EEG Errores Pitfalls Mala Interpretacion Misdiagnosis Mal Diagnostico Neurology 2013

Pitfalls in ictal EEG interpretation Critical care and intracranial recordings

Nicolas Gaspard, MD, PhD Lawrence Lawre nce J. Hirsc Hirsch, h, MD

Correspondence to Dr. Hirsch: [email protected]

ABSTRACT

EEG is the cornerstone examination examination for seizure diagnosis, diagnosis, especially nonconvulsive nonconvulsive seizures in the critically ill, but is still subjec subjectt to many errors that can lead to t o a wrong diagnosis diagnosis and unnecessary or inadequate treatment. Many of these pitfalls to EEG interpretation are avoidable. This article reviews vie ws com commo mon n err errors ors in EE EEG G int inter erpre pretat tation ion,, foc focusi using ng on ic ictal tal or pot poten entia tially lly ict ictal al rec record ording ings s ob obtai taine ned d in critically ill patients. Issues discussed include artifacts, nonepileptic events, equivocal EEG patterns seen in comatose patients, and quantitative EEG artifacts. This review also covers some difficulties encountered with intracranial EEG recordings in patients undergoing epilepsy surgery, including issues related to display resolution. Neurology  2013;80 (Suppl 1):S26–S42 GLOSSARY AED

5 antiepileptic drug; AEEG 5 amplitude-integrated EEG; CEEG 5 continuous EEG; GPD 5 generalized periodic discharge; ICE 5 intracortical EEG; ICU 5 intensive care unit; NCSE 5 nonconvulsive status epilepticus; NSE 5 neuronspecific enolase; QEEG 5 quantitative EEG; SE 5 status epilepticus.

The diagnosis of seizures and epilepsy often depends on the correct interpretation of EEG studies. Diagnosis Diagno sis almost compl completely etely relies on EEG for nonconvu nonconvulsive lsive seizures in the critically ill. Overinterpretation of an EEG is frequent and can lead to serious adverse consequences. 1,2 This is particularly true for continuous EEG (CEEG) monitoring in the intensive care unit (ICU), where artifacts are more mo re ab abun unda dant nt an and d di dive vers rsee an and d ca can n at ti time mess be ve very ry mi misl slea eadi ding ng.. Th Thee EE EEG G ba back ckgr grou ound nd in cr crit itic ical ally ly il illl and comatose patients differs greatly from the background in alert individuals, and many patterns frequently encountered in these patients are difficult to classify into ictal and nonictal categories. Technological advances, such as improved quantitative EEG (QEEG) techniques, networking, and invasive intracortical EEG (ICE) monitoring have improved the performance and feasibility of CEEG but they are not by any means immune to artifacts and misinterpretation. Herein, Here in, we addr address ess some some of the most comm common on pitfall pitfallss that should be avoide avoided d while read reading ing ICU ICU EEGs and CEEGs, in order to avoid over- and underinterpretation and inappropriate treatment. ARTIFACTS The ICU can be considered a hostile environment environment for EEG recording. Many sources of extracerebral

Supplemental data at www.neurology.org

signals can interfere with the cerebral activity, and obtaining a study not contaminated by artifact is a challenging and often impossible task. Some artifacts are common to all EEG recordings (EKG, eye movements, muscle activity, sweating, electrode instability, etc.) (figures 1 and 2 and table 1) but prolonged recordings are more prone to technical issues than shorter ones. The ICU environment also significantly differs from the EEG lab or the epilepsy monitoring unit because of the presence of numerous electrical signal generators that can produce peculiar artifacts that require some experience to recognize. Examples include mechanical ventilation, ventricular assist devices, oscillating beds, and dialysis and patient care, especially chest percussion by respiratory therapists, a notorious seizure-mimicker (figures 3 –5 and table 1).3,4 Some So me EEG wa wave vefo form rm fe featu atures res,, wh when en pre prese sent, nt, sh shou ould ld rai raise se sus suspi pici cion on of the art artif ifact actua uall na natur turee of a pat patter tern n (ta (tabl blee 2), altho al thoug ugh h they are no nott ab abso solu lute te and can als alsoo be see seen n wi with th cer cerebr ebral al act activ ivit ity. y. Si Simu mult ltane aneou ouss vi video deo rec recor ordin dingg and no notes tes of the technologists, nurses, or others may be of great help in case of doubt. We strongly encourage frequent entry of comments into the EEG record at the bedside by any caregiver because this aids communication greatly. When dealing dealing with artifacts, artifacts, it is tempting to make excess use of filters, especially especially the high high-freq -frequenc uencyy (a.k (a.k.a., .a., low-pass) filter to reduce muscle activity and the notch filter to hide 60-Hz electrical noise. However, setting From Yale University, School of Medicine, Neurology Department and Comprehensive Epilepsy Center, New Haven, CT. Go to Neurology.org for full disclosures. Funding information and disclosures deemed relevant by the authors, if any, are provided at the end of the article.

S26

© 2012 American Academy of Neurology

Figure 1

Facial-twitching artifact mimicking periodic lateralized epileptiform discharges (PLEDs)

(A) The EEG in this 39-year-old woman shows periodic spike-wave-like or polyspike-wave-like potentials over the right hemisphere (boxes). Lower voltage periodic slow waves (blunt PLEDs) are present on the left (underlined). (B) After the administration of vecuronium, the right-sided “spikes” are no longer present. They were attributable to muscle artifact associated with twitching movements on the right side of the face. The movements were associated with the low-voltage PLEDs present over the left hemisphere (now in boxes), maximal in the parasagittal region. Thus, the left PLEDs were real (and ictal in this case), but the right “PLEDs” were artifact. (Reproduced from Brenner and Hirsch, 20 with permission.)

Neurology 80 (Suppl 1) January 1, 2013

S27

Figure 2

Chewing artifact mimicking seizures

TheEEG shows a bilateral sharply contouredrhythmicdelta activity more prominent anteriorly with some degreeof evolution in frequency, morphology, and distribution, thus qualifying for a seizure. The chewing movements of this awake patient while eating caused this activity.

the high-frequency (low-pass) filter at a frequency at #15 Hz may affect the morphology of artifacts to the point of disguising them in waves that appear like abnormal cerebral activity, including epileptiform discharges and seizures (see figure 6).

S28

aberrant morphology or polarity, or with very restricted fields over a skull defect. The reader must be aware of this situation, either by reviewing the patient history or by recognizing other EEG features, to make a proper interpretation. Technologists should also record skull defects carefully.

THE OPERATED AND INJURED BRAIN AND SKULL A skull defect, such as a bur hole or craniot-

NONEPILEPTIC MOTOR MANIFESTATIONS CEEG

omy, results in an increase in the voltage and the sharpness of cerebral activity and an accentuation of faster frequencies (referred to as the “breach rhythm” or “breach effect”). A small defect, such as after the insertion of an intraventricular catheter, can cause very focal distortion, located over one electrode only (figure 7). Care must be taken not to overinterpret sharply contoured waveforms within this breach rhythm as epileptiform discharges or a sign of dysfunction on the opposite hemisphere. However, the alteration of cortical anatomy after brain injury or surgery affects the spatial distribution of electric dipoles. Spikes and sharp waves may present with

studies are often requested because of transient spontaneous motor spells that are ascribed to seizures. In fact, there are many movements in critically ill patients that are not epileptic. Up to 10% of presumed motor seizures in the ICU for which CEEG is requested are not seizures.5 These movements include myoclonus, asterixis, tremor, shivering, semipurposeful movements, posturing due to pain or herniation, and deep tendon reflex clonus (which can mimic stimulus-induced seizures).6 During these nonepileptic spells, the absence of ictal activity supports the diagnosis. Sometimes, however, movement artifacts may obscure the EEG. In this case, the diagnosis has to be made solely on clinical


Table 1

Potentialsourcesof artifactwhen recordingEEG in theintensive care unit

Patient Eyes and eyelids (eye movement, eyelid flutter, blinking, nystagmus, bobbing, etc.) Orolingual movements (glossokinetic potential, chewing, etc.) Muscle activity (myoclonus, micro-shivering, jaw clenching, tremor, etc.) Cardiovascular activity (EKG, pulse artifact, etc.) Respiration Sweat Patting/rocking (especially in infants) Continuous EEG setup Electrodes (instability, electrode pop, unequal impedances) Wires Jacks/jackboxes Monitoring and life-support devices 60-Hz noise (or 50-Hz in some countries) Mechanical ventilation (including water condensation in the ventilation tubing, extracorporeal membrane oxygenation, rapid oscillation ventilators IV drip Hemofiltration, hemodialysis Pacemaker Implanted ventricular assist device Oscillating bed Staff Chest percussion (for pulmonary care): most common mimic of seizures Suctioning

interpretation, including video review. It should also be noted that the absence of ictal activity on scalp EEG does not rule out seizures because many focal seizures, including the majority of simple partial seizures in patients with epilepsy, do not have a clear scalp EEG correlate; this may occur more often in the critically ill (see below). Video recording is very helpful in providing additional information about the semiology of the spell. Bedside examination can also help at times. If the motor activity reliably ceases after repositioning the involved limb, it is most likely not a seizure. However, if the activity is induced by stimulation, including repositioning the patient, it could still be a seizure. The type of stimulation may sometimes point to the nature of the activity. Reflex movements provoked only by a specific maneuver (deep tendon percussion, passive extension of a limb) rather than by a broad array of stimuli are most likely not seizures, although exceptions occur, such as parietal lobe reflex seizures. THE (MIS)DIAGNOSIS OF NONCONVULSIVE STATUS EPILEPTICUS IN COMATOSE PATIENTS The EEG

background in comatose and critically ill patients differs widely from common EEG backgrounds seen in alert individuals. With the increasing use of CEEG, it has

become clear that it is often difficult, and occasionally impossible, to distinguish ictal, interictal, and nonictal patterns in encephalopathic patients. The interpretation of these periodic and rhythmic patterns is still a subject of controversy and different viewpoints exist. More clinical and animal studies are required to clarify their nature. Generalized periodic discharges (GPDs) at 1 to 2 Hz can be seen in metabolic encephalopathy and postanoxic coma, as well as during or after the course of nonconvulsive seizures and nonconvulsive status epilepticus (NCSE), even if they do not appear “epileptiform.” It is virtually impossible to reliably discriminate between encephalopathy and status epilepticus (SE)-associated GPDs in a given individual although some group differences exist: GPDs associated with seizures and SE tend to be sharper (higher amplitude and shorter duration) and appear on an interdischarge background of lower amplitude than GPDs associated with encephalopathy.7 However, there is too much overlap for this to be relied on for a given individual. Terms such as “triphasic waves” or the presence of an “anterior-posterior lag ” carry an etiologic connotation (of toxic or metabolic encephalopathy) and are often thought to be specific; they are not specific and can be seen during or after seizure and SE. To add to the confusion, the morphology and frequency of periodic discharges usually vary in the same patient, appearing epileptiform at one time and not at other times. Whether periodic lateralized epileptiform discharges represent an ictal or interictal phenomenon is probably variable. Rarely, they are clearly ictal and associated, for instance, with contralateral synchronous periodic focal motor activity. In most cases, however, they are devoid of any clinical manifestation and assumed to be “ nonictal”—either interictal, or on an interictal-ictal continuum.8 Regardless, it should be remembered that up to 80% of patients with periodic lateralized epileptiform discharges have seizures during the acute course of their illness8,9; thus, we believe all of these patients should be receiving antiepileptic medication, especially if CEEG is not being performed and closely monitored. Another frequent misconception is that if an EEG pattern is induced or accentuated by stimulation it is not ictal. It is now well recognized that alerting stimuli in comatose patients can repeatedly elicit periodic, rhythmic, or ictal discharges (globally referred to under the acronym SIRPIDs: Stimulus-Induced Rhythmic, Periodic, or Ictal Discharges; see figure 8), 10 typically with no clinical correlate, but sometimes with focal motor seizures (see figure 9 and video). 11 Overall, it is crucial to recognize that such patterns belong to the same continuum of activities that may be ictal at times and nonictal at others, including in the same patient, fluctuating between the 2 or remaining Neurology 80 (Suppl 1) January 1, 2013

S29

Figure 3

Mechanical ventilation artifact mimicking generalized periodic epileptiform discharges

This EEG shows generalized periodic polyspike-wavedischarges (box). These discharges were synchronous to mechanical ventilation and were not cerebral; they resolved when fluid was removed from the ventilator tubing.

Figure 4

Dialysis artifact

The EEGin this 92-year-old manwithmental status changes and renal failure showsrhythmicartifact(boxes),predominantlyinvolving the anteriorheadregions (electrodes Fp1 and Fp2), more marked on the right. The discharges are also present in the T4-T6 derivation, which provides evidence that this could not represent eyemovement artifact. The patient was being dialyzed utilizing slow continuous ultrafiltration that resulted in this artifact. (Reproduced from Brenner and Hirsch, 20 with permission.) S30


Figure 5

Chest percussion artifact mimicking a seizure

These 2 contiguous EEG pages show a rhythmic sharply contoured delta activity in the left temporoparietal region (box). There is evolution in amplitude, morphology, andlocation. A physical therapist was performing chest percussion with thepatient on their left side, explaining thepotentiallyphysiologic field. Use of video allows rapid detection of this pattern, which could be misinterpreted as seizure otherwise.

equivocal, lying on what has been coined the ictal- when modification in the EEG is accompanied by clininterictal continuum (figures 10 and 11). ical improvement. This improvement is often not It is thus important to recognize this lack of certainty concomitant to the EEG changes but when it occurs, it and to avoid dogmatic EEG interpretations that falsely is usually within 24 hours after the trial. 15 It is important suggest more EEG specificity than exists. EEG reports to note that the absence of clinical improvement does in the critically ill often need to stress this uncertainty not rule out NCSE; unfortunately, most of these trials and lack of specificity. are equivocal in the end. Trying nonsedating IV AEDs EEG criteria for the diagnosis of NCSE have been (valproate, fosphenytoin, levetiracetam, or lacosamide) proposed (table 3),12,13 although their validity has never may give the best chance of successfully terminating a been prospectively investigated. When confronted with seizure and showing clinical improvement. a pattern belonging to “the ictal-interictal continuum,” Another possibility when confronted with equivocal there are several pragmatic approaches. A common prac- EEG patterns is to investigate the metabolic/physiologic tice used to distinguish ictal from nonictal EEG patterns impact of these discharges. Perfusion imaging with is to determine whether they can be abolished by a trial SPECT, CT, or MRI and functional imaging with of short-acting antiepileptic drug (AED), usually benzo- FDG-PET, MR spectroscopy, or BOLD fMRI can diazepines (table 4 and figure 12). However, most peri- reveal areas of hyperperfusion, hypermetabolism, lactate odic discharges, including triphasic waves in metabolic production, glutamate increase, etc., that would suggest encephalopathy, can attenuate or disappear after ben- that the pattern is more likely to represent ictal activity, zodiazepine injection. 14 The trial is thus helpful only or, more importantly, that it may be causing metabolic stress and possibly secondary neuronal damage. 16 More invasive monitoring with intracerebral microdialysis can provide additional evidence regarding Table 2 Features that may suggest artifacts rather than cerebral activity whether or not an EEG pattern is associated with neuronal stress/injury: increased lactate/pyruvate Distribution of the activity over multiple electrodes without a physiologic electrical field ratio, glutamate, and glycerol are all suggestive of seizureAtypical multiple phase reversals related neuronal injury. Neuron-specific enolase (NSE) Activity localized to a single electrode levels in blood and CSF also reflect the extent of neuroHighly stereotyped or very monomorphic pattern nal injury, for instance after traumatic brain injury, 17 but Periodic pattern with perfect regularity also after seizures and SE.18,19 We sometimes use serial Evidence from the video recording pointing at the source of the artifact serum NSE to determine the potential harm caused by a (chewing, toothbrushing, patting, chest percussion, etc.) prolonged but equivocal pattern; a transient increase in Neurology 80 (Suppl 1) January 1, 2013

S31

Figure 6

Filtered muscle mimicking brain activity

(A) Fasterfrequency activityis present on the left (boxes) in this 79-year-old man. Thehigh-frequencyfilter (HFF; a.k.a., low-pass filter) is setat a lowsetting of 15 Hz.(B) TheHFF is nowset at a more standard 70 Hz.The fast activity on theleft is attributableto unilateral muscleartifact. The15-Hzfilter decreases muscle artifact, which is in the faster frequency range. With the 15-Hz filter, muscle artifact can be mistaken for cerebral beta activity or even epileptiform discharges. Filters do not distinguish between artifact or cerebral activity, and inappropriate use of filters can often lead to misinterpretation. (Reproduced from Brenner and Hirsch, 20 with permission.)

S32


Figure 7

Breach rhythm

The EEG shows high-voltage beta activity, particularly in the right central region (long box). Activity is also of higher voltage and slower over the right side, particularly in the frontal temporal area. The patient had a right-sided craniotomy. This is a breach rhythm (enhanced fast activity because of a skull defect, most marked at C4) as well as underlying dysfunction as manifest by the focal slowing (2 smaller boxes). (Reproduced from Brenner and Hirsch, 20 with permission.)

Figure 8

SIRPIDs, ictal-appearing without clinical correlate

Three consecutive EEGpages(20 seconds perpage)displaying a focal ictal-appearing dischargein theleft hemisphere that was consistently elicited by stimulation. (A) The EEG initially shows diffuse background slowing, most prominent in the left hemisphere; someone approaches the bedside at second 12 (arrow); this is followed by the onset of sharply contoured rhythmic delta activity mixed with fasterfrequencies in theleft hemisphere, already visible in thelast 3 seconds of thepage (box).(B) and(C) There is evolution of thedischarge over thenext 30 seconds, with changein amplitude, frequency, andmorphology (presence of intermixed spikes and faster frequencies). This pattern thus qualifies for a stimulus-induced ictal-appearing discharge. There was no clinical correlate. Neurology 80 (Suppl 1) January 1, 2013

S33

Figure 9

SIRPIDs, with clinical correlate: Stimulus-induced focal motor seizure

(A)The patient was stimulatedwith nostril tickle(arrow).This elicited theonset of bilateralalphaand beta activity,whichthen evolved in amplitude, frequency, and morphology into unequivocal electrographic seizure (B–D) Clinically, there were clonic movements of the left fingers (first arrow in C) and the patient ’ s eyes opened wide and deviated upward (second arrow in C) (see video on the Neurology® Web site at www.neurology.org). (Reproduced from Hirsch, 11 with permission from John Wiley & Sons.)

Figure 10

Gradual resolution of nonconvulsive status epilepticus (NCSE): The ictal-interictal continuum

(A) The EEG shows posterior-predominant, approximately 1.5-Hz periodic epileptiform discharges, mostly but not always bisynchronous, often polyspikes, superimposed on a background of rhythmic delta. This was interpreted as ictal at this point. (B) The EEG shows a similar pattern, but a bit slower, with brief breaks in the rhythmicity for half a second or so, and with more restricted field and more evidence of a bilateral independent pattern. This is on the ictal-interictal continuum and was interpreted as bilateral independent posterior-predominant periodic lateralized epileptiform discharges (BIPLEDs)plus, more prominent on the right. (C) BIPLEDs, slower than 1 Hz and probably not ictal at this point. (D) Twelve-hour spectrogram showing the gradual resolution of NCSE. This example also supports the concept of an ictal-interictal continuum because this patient has gradual transition for ictal to interictal, with a necessarily arbitrary cutoff point if trying to dichotomize. (Reproduced from Brenner and Hirsch, 20 with permission.) S34


Figure 11

Fluctuations on the ictal-interictal continuum

Six EEG pages of the same patient over 2 consecutive days showing a fluctuation of EEG patterns between ictal (D –F; probably A, and possibly C) and nonictal-appearing (B; possibly C) patterns within an 18-hour period. There was no clinical correlate.

Table 3

Criteria for the diagnosis of nonconvulsive seizures and nonconvulsive status epilepticus a,b

Any pattern satisfying any of the primary criteria and lasting ‡ 10 s (for nonconvulsive seizures) or ‡ 30 min (for nonconvulsive status epilepticus) Primary criteria 1. Repetitive generalized or focal spikes, sharp waves, spike-and-wave complexes at $3/s 2. Repetitive generalized or focal spikes, sharp waves, spike-and-wave or sharp-and-slow wave complexes at ,3/s and the secondary criterion 3. Sequential rhythmic, periodic, or quasi-periodic waves at $1/s and unequivocal evolution in frequency (gradually increasing or decreasing by at least 1/s, e.g., 2 to 3/s), morphology, or location (gradual spread into or out of a region involving at least 2 electrodes). Evolution in amplitude alone is not sufficient. Secondary criterion 1. Significant improvement in clinical stateor appearance of previously absent normal EEG patterns (suchas posterior-dominant “alpha” rhythm) temporally coupled to acute administration of a rapidly acting antiepileptic drug. Resolution of the “ epileptiform” discharges leaving diffuse slowing without clinical improvement and without appearance of previously absent normal EEG patterns would not satisfy the secondary criterion. a

It is important to note that when these criteria are not fulfilled, nonconvulsive status epilepticus has not been excluded; it simply cannot be ruled in definitively. b Adapted from Young et al. 12 and Chong and Hirsch. 13 Neurology 80 (Suppl 1) January 1, 2013

S35

Table 4

Antiepileptic drug trial for the diagnosis of nonconvulsive status epilepticusa,b

Indication Rhythmic or periodic focal or generalized epileptiform discharges on EEG with neurologic impairment Contraindication Patients who are heavily sedated/paralyzed Monitoring EEG, pulse oximetry, blood pressure, electrocardiography, respiratory rate with dedicated nurse Antiepileptic drug trial Sequential small doses of rapidly acting, short-duration benzodiazepine such as midazolam at 1 mg or nonsedating IV antiepileptic drug such as levetiracetam, valproate, fosphenytoin, or lacosamide Between doses, repeated clinical and EEG assessment Trial is stopped after any of the following: Persistent resolution of the EEG pattern (and examination repeated) Definite clinical improvement Respiratory depression, hypotension, or other adverse effect A maximum dose is reached (such as 0.2 mg/kg midazolam, although higher may be needed if taking chronic benzodiazepines) Test is considered positive if there is resolution of thepotentially ictal EEGpatternand eitheran improvement in theclinicalstate or theappearance of previously absent normal EEG patterns (e.g., posterior-dominant ‘‘ alpha’’ rhythm). If EEG improves but patient does not, the result is equivocal. a b

A negative or equivocal result does not rule out NCSE. Adapted from Foreman and Hirsch, 26 with permission from Elsevier.

NSE after the occurrence of the pattern without an alternative explanation suggests secondary damage and may warrant more aggressive treatment. However, this needs to be investigated in controlled trials. When confronted with equivocal EEG patterns, it is probably reasonable to start treatment with an AED, but it is best to avoid prolonged anesthetic doses of sedative medications. In these instances, IV fosphenytoin, valproate, levetiracetam, or lacosamide are good options. In addition, it is also probably useful to optimize patient condition such as fever, and avoid proseizure drugs and metabolic imbalances, including alkalosis; withdrawal from ethanol, barbiturates, or benzodiazepines needs to be avoided as well. If all of this fails and there is some confidence that the EEG pattern is contributing to the patient’s altered mental status or is causing neuronal injury, a 24-hour trial of suppression with midazolam or propofol is reasonable. However, prolonged aggressive treatment should probably be avoided with equivocal EEG patterns, because the definite risks of prolonged intubation and sedation will often outweigh the possible benefit of seizure cessation; obviously, this needs to be assessed on a case-by-case basis, and there is plenty of room for clinical judgment given the lack of definitive evidence. QUANTITATIVE EEG QEEG is increasingly used to

monitor and trend CEEG data. QEEG analysis has proven to be useful for detection of nonconvulsive seizures and delayed cerebral ischemia. It can also detect other acute brain events, including raised intracranial pressure, rebleeding, hypoxemia, etc. 20 S36


Algorithms that transform and compress the raw EEG signal in time-amplitude graphs (amplitude-integrated EEG or AEEG) or time-frequency spectra (fast-Fourier transformation) allow the graphic display of long periods of recordings (from several hours to days) on a single computer screen, for faster reviewing and appreciation of long-term trends. QEEG can measure asymmetries, amplitudes, rhythmicity, power at specific frequencies, and can be run on individual channels or many channels combined. Although this has immense potential, artifacts captured during EEG recording are incorporated in the analysis and can generate graphic patterns that mimic seizures or ischemia (figure 13A). These QEEG displays should never be interpreted without review of the underlying raw EEG tracing, preferably by a boardcertified electroencephalographer. In particular, we have seen repeated examples both clinically and in the literature of AEEGoverinterpretation;it is virtually impossible to tell increased amplitude due to artifact from a similar increase in amplitude due to seizure without review of the raw EEG (figure 13, B and C). Furthermore, it can be almost impossible to distinguish seizure from artifact even with review of the raw EEG when there are only a couple channels of raw EEG recorded, as is standard with these bedside devices. Thus, although AEEG can be very useful for assessment of background EEG and for screening for possible seizures, it has only a moderate sensitivity and specificity for seizures.21,22 Traditional complete EEG should be obtained whenever abnormalities are suggested on the AEEG.

Figure 12

Benzodiazepine trial

(A) EEG from a 20-year-old man who was thought to be in possible nonconvulsive status epilepticus (NCSE) associated with continual, widespread epileptiform activity (boxes). The patient was able to answer many questions correctly, although he was frequently slow in his responses. (B) His clinical state and EEG improved after the administration of lorazepam confirming the diagnosis of NCSE. (Reproduced from Brenner and Hirsch, 20 with permission.)


S37

Figure 13

QEEG: Multiple seizures and identical-appearing false positives on amplitude-integrated EEG (AEEG)

(A) Three to four hours of quantitative EEG (QEEG) from a man in his 60s with a left-hemisphere brain tumor, presenting with worsening memory and language. Multiplenonconvulsiveseizures were recorded (labeled),maximal on the left as evident on theAEEG (higheramplitudes on left) andthe relative asymmetryindex, going sharplydownward (more power on left) witheachseizure. Thestandard spectrogram andthe asymmetryspectrogram bothdemonstrate involvement of all frequencies, and the rhythmic run detector shows a burst of rhythmicity with most of them. Note the 2 episodes labeled “not seizure” (and with dashed lines) in which theAEEG tracingjumps up in a manner almost identical to theprior and subsequent seizures. However, these are dueto muscle artifact. Note that the 2 asymmetry panels do not showthe typical seizure pattern with these artifactual increases in amplitude. This example shows the benefit of using multiple QEEG measures simultaneously, and again stresses the importance of not relying on 1 measure alone without reviewing the raw EEG. (B) EEG at “B” blinking, movement, and muscle artifact only. No seizure. (C) EEG at “ C” , left-sided seizure. (Reproduced from Brenner and Hirsch, 20 with permission.)

INTRACORTICAL EEG A negative EEG never rules out

In addition to recording unrecognized seizure seizure, including during CEEG in the ICU. The use of activity, ICE is less prone to electrode artifacts and ICE in severe acute brain injury, obtained via bedside offers a higher signal:noise ratio than scalp EEG. placement of a mini-depth electrode through a bur This is useful for computerized detection of ischehole,23 has demonstrated the existence of small-scale mia or other secondary events, including with alarms intracortical seizures with no or poor correlation at with rare false positives. 23 However, the extracranial the scalp (figure 14). This is likely attributable to mul- part of the recording setup (wires, connections, amtifocal, asynchronous, mini-seizures that are not ade- plifiers, etc.) is still susceptible to interference with quately synchronized to be seen on scalp EEG. artifact-generating sources. This applies to intracra Whether or not these contribute to deeper coma or nial recordings in patients with epilepsy as well secondary neuronal injury remains unclear. (figure 15). S38


Figure 14

Seizures detected by intracortical EEG (ICE) without correlate on scalp EEG

A 74-year-old woman with subarachnoid hemorrhage grade III and receiving multimodality monitoring, including ICE with mini-depth electrode located in the right frontal cortex. The bottom 6 channels are from the mini-depth (ICE), and the remainder are from standard scalp EEG. ICE shows rhythmic 3-Hz spike-and-wave complexes maximal at D3-D4 with decrease in frequencyand evolutionin amplitudeand morphology. This is theoffset of oneof hertypical seizures.Therewas no correlate on the scalp EEG despite a high-quality recording. (Reproduced from Brenner and Hirsch, 20 with permission.)

Figure 15

Toothbrushing artifact during intracranial EEG recording mimicking seizure

This EEG shows a nonevolving, rhythmic, 5-Hz activity. This was induced by the patient brushing his teeth, causing movement of jackbox. (Reproduced from Goodkin and Quigg, 27 with permission from Wolters Kluwer Health.) Neurology 80 (Suppl 1) January 1, 2013

S39

Figure 16

Low display resolution affecting the representation of higher frequencies in intracranial EEG (ICEEG) recording

(A)ICEEG at theseizure onset viewed with a time base of 30 mm/s.The earliest sustained ictalactivity appears to be in theLMT (left mesial temporal) channels 6 to 8. (B) ICEEG at the seizure onset at a time base of 60 mm/s. At this setting, the low-amplitude fast activity in the LP (left parietal) channels is clearly visible as the earliest sustained ictal activity (box). (C) Power spectral analysis of 2 electrode channels, LP15 and LMT7, for the same 1-second epoch at the seizure onset (represented by the black bar in A and B). Powers in the 10- to 120-Hz frequency range are shown for each channel. Note the activity at 70 to 85 Hz in LP15. (D) Frequency aliasing of a 30-Hz signal at a screen resolution of 95 pixels per second horizontally. Compare with thesame signal viewed at a resolution of 190 pixels per second. (Reproduced from Schevon et al., 24 with permission from John Wiley & Sons.)

DISPLAY RESOLUTION FOR VIEWING INTRACRANIAL EEG Misinterpretation can arise from inadequate dis-

playing of the EEG, particularly when faster frequencies are involved. It is well known that during analog-to-digital conversion of the EEG signal, a sampling rate of at least twice the highest frequency component (referred to as the Nyquist frequency) has to be used to avoid frequency aliasing; a rate at least 5 times is recommended, because this is about what is needed for reliable reproduction of complex waveforms. It is less frequently appreciated that the same rule also applies when the digitized EEG signal is displayed on a monitor screen. Using a screen resolution too low is a form of downsampling and can lead to the obliteration of higher frequencies or aliasing (appearance of false frequencies), with possible adverse consequences, such as the erroneous localization of the seizure-onset zone (figure 16). 24 S40


If one hopes to visualize up to 100-Hz activity on a typical 21-inch monitor with 1280 3 1024 resolution, only 2.5 seconds should be displayed on the screen at a time. Similar issues can occur with vertical resolution, and too many channels displayed at once should be avoided. Computer-aided analysis of intracranial EEG will become essential as broader band EEG (from DC to several hundred Hz or more) is used more frequently, especially if clinical utility of high-frequency oscillations is confirmed.25 CONCLUSION Every EEG should be interpreted with

care and caution to avoid pitfalls (table 5). This is especially true for studies recorded in the ICU where artifacts are numerous and many EEG patterns may reflect different processes, including ictal, interictal, and metabolic, often combined simultaneously and varying over time.

Table 5

Some common errors related to interpreting intensive care unit EEG

Misinterpreting artifact as seizures

4.

5.

Assuming there is a clear dichotomy between ictal and interictal EEG patterns in encephalopathic patients (there is not) Underdiagnosing nonconvulsive seizures/status epilepticus on EEG Believing that because some patterns can be ictal at times implies that they are always, often, or usually ictal

6.

Assuming a comatose patient in nonconvulsive status will wake up immediately if successfully treated Corollary error: If they don’ t improve clinically, concluding it was not nonconvulsive status epilepticus (it still could be, just not proven)

7.

Related error: Concluding that if an EEG pattern resolves with an antiepileptic drug, that proves it was nonconvulsive status (might have been, but need clinical improvement to prove it) Also related error: When doing a diagnostic benzodiazepine treatment trial, using too high of a dose (and putting the patient into deep sleep/coma) Concluding that if a pattern is induced or exacerbated by alerting or stimulation, it is not ictal” (it still can be)

“

8.

9.

Interpreting quantitative EEG, especially amplitude-integrated EEG, without the raw EEG or without an electroencephalographer Assuming that a negative scalp EEG rules out seizure (it does not) Calling clinical spells seizures when not

10.

Assuming intracranial EEG recordings have no artifact Overuse of filters (especially the high-frequency and notch filters)

11.

There are ways of trying to clarify their significance, including AED trials, but this is often inconclusive. In case of doubt, one has to avoid overinterpretation and unnecessarily aggressive treatment. Newer methods of EEG analysis are useful and improve the yield of EEG monitoring but they are themselves subject to artifact and misinterpretation. Proper training is a crucial aspect of minimizing as many of the errors as possible.

12.

13.

14.

AUTHOR CONTRIBUTIONS N. Gaspard and L.J. Hirsch drafted the article. L.J. Hirsch critically revised the manuscript for intellectual content. Both gave their final approval of the article.

15.

16.

DISCLOSURE N. Gaspard reports no disclosures relevant to the manuscript. L. Hirsch has received research support for investigator-initiated studies from Eisai, Pfizer, UCB-Pharma, Lundbeck, and Upsher-Smith and consultation fees for advising from Lundbeck, Upsher-Smith, and GlaxoSmithKline. Go to Neurology.org for full disclosures.

17.

18.

Received January 11, 2012. Accepted in final form May 1, 2012.

REFERENCES 1.

2.

3.

19.

Hernandez-Frau PE, Benbadis SR. Pearls & oy-sters: errors in EEG interpretations: what is misinterpreted besides temporal sharp transients? Neurology 2011;76: e57–e59. Benbadis SR. Errors in EEGs and the misdiagnosis of epilepsy: importance, causes, consequences, and proposed remedies. Epilepsy Behav 2007;11:257–262. Hirsch LJ. Continuous EEG monitoring in the intensive care unit: an overview. J Clin Neurophysiol 2004;21:332–340.

20. 21.

22.

Young GB, Campbell VC. EEG monitoring in the intensive care unit: pitfalls and caveats. J Clin Neurophysiol 1999;16: 40–45. Pandian JD, Cascino GD, So EL, Manno E, Fulgham JR. Digital video-electroencephalographic monitoring in the neurological-neurosurgical intensive care unit: clinical features and outcome. Arch Neurol 2004; 61:1090–1094. Benbadis SR, Chen S, Melo M. What’s shaking in the ICU? The differential diagnosis of seizures in the intensive care setting. Epilepsia 2010;51:2338–2340. Husain AM, Mebust KA, Radtke RA. Generalized periodic epileptiform discharges: etiologies, relationship to status epilepticus, and prognosis. J Clin Neurophysiol 1999; 16:51–58. Pohlmann-Eden B, Hoch DB, Cochius JI, Chiappa KH. Periodic lateralized epileptiform discharges: a critical review. J Clin Neurophysiol 1996;13:519–530. Snodgrass SM, Tsuburaya K, Ajmone-Marsan C. Clinical significance of periodic lateralized epileptiform discharges: relationship with status epilepticus. J Clin Neurophysiol 1989;6: 159–172. Hirsch LJ, Claassen J, Mayer SA, Emerson RG. Stimulusinduced rhythmic, periodic, or ictal discharges (SIRPIDs): a common EEG phenomenon in the critically ill. Epilepsia 2004;45:109–123. Hirsch LJ, Pang T, Claassen J, et al. Focal motor seizures induced by alerting stimuli in critically ill patients. Epilepsia 2008;49:968–973. Young GB, Jordan KG, Doig GS. An assessment of nonconvulsive seizures in the intensive care unit using continuous EEG monitoring: an investigation of variables associated with mortality. Neurology 1996;47:83–89. Chong DJ, Hirsch LJ. Which EEG patterns warrant treatment in the critically ill? Reviewing the evidence for treatment of periodic epileptiform discharges and related patterns. J Clin Neurophysiol 2005;22:79 –91. Fountain NB, Waldman WA. Effects of benzodiazepines on triphasic waves: implications for nonconvulsive status epilepticus. J Clin Neurophysiol 2001;18: 345 –352. Drislane FW, Lopez MR, Blum AS, Schomer DL. Detection and treatment of refractory status epilepticus in the intensive care unit. J Clin Neurophysiol 2008;25:181–186. Claassen J. How I treat patients with EEG patterns on the ictal-interictal continuum in the neuro ICU. Neurocrit Care 2009;11:437–444. Chiaretti A, Barone G, Riccardi R, et al. NGF, DCX, and NSE upregulation correlates with severity and outcome of head trauma in children. Neurology 2009;72:609–616. DeGiorgio CM, Correale JD, Gott PS, et al. Serum neuronspecific enolase in human status epilepticus. Neurology 1995;45:1134–1137. Sankar R, Shin DH, Wasterlain CG. Serum neuronspecific enolase is a marker for neuronal damage following status epilepticus in the rat. Epilepsy Res 1997;28: 129 –136. Brenner RP, Hirsch LJ. Atlas of EEG in Critical Care. Chichester, UK: John Wiley & Sons; 2010:1–348. Evans E, Koh S, Lerner J, Sankar R, Garg M. Accuracy of amplitude integrated EEG in a neonatal cohort. Arch Dis Child Fetal Neonatal Ed 2010;95:F169–F173. Shellhaas RA, Gallagher PR, Clancy RR. Assessment of neonatal electroencephalography (EEG) background by


S41

23.

24.

S42

conventional and two amplitude-integrated EEG classification systems. J Pediatr 2008;153:369–374. Waziri A, Claassen J, Stuart RM, et al. Intracortical electroencephalography in acute brain injury. Ann Neurol 2009;66:366–377. Schevon CA, Thompson T, Hirsch LJ, Emerson RG. Inadequacy of standard screen resolution for localization of seizures recorded from intracranial electrodes. Epilepsia 2004;45:1453– 1458.


25.

26. 27.

Jacobs J, Zijlmans M, Zelmann R, et al. High-frequency electroencephalographic oscillations correlate with outcome of epilepsy surgery. Ann Neurol 2010;67:209– 220. Foreman B, Hirsch LJ. Epilepsy emergencies: diagnosis and management. Neurol Clin 2012;30:11–41. GoodkinHP, Quigg M. Toothbrushing EEG artifact recorded from chronically implanted subdural electrodes. Neurology 2010;75:1850.

EEG Errores Pitfalls Mala Interpretacion Misdiagnosis Mal Diagnostico Neurology 2013

Recommend Documents