Dr Ericka Fink at Children’s Hospital of Pittsburgh of UPMC, 4401 Penn Avenue, Faculty Pavilion, 2nd Floor, Pittsburgh, PA 15224, USA. E-mail: email@example.com
Aim Dysautonomia after brain injury is a diagnosis based on fever, tachypnea, hypertension, tachycardia, diaphoresis, and/or dystonia. It occurs in 8 to 33% of adults with brain injury and is associated with poor outcome. We hypothesized that children with brain injury with dysautonomia have worse outcomes and prolonged rehabilitation, and sought to determine the prevalence of dysautonomia in children and to characterize its clinical features.
Method We developed a database of children (n=249, 154 males, 95 females; mean [SD] age 11 years 10 months [5y 7mo]) with traumatic brain injury, cardiac arrest, stroke, infection of the central nervous system, or brain neoplasm admitted for rehabilitation to The Children’s Institute of Pittsburgh between 2002 and 2009. Dysautonomia diagnosis, injury type, clinical signs, length of stay, and Functional Independence Measure for Children (WeeFIM) testing were extracted from medical records, and analysed for differences between groups with and without dysautonomia.
Results Dysautonomia occurred in 13% of children with brain injury (95% confidence interval 9.3–18.0%), occurring in 10% after traumatic brain injury and 31% after cardiac arrest. The combination of hypertension, diaphoresis, and dystonia best predicted a diagnosis of dysautonomia (area under the curve=0.92). Children with dysautonomia had longer stays, worse WeeFIM scores, and improved less on the score’s motor component (all p≤0.001).
Interpretation Dysautonomia is common in children with brain injury and is associated with prolonged rehabilitation. Prospective study and standardized diagnostic approaches are needed to maximize outcomes.
• Dysautonomia occurred in 13% of children with brain injury (95% CI 9.3–18.0).
• Hypertension, diaphoresis, and dystonia best predicted dysautonomia among clinical signs.
• Children with dysautonomia experienced longer rehabilitation and worse functional scores.
Dysautonomia is a severe, debilitating sequela of acquired brain injury (ABI). It is characterized by dysfunction of the autonomic nervous system with resultant disturbances in temperature and hemodynamic homeostasis, and dystonic muscle contractions. Myriad names in the literature (including ‘paroxysmal autonomic instability with dystonia’ and ‘paroxysmal sympathetic hyperactivity,’ among others) and the lack of generalized agreement over precise symptomatology required for diagnosis complicate the development of an organized approach to this disorder.1–3 The most common constellation of clinical signs, however, includes a combination of fever, tachypnea, hypertension, tachycardia, diaphoresis, and dystonia.1,2,4,5 Over a dozen names have been given to this constellation of signs and symptoms (‘sympathetic storms’, ‘thalamic storms’, and others) but for this study we have chosen to adhere to the most predominant term, ‘dysautonomia’.6,7 After injury, symptom onset has been observed between 1 and 60 days, but most reports focus on the first several weeks after injury.1 One prospective study observed widespread autonomic arousal (24%) among its participants within 7 days of injury, but only 8% qualified as having dysautonomia by day 14.8 The duration of dysautonomia symptoms is variable, with some groups proposing two or three distinct phases, or even distinct disorders: a short-duration variant (lasting through the acute recovery stage), as well as a prolonged variant (lasting weeks to 6 or more months).4,6,9 The pathophysiology of dysautonomia is not well understood but may arise from disinhibition of diencephalic autonomic centers, which, in turn, leads to over-responsiveness to innocuous and/or nociceptive sensory stimuli.10
The prevalence and overall outcomes of dysautonomia vary by injury type. In previous studies, the prevalence of dysautonomia has ranged from 8 to 33% after traumatic brain injury (TBI) and from 6 to 29% after non-traumatic or anoxic brain injury.2,5,8,11–13 There is very little in the literature on dysautonomia in children, but its prevalence after TBI has been reported to be 12%, whereas a prevalence of 29% has been observed after anoxic brain injury in a small series.5 Dysautonomia has been prospectively associated with worse outcomes after TBI in adults: patients had longer stays in intensive care units, worse Glasgow Outcome Scale scores, increased frequency of infection, need for tracheotomy, longer duration of post-traumatic amnesia, and increased estimated hospital costs.8,13 Importantly, dysautonomia has also recently been associated with increased mortality in adults after TBI and diffuse axonal injury.14
In the pediatric population, clinical reports about dysautonomia have infrequently extended beyond case reports15,16 despite the fact that brain injury is a leading cause of morbidity and mortality in children. Thus, our aim in this study was to define the epidemiology of dysautonomia in children recovering from ABI, including prevalence, signs, and symptoms used to make the diagnosis, brain injury etiologies, and outcomes. We audited medical records from a regional rehabilitation center for children to accomplish these objectives. In addition, we extracted from these records various medical aspects of the care that these children required during their post-injury period. We hypothesized that after ABI, children with dysautonomia have longer rehabilitation courses and worse functional outcomes than children without it.
Design and setting
The institutional review boards at the University of Pittsburgh and at The Children’s Institute approved this study. The Children’s Institute is the regional pediatric rehabilitation hospital for children in western Pennsylvania, northern West Virginia, and southern Ohio. We performed a retrospective cohort chart review of children admitted to The Children’s Institute between October 1, 2002 and October 1, 2009. Children between 1 week and 18 years of age who were admitted to The Children’s Institute with an acute ABI (defined as TBI, cardiac arrest, stroke [hemorrhagic and ischemic], infection of the central nervous system [CNS], or brain neoplasm) were studied. To obtain this cohort, electronic medical records were queried for admission or discharge diagnoses of any of the following: TBI, cardiac arrest, cardiopulmonary resuscitation, CNS infection, stroke, hyperthermia, diencephalic seizure, thalamic storm, hypertension, dysautonomia, and dystonia. Children were excluded if there was a history of congenital heart defects, congenital dystonia or spasticity, admissions for non-CNS-related rehabilitation, and lack of confirmation of primary ABI in the medical chart (Fig. 1).
Definition of dysautonomia and data collection
For this report, we relied on the clinical diagnosis of dysautonomia from the medical record, as documented by physicians. Any of the following terms (at admission or discharge) were sufficient for inclusion within the dysautonomia group: dysautonomia, thalamic storms, hypothalamic storms, adrenergic storms, sympathetic storms, adrenergic surges, hyperadrenergic state, sympathetic hyperactivity, or autonomic instability.
Demographic data were abstracted from the medical records including age, sex, race, and primary etiology of brain injury. Signs of dysautonomia from the medical records were compiled by examining all recorded vital signs during the rehabilitative stay and clinical notes. A priori definitions of abnormal vital signs were adopted (fever defined as temperature >38°C; tachypnea defined as respiratory rate more than two times normal for age; hypertension defined as systolic blood pressure greater than the 95th centile for age; tachycardia defined as heart rate >2SD over normal reference range for age, and dystonia and diaphoresis diagnosed by treating physicians in medical records).17,18 Aforementioned ‘rehabilitation stay’ refers to one course per patient, including all interrupting acute care admissions and readmissions to the rehabilitation center. Other data relevant to the treatment course were also collected including seizure diagnosis (both pre- and post-injury), evidence of infections, duration of in-patient rehabilitation, number of readmissions to an acute care facility, medications, patient disposition at discharge, and admission and discharge Functional Independence Measure for Children (WeeFIM) scores. We then retrospectively correlated these clinical data with the dysautonomia diagnoses extracted from the record.
The WeeFIM assesses a variety of functional motor and cognitive tasks on a scale from 1 to 7 (total motor range 13–91; total cognitive range 5–35).19 Though historically limited to the age range 6 months to 7 years, the WeeFIM has been standardized for use (and compared with the adult FIM) in the adolescent population with developmental disability,20 and used in recent literature to assess a nationwide TBI sample age 0 to 21 years.21 Only one patient in our population fell below 6 months of age; this patient was not assessed by the rehabilitation center by WeeFIM, and was therefore excluded from the WeeFIM analysis.
The prevalence of dysautonomia in children with ABI at the rehabilitation institution was calculated. Additionally, the frequency of abnormal vital signs (defined above and consistent with signs of dysautonomia) and the length of hospital stay in children with and without the diagnosis of dysautonomia was calculated. Statistical comparisons between children with and without dysautonomia were performed, using Fisher’s exact tests (for all categorical variables), t-tests (for normally distributed continuous variables), Mann–Whitney rank sum tests (for non-parametric continuous variables), and logistic regression (to obtain odds ratios and 95% confidence intervals [CIs]), as well as the binomial proportion CI. Multivariate logistic regression (using characteristics that had a p value of <0.001 on univariate analysis) and receiver operating characteristic analysis were used to build the best prediction model for dysautonomia diagnosis. All p values were two-sided. Missing data were excluded from the analysis. Data are presented as mean (SD). Data analysis was performed using Stata software version 10 (StataCorp, College Station, TX, USA).
Dysautonomia was diagnosed in 33 out of 249 children (13.3%, 95% CI 9.3–18.0%, see Table I) meeting our inclusion criteria. Children with TBI represented 78.3% of the overall study population and the prevalence of dysautonomia in this subgroup was 9.7% (19/195; Table SI, supporting information online). Children with cardiac arrest (most often resulting from drowning, shock, trauma, or aspiration) represented 10.4% of the study population and the prevalence of dysautonomia in this subgroup was 30.8% (8/26; see Table I). Temporally, dysautonomia was diagnosed before admission to the rehabilitation facility in 21 out of 33 (63.6%) children, during the rehabilitation admission in 9 out of 33 (27.3%), and during an acute care readmission in 3 out of 33 (9.1%) children. Mean time from injury to rehabilitation admission was 2.95 weeks (SD 8.42).
Table I. Demographic characteristics of participants
No dysautonomia (n=216)
aStatistically significant. CNS, central nervous system.
Age, mean (SD)
11y 4mo(6y 1mo)
11y 11mo(5y 6mo)
Race, n (%)
Primary etiology of brain injury, n (%)
Traumatic brain injury
Based on our review of all vital signs obtained during the rehabilitation stay, fever, tachypnea, hypertension, tachycardia, diaphoresis, and dystonia all occurred with increased frequency in children diagnosed with dysautonomia compared with those without dysautonomia (all p<0.001; Table SII, supporting information online). Tachycardia and hypertension were the most common abnormalities observed. Children with dysautonomia had a greater number of combinations of these vital-sign abnormalities on a single day during their rehabilitation course than those without dysautonomia (all p<0.001). On multivariate logistic regression, only hypertension (odds ratio 9.8, 95% CI 3.0–31.9), diaphoresis (27.5, 2.8–264.8), and dystonia (7.9, 2.9–21.6) remained significant (p values 0.006, 0.009, and <0.001 respectively). Upon receiver operating characteristic analysis of this combination to predict the diagnosis of dysautonomia, the area under the curve was 0.92 (Fig. 2). The sensitivity of this combination was 73% and the specificity 93%.
Children with dysautonomia were treated more frequently with benzodiazepines (odds ratio 35.2, 95% CI 8.2–151.5), baclofen (29.0, 11.0–76.3), clonidine (8.5, 3.8–19.0), beta-blockers (18.2, 7.6–43.3), and other antihypertensives (including calcium channel blockers and angiotensin-converting enzyme inhibitors; 5.4, 2.1–13.8) than children without dysautonomia (all p≤0.001; Table SIII, supporting information online). Seizure diagnoses (before rehabilitation admission) and bacterial infections during the rehabilitation stay (primarily respiratory) were more frequent among children with dysautonomia (p=0.040 and p<0.001 respectively). Children with dysautonomia had several factors indicating a greater severity of injury including (1) increased length of stay at the rehabilitation facility (mean [SD] 114d  vs 47d ; p<0.001), (2) more readmissions to acute care hospitals (1.52 readmissions [1.33] vs 0.32 readmissions [0.80]; p<0.001), and (3) a decreased frequency of discharges to home (66.7% [22/33] vs 87.5% [189/216], see Table II). Children with dysautonomia had worse motor and cognitive scores at admission to the rehabilitation facility (mean [SD] motor score 13.5 [1.8] vs 31.0 [18.9]; cognitive scores 6.1 [3.4] vs 16.9 [10.1]; both p<0.001) as well as at discharge (motor scores 30.1 [24.1] vs 60.8 [25.0]; cognitive scores 12.0 [8.6] vs 23.9 [9.6]; both p<0.001; Fig. S1, supporting information online). Finally, children with dysautonomia had less improvement in motor function over the course of rehabilitation than those without dysautonomia (mean motor change 16.5 [23.6] vs 29.6 [20.7]; p=0.001). There was no difference in the degree of cognitive improvement between the groups (p=0.416).
Table II. Outcomes for children with and without dysautonomia
aCorrected for days spent away from The Children’s Institute during acute care readmissions.
Length of stay, d
Length of stay correcteda, d
Number of readmissions to acute care facilities
Disposition, n (%)
Procedure, then home
Alternative rehab facility
Long-term nursing care
In this comprehensive analysis of children who were diagnosed with dysautonomia after acute brain injury, we found that dysautonomia affects a significant proportion of injured children. It occurs after cardiac arrest, TBI, brain tumors, and CNS infections, with the highest incidence after global hypoxia–ischemia from cardiac arrest. The constellation of non-specific vital-sign abnormalities was more prevalent in children diagnosed with dysautonomia than in the population without dysautonomia, and combinations of symptoms were highly predictive in distinguishing the populations. Lastly, we found a negative association between dysautonomia and outcome from the time of rehabilitation admission until discharge. This association indicates that novel rehabilitative strategies targeted specifically to children with dysautonomia after brain injury are required to improve outcomes.
The burden of dysautonomia within the population of children who have sustained ABI is understudied. We observed a prevalence of 13% (95% CI 9.3–18.0), which is in accord with similar studies of adults and children with brain injury.2,5,6,8,11–13 In children, Krach et al.5 performed a comprehensive study similar to ours and demonstrated a prevalence of 14%. In their study, they relied on signs and symptoms from the medical record to diagnose dysautonomia rather than our method of using the treating clinician’s diagnoses from the medical record.
The precise cause of dysautonomia is unknown, although some have suggested hypoxia/ischemia as a contributor to development of this uncontrolled dysfunction of the neurological system.6 Although our study was not designed to test this hypothesis, we did observe that children who experienced global cerebral ischemia after cardiac arrest had a high prevalence of dysautonomia (31%, 8/26). It is likely that the overall increased societal incidence of pediatric TBI accounts for the greater proportion of post-TBI dysautonomia.22 Others have suggested that injury to white matter might be an important determinant in development of dysautonomia, as studies in adults with TBI found an association between the presence of dysautonomia and diffuse axonal injury, pre-admission hypoxia, and evidence from magnetic resonance imaging of injuries to deep-brain structures.4,12,13 A prospective study in children with ABI that includes early magnetic resonance images (and possibly the use of serum neurological markers) and comprehensive long-term assessments of dysautonomia symptoms could effectively test this hypothesis.
Lack of specific symptoms impedes early diagnosis and treatment of dysautonomia.3 For example, patients with ABI are often at increased risk of seizures and withdrawal from narcotics, both often presenting with tachycardia, tachypnea, and diaphoresis. Effective, specific treatment for dysautonomia clearly requires more precise delineations of symptoms. Our study uniquely validated which common symptoms associated with dysautonomia were observed in children during their rehabilitation stay. Many of the signs and symptoms in our children were similar to those observed in adult populations with TBI having dysautonomia. Importantly, the combination of hypertension, diaphoresis, and dystonia predicted a diagnosis of dysautonomia with the greatest accuracy. These clinical findings also complement prevailing pathophysiological theories involving both a ‘release’ of diencephalic autonomic centers, as well as brainstem lesions leading to decerebrate posturing.10 The clinical signs of hypertension, diaphoresis, and dystonia used in combination may potentially provide a screening tool for early diagnosis, treatment, and comparative effectiveness research in dysautonomia.
Dysautonomia has been treated with a wide variety of agents in published reports. These include gabapentin, opioid agonists, GABAergic agonists such as benzodiazepines, and baclofen, sympatholytics such as centrally acting alpha-agonists and beta-blockers, and dopamine-modulating agents such as chlorpromazine and bromocriptine.1,2,4,6,23,24 However, the optimal strategy for treating this life-threatening condition remains elusive. Most clinical protocols for adults with dysautonomia are directed at amelioration of the overactive sympathetic autonomic system and the motor symptoms of dystonia, spasticity, and agitation. Benzodiazepines, baclofen, clonidine, and beta-blockers were the most common drugs used in our study population. In contrast, Krach et al.5 predominantly observed the use of chlorpromazine and bromocriptine, a central dopamine antagonist and agonist respectively. In our population, the clinical protocol included pharmacotherapy as well as a low-stimulation environment to decrease stimulation from noise and light to attenuate patient hyperreactivity to external stimuli.9,25 Future studies, potentially using our symptom complex for accurate diagnosis, will be needed to optimize treatment of dysautonomia.
The relationship between the development of dysautonomia and overall outcomes has been infrequently studied. In one pediatric retrospective study, children with dysautonomia had more severe ABI, more rehabilitation complications, and less favorable outcome.5 In our study, we found a similar trend toward prolonged rehabilitation stay and increased number of readmissions in children with dysautonomia. Most readmissions to acute-care hospitals occurred because of neurological or respiratory causes, or because of unrelenting dysautonomia. Importantly, assessment using the WeeFIM, among the most commonly used neuropsychological tests of rehabilitation recovery, is standard practice at our institution. As is evident in our data, children with dysautonomia faced more severe impairment upon entering rehabilitation. Because of this, and because children with dysautonomia have a blunted response to currently used rehabilitation techniques, optimal neurological recovery in this vulnerable population will probably require novel rehabilitative strategies. It is possible that early advanced neuromonitoring and neuroimaging techniques could lead to a better understanding of the specific brain regions associated with risk of dysautonomia, and thereby allow earlier intervention to enhance brain recovery before children reach the rehabilitation setting.11,13
Our study has several limitations. First, as it is retrospective in nature, the diagnosis of dysautonomia was subject to the clinical judgment of the care team. We chose to define our study population by the clinical diagnosis given by the treating team. This method probably underestimates the true prevalence of dysautonomia in this population, as the recognition and presentation of dysautonomia, and the therapeutic interventions for dysautonomia, are variable. Second, we were unable to link the acute clinical events of the ABI with the rehabilitation outcomes directly because the referring hospital charts were not part of this study. Third, within our vital-sign analysis, we were unable to differentiate between fever and hyperthermia in the comparison groups, again because of our data’s retrospective nature, and this may have limited the use of temperature as a distinguishing factor. Last, the vital-sign abnormalities we observed were simply extracted from the medical records and an exhaustive, prospective collection of data was not possible in this review. The prediction of dysautonomia might have been improved by a prospective assessment of these critical data. Yet our combination of factors still demonstrated high sensitivity and specificity.
In conclusion, dysautonomia occurs with significant frequency after ABI in children. Diagnosis of dysautonomia portends a worse neurological outcome after rehabilitation. Current literature on dysautonomia notably lacks a prospective determination of optimal diagnostic and therapeutic strategies, especially for children. Thus, development of improved tools in prospective studies would significantly advance the field. Moreover, development of novel strategies to maximize neurological outcome in the rehabilitation setting is also needed.
We thank Sharon Dorogy, Pat Wotherspoon, Ron Reeher, and Christopher Joseph from The Children’s Institute for assistance in data collection and medical record access. We acknowledge our funding sources, National Institutes of Health grant 1K23NS065132 and University of Pittsburgh School of Medicine Dean’s Summer Research Program.