729CHAPTER 60 Neurologic Assessment and Monitoring authors selected a target ICP of 20 mm Hg or less and CPP of 60 mm Hg or higher, with PbtO2 of 20 mm Hg or greater Despite achieving these values for[.]
CHAPTER 60 Neurologic Assessment and Monitoring Electroencephalography Monitoring The importance of the EEG as a monitoring tool in the PICU derives both from its role in the detection of electrographic seizures as a noninvasive and continuous monitor for new neurologic injury in the critically ill patient and as an adjunct to provide prognostic information after neurologic injury The standard placement of EEG electrodes uses 20 leads placed over the scalp to record the electrical activity of the surface of the brain (cortex) and can provide continuous tracings (continuous EEG [cEEG]) with concomitant video recording of the patient (video EEG) for prolonged periods of time (hours to days) It can also provide a snapshot of cerebral electrical activity with a technician at bedside to perform various provocative maneuvers and document patient behavior (routine EEG) for approximately hour Until recently, the use of cEEG in pediatric critical care was hampered by lack of availability, delays in timely interpretation of the EEG tracing, and the absence of data to show that treating seizures in some critically ill populations improves outcome There is now greater availability of this technology in many PICUs,79 with recognition of both the incidence of seizures80,81 and data showing the impact that seizures can have on long-term outcome from critical illness in children.82–85 Consensus criteria for the use of EEG in the ICU have been established.86,87 cEEG is indicated for the detection of nonconvulsive seizures (NCSs; e.g., in the patient not awakening after status A B • Fig 60.2 epilepticus, with unexplained encephalopathy, or the paralyzed and sedated patient), as an adjunct for detection of cerebral is chemia in patients at high risk, assessment of level of consciousness in sedated or comatose patients, and for prognostication in patients after severe TBI, cardiac arrest, and subarachnoid hemorrhage.86,87 In patients in which NCSs are suspected, cEEG monitoring for at least 12 hours is required to detect NCSs.88–90 Data in support of these recommendations are robust NCSs are a common phenomenon for patients who are at risk for or have suffered from recent brain injury, who have had a recent clinical seizure, or who are encephalopathic Among critically ill children who undergo cEEG for any reason, NCSs occur in nearly 30%.91 NCSs occur in up to 23% of children supported with extracorporeal membrane oxygenation (ECMO)92; 8% of neonates after cardiac surgery with cardiopulmonary bypass80; and approximately one-third of children with structural brain injuries, prior in-hospital convulsive seizures, or with interictal EEG abnormalities.93 In children with acute encephalopathy, NCSs occurred in 36% to 46%.84,94 In a prospective study of 87 children with mild to severe TBI, 43% had early posttraumatic seizures,95 among which 16% were electrographic only.95 EEG backgrounds have also been associated with outcomes after pediatric cardiac arrest, with severely abnormal background patterns (burst-suppression, attenuation) being associated with death and unfavorable neurologic outcome96 and the presence of normal waveforms such as a normal background or sleep spindles being associated with favorable outcomes.97 Fig 60.2 is an example of cEEG obtained post–cardiac arrest in a 2-year-old boy with neuromuscular weakness The deterioration in the EEG background over 24 hours correlated with his progression to meeting most criteria for brain death The gold standard for seizure detection using EEG is interpretation of the EEG tracing by an electroencephalographer However, emerging data using quantitative EEG (qEEG) has shown potential for seizure detection by non-EEG-expert critical care providers qEEG computes features of the different power band frequencies to display trends in the EEG signal in a numerical format (Fig 60.3) Precedent for use of qEEG—specifically, amplitude integrated EEG (aEEG)—has been described in neonatal ICUs for seizure detection and background assessment in hypoxemic-ischemic encephalopathy from birth asphyxia.98,99 Pilot authors selected a target ICP of 20 mm Hg or less and CPP of 60 mm Hg or higher, with PbtO2 of 20 mm Hg or greater Despite achieving these values for ICP and CPP, there were periods when PbtO2 was below the target value The study provides a precedent for combining ICP-directed therapy linked to preventing subthreshold PbtO2 Whether this approach results in a reduction of neurologic complications of DKA-associated cerebral edema is not known There is potential for the use of ICP monitoring to direct management of other pediatric CNS injuries beyond severe TBI, but there remains insufficient data at present to justify its use as a standard of care Precedent from these and other reports suggest that its use in selected patients may be safe and could provide meaningful data to direct care of the injured brain.77,78 C Longitudinal bipolar continuous electroencephalogram montage of a 2-year-old boy after cardiac arrest Note the deterioration in background over a 24-hour monitoring period (A) First epoch: discontinuity (B) Second epoch: slowing with epileptiform discharges (C) Last epoch: severe attenuation 729 730 S E C T I O N V I Pediatric Critical Care: Neurologic A 600 400 200 normAD F4 200 normAD C3 400 normAD C4 200 normAD P3 400 normAD P4 0 200 400 600 600 200 0 400 600 600 200 0 400 600 –2.0 –1.8 –1.6 –1.4 –1.2 –1.0 –0.8 –0.6 –0.4 –0.2 B 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 Time (days) • Fig 60.3 (A) Representative longitudinal bipolar electroencephalogram (EEG) montage (B) Examples of quantitative EEG recordings of three patients before and after cardiac arrest The EEG tracing has been processed to show changes in the ratio of the a and d power bands (AD ratio) studies in adult and pediatric ICUs using aEEG and density spectral array (DSA) for seizure detection show a reasonable sensitivity and specificity and could be a future direction to enhance seizure detection in settings with limited access to an electroencephalographer.100 A single-center study of 87 children resuscitated after cardiac arrest also identified specific qEEG features (g, d, and b band power) in random forest models that predicted poor neurologic outcome.101 The trends in the use of the qEEG in pediatric critical care are toward automated detection and prediction of seizures, identification of pathologic qEEG features, and incorporation of these data into models that enable earlier and more accurate neuroprognostication CHAPTER 60 Neurologic Assessment and Monitoring A number of studies now indicate that treatment of NCSs improves outcome, including a study of 259 patients with a variety of neurologic insults admitted to the PICU and CICU with cEEG monitoring.84 Seizure burden was defined as the maximum percentage of any given hour occupied by electrographic seizures The maximum total seizure burden was much greater in patients with neurologic decline (15.7%/h) than in patients without such alterations (1.8%) The odds of neurologic decline increased by 1.13 (95% confidence interval [CI], 1.05–1.21; P , 0016) for every 1% increase in maximum hourly seizure burden In a study of 300 children with an acute neurologic disorder (137 of whom were neurodevelopmentally normal) who underwent clinically indicated cEEG in the PICU, electrographic seizures and status epilepticus were associated with worse adaptive behavior after discharge in the previously healthy children.82 Collectively, these studies suggest that seizures in critically ill children are associated with neurologic decline and may be a therapeutic target A Transcranial Doppler Measurement of Cerebral Blood Flow TCD uses ultrasonography to measure CBFV in the anterior and posterior circulation.102 Normal values have been published for healthy neonatal, pediatric, and adult populations.103,104 CBFV measured by TCD is lower in sedated critically ill children compared with age-matched healthy children.105 CBFV is decreased in children supported on ECMO compared with other mechanically ventilated critically ill children.106,107 There has been growing use of TCD in pediatric neurocritical care during the management of TBI, hypoxic-ischemic encephalopathy, stroke, CNS infection, DKA, ECMO, and vasospasm.106–112 The availability of TCD as a noninvasive neuromonitoring modality for serial surveillance for key injury mechanisms— including thrombosis, ICP, and vasospasm—is promising, but there have been no standards defined for its use in pediatric critical care (Fig 60.4) A survey of North American centers identified 27 B PS 159 cm/s, ED 55 cm/s TAMAX 96 cm/s PI 1.08 RI 0.65 D C PS 168 cm/s, ED 94 cm/s TAMAX 133 cm/s PI 0.56 RI 0.44 E PS 127 cm/s, ED 25 cm/s TAMAX 44 cm/s PI 0.47 RI 0.37 731 PS 38 cm/s, ED 24 cm/s TAMAX 31 cm/s PI 0.47 RI 0.37 F PS 56 cm/s, ED 46 cm/s TAMAX 51 cm/s PI 0.22 RI 0.19 PS 46 cm/s, ED 11 cm/s TAMAX cm/s PI 35 RI 1.24 • Fig 60.4 Representative transcranial Doppler (TCD) images of various patterns of cerebral blood flow in the middle cerebral artery (A) Normal TCD pattern from a 17-year-old male (B) Hyperemic TCD pattern from a 2-year-old boy after cardiac arrest Note the elevated peak systolic and end diastolic velocities. (C) Low flow pattern from a 6-week-old male after cardiac arrest Note the decreased peak systolic and end-diastolic velocities (D) Increased intracranial pressure pattern from a 16-year-old female with hepatic encephalopathy Note the elevated pulsatility index (E) Lack of pulsatility with maintained cerebral blood flow from a 13-year-old male with a ventricular assist device (F) Reversal of diastolic flow from an 8-month-old male with severe traumatic brain injury ED, End-diastolic velocity; PI, pulsatility index; PS, peak systolic velocity; RI, resistive index; TAMAX, mean velocity 732 S E C T I O N V I Pediatric Critical Care: Neurologic performing TCDs as part of routine clinical care and research, of which 15 (56%) also had a dedicated pediatric neurocritical care service.113 The most common indications for use were for the management of TBI, stroke, and intracranial hemorrhage However, only 30% of the respondents had a standardized, written protocol for the performance and interpretation of TCD studies An expert panel, using a modified Delphi process, has now created recommendations for the technical performance, interpretation, documentation, and reporting of TCD examinations in critically ill children.114 These are available at www.pncrg.org Near Infrared Spectroscopy Near infrared spectroscopy (NIRS) measures tissue oxygen saturation by determining the difference in intensity between transmitted and received light delivered at specific wavelengths.115 Commercial cerebral NIRS products measure oxygenation using wavelengths of 700 and 1000 nm to measure transparency of tissue to light Since hemoglobin differentially absorbs near infrared light depending on its oxygenation state, changes in hemoglobin oxygenation can be quantified using the modified Lambert-Beer law NIRS may serve as a noninvasive method for detection of cerebral or somatic hypoxia In practice, the reliability of cerebral NIRS has been questioned, including a study showing that in infants up to 190 days old cerebral NIRS values were not reproducible.116 In adult human cadavers, the cerebral tissue oxygenation measured by NIRS in one-third of these subjects was higher than the lowest value of normal controls.117 The adoption of NIRS as a standard of care has been limited by the retrospective design of many of these single-center studies and the lack of evidence to show that targeting specific NIRS values improves organ function or outcome.118 Studies in children with corrected congenital heart disease suggest that NIRS can detect perioperative cerebral hypoxia that is associated with increased risk for compromised neurodevelopmental outcome.119 In a study of postoperative congenital heart disease patients, cerebral and flank regional oxygen saturation was correlated with central venous oxygen saturation in both cyanotic or acyanotic patients and single- or two-ventricle physiology.120 Change in partial pressure of arterial carbon dioxide was associated with change in cerebral, but not flank, regional oxygen saturation, suggesting that cerebral NIRS is detecting changes in cerebral oxygenation A reduction in cerebral oxygen uptake during deep hypothermic cardiac arrest may not be associated with white matter injury, however.121 NIRS may emerge as an effective tool for mitigating perioperative neurologic injury in children with congenital heart disease.122 A case series of four neonates with hypoxic ischemic encephalopathy, seizures, and hemodynamic instability that combined simultaneous aEEG and cerebral NIRS monitoring found abnormal NIRS values preceding aEEG changes.123 The validity of NIRS for the detection and management of neurologic insults in children is also not established A higher regional oxygen saturation may be associated with greater chance of achieving return of spontaneous circulation in adults with inhospital or out-of-hospital cardiac arrest,124 but such data in children are lacking In a single-center observational study of infants with cerebral circulatory arrest, cerebral NIRS was significantly reduced in the patients with arrest compared with those with preserved cerebral perfusion.125 However, the specificity of this difference was low.124 In a pilot study of 28 children, NIRS was used in combination with head CT to detect the presence of intracerebral hemorrhage Using a difference in optical density of 0.2 between hemispheres or scalp locations, NIRS correctly identified all 12 intracerebral hemorrhage cases.126 An exciting development in the use of NIRS involves the autocorrelation of cerebral NIRS and mean arterial pressure (MAP) as a noninvasive measurement of CBF autoregulation.127 A feasibility study of 36 children used autocorrelation of cerebral NIRS and MAP to identify the optimal MAP for preservation of cerebrovascular reactivity after cardiac arrest.128 A similar approach to measure autoregulation was used in a study of 15 children with moyamoya vasculopathy to measure autoregulation during and after surgical revascularization.129 This study suggested that poorer autoregulation during surgery was associated with increased risk for cerebral ischemia Evidence also suggests that there may be a relationship between NIRS and intracranial hypertension, although the directionality of this relationship was dependent on diagnosis.130 Brain Tissue Oxygen Monitoring The Licox system (Integra Neurosciences) uses a Clark-type electrode and measures both PbtO2 and temperature The indications for placing the PbtO2 monitor are the same as those for patients requiring ICP monitoring following TBI, which is its most common use.131,132 Typically, the probe is placed in the white matter of the brain, because it is more metabolically stable than the gray matter If the region of injury is to be monitored, the catheter is placed adjacent to the contusion or in the ischemic penumbra but not directly in the contusion or infarct A retrospective analysis of 150 evaluable patients with severe TBI found that brain hypoxia (PbtO2 ,10 mm Hg) was associated with worse outcome and mortality.133 There is debate as to whether the PbtO2 monitor is estimating oxygen extraction or CBF.134–136 With the greater adoption of this technology, the current guidelines for the management of severe TBI include a pathway incorporating its use and recommend a minimum target level of 10 mm Hg.11 In children, experience with PbtO2 monitoring is increasing, particularly with severe TBI.137–139 In a single-center study of 46 children with severe TBI in which the probe was placed in the uninjured frontal cortex, PbtO2 was preserved in the goal range in some cases but outcome was poor.76 CPP was the only factor that was independently associated with a favorable outcome at months Other pediatric studies have also reported a complex interaction between PbtO2 and other clinical parameters.132,140 In each of these studies, the analyses were based on hourly values for PbtO2 and daily averages for PbtO2 as opposed to episodes of PbtO2 below predefined thresholds,138 making comparisons between these studies difficult Small series at single centers have suggested that PbtO2 monitoring may augment the detection of CNS metabolic crisis in children with stroke141 and DKA.76 Differences in the site for probe placement and the thresholds selected for treatment limit meaningful comparisons between these studies Further, PbtO2 values may reflect both the presence of ischemia and other factors, such as partial pressure of arterial oxygen Further, PbtO2 measures local tissue conditions and would fail to detect hypoxic-ischemic insults in remote brain regions The promise of this technology for pediatric neuromonitoring is considerable but a number of caveats remain For example, the threshold (10–15 mm Hg) for poor outcome is based on mostly adult data, and the dose dependence of an exposure to cerebral hypoxia and the disease specificity for insults other than TBI remain to be determined CHAPTER 60 Neurologic Assessment and Monitoring Optic Nerve Sheath Diameter Measurement The optic nerve is enveloped by a dural sheath that is continuous with the intracranial dura mater Increased ICP can displace CSF from the intracranial space and expand the subarachnoid space surrounding the optic nerve This perineural space can be visualized and measured using transorbital ultrasound and is called the optic nerve sheath diameter (ONSD) ONSD measurement is proposed as a noninvasive means to detect increased ICP.142 A study of 103 healthy volunteers, 13% of whom were younger than 16 years of age, found a median ONSD of 4.41 (range, 4.24–4.83) mm.143 Results of studies using ONSD measurement to detect increased ICP in children are equivocal, with data suggesting that it may be more sensitive and specific in certain intracranial pathologies versus others.144 ONSD measurement had good sensitivity and specificity for detecting elevated ICP due to idiopathic intracranial hypertension145 but performed less well for the detection of ventriculoperitoneal shunt malfunction.146 The temporal relationship between ONSD fluctuations and ICP spikes is also unclear, with data showing both a rapid change in ONSD correlating with invasively measured ICP changes147 and delayed reversal of ONSD distension after normalization of ICP.148 Further investigation correlating invasively measured ICP with ONSD measurements needs to be performed to find the clinical conditions under which it is a reliable predictor of increased ICP Tympanometry Tympanometry has been proposed as a noninvasive measure of ICP but is subject to critiques of lack of accuracy and lack of availability for continuous monitoring.149 The largest studies of its clinical application in critically ill children have been carried out in Africa in children with acute, nontraumatic coma and excluded children with epilepsy, neurodevelopmental delay, and sickle cell disease.150–152 Children were more likely to die if tympanometry was abnormal (adjusted odds ratio, 17.0; 95% CI, 1.9–152.4; P 01) The majority of the cases (53%) were caused by cerebral malaria and the remainder caused by either bacterial meningitis, sepsis, or no identifiable cause (30%) This technology needs further evaluation in studies that incorporate physiologic monitoring and brain imaging to validate this measurement Nevertheless, the ease of use and feasibility in resource-poor countries justify further evaluation Cerebral Microdialysis Cerebral microdialysis uses the capillary technique to measure the concentration of key metabolites (glucose, pyruvate, lactate), excitotoxic amino acids (glutamate and aspartate) and membrane breakdown products (glycerol), in the brain parenchyma in order to detect evolution of primary and development of secondary injury.153 This technique is currently employed at approximately 75 centers worldwide, although the number of pediatric centers at which it is employed is not known.154 The microdialysis catheter is a fine tube, placed via a burr hole, within a semipermeable dialysis membrane that permits diffusion of molecules from the extracellular space along the catheter and into a vial This vial is then placed in the microdialysis analyzer In general, the catheter is placed via the same burr hole as the PbtO2 and ICP monitors into the uninjured brain or penumbra of the injured brain The most common system is the CMA600 microdialysis analyzer As 733 with PbtO2 monitoring, results vary depending on the placement of the probe in healthy tissue, tissue adjacent to the primary site of injury (penumbra), or in the injured tissue.155 A number of adult studies have shown an association between metabolic stress (lactate/pyruvate ratio 40) and poor outcome in TBI and SAH cases.156 In a retrospective analysis of 20 adult patients with severe brain injury, tight systemic glucose control was associated with reduced cerebral glucose and with corresponding evidence (increased lactate/pyruvate ratio) of compromised brain metabolism.157 Data from adult studies have supported the use of cerebral microdialysis primarily for the management of TBI and SAH and have converged on specific values for each component of brain chemistry.158 In adult studies, data linking these metabolic derangements to long-term functional outcome are lacking, with the strongest evidence showing an association between lactate/ pyruvate ratio and frontal lobe atrophy at months’ recovery.159 The 2004 consensus statement155 on its use in neurocritical care was updated in 2015.160 This expert opinion contains no recommendations about use in children, and there are no recent published data to assess its risks and utility in the practice of pediatric neurocritical care Integrating Neurologic Monitoring Data Neuromonitoring begins with the neurologic examination and incorporates the patient’s medical history together with data from the medical devices (including all organ systems in addition to brain monitoring) and laboratory results These data are analyzed to identify a mechanism of brain injury, predict the risk for further injury, and select the optimal treatment of this specific patient at that stage of the brain injury to achieve the best long-term outcome This is standard medical practice—the clinician incorporates these data into the standards of care recommended for the brain injury in question, such as TBI, stroke, cardiac arrest, or status epilepticus.9,10,161 For brain-directed critical care in children, the growing volume of available neuromonitoring data, gaps in data supporting current treatment guidelines, and risk for longterm neurologic morbidities for children and their families after critical illness162 highlight the need for more effective ways to use these data and to link their use to long-term outcomes The scale and complexity of these data mean that these goals of meaningful use are contingent on solving the technical barriers to their integration This is a significant hurdle that first requires extraction and validation of data from the electronic health record (EHR).163,164 Using data largely available from the EHR, a realtime tool for predicting a decline in neurologic function at the time of discharge from the PICU has been reported.165 These types of monitor data also may appear appropriate for the application of machine-learning algorithms to detect hidden patterns While the greater volume of data lends itself to machine learning, predicting an event such as stroke, status epilepticus, or cardiac arrest in the ICU is not synonymous with explaining why that event occurred.166 Machine learning may help with prediction but the adoption and clinical use of this type of analysis requires a biologically plausible and transparent mathematical model without which they will not be adopted.167,168 Modeling brain and other organ function at a level of complexity that achieves the goals of personalized care and meaningful prediction of deterioration and response to individualized therapy cannot be achieved using low-frequency EHR data.169,170 A trajectory analysis of ICP changes over days following TBI identified ... sepsis, or no identifiable cause (30%) This technology needs further evaluation in studies that incorporate physiologic monitoring and brain imaging to validate this measurement Nevertheless, the... placed via a burr hole, within a semipermeable dialysis membrane that permits diffusion of molecules from the extracellular space along the catheter and into a vial This vial is then placed in... further injury, and select the optimal treatment of this specific patient at that stage of the brain injury to achieve the best long-term outcome This is standard medical practice—the clinician incorporates