International Classification of Diseases, Ninth Revision (ICD-9) and International Classification of Diseases, Tenth Revision (ICD-10) code searching was performed by 2 independent biomedical informatics experts separately accessing the universal provincial hospital inpatient and outpatient systems respectively. Analysts were blinded to each other’s findings, which were then cross-referenced to exclude duplicates and create the final sample. APrON indicates Alberta Pregnancy Outcomes and Nutrition; APSP, Alberta Perinatal Stroke Program; CSVT+HT, hemorrhagic transformation of cerebral sinovenous thrombosis; HIE+HT, hemorrhagic transformation of hypoxic ischemic encephalopathy; HT, hemorrhagic transformation; NAIS+HT, hemorrhagic transformation of neonatal arterial ischemic stroke; NHS, neonatal hemorrhagic stroke; PI, principal investigator; PPHS, presumed perinatal hemorrhagic stroke; and REDCap, Research Electronic Data Capture.
Arrowheads designate location of injury. A, Neonatal hemorrhagic stroke is a hematoma in the brain parenchyma (gradient echocardiogram shown). B, Neonatal arterial ischemic stroke with hemorrhagic transformation is blood within an area of arterial territory infarction (T1 shown). C, Cerebral sinovenous thrombosis with hemorrhagic transformation is blood within an area of venous infarction confirmed by parenchymal and venous imaging (diffusion image shown). D, Hypoxic-ischemic encephalopathy with hemorrhagic transformation is clinical and radiographic confirmation of blood within areas of global infarction (T1 shown). E, Presumed perinatal hemorrhagic stroke is a focal area of remote parenchymal damage with demonstration of hemorrhage (susceptibility shown).
Long-term neurological outcome categories by Pediatric Stroke Outcome Measure were available for 50 cases. Mean Pediatric Stroke Outcome Measure scores varied widely across subgroups (A). The error bars indicate the greatest and lowest value excluding outliers, the horizontal lines within the boxes are median lines, and the circles indicate outliers (> or < 1.5 times the quartile value). Classification of poor outcome by Pediatric Stroke Outcome Measure >1 or ≥1 was generally comparable across groups with hemorrhagic transformation of neonatal arterial ischemic stroke (HT+NAIS) and hemorrhagic transformation of cerebral sinovenous thrombosis (HT+CSVT) having the highest and lowest proportions, respectively (B). For the incidence rate of poor outcome, HT+NAIS, 80% had Pediatric Stroke Outcome Measure >1 and 100% had ≥1 and HT+CSVT, 20% had Pediatric Stroke Outcome Measure >1 and 60% ≥1. Examination of Pediatric Stroke Outcome Measure subcategories for the neonatal hemorrhagic stroke (NHS) group demonstrated wide-ranging morbidity with sensorimotor deficits being most common (C). PPHS indicates presumed perinatal hemorrhagic stroke; HIE+HT, hemorrhagic transformation of hypoxic ischemic encephalopathy.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Cole L, Dewey D, Letourneau N, et al. Clinical Characteristics, Risk Factors, and Outcomes Associated With Neonatal Hemorrhagic Stroke: A Population-Based Case-Control Study. JAMA Pediatr. 2017;171(3):230–238. doi:10.1001/jamapediatrics.2016.4151
Copyright 2017 American Medical Association. All Rights Reserved.
What are the incidence, types, presentations, associated factors, and long-term outcomes of neonatal hemorrhagic stroke?
This population-based case-control study found a neonatal hemorrhagic stroke incidence of 1 in 6300 live births and independent associations between idiopathic neonatal hemorrhagic stroke and lower maternal age, primiparity, prior spontaneous abortion, difficult fetal transition, and small for gestational age. Outcomes were poor in approximately 50% of infants including sensorimotor delays and epilepsy.
Neonatal hemorrhagic stroke is more common than previously reported, and clinical risk factors associated with idiopathic neonatal hemorrhagic stroke do not support a primary acquired pathophysiology.
Hemorrhage into the brain of term newborns often results in major injury and lifelong disability. The clinical epidemiology of neonatal hemorrhagic stroke (NHS) remains undefined, hindering the development of strategies to improve outcomes.
To characterize the incidence, types, presentations, associated factors, and outcomes of neonatal hemorrhagic stroke.
Design, Setting, and Participants
Population-based, nested case-control study. The Alberta Perinatal Stroke Project, a provincial registry, ascertained NHS cases using exhaustive diagnostic code searching (1992-2010, >2500 medical record reviews). Prospective cases were captured through the Calgary Pediatric Stroke Program (2007-2014). Participants included term neonates with magnetic resonance imaging–confirmed NHS including primary and secondary intracerebral hemorrhage, hemorrhagic transformation of ischemic injury, and presumed perinatal hemorrhagic stroke. Control infants with common data were recruited from a population-based study (4 to 1 ratio).
Main Outcomes and Measures
Infants with NHS underwent structured medical record review using data-capture forms and blinded scoring of neuroimaging. Clinical risk factor common data elements were explored using logistic regression. Provincial live births were obtained from Statistics Canada. Outcomes were extrapolated to the Pediatric Stroke Outcome Measure.
We identified 86 cases: 51 infants (59%) with NHS, of which 32 (67%) were idiopathic, 30 (35%) were hemorrhagic transformation of primary ischemic injuries (14 with neonatal cerebral sinovenous thrombosis, 11 with hypoxic ischemic encephalopathy, and 5 with neonatal arterial ischemic stroke), and 5 were presumed perinatal hemorrhagic stroke. Sixty-two percent were male. Incidence of pure NHS was 1 in 9500 live births and 1 in 6300 for all forms. Most presented in the first week of life with seizures and encephalopathy. Acute neurosurgical intervention was rare (3 of 86 total cases; 3.5%). Temporal lobe was the most common NHS location (16 of 51 pure NHS cases; 31%). A primary cause was evident in 19 of the 51 cases of non–hemorrhagic transformation NHS (37%). Idiopathic NHS was independently associated with lower maternal age (odds ratio [OR], 0.87; 95% CI, 0.78-0.94), primiparity (OR, 2.98; 95% CI, 1.18-7.50), prior spontaneous abortion (OR, 0.11; 95% CI, 0.02-0.53), difficult fetal transition (bradycardia [OR, 15.0; 95% CI, 2.19-101.9] and low Apgar [OR, 14.3; 95% CI, 2.77-73.5]), and small for gestational age (OR, 14.3; 95% CI, 1.62-126.1). Follow-up of 50 cases at a median of 37 months demonstrated poor neurological outcomes in 21 patients (44%).
Conclusions and Relevance
Neonatal hemorrhagic stroke is more common than previously reported, occurring in at least 1 in 6300 live births. Etiologies are approximately equally distributed between idiopathic, secondary, and hemorrhagic transformation. Clinical associations do not suggest a common mechanism or predictability of NHS. Recurrence is rare. Outcomes are often poor, mandating attention to prevention and rehabilitation.
Perinatal stroke is a leading cause of cerebral palsy and lifelong neurological morbidity. Compared with perinatal ischemic stroke diseases, which are now well defined, term neonatal hemorrhagic stroke (NHS) research has been limited. Neonatal hemorrhagic stroke is frequently encountered clinically, but only 1 retrospective study of 20 patients could suggest an incidence of 6.2 per 100 000 live births.1 The lack of well-powered, population-based studies has limited understanding of presentations, possible risk factors, treatment strategies, and long-term outcomes.
Classification of perinatal intracerebral hemorrhage is challenging, and variable terminology represents another barrier to progress. Perinatal intracerebral hemorrhage is defined within the National Institutes of Health Common Data Elements as a term neonate with encephalopathy, seizures, altered mental status, and/or neurological deficit within the first 28 days of life with a focal collection of blood within the brain parenchyma confirmed by neuroimaging or autopsy.2 This is distinct from intracranial hemorrhage in preterm infants, where germinal matrix hemorrhages are common.3 The definition does not address overlapping hemorrhages (intraventricular and subarachnoid), does not capture nongerminal matrix hemorrhages in the fetus, and excludes presentation beyond the neonatal period.
The epidemiology of NHS is poorly described. Modestly powered, single-center, uncontrolled, retrospective, non–population-based studies constitute the bulk of current knowledge.1,4-6 Studies report variable proportions of different syndromes.1,6-8 Minimal case-control data limit the exploration of risk factors, a successful approach in perinatal ischemic stroke.9,10 Suggested risk factors include vascular malformations, bleeding diatheses, and possibly trauma, although these account for few cases. “Excessive” trauma is notoriously difficult to define, and studies suggesting associations with NHS are uncontrolled.11,12 Well-powered, population-based, case-control data are required to explore potential risk factors.
Modern neuroimaging has facilitated the classification and study of specific perinatal stroke disease states. Hemorrhagic transformation (HT) of ischemic injuries, such as neonatal arterial ischemic stroke (NAIS), neonatal cerebral sinovenous thrombosis (CSVT), and hypoxic ischemic encephalopathy (HIE), also creates intraparenchymal hemorrhage. Magnetic resonance imaging sequences are exquisitely sensitive to hemorrhage, including detection of hemosiderin deposition years later, creating the ability to retrospectively diagnose NHS that was asymptomatic at birth, analogous to presumed perinatal ischemic stroke.9,13
With no controlled population-based studies to our knowledge, most cases unexplained, poor outcomes lasting decades, and no treatment or prevention strategies, there is a clear need for improved understanding of NHS epidemiology. We used a large, population-based perinatal stroke cohort to perform a nested case-control study to define NHS incidence, classification, presentations, possible risk factors, and outcomes.
This was a population-based, nested case-control study. Neonatal hemorrhagic stroke cases were identified via retrospective and prospective methods within the Alberta Perinatal Stroke Project. This research cohort was designed to capture all perinatal stroke cases in Southern Alberta, Canada (population of approximately 2.1 million), leveraging universal health care at a single tertiary care pediatric center. Established in 2008, the Alberta Perinatal Stroke Project includes more than 900 magnetic resonance imaging–confirmed perinatal stroke cases. Methods were approved by the institutional research ethics board at University of Calgary and written informed consent was obtained.
Case ascertainment is outlined in Figure 1. Exhaustive retrospective analysis of at least 140 International Classification of Diseases, Ninth Revision and International Classification of Diseases, Tenth Revision codes was performed across inpatient and outpatient populations for 1992-2010. A standardized data collection form extracted medical records and imaging report data. Imaging was then reviewed in person with an expert investigator (A.K.) to confirm perinatal stroke diagnosis and subtype. Inclusion criteria were (1) radiographic NHS, (2) term birth (≥35 weeks to exclude germinal matrix bleeding), and (3) resident of Southern Alberta. Prospectively, all cases were identified through active Alberta Perinatal Stroke Project case surveillance from 2007 through June 2015.
Neonatal hemorrhagic stroke was defined as imaging evidence of blood within the brain parenchyma with or without intraventricular or subarachnoid blood (Figure 2). Cases where blood was exclusively extra-axial were not included. Imaging confirmed concurrent neonatal arterial ischemic stroke (NAIS), neonatal cerebral sinovenous thrombosis (CSVT), and global hypoxic-ischemic encephalopathy (HIE). Neonatal hemorrhagic stroke categories were hemorrhagic transformation (HT) of a primary ischemic injury (NAIS+HT, CSVT+HT, or HIE+HT),14 secondary NHS (sNHS) explained by a highly probably cause (vascular malformation and bleeding diathesis), or idiopathic NHS (iNHS). Major trauma was defined by concurrent skull fracture or severe bruising/soft tissue injury. Presumed perinatal hemorrhagic stroke was defined as a term child with no perinatal neurological history demonstrating remote NHS on imaging performed in childhood.
Using a standardized data capture form, medical and birth record review collected the following: (1) demographics and clinical presentation (age, sex, presenting signs, and acute neurosurgical intervention); (2) potential risk factors classified using common data elements including maternal health, pregnancy, labor and delivery, and neonatal factors; (3) neuroimaging analysis using a standardized scoring tool (location, mass effect, restricted diffusion, hydrocephalus, or concurrent NAIS/CSVT/HIE)15,16; and (4) neurological outcomes measured by the Pediatric Stroke Outcome Measure13,17 and epilepsy outcomes according to the Modified Engel Classification.18 The Pediatric Stroke Outcome Measure, based on the most recent clinical assessment at older than 12 months of age by the same pediatric neurologist (A.K.), generated summative scores (normal/mild/moderate/severe) for sensorimotor, language production, language comprehension, and cognitive deficits. Total scores were dichotomized into good vs poor based on cutoffs of more than 1 and at least 1. The Modified Engel Classification classified epilepsy outcomes as follows: 0 = seizure free and no antiepileptic drugs for 6 months, 1 = seizure free for 6 months with medication or seizure free without medication less than 6 months, 2 = less than 1 seizure/mo with medication, 3 = 1 to 4 seizures/mo with medication, 4 = 5 to 30 seizures/mo with medication, and 5 = 30 or more seizures/mo.
Control data were obtained from the Alberta Pregnancy Outcomes and Nutrition study, a prospective cohort study of more than 2000 mother-infant dyads followed up from early pregnancy beyond the perinatal period (Figure 1).19 This sample was optimal because it was population-based from the same geographical region with precise matching of common data elements. A blinded coordinator randomly selected 4 control dyads per NHS case (n = 204).
Data were entered and managed through Research Electronic Data Capture hosted at the University of Calgary. Incidence rates were calculated annually by dividing the number of cases into the number of live births in Southern Alberta (Statistics Canada). Categorical and continuous variables were compared using χ2, Fisher exact, and t tests. Logistic regression modeling used a reverse elimination method where the maximum number of variables was entered into the model, and variables not associated with iNHS were removed one by one until only significant variables (or confounders) remained.20 Variables were grouped into blocks according to clinical significance. Significant associations were expressed as odds ratios (ORs) with 95% CIs. Analysis used Stata, version 13 (StataCorp).
The final sample consisted of 86 cases. Fifty-one (59%) were pure NHS (iNHS/sNHS), with 32 (67%) being idiopathic. Thirty (35%) were HT including CSVT (14), HIE (11), and NAIS (5). Five cases were presumed perinatal hemorrhagic stroke. A significant male predominance was observed (62%). Additional population characteristics are available in Table 1.
Mean live births in Southern Alberta were 24 650 per year, increasing steadily during the study period. This yielded an incidence for all forms of NHS of 15.9 in 100 000 per year or approximately 1 in 6300 live births. For pure NHS and presumed perinatal hemorrhagic stroke, the incidence was 10.5 in 100 000 per year or approximately 1 in 9500 live births. Incidence rates did not differ between the retrospective and prospective ascertainment periods. Prevalence estimates suggest there are currently at least 1000 Canadian children living with NHS.
Most presented within the first 28 days of life (n = 81; 94%) and were imaged in the first week (n = 59; 69%). Median age at diagnosis or first imaging did not differ between subgroups. Common clinical presentations consisted of seizures (n = 54; 67%) and/or encephalopathy (n = 67; 83%) and/or hypotonia (n = 34; 42%). Apgar scores were less than 5 at 1 minute in 36 cases (47%) and at 5 minutes in 12 cases (16%). Neonatal intensive care unit admission was common (88%). There were no differences in proportions of clinical presentations between subgroups. No cases of neonatal vitamin K refusal were documented. Given the decades of study, neonatal electroencephalogram and cerebral monitoring were variable and data were not consistently available. Neonatal neurosurgery was rare, with 1 infant having a hematoma evacuation and aneurysm clipping and another requiring external ventricular drain placement.
Imaging results are summarized in Table 2. Initial modalities included computed tomography in 31 patients (36%) and magnetic resonance imaging in 55 patients (64%). No difference in median age at imaging was seen between subgroups aside from presumed perinatal hemorrhagic stroke. The temporal lobe was the most common location (31% compared with <20% for other locations). Infratentorial hemorrhage was uncommon (cerebellar in 11 patients [13%]). Multifocal hemorrhage was described in 22% and comparable across subgroups. Mass effect was documented in a minority (24%). Restricted diffusion within and surrounding hemorrhage was observed in 30% (n = 6 iNHS, n = 4 sNHS, n = 3 nAIS+HT, n = 5 CSVT+HT, and n = 8 HIE+HT). Intraventricular hemorrhage was described in 49% but acute hydrocephalus was uncommon (11%). Cerebral sinovenous thrombosis + HT and HIE+HT demonstrated higher rates of thalamic involvement, intraventricular hemorrhage, and acute hydrocephalus.
In the 51 patients with NHS, 19 (37%) had a highly probable primary etiology. Blood disorders were most common including 7 cases of severe thrombocytopenia, 1 of unspecified coagulopathy, and 1 of hemophilia A. Vascular abnormalities in 6 cases included arteriovenous malformation (n = 3), cavernoma (n = 2), and aneurysm (n = 1). Major trauma was identified by skull fracture, soft tissue injury, and extra-axial hemorrhage in 4 cases. When these 19 sNHS cases were compared with the remaining 32 idiopathic cases, the only differences observed involved only 1 to 2 patients and were not considered clinically significant (Table 2).
The maternal and infant variable exploratory logistic regression model demonstrated independent associations between maternal age, spontaneous abortion, and primiparity with iNHS. The odds of iNHS were reduced by a factor of 0.87 (95% CI, 0.78-0.94) for every year of increase in maternal age. When controlling for age and spontaneous abortion, the odds of iNHS were increased with primiparity by 2.98 times (95% CI, 1.18-7.50). Controlling for age and parity, the odds of iNHS among mothers with prior spontaneous abortion were 0.11 times the odds of those without (95% CI, 0.02-0.53). The labor and delivery model found that bradycardia or variable decelerations were both independently associated with iNHS, controlling for maternal age (OR, 15.0; 95% CI, 2.19-101.9 and OR, 14.3; 95% CI, 2.77-73.5, respectively). Among neonatal factors, the odds of being small for gestational age were 14.3 times higher (95% CI, 1.62-126.1) in iNHS compared with control infants when controlling for maternal age and 5-minute Apgar score. Apgar scores were independently associated with iNHS, controlling for maternal age and small for gestational age. For every unit increase in 5-minute Apgars, the odds of iNHS were 86% lower (OR, 0.24; 95% CI, 0.12-0.48).
There were 50 cases of infants with Pediatric Stroke Outcome Measure at older than 12 months of age. Median (SD) age at last Pediatric Stroke Outcome Measure was 37 (51) months with follow-up ranging from 1 to 15 years. There were no recurrences of hemorrhagic stroke. Neurosurgical interventions outside the neonatal period included ventriculoperitoneal shunt insertion (n = 4), epilepsy surgery (n = 3), and arteriovenous malformation embolization (n = 1). There were 3 deaths (4%). All occurred beyond the neonatal period including 5 weeks (sNHS with malignant arteriovenous malformation), 5 years (sNHS with shunt failure), and 7 years (HIE with severe disability and intercurrent illness).
Outcomes are summarized in Figure 3. Mean total Pediatric Stroke Outcome Measure scores varied across subgroups (Figure 3A). Mean Pediatric Stroke Outcome Measure scores were significantly higher in NAIS+HT compared with CSVT+HT. Proportions of poor outcome (Pediatric Stroke Outcome Measure ≥1 or >1) are compared in Figure 3B. Outcomes were categorized as poor (Pediatric Stroke Outcome Measure >1) in 21 cases (42%) compared with 28 cases (56%) for Pediatric Stroke Outcome Measure of at least 1. Both poor outcome categories were significantly different, with NAIS+HT having higher proportions compared with CSVT+HT. In the iNHS and sNHS groups, approximately 40% had poor outcome regardless of the cutoff value. Breakdown of Pediatric Stroke Outcome Measure subcategories in the iNHS and sNHS groups demonstrated a wide range of morbidities with sensorimotor deficits being most common (Figure 2C). The iNHS group had 16 cases with follow-up at greater than 12 months of age, and there were 6 cases (38%) with poor outcome. Thirty-nine case patients (78%) required physical, occupational, and/or speech therapy at last follow-up.
Epilepsy occurred in 11 cases (13%). Rates of epilepsy by subgroup were 3 with iNHS or sNHS (6%), 1 with NAIS+HT (20%), 3 with CSVT+HT (21%), 2 with HIE+HT (18%), and 2 with presumed perinatal hemorrhagic stroke (40%). Mean (SD) age at seizure onset was 2.5 (3.2) years (range, 0.1-13 years). The mean (SD) number of anticonvulsants was 1.3 (1.4) (range, 0-8). Outcomes according to modified Engel Classification were 0 (n = 3; 27%), 1 (n = 3; 27%), 2 (n = 1; 9%), 3 (n = 2; 18%), 4 (n = 1; 9%), and 5 (n = 1; 9%). Three (23%) had epilepsy surgery with postoperative Engel classifications of 0, 1, and 5.
Five cases presented outside the first 28 days of life with imaging confirming remote focal hemorrhage consistent with presumed perinatal hemorrhagic stroke. Clinical presentations included seizures and developmental delay in 2 cases and early hand preference or motor asymmetry with developmental delay in 2 others. Positive family history led to imaging and diagnosis without symptoms in the fifth case. Etiologies included hereditary hemorrhagic telangiectasia in 2 and hemophilia B in 1, while the other 2 were idiopathic.
We provide population-based, controlled data, including original imaging classification, common data elements, and standardized outcomes, for children with NHS. The incidence of all forms of NHS approximates 1 in 6000 live births. Cases are often explained by primary conditions or hemorrhagic transformation of ischemic injury, but many remain idiopathic. Neonatal hemorrhagic stroke is associated with difficult neonatal transition, but no primary causative associations are suggested for idiopathic NHS. Outcomes are poor, with long-term morbidity in most.
Our methods advance the levels of evidence for NHS. Most previous studies have been uncontrolled case series with inconsistent classification, variable inclusion criteria, and other limitations.1,6-8 A notable exception is a case-control study using administrative data from Northern California.1 We have tried to overcome some of the limitations of that study with truly population-based data, broader International Classification of Diseases code searching, and original neuroimaging review. Many findings appear comparable, including clinical presentations and associated factors, but notable discrepancies are discussed here.
Inconsistent terminology and classification has complicated NHS studies to date.1,6-8 We focused on NHS with direct parenchymal brain injury6 and maximal risk of long-term morbidity. Isolated hemorrhage in the subdural, subarachnoid, and epidural spaces does not usually cause brain injury and is common in normal newborns.21 Classification must also consider hemorrhagic transformation of ischemic injuries with our findings supporting previous evidence7 that this is a common cause of intraparenchymal blood. A final classification advance is our description of presumed perinatal hemorrhagic stroke, analogous to its ischemic counterpart where such terminology has advanced studies.9,13
Precise incidence and prevalence estimates for term intracerebral hemorrhage are not established. In the retrospective, 20-case, administrative study, the estimated incidence was 6.2 per 100 000 (approximately 1 in 16 000) live births for combined intraparenchymal and subarachnoid hemorrhage.1 Our population-based calculations, including prospective ascertainment and original imaging, may be more accurate. Our rate of 1 in 9000 live births for pure NHS (iNHS and sNHS) and 1 in 6000 for all NHS is at least 3 times higher. While we are confident our methods captured a very high proportion of cases without false positives, this still reflects a minimum incidence.
Acute clinical presentations of seizures and encephalopathy are consistent with previous studies,7,8,22 and proportions were similar across groups compared with ischemic perinatal stroke.23 This suggests NHS should be considered in any neonate with encephalopathy, but that specific clinical predictors are unlikely. Although we could not investigate seizure features in detail, previous studies suggest lesion-specific (eg, temporal) and subclinical seizures often occur, suggesting a role for electroencephalogram monitoring.24 Precise diagnosis and assessment of cause requires magnetic resonance imaging.
Our high proportion of idiopathic cases is consistent with existing literature.1,22 Our exploratory regression analysis for possible associations in this subset was an advantage over previous studies that often considered all types together. Consistent with previous studies1 was an association between difficult transition including fetal bradycardia, decelerations, and low Apgar scores. As suggested by NHS7 and perinatal ischemic stroke25-27 studies, such difficulty may be secondary to brain injury that has already occurred. Associations between maternal factors including age, primiparity, and abortion have only been inconsistently described in perinatal ischemic stroke. Additional possible risk factors not supported by our data include gestational age and postmaturity.1 Trauma, an often-suggested but usually unproven consideration, was rare. Uncontrolled studies have suggested increased rates of assisted deliveries in NHS, but we did not observe this.4,6 Our results support long-standing evidence dissociating routine trauma, assisted delivery, and neonatal hemorrhage.12
Long-term outcomes were often poor and comparable with previous reports.6,28,29 Our length of follow-up was long, although young children may grow into their deficits over decades. One study found that 56% of NHS children required physical, occupational, and/or speech therapy services, a rate comparable with our population.7 Careful clinical and imaging classification may assist in predicting some outcomes. For example, hemorrhage associated with deep CSVT was associated with hemorrhage pattern (thalamic and intraventricular extension) while thalamic hemorrhage has been associated with long-term epilepsy.30-32 The recurrence risk of NHS appears to be very low.
Limitations include potential reporter bias during medical record review, although these were completed prior to final imaging confirmation by the lead investigator. Completeness and consistency of data collection across multiple Southern Alberta centers was challenging but facilitated by universal health care, unified medical records systems across sites, and tertiary level neonatal and pediatric care at only a single center. Neurological outcome data were only available from the most recent clinic visit and more complete data (eg, neuropsychological testing) at longer follow-up intervals may demonstrate additional morbidities that we could not measure. Our strict case criteria that excluded subdural and subarachnoid hemorrhages were considered more specific but limits comparability with some previous studies.
What causes idiopathic NHS? A small weakness in an artery that ruptures with the large surge in blood pressure that accompanies transition to extrauterine life is purely speculative. However, this would be consistent with existing evidence including the lack of associations mentioned in previous paragraphs, known risk of larger arteriovenous malformations, lack of lesions on follow-up imaging, and minimal recurrence. Why this would occur in 1 of every 6000 to 10 000 live births may relate to unique vascular development or anatomy. An increasing number of genetic disorders may lead to perinatal hemorrhage.33 Quantified arterial tortuosity may be an imaging biomarker of vascular biology in children with stroke,34 and we have recently demonstrated increased variability of tortuosity in NHS compared with controls (unpublished). Collectively, evidence suggests NHS pathophysiology involves rare events occurring in uniquely susceptible individuals rather than any controllable external factors, limiting opportunities for prevention.
Corresponding Author: Adam Kirton, MD, Alberta Children’s Hospital, University of Calgary, 2888 Shaganappi Trail NW, Calgary, AB T3B6A8, Canada (email@example.com).
Accepted for Publication: October 19, 2016.
Correction: This article was corrected on April 3, 2017, to correct errors in Figure 1 and Table 1.
Published Online: January 17, 2017. doi:10.1001/jamapediatrics.2016.4151
Author Contributions: Dr Kirton had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Cole, Letourneau, Kirton.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Cole, Letourneau, Kirton.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Cole, Letourneau, Chaput, Kirton.
Obtained funding: Cole, Dewey, Letourneau, Kaplan, Kirton.
Administrative, technical, or material support: Cole, Dewey, Letourneau, Kaplan, Hodge, Floer, Kirton.
Study supervision: Letourneau, Hodge, Kirton.
Conflict of Interest Disclosures: None reported.
Funding/Support: Funding was provided by Alberta Innovates Health Solutions.
Role of the Funder/Sponsor: The funding source did not influence the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review or approval of the manuscript; or decision to submit the manuscript for publication.
Create a personal account or sign in to: