postnatal day 28; it can be confirmed by neuroimaging.2, 3 This type of stroke can be further subclassified as neonatal or presumed perinatal stroke. The onset of neonatal stroke can occur from around the time of birth to 28 days of age. A diagnosis of presumed perinatal stroke is made when an infant has a typical neonatal course, but presents in the first 18 months of life with focal signs of hemiplegia. If post-perinatal neuroimaging reveals chronic changes suggestive of a vascular event involving one of the middle cerebral arteries or the periventricular region, and there has been no history suggestive of a stroke during the infant's life, a stroke is ‘presumed’ to have occurred prior to birth. Childhood stroke can occur between 28 days and 18 years of age.1
Since pediatric stroke can occur during different phases of brain development, the age at onset may affect the mode and extent of brain damage, influencing the clinical outcome and, therefore, the health-related quality of life (HRQL). Some authors support a notion of higher vulnerability (i.e. that the immature brain is more sensitive to insults), while others favor a hypothesis of plasticity (i.e. the immature brain has a greater chance of recovery after an early injury).4, 5 The literature to date suggest that children with an early brain injury (before 1–2y of age) demonstrate poor neurocognitive and motor outcomes.6-9 In contrast, others indicate a better outcome after neonatal than childhood stroke.10 However, none of these studies has differentiated individuals who have suffered an acute neonatal stroke from those who have experienced a presumed perinatal stroke, and often report on heterogeneous populations with variable underlying brain pathologies.
Currently, the effect of age at onset of stroke on the long-term HRQL for pediatric stroke survivors is not well understood. HRQL is acknowledged as an important outcome indicator for chronic diseases, including stroke. In comparison with the traditional outcome measures of morbidity, HRQL provides a comprehensive assessment of the health status of an individual, and includes their physical, mental, and social well-being, a concept that conforms to the World Health Organization's definition of health.11 HRQL is a patient-reported health status measure and, therefore, plays a vital role in the assessment of the health care needs of communities and has important implications for health care delivery and resource allocation.12
We sought to determine the influence of age at onset of pediatric ischemic stroke on neurological outcomes and HRQL using the consensus-based definitions for the classification of pediatric stroke.
Method
Participants
A cross-sectional study was conducted at the Stollery Children's Hospital, Edmonton, Alberta, Canada. This is a tertiary-care pediatric hospital with referrals from a major region in western Canada (northern Alberta, parts of Saskatchewan, and Manitoba). Children diagnosed with pediatric ischemic stroke between January 2003 and June 2012 were considered for participation if they were aged 2 to 18 years upon assessment, and had received at least 1-year of follow-up care after the childhood stroke. Diagnosis of ischemic stroke was confirmed by magnetic resonance imaging (MRI) or computed tomography in all participants at the time of initial presentation. A lower cut-off age of 2 years was selected, as HRQL is difficult to evaluate in very young children and there are very few HRQL tools available for infants and young children. A minimum of 1-year of follow-up allowed for peak neurological recovery to occur after the onset of stroke. For this study, children with global brain injury, watershed infarcts, intracranial hemorrhage, cerebral sinovenous thrombosis, transient ischemic attacks, underlying genetic syndromes, or other associated brain pathologies or neurological comorbidities (e.g. autism) were excluded. The study was approved by the Health Research Ethics Board, University of Alberta, Canada.
Parents of all eligible children were recruited either during routine pediatric stroke clinics or by telephone. After obtaining an informed consent, all study participants were provided with a set of standardized questionnaires, as outlined below. To ensure a uniform assessment and to prevent bias related to the mode of administration, all participants were given a paper copy of the survey, and advised to complete it at home then return it by post.
Neurological outcomes
Neurological outcomes and the need for ongoing rehabilitative health care services were assessed at the same time as the HRQL assessment was performed, by using the Pediatric Stroke Recurrence and Recovery Questionnaire (RRQ) through parent report.13 The recently validated RRQ has been adapted from the Pediatric Stroke Outcome Measure to assess post-stroke neurological function. The RRQ measures neurological impairments across five categories: right sensorimotor, left sensorimotor, cognition/behavior, language production, and language comprehension. The neurological deficits in each category were scored as 0 (no deficit), 0.5 (mild deficit with normal function), 1 (moderate deficit with decreased function), or 2 (severe deficit with complete absence of function). The global neurological outcome was also reported for each participant as ‘normal’ (score of 0 in all five categories), as a ‘mild’ deficit (score of 0.5 in only one category), as a ‘moderate’ deficit (score of 0.5 in two, three, or four categories; score of 1 in one category; or score of 1 in one category and 0.5 in one category), or as a ‘severe’ deficit (score of 0.5 in all five categories; score of 1 in two categories; score of 1 in one category and 0.5 in two categories; or score of 2 in one category).14
Health-related quality of life assessment
HRQL was evaluated using proxy report versions of the Pediatric Quality of Life Inventory (PedsQL 4.0). PedsQL is the most widely used generic HRQL measure for children. The instrument contains 23 items encompassing four areas: physical, emotional, social, and school. Participants respond on a Likert scale from 0 to 4. Items are reverse scored and linearly transformed to a 0 to 100 scale, with higher converted scores indicating better HRQL. The instrument provides four domain scores, two summary scores (physical and psychosocial functioning), and a total HRQL score. The scale has adequate psychometric properties of reliability and validity.15
Sociodemographic, clinical, and radiological data
Data on the demographic, clinical (i.e. age at presentation, side and clinical signs at presentation, and underlying risk factors), and neuroimaging features of each participant were retrieved from medical records. Neuroimaging, performed as a part of a routine clinical procedure for all participants, comprised sagittal T1-weighted, axial T2-weighted, diffusion weighted, and fluid-attenuated inversion recovery imaging and was reported by a radiologist and independently reviewed by a pediatric neurologist (JYY). The first MRI obtained after the clinical suspicion of stroke was used to determine the radiological parameters (i.e. vascular territory involved, and the size, site, and lateralization of the stroke). Vascular territories were defined using published resources.16 Size was described as an infarct involving less than one-third, one- to two-thirds, or more than two-thirds of the ipsilateral cerebral hemisphere. Children were categorized according to the type of stroke (presumed perinatal, neonatal, or childhood), based on historical features and neuroimaging, by a pediatric neurologist specialized in stroke. A four-factor index of social status was used to determine socio-economic status through the information obtained via a parental questionnaire on parents' education, occupation, and marital status.17 Information on place of residence was also obtained.
Statistical analysis
The data was checked for normality and found to be normally distributed. Categorical and continuous variables were expressed as proportions and means respectively. χ2/Fisher's exact tests were used to compare the neurological outcomes and other categorical variables across the three age groups. Linear regression was used to compare HRQL data (continuous variable) across the three groups. HRQL scores were also compared with the published reference ‘norms’ using unpaired Student's t-tests, z-scores, and effect sizes. The z-scores and effect sizes were computed to assess the magnitude of the difference between the study population and reference population scores, and the clinical importance of this difference. Effect sizes were interpreted as small (0.20–0.49), moderate (0.50–0.79), or large (>0.80), in accordance with Cohen's recommendations.18 Univariate logistic regression analysis was performed to identify predictors of global neurological outcome (which was dichotomized as good [none or mild deficit] or poor [moderate or severe deficit]). Results were expressed as odds ratios with 95% confidence intervals. Variables showing p values of <0.1 on univariate analysis, and on the basis of clinical importance, were entered into a multivariable logistic regression model with neurological condition (good or poor) as the outcome. A backward elimination procedure was used to generate the final model.
The prognostic accuracy of the model was estimated using the area under the receiver operating characteristic curve. Statistical analysis was performed using stata version 12.0 (StataCorp LP, College station, TX, USA). All tests of significance were two-sided and statistical significance was defined as a p value of <0.05.
Results
Ninety participants who had suffered the following types of pediatric ischemic stroke were enrolled: presumed perinatal (n=31; 34%), neonatal (n=36; 40%), and childhood (n=23; 26%). Baseline demographic, clinical, and radiological parameters for all participants are presented in Table 1. Since the exact age at onset of stroke could not be ascertained in those who had experienced presumed perinatal stroke, we assigned an age of 0 days. The median time elapsed from stroke onset was similar in all three categories (p=0.74). More males than females were classified as having had neonatal (M:F=1.4:1) and presumed perinatal (M:F=1.8:1) strokes, while more females were classified as having had a childhood (M:F=0.8:1) stroke (p=0.29).
| Parameter | Presumed perinatal stroke (n=31) | Neonatal stroke (n=36) | Childhood stroke (n=23) | p value |
|---|---|---|---|---|
| Mean age at stroke onset (y:mo), SD (range) | ND | 2 days, 4 (0–21) days | 5:2 4.9 (2mo–14y) | |
| Male to female ratio | 1.8:1 | 1.4:1 | 0.8:1 | 0.29 |
| Mean SES score, SD (range) | 44.1, 11.6 (15–66) | 42.4, 13.1 (9–63.5) | 38.7, 12.9 (13–61) | 0.48 |
| Mean time elapsed since stroke (y:mo), SD (range) | 5:2, 3.4 (2:1–15:2) | 5:2, 3 (2:5–15:1) | 4:7, 2.9 (1:0–10:8) | 0.74 |
| Mean age (y:mo) at assessment, SD (range) | 5:2, 3.4 (2:1–15:2) | 5:2, 3 (2:5–15:1) | 9:10, 5.8 (2:2–17:10) | <0.001 |
| Side of clinical presentation | ||||
| Right, % (n) | 58.1 (18) | 47.2 (17) | 26.1 (6) | 0.16 |
| Left, % (n) | 32.3 (10) | 25 (9) | 47.8 (11) | |
| Bilateral, % (n) | 6.5 (2) | 19.4 (7) | 21.7 (5) | |
| Asymptomatic, % (n) | 3.2 (1) | 8.3 (3) | 4.4 (1) | |
| Clinical signs at presentation | ||||
| Motor deficit, % (n) | 90.3 (28) | 0 | 43.5 (10) | |
| Seizures, % (n) | 0 | 91.7 (33) | 21.7 (5) | <0.001 |
| Other, % (n) | 6.5 (2) | 5.5 (2) | 30.4 (7) | |
| Asymptomatic, % (n) | 3.2 (1) | 2.8 (1) | 4.4 (1) | |
| Risk factors | ||||
| Maternal, % (n) | 35.5 (11) | 22.2 (8) | 0 | |
| Child, % (n) | 12.9 (4) | 52.8 (19) | 87.0 (20) | <0.001 |
| None, % (n) | 51.6 (16) | 25.0 (9) | 13.0 (3) | |
| Congenital heart disease, % (n) | 6.5 (2) | 11.1 (4) | 43.5 (10) | 0.002 |
| Lateralization of stroke | ||||
| Right, % (n) | 35.5 (11) | 36.1 (13) | 56.5 (13) | 0.06 |
| Left, % (n) | 61.3 (19) | 47.2 (17) | 26.1 (6) | |
| Bilateral, % (n) | 3.2 (1) | 16.7 (6) | 17.4 (4) | |
| Size of infarction | ||||
| <1/3 of cerebral hemisphere, % (n) | 64.5 (20) | 80.6 (29) | 82.6 (19) | 0.04 |
| 1/3–2/3 of cerebral hemisphere, % (n) | 32.3 (10) | 16.7 (6) | 4.4 (1) | |
| >2/3 of cerebral hemisphere, % (n) | 3.2 (1) | 2.8 (1) | 13 (3) | |
| Site of infarction | ||||
| BG and PV involvement, % (n) | 61.3 (19) | 25 (9) | 43.5 (10) | 0.003 |
| Cerebral with no BG and PV involvement, % (n) | 38.7 (12) | 63.9 (23) | 34.8 (8) | |
| Non-cerebral, % (n) | 0 | 11.1 (4) | 21.7 (5) | |
| Cortical, % (n) | 64.5 (20) | 80.6 (29) | 56.5 (13) | 0.12 |
| Non-cortical, % (n) | 35.5 (11) | 19.4 (7) | 43.5 (10) | |
| Vascular territorya involved | ||||
| PM1, % (n) | 25.8 (8) | 16.7 (6) | 21.7 (5) | 0.002 |
| DM1, % (n) | 9.7 (3) | 8.3 (3) | 4.4 (1) | |
| AT, % (n) | 3.2 (1) | 16.7 (6) | 21.7 (5) | |
| PT, % (n) | 25.8 (8) | 38.9 (14) | 8.7 (2) | |
| LLS, % (n) | 12.9 (4) | 5.6 (2) | 21.7 (5) | |
| PVI, % (n) | 22.6 (7) | 2.8 (1) | 0 | |
| Non-cerebral, % (n) | 0 | 11.1 (4) | 21.7 (5) | |
- a Vascular territories were defined according to published resources.16 ND, could not be determined; SD, Standard deviation; SES, socio-economic status; BG, basal ganglion; PV, periventricular; PM1, proximal middle cerebral artery (MCA); DM1, distal MCA; AT, anterior trunk (superior MCA division); PT, posterior trunk (inferior MCA division); LLS, lateral lenticulostriate; PVI, periventricular venous infarction. p value indicate presence or absence of statistical significance among three groups.
The three groups varied in terms of their clinical presentation, underlying risk factors, size and site of stroke, and the vascular territory involved. The mode of clinical presentation was significantly different across the three categories; the majority (92%) of participants with neonatal stroke presented with seizures, the majority (91%) of those with presumed perinatal stroke presented with motor deficits, and those with childhood stroke had a variable presentation, with 44% having motor deficits and 22% having seizures (p<0.001) (see Table 1).3
Potential risk factors for stroke were identified in 75% of the participants with neonatal stroke and included labor and delivery complications (n=6), thrombophilia (n=6), congenital heart disease (n=4), meconium aspiration (n=3), birth asphyxia (n=2), twins (n=2), pre-eclampsia (n=2), maternal smoking (n=1), maternal diabetes (n=1), intrauterine growth retardation (n=1), and infection (n=1). In the participants with presumed perinatal stroke, risk factors included antenatal bleeding (n=3), thrombophilia (n=2), twins (n=2), pre-eclampsia (n=2), maternal diabetes (n=2), smoking (n=1), oligohydraminos (n=1), and antenatal trauma (n=1). However, no etiologies could be documented in over half (52%) of the participants with presumed perinatal stroke. About 87% of the participants with childhood stroke had an underlying condition which contributed to the stroke, with congenital heart disease being reported in up to 44% (n=10) of these participants, and others having arteriopathy (n=4), infection (n=3), arrhythmia (n=1), and thrombophilia (n=1).
Over 80% of participants from the neonatal and childhood stroke groups had an infarction involving less than one-third of the cerebral hemisphere, whereas participants from the presumed perinatal stroke group, in general, had larger infarct sizes (p=0.04). The basal ganglia and periventricular involvements were also more evident in the presumed perinatal stroke group (61%) than in the childhood (44%) or neonatal stroke (25%) groups (p=0.003). In terms of vascular involvement, the presumed perinatal stroke group tended to demonstrate involvement of the proximal segment of the middle cerebral artery and periventricular regions. In participants with neonatal stroke, the most commonly involved vascular territory was the posterior trunk of the middle cerebral artery (in 39% of these participants) (p=0.002). Cortical involvement was similar in all three groups (p=0.12).
Almost half (47%) of the neonatal stroke and one-third (30%) of the participants with childhood stroke had no residual impairments. In contrast, 97% of participants with presumed perinatal stroke demonstrated long-term morbidity. Of the three groups, participants from the participants with presumed perinatal stroke group demonstrated the worst global (p=0.002) and sensorimotor (p=0.001) outcomes and the least independence in daily activities (p=0.001) as shown in Table 2. Furthermore, parents of the participants with presumed perinatal stroke identified the maximum need for rehabilitative services after the onset of stroke (p<0.001).
| Parameter | Presumed perinatal Stroke (n=31) | Neonatal stroke (n=36) | Childhood stroke (n=23) | p value |
|---|---|---|---|---|
| Global outcome | ||||
| Normal, % (n) | 3.2 (1) | 47.2 (17) | 30.4 (7) | <0.001 |
| Mild deficit, % (n) | 12.9 (4) | 11.1 (4) | 4.4 (1) | |
| Moderate deficit, % (n) | 41.9 (13) | 8.3 (3) | 34.8 (8) | |
| Severe deficit, % (n) | 41.9 (13) | 33.3 (12) | 30.4 (7) | |
| Sensorimotor outcome | ||||
| No impairment, % (n) | 3.2 (1) | 55.6 (20) | 43.5 (10) | <0.001 |
| Mild impairment, % (n) | 19.4 (6) | 13.9 (5) | 13 (3) | |
| Moderate impairment, % (n) | 54.8 (17) | 22.2 (8) | 21.7 (5) | |
| Severe impairment, % (n) | 22.6 (7) | 8.3 (3) | 21.7 (5) | |
| Cognition and behavior outcome | ||||
| No impairment, % (n) | 58.1 (18) | 66.7 (24) | 60.9 (14) | 0.92 |
| Mild impairment, % (n) | 22.6 (7) | 19.4 (7) | 17.4 (4) | |
| Moderate impairment, % (n) | 16.1 (5) | 13.9 (5) | 21.7 (5) | |
| Severe impairment, % (n) | 3.2 (1) | 0 | 0 | |
| Language production outcome | ||||
| No impairment, % (n) | 54.8 (17) | 63.9 (23) | 52.2 (12) | 0.88 |
| Mild impairment, % (n) | 19.4 (6) | 19.4 (7) | 17.4 (4) | |
| Moderate impairment, % (n) | 22.6 (7) | 13.9 (5) | 21.7 (5) | |
| Severe impairment, % (n) | 3.2 (1) | 2.8 (1) | 8.7 (2) | |
| Language comprehension outcome | ||||
| No impairment, % (n) | 74.2 (23) | 80.6 (29) | 65.2 (15) | 0.77 |
| Mild impairment, % (n) | 12.9 (4) | 13.9 (5) | 17.4 (4) | |
| Moderate impairment, % (n) | 9.7 (3) | 2.8 (1) | 13 (3) | |
| Severe impairment, % (n) | 3.2 (1) | 2.8 (1) | 4.4 (1) | |
| Independence, % (n) | 29 (9) | 75 (27) | 61 (14) | 0.001 |
| Ongoing need of AEDs, % (n) | 12.9 (4) | 11.1 (4) | 8.7 (2) | 1.0 |
| Ongoing rehabilitation | ||||
| Physical therapy, % (n) | 77.4 (24) | 19.4 (7) | 26.1 (6) | <0.001 |
| Occupational therapy, % (n) | 67.7 (21) | 27.8 (10) | 26.1 (6) | 0.001 |
| Speech therapy, % (n) | 38.7 (12) | 13.9 (5) | 17.4 (4) | 0.05 |
| Special education services, % (n) | 29 (9) | 13.9 (5) | 8.7 (2) | 0.15 |
- AED, antiepileptic drug; p value indicate presence or absence of statistical significance among three groups.
Overall, HRQL scores were significantly higher for participants from the neonatal stroke group than for those from the other two groups (p=0.007; Table 3). Physical (p=0.001) and social (p=0.02) functioning were also better in the participants with neonatal stroke than in the participants with presumed perinatal or childhood stroke. Scores for psychological and school functioning were also higher for participants with neonatal stroke; however, these results did not reach statistical significance. Emotional functioning was similar in all three groups (p=0.52). Participants with presumed perinatal and childhood stroke had similar scores in all HRQL domains. In comparison with published reference scores, participants from the neonatal stroke group achieved similar scores in all domains except emotional functioning, where they demonstrated significantly lower scores (p=0.03) with a moderate effect size (ES=-0.52) (Table 4). Both presumed perinatal and childhood stroke participants had significantly lower scores in all HRQL domains than the reference population. Parents reported relatively larger deficits after childhood stroke than after presumed perinatal stroke, indicating more impairment in HRQL after childhood stroke.
| HRQL parameter mean scores (SD) | Presumed perinatal stroke (n=31) | Neonatal stroke (n=36) | Childhood stroke (n=23) | p value |
|---|---|---|---|---|
| Total HRQL score | 68.8 (15.4) | 80.0 (17.8) | 66.1 (22.3) | 0.007 |
| Physical functioning score | 67.1 (20.3) | 85.1 (21.5) | 64.8 (34.0) | 0.001 |
| Psychological functioning score | 70.3 (14.6) | 77.3 (18.5) | 67.3 (19.8) | 0.07 |
| Emotional functioning score | 71.1 (20.3) | 71.5 (22.5) | 65.4 (20.7) | 0.52 |
| Social functioning score | 70.2 (16.2) | 81.0 (19.7) | 68.3 (22.0) | 0.02 |
| School functioning score | 68.1 (16.9) | 79.3 (20.7) | 67.8 (25.6) | 0.08 |
- HRQL, health-related quality of life.
| HRQL domains | Reference scores mean (SD) | Neonatal stroke vs reference | Presumed perinatal stroke vs reference | Childhood stroke vs reference | ||||||
|---|---|---|---|---|---|---|---|---|---|---|
| p value | z-score | ES | p-value | z-score | ES | p value | z-score | ES | ||
| Total HRQL | 81.34 (15.92) | 0.61 | –0.08 | S | <0.001 | –0.79 | L | 0.003 | –0.96 | L |
| Physical functioning | 83.26 (19.98) | 0.58 | 0.09 | S | <0.001 | –0.81 | L | 0.01 | –0.92 | L |
| Psychological functioning | 80.22 (15.84) | 0.27 | –0.18 | S | 0.005 | –0.63 | M | 0.001 | –0.82 | L |
| Emotional functioning | 80.28 (16.99) | 0.03 | –0.52 | M | 0.003 | –0.54 | M | <0.001 | –0.88 | L |
| Social functioning | 82.15 (20.08) | 0.73 | –0.06 | S | 0.009 | –0.60 | M | 0.001 | –0.69 | M–L |
| School functioning | 76.91 (20.16) | 0.48 | 0.12 | S | 0.02 | –0.44 | S–M | 0.03 | –0.45 | S–M |
- HRQL, health-related quality of life; ES, effect size; S, small; M, moderate; L, large; S–M, small to moderate; M–L, moderate to large.
Of 90 participants, 34 (38%) had good and 56 (62%) had poor neurological outcomes. In univariate analysis, age, clinical signs, risk factors, size, site, vascular territory, socio-economic status, and residence were found to be significantly associated with neurological outcome (Table 5). Since age at onset of stroke correlates with clinical signs and risk factors, in a virtually linear fashion, only age was included in the multivariable model. Vascular territory was not included because of sample size limitations. In the multivariable analysis, age at stroke onset (p=0.02) and size of stroke (p=0.04) were found to be predictors of neurological outcome. The area under the receiver operating characteristic curve was 0.80, indicating strong predictive ability.
| Variable | Univariate analysis | Variables selected for multivariate analysis | Multivariate analysis | ||
|---|---|---|---|---|---|
| Odds ratio (95% CI) | p value | Odds ratio (95% CI) | p value | ||
| Age at stroke onset | In | … | … | ||
| Neonatal (n=36) | Ref | … | |||
| Presumed perinatal (n=31) | 7.28 (2.3–23.3) | 0.001 | 5.1 (1.4–19.0) | 0.02 | |
| Childhood (n=23) | 2.63 (0.9–7.8) | 0.08 | 1.9 (0.6–6.5) | 0.3 | |
| Sex | Out | ||||
| Male (n=51) | Ref | … | |||
| Female (n=39) | 0.53 (0.2–1.3) | 0.15 | |||
| SES score | In | 0.16 | |||
| <42.75 (median score) | Ref | … | |||
| ≥42.75 | 0.46 (0.2–1.1) | 0.08 | |||
| Residence | In | 0.18 | |||
| Urban (n=45) | Ref | … | |||
| Rural (n=45) | 2.6 (1.1–6.4) | 0.03 | |||
| Clinical signs | Out | ||||
| Motor deficit (n=38) | Ref | … | |||
| Seizures (n=38) | 0.06 (0.02–0.2 ) | <0.001 | |||
| Other + asymptomatic (n=14) | 0.05 (0.01–0.2) | <0.001 | |||
| Side of clinical presentation | Out | ||||
| Right (n=41) | Ref | … | |||
| Left (n=30) | 1.08 (0.4–3.0) | 0.9 | |||
| Bilateral (n=14) | 0.27 (0.09–0.9) | 0.03 | |||
| Risk factors | Out | ||||
| Maternal (n=19) | Ref | … | |||
| Child (n=43) | 0.25 (0.06–0.9) | 0.05 | |||
| None (n=28) | 0.23 (0.06–0.9) | 0.05 | |||
| Size of infarction | In | ||||
| ≤1/3 of cerebral hemisphere (n=68) | Ref | … | … | … | |
| >1/3 of cerebral hemisphere (n=22) | 5.3 (1.4–19.6) | 0.01 | 4.8 (1.1–21.2) | 0.04 | |
| Site | In | … | |||
| BG and PV involved (n=38) | Ref | … | … | ||
| No BG and PV involvement (n=52) | 0.27 (0.1–0.7) | 0.006 | 0.49 (0.2–1.5) | 0.17 | |
| Vascular territorya | Out | ||||
| PM1 (n=19) | Ref | … | |||
| DM1 (n=7) | 0.04 (0–0.5) | 0.01 | |||
| AT (n=12) | 0.01 (0–0.1) | 0.00 | |||
| PT (n=24) | 0.09 (0.01–0.8) | 0.03 | |||
| LLS (n=11) | 0.07 (0.01–0.7) | 0.02 | |||
| PVI (n=8) | 0.17 (0.01–2.2) | 0.17 | |||
| Non-cerebral (n=9) | 0.11 (0.01–1.3) | 0.08 | |||
- a Vascular territories were defined according to published resources.16 Area under the receiver operating characteristic curve = 0.80. Lateralization, age at assessment, and time elapsed since stroke were not found to be significant on univariate analysis. CI, confidence interval; Ref, reference category; SES, socio-economic status; BG, basal ganglion; PV, periventricular; PM1, proximal middle cerebral artery (MCA); DM1, distal MCA; AT, anterior trunk (superior MCA division); PT, posterior trunk (inferior MCA division); LLS, lateral lenticulostriate; PVI, periventricular venous infarction. p value indicate presence or absence of statistical significance compared to reference category. …, no p value available, this value was used as the reference against which statistics were performed.
Discussion
Our study assessed the impact of age at pediatric stroke onset on clinical outcomes and HRQL using a consensus classification system and validated outcome measures. Age at stroke onset was an important prognostic marker of long-term neurological outcomes, as assessed by multivariable analysis of participants with pediatric ischemic stroke. Individuals who had experienced presumed perinatal stroke had the poorest long-term clinical outcomes, while those who had suffered neonatal stroke exhibited the best long-term HRQL.
Almost half of the neonatal stroke and one-third of the participants with childhood stroke had typical global neurological outcomes. In contrast, participants with presumed perinatal stroke demonstrated much higher levels of morbidity. In addition, these participants exhibited the lowest levels of independence with regard to daily activities, and had a greater perceived need for rehabilitative health care services. Since these morbidities were assessed after a median time of 4 years, it is likely that they will persist throughout life and, therefore, will amplify the burdens on family, society, and health care.
In the current study, HRQL assessment was variably impacted across three ‘age at stroke onset’ categories. According to the parent proxy reports, participants with neonatal stroke demonstrated the highest HRQL for multiple domains. Parents of participants with presumed perinatal and childhood stroke perceived a lower HRQL for their children than the parents of the neonatal stroke population and reference population. Parents also reported a lower HRQL after childhood stroke than after presumed perinatal stroke. However, the interpretation of these observations is limited by the lack of age-matched controls in the reference population and future research should explore these findings further.
Previous literature has described associations between the outcome of stroke and stroke size and location. Large stroke size, bilateral infarcts, and the involvement of basal ganglia have been linked with poor outcomes.19, 20 We established a host of predictors of long-term neurological outcomes in our participants with pediatric ischemic stroke. In the multivariable analysis, we found that age at stroke remained an important predictor of long-term outcomes even when adjusted for the location and size of the lesion.
A bimodal relationship between the age at onset of stroke and the neurological outcomes and HRQL of individuals affected by pediatric stroke was also observed. Clinical outcomes and HRQL were worse after presumed perinatal stroke and childhood stroke than after neonatal injury. This bimodal relationship is contradictory to published reports suggesting more damage after an early brain injury and a linear relationship between the age at brain injury and the neurocognitive outcome.6-9 However, the results of the current study are consistent with the process of brain maturation, which has been described to occur in a step-wise manner with critical maturation periods separated by more stable periods.21, 22 This phenomenon is supported by animal models of perinatal brain injury, which propose a non-linear relationship between the age at brain injury and recovery.5
Of particular note is the fact that the differential outcomes across the three age categories were only documented for the global and sensorimotor domains, with no differences apparent in the cognition, behavior, or language sectors. This may be attributable to the similar levels of cortical involvement in all three types of stroke. The finding of similar cognitive outcomes across the different age categories demonstrated here is in contrast to previous reports suggesting more cognitive impairment after an early brain injury.6 However, a direct comparison of different studies is difficult because of the heterogeneous patient populations and different study characteristics. Nonetheless, the lack of formal cognitive testing may have underestimated the extent of cognitive and language impairments in the current study.
Age at stroke onset was associated with clinical recovery, as well as with the distribution and severity of the brain injury and, hence, the clinical presentation. Participants with presumed perinatal stroke tended to have a relatively large size of stroke with an involvement of deep grey matter structures and corresponding vascular territories. This could account for the pronounced motor disabilities documented in these participants. However, the underlying mechanism accounting for differential brain involvement remains unclear. One plausible hypothesis is the disruption of critical neurodevelopmental processes and pathways responsible for motor development after in-utero stroke.
We also found that the underlying risk factors vary according to the age at stroke onset. Potential risk factors were identified in 75% of the neonatal and 87% of the participants with childhood stroke. In contrast, in more than half of the participants with presumed perinatal stroke, no underlying etiologies were apparent. A number of risk factors have been linked to presumed perinatal stroke in the literature.23, 24 However, the interpretation of the results of these studies is limited by the retrospective nature of the study designs. Clearly, there is a need for prospective and case–control studies to explore and provide more conclusive evidence regarding the etiological factors relating to the different types of perinatal and childhood stroke.
The limitations of our study included a modest sample size, and the retrospective nature of identifying our population. This is not unusual, however, when determining long-term outcomes and HRQL in a relatively uncommon population with a survey-based study design. Nevertheless, the data on morbidity, clinical, and radiological characteristics are similar to those reported in previously published literature.8, 25, 26 Another limitation of the study is that, since participants with presumed perinatal stroke were more likely to be picked up because of their motor presentations, they may be more likely to have worse motor outcomes. This may bias the study findings. In addition, the different pattern of injury across the three groups might have contributed to the different outcomes among study groups. However, age at onset of stroke was still an important prognostic marker after adjusting for the location and size of the lesion. Vascular territory could not be included in the multivariable analysis because of sample size limitations.
One of the strengths of our study is a median follow-up of more than 4 years. Thus, our observations provide a realistic assessment of the long-term outcomes and HRQL after pediatric ischemic stroke. In addition, proxy measures are being increasingly recognized as important instruments in HRQL and health outcomes research, and for designing health care plans and models, as these tools provide parents' perspectives about their child's health status. In addition, while proxy measures may be considered a limitation, in this case, they are actually a strength given the validated approach of the PedsQL 4.0 instrument; furthermore, because some children would not be able to provide self-reports, the consistent measurement approach used provided a more valid comparison across age groups. In this study, we were successful in achieving a parent perspective, which included both objective and subjective assessments of their child's health status and HRQL. Furthermore, our study gives rise to questions regarding earlier presumptions of an inverse linear relationship between recovery, age at stroke onset, and outcome.
Acknowledgements
The authors wish to thank the parents and children who participated with their time and patience to provide the information required for the quality of life measures. We also thank our funding agencies, including NeuroDevNet and the Alva Foundation, as well as Alberta Innovates Health Solutions (AIHS) and the Women's and Children's Health Research Institute (WCHRI), for their support to SKG. The authors have stated that they had no interests that might be perceived as posing a conflict or bias.
