The EEG in Coma

G. Bryan Young

Department of Clinical Neurological Sciences, The University of Western Ontario, London, Ontario, Canada

Summary: The EEG allows insight into thalamocortical function in comatose patients when this is inaccessible clinically. A single EEG can help with broad diagnostic categorization whereas continuous or serial EEG provides monitoring for unstable and potentially treatable conditions and for monitoring the effects of therapy. The EEG plays a supplemental role in establishing the prognosis in disease states that are capable of causing neuronal death. The most prevalent and problematic of these conditions involves survivors of cardiac arrest who are initially in coma with intact brainstem reflexes. In such patients single EEGs are of 100% specificity for no possibility of recovery of consciousness only for essentially complete generalized suppression (Ͻ10 ␮V) after the first day of the arrest. Several other generalized patterns, including less marked suppression, burst–suppression, epileptiform activity, periodic complexes, and ␣–⌰ coma patterns, usually but not invariably indicate a poor outcome. Serial EEGs, continuous raw and automated “trending,” testing of reactivity, and the inclusion of multiple variables hold promise for an improved role in the prognostic determination in these patients. Key Words: EEG—Coma—Prognosis—Anoxic-ischemic encephalopathy.

The EEG is underused in coma. Although thalamocor- or drug-induced coma). When the etiology of the coma is tical function cannot be adequately assessed clinically in known, the EEG can be very useful in determining the the comatose patient, EEG allows for an immediate prognosis in some conditions, especially when combined examination of cortical or cortical–subcortical dysfunc- with the neurologic examination (Sharbrough, 1981). tion in an inexpensive, safe, and readily available manner The EEG is usually dynamic: If variation does not occur in the intensive care unit. Although patients may show within a given recording, it usually will over time. The the same clinical features, especially when brainstem variety and complexity of possible rhythms is inversely functions are clinically intact, the EEG can reveal a wide related to the severity of the dysfunction (Rumpl, 1987). spectrum of abnormalities. EEG is especially sensitive to Absolutely invariant EEGs usually indicate a severe the graded severity of brain dysfunction and the direction encephalopathy and often a poor prognosis (Cant and of the process if serial tracings or continuous recordings Shaw, 1984; Karnaze et al., 1982). are used (Chiappa and Ropper, 1984). Although rarely The clinical importance of various EEG patterns in specific for the etiology of coma, the EEG may help coma must take into consideration the clinical picture, determine the class or general category of disease pro- age, etiology, acuity, and the integrity of brainstem cess (e.g., the differentiation of seizures from metabolic reflexes. Most EEG patterns are nonspecific, but some findings lead to productive intervention (Jordan, 1993). EEG allows for the recognition and monitoring of treat- Presented at the 1999 annual meeting of the American Neurophys- ment of seizures. In patients with vasospasm, the EEG iology Society; St. Louis, MO; October 29–November 1, 1999. can be used to note the response to adjustment of perfu- Address correspondence and reprint requests to Dr. G. B. Young, Department of Clinical Neurological Sciences, London Health Sciences sion pressure. Certain “metabolic patterns” (e.g., tripha- Centre, 375 South Street, London, Ontario, Canada N6A 4G5. sic waves or intermittent rhythmic ␦-waves with an

473 474 G. B. YOUNG abnormal background) may trigger searches for systemic epilepticus, recurring seizures, and conditions in which disorders or exogenous toxins. Lateralized abnormalities intracranial pressure is elevated or cerebral perfusion is prompt a scan to detect a treatable structural lesion variable (e.g., various neurosurgical conditions including (Jordan, 1993). post-traumatic coma, subarachnoid hemorrhage, tumor, It is valuable to assess reactivity or EEG change after postoperative cases, and hydrocephalus). Changes in sensory stimulation in states of decreased consciousness frequency and amplitude, new focal features, or epilep- (Fishgold and Mathis, 1959). In general, reactivity is a tiform activity have diagnostic and therapeutic implica- feature of lighter stages of coma, but it is inconsistently tions that can influence management and outcome (Jor- associated with clinical arousability. Reactivity may ex- dan, 1990, 1992). Continuous EEG monitoring is the best ist in a variety of forms: an increase in amplitude, method of monitoring the depth of anesthesia achieved frequency, or even slowing (often as bisynchronous with barbiturates used in the treatment of status epilep- rhythmic ␦-activity), or a decrease in amplitude of back- ticus, or even of monitoring the depth of sedation in other ground rhythms. In general, EEG responsiveness is as- patients (Van Ness, 1990). This technology may be sociated with a greater chance of recovery than the lack valuable in monitoring those with central nervous system of reactivity (Fishgold and Mathis, 1959; Young et al., disease who cannot be assessed clinically, especially 1999). Reactivity should be tested in all comatose pa- patients on neuromuscular blocking agents. tients, unless contraindicated because of concerns re- Although this review is restricted to EEG, comprehen- garding raised intracranial pressure. We usually test for sive neurophysiologic monitoring includes sensory auditory reactivity by clapping or shouting in the pa- evoked potentials (Chiappa and Hoch, 1993). These and, tient’s ears. Somatosensory stimulation should be tested potentially, motor evoked responses, add new dimen- by applying pressure to the nail bed of each hand and to sions to the examination of integrity of the anatomic– the supraorbital nerve above the medial third of the physiologic subcortical pathways and primary sensory eyebrow. Passive eye opening is recommended in sus- and motor cortices. The early components are not af- pected ␣-coma (no change in ␣) as well as a stimulus for fected substantially by drugs or reversible metabolic reactivity. Because some patients may respond to only abnormalities, whereas EEGs are. These techniques alone of these stimuli, all should be tried. low for the assessment of prognosis in a more robust way There are several scoring systems for grading the than has been possible with EEG alone (Chiappa and severity of EEG abnormalities: for anoxic, see Hockaday Hoch, 1993; Goldie et al., 1981; Rothstein et al., 1991). et al. (1965); traumatic, Rae–Grant et al. (1991); or both Any review of EEG in coma is hampered by the varied anoxic and traumatic, Hughes et al. (1976), Synek causes of coma, the concomitant use of sedative medi- (1988), and Young et al. (1997); septic encephalopathy, cations, the small number of patients in each study, the Young et al. (1990a); and Reye’s syndrome, Aoki and paucity of prospective studies, the nonuniform criteria Lombroso (1973). In attempting to determine a progno- for coma, the different EEG classification systems, and sis, however, it is best to relate the EEG to the entire the absence of validation (insufficient survival times, clinical picture and to consider whether potentially re- lack of follow-up studies or postmortem correlation). versible factors may be operative (Hansotia et al., 1981). With these caveats in mind, the EEG in several disease The less ambiguous the categories, the better the inter- categories is reviewed, with emphasis on anoxic–isch- and intrarater reliability (Rae–Grant et al., 1991; Young emic encephalopathy. et al., 1999). Quantitative EEG, notably power spectral analysis, STRUCTURAL LESIONS allows for trending over longer periods of time. This is useful in examining variability objectively, in charting The role of the EEG as a primary diagnostic test for the course of encephalopathies, and in detecting sudden structural brain lesions has been greatly modified by the changes such as those produced by seizures (Nuwer, widespread use of computed tomography and MRI. 1988a, 1988b). Purely quantitative studies may miss However, focal EEG abnormalities should raise the sus- major morphologic phenomena in EEG (e.g., triphasic picion of structural cerebral lesions. Furthermore, the waves, some seizures). Hence, the “raw EEG” should EEG reveals the functional disturbance caused by the always be available along with the spectral plot or brain structural abnormality and thus plays a complementary map. role to imaging tests. Continuous EEG monitoring might best be applied to Single structural lesions in the supratentorial compart- cases in which the patient is in coma with an unstable but ment produce coma mainly by compression or displace- potentially treatable condition. Examples include status ment of the diencephalon or mid-brain tegmentum. EEG

J Clin Neurophysiol, Vol. 17, No. 5, 2000 EEG IN COMA 475 manifestations may be diffuse by the time the patient is rhage into the brain substance or subarachnoid, subdural, in coma, but one should suspect a supratentorial lesion if or epidural space; ischemic infarction (e.g., from vascu- there is a marked and consistent asymmetry of voltage lar injury, arterial spasm, or decreased cerebral perfusion between the two hemispheres or if intermittent rhythmic pressure); axonal (shearing) injury in the white matter; ␦-activity is exclusively frontal (Harner and Naquet, associated metabolic derangements: hypoxemia, fluid, 1975). Another mechanism by which supratentorial le- and electrolyte disturbances and sepsis; and symptomatic sions may produce coma is by causing seizures. Epilep- seizures. It is thus not surprising that there is a rich tiform activity on EEG may suggest that an earlier variety of EEG abnormalities. Their prognostic impor- seizure was missed clinically and that the patient is in a tance rests not only on their severity but also on the postictal state. Alternatively, nonconvulsive status epi- reversibility of the pathogenetic mechanisms responsi- lepticus may be revealed. ble. Lesions limited to the brainstem often do not produce The following patterns have been described: slowing much EEG slowing; however, sleep potentials may be in the ⌰-range (Courjon and Scherzer, 1972); ␦-activity, ␣ found in some cases. Often the EEG shows an -pattern including continuous focal arrhythmic, diffuse rhythmic, coma that may be more marked posteriorly (Chase et al., or arrhythmic or intermittent rhythmic patterns (Dusser 1968). Care should be taken that the patient is in coma et al., 1989; Stockard et al., 1975); faster frequencies and not in a locked-in state (Harner and Naquet, 1975). (Loeb et al., 1959); epileptiform activity, including focal As a corollary, posterior fossa lesions with marked spikes or seizures, either clinical or electrographic (Sil- EEG slowing probably also have dysfunction in the verman, 1963; Synek, 1988; Williams, 1941a, 1941b); supratentorial compartment, especially the diencephalon, synchronous periodic slow-wave bursts (Fishgold and because of hydrocephalus (Harner and Naquet, 1975). Mathis, 1959); triphasic waves (Stockard and Bickford, 1975), focal suppression (Dow et al., 1961); other asym- CEREBROVASCULAR DISEASES metries (Courjon and Scherzer, 1972); and ␣-, ⌰-, or spindle coma patterns (Chatrian et al., 1963; Rumpl et Cerebrovascular disorders typically cause structural al., 1983; Synek and Synek, 1984; Westmoreland et al., brain lesions. Statements in the previous section regard- 1975). ing structural lesions apply to cerebrovascular disease. The timing of the EEG in head injury is very important Sometimes, however, the lesions are not single (e.g., in with respect to the types of abnormalities and the asso- subarachnoid hemorrhage, multifocal vasospasm may ciated etiologic and management implications as well as cause multifocal ␦-waves or epileptiform activity [Van the prognosis. The comprehensive reviews by Courjon der Drift and Kok, 1972]). Watershed ischemia also shows bilateral abnormalities, but with maximal local- and Scherzer (1972), Stockard et al. (1975), Rumpl ization over the posterior temporo-occipitoparietal re- (1987), and Rath and Klein (1991) are recommended. gions. The background is slow and often accompanied by Localized abnormalities at any stage should raise the periodic epileptiform activity (Niedermeyer, 1987). possibility of structural lesions. Changes of amplitude, Thrombosis of the superior sagittal sinus and draining especially localized reduced voltage, should be consid- veins would be expected to produce a somewhat similar ered to reflect an intracranial hematoma until proved picture of bilateral arrhythmic ␦- and epileptiform activ- otherwise, provided local scalp swelling over the region ity, but may not be consistently maximal in the posterior is excluded (Courjon and Scherzer, 1972). Focal slowing head. Brainstem stroke may produce a normal EEG in is sometimes associated with such lesions but can occur the locked-in syndrome or ␣-pattern coma or diffuse as a transient phenomenon and may not always prove to slow-wave frequencies with reduced reactivity in rostral be important. This is not uncommon in children with brainstem tegmental lesions. occipital ␦-activity. Electrographic seizures usually re- Metabolic consequences of cerebrovascular diseases flect severe brain injury, are of reliable lateralizing value, (e.g., inappropriate antidiuretic hormone secretion and but should prompt urgent brain imaging and treatment hyponatremia) may alter the EEG diffusely, as does (Courjon and Scherzer, 1972). hydrocephalus with coma. When the recording shows continuous rhythmic activity, including sleep phenomena, the presence of variabil- TRAUMA ity in frequencies and amplitude is more favorable for outcome than recordings showing no variability (i.e., a Trauma can affect the brain by a number of mecha- monophasic picture [Bergamasco et al., 1968; Chatrian nisms either singly or in different combinations: hemor- et al., 1963]). Any reactivity is associated with a better

J Clin Neurophysiol, Vol. 17, No. 5, 2000 476 G. B. YOUNG prognosis than no reactivity (Stockard and Bickford, weeks suggests a complication such as infarction from 1975). vasculitis, hydrocephalus, or subdural hygromas (the Sometimes trauma affects the brain indirectly. Fat latter in young children). Marked EEG abnormalities are embolism to the brain and lungs is a complication of the associated with severe neurologic sequelae (Chequer et fractures of long bones. The patient may be asymptom- al., 1992). atic for 12 to 36 hours after the injury and then may Encephalitis also usually produces ␦-waves as the become obtunded in association with pulmonary symp- most predominant abnormality (Vas and Cracco, 1990). toms and tachycardia. The EEG shows the replacement Multifocal epileptiform activity may occur, reflecting of normal rhythms with diffuse polymorphic ␦- and cortical gray matter disease (Gloor et al., 1968). In ⌰-activity, accentuated over the frontal regions (Courjon herpes simplex encephalitis, one or both temporal re- and Scherzer, 1972). Trauma to the neck may cause gions are the site of maximal slowing, and between the delayed occlusion of carotid or vertebral arteries, and second and fourteenth day of illness these sites often clinical and EEG features of stroke in these territories. demonstrate periodic, lateralized epileptiform discharges The results of serial EEGs or continuous monitoring at a repetition rate of 0.5 to 2 per second (Fig. 1; Gupta assist in determining the prognosis and in detecting and Seth, 1973; Illis and Taylor, 1972). This diagnostic potentially treatable conditions such as seizure activity, feature is especially important because herpes simplex brain herniation, hydrocephalus, and intracerebral and encephalitis is specifically treatable with the antiviral subdural hematomas (Rumpl et al., 1979). The latter agent acyclovir (Skoldenberg et al., 1984). condition should be suspected if the EEG shows increas- Multifocal ␦-activity is expected in acute disseminated ing slow-wave activity or focal suppression, increasing encephalomyelitis—a postinfectious immunologically generalized slow frequencies, or chronically persisting mediated syndrome that is characterized by perivenous focal abnormalities along with intermittent diffuse rhyth- demyelination (Gloor et al., 1968). mic ␦-waves (Stockard et al., 1975). Vespa and col- leagues (1999) detected seizures in 22% of patients with STATUS EPILEPTICUS moderate to severe brain injury using continuous EEG monitoring. Half the seizures were nonconvulsive and Nonconvulsive seizures (NCSs) are epileptic attacks would not have been noted clinically. Because there is a without a tonic, clonic, or tonic–clonic component. The strong association of prolonged nonconvulsive seizures transition from a convulsive seizure or an NCS to non- with greater mortality (Vespa et al., 1999; Young et al., convulsive status epilepticus (NCSE) or “subtle status” is 1996), continuous EEG monitoring may prove to be a the change of a self-limited epileptic attack to a persis- valuable component of intensive care unit care. Addi- tent or protracted epileptic state. Nonconvulsive seizures tional work to show improved outcomes (lessened mor- or nonconvulsive status epilepticus is not always gener- tality and shorter intensive care unit length of stay) alized from the onset. Some patients with focally origi- would provide greater justification. nating seizures may develop the clinical and EEG ap- These EEG findings and their timing need to be pearance of generalized epileptiform activity (Fagan and considered along with the integrity of brainstem function Lee, 1990.) Thus, generalized nonconvulsive status epi- and clinical responsiveness in formulating a prognosis lepticus is heterogenous. for a particular patient (Hansotia et al., 1981; Shar- Nonconvulsive status epilepticus may produce stupor brough, 1981). or coma. Although patients with complex partial seizures are more prevalent than those with absence attacks, CENTRAL NERVOUS SYSTEM INFECTIONS absence status is more common than complex partial status epilepticus. Most often these occur in patients with Although viral meningitis should not affect the EEG, known seizure disorders. Eyelid flutter and rhythmic bacterial and tuberculous meningitis usually produce myoclonus of the face or extremities may suggest ab- irregular, widespread ␦-waves as the predominant abnor- sence status, but an EEG is almost always necessary for mality. Background activity is usually better preserved in confirmation. Absence status may occur in the elderly meningitis than in encephalitis (Stockard and Bickford, without a previous history of seizures, and can be missed 1975). The improvement in EEG frequencies parallels if the diagnosis is not considered or if an EEG is not clinical resolution (Turrell and Roseman, 1955). In pa- performed (Ellis and Lee, 1978). The EEG usually shows tients with a good prognosis, the EEG shows rapid bilaterally synchronous paroxysmal discharges, often in normalization within 6 to 9 days (Turrell and Roseman, the form of generalized spikes and waves. The repetition 1955). Prolongation of EEG abnormalities beyond 2 rate in status epilepticus may be slower (e.g., two sei-

J Clin Neurophysiol, Vol. 17, No. 5, 2000 EEG IN COMA 477

FIG. 1. The EEG was performed on the fourth day of an illness characterized by fever, headache, confusion, and recurring complex partial seizures. There is bilateral slowing with medium-voltage ␦-waves in the anterior head bilaterally. Periodic lateralized epileptiform discharges (PLEDs at asterisks) occur over the right anterior–midtemporal region at intervals of just more than 1 second. (Reprinted with permission from Young GB, Ropper AH, Bolton CF, eds. Coma and impaired consciousness: a clinical perspective. New York: McGraw–Hill, 1998:250.) zures per second) compared with brief absence attacks pression is reversible. However, in some clinical condi- (three seizures or more per second). Nonconvulsive sei- tions, such as anoxic–ischemic encephalopathy after car- zures may follow generalized convulsive seizures, and diac arrest, complete suppression signifies death of account for what is often mistaken clinically for a pro- neocortical neurons, and the suppression is permanent. longed “postictal state.” The EEG is the only reliable means of detecting this phenomenon, and it is valuable diagnostically and prognostically (DeLorenzo et al., 1998).

METABOLIC ENCEPHALOPATHIES The graded EEG response to anesthetics is similar to the evolutionary changes seen with increasing severity of metabolic derangement. There is an inverse linear relationship of EEG-dominant frequency and a direct linear relationship of amplitude (resulting from slower frequencies) with anesthetic dose (Glaser, 1972–1977). Such slowing accompanies decreased unitary activity of cerebral cortical neurons (Creutzfeldt and Meisch, 1963). This linearity breaks down at high doses with the appearance of a burst–suppression pattern or epileptiform activity. The latter may occur with some agents that FIG. 2. ␦-Waves constitute the principal finding in a patient with renal produce a mixture of excitation and inhibition (e.g., failure and sepsis. Faster frequencies are better developed on the left diethyl ether and cyclopropane; Stockard and Bickford, because of an earlier infarction in the right cerebral hemisphere. The 1975). patient recovered. (Reprinted with permission from Young GB, Kreeft JH, McLachlan RS, Demelo J. EEG and clinical associations with With anesthetic agents and some metabolic conditions mortality in comatose patients in a general intensive care unit. J Clin (e.g., drug overdose and septic encephalopathy), sup- Neurophysiol 1999;16:354–460.)

J Clin Neurophysiol, Vol. 17, No. 5, 2000 478 G. B. YOUNG

FIG. 3. (A) After cardiac arrest, this patient had a burst–suppression pattern with bursts containing epileptiform discharges in addition to waves of various frequencies. He died without recovery of awareness. (B) This patient had a burst–suppression pattern with no epileptiform discharges within the bursts. Such patients may have a better prognosis than those with the pattern shown in A. (Reprinted with permission from Young GB, Kreeft JH, McLachlan RS, Demelo J. EEG and clinical associations with mortality in comatose patients in a general intensive care unit. J Clin Neurophysiol 1999;16:354–460.)

Thus, knowledge of etiology is vital in understanding the neuronal death (Lindenberg, 1982; Pearigen et al., 1996). importance of the suppression in clinical settings. Thus, the term generalized ischemic encephalopathy is Although many destructive, infectious, and degenera- more accurate pathophysiologically. The mechanisms for tive diseases, and end-stage storage diseases involving neuronal death include release of excitotoxic neurotrans- the brain are not considered to be acute metabolic en- mitters, activation of N-methyl-D-aspartate receptors cephalopathies, the EEG may show many of the features with calcium influx into neurons, peroxynitrile produc- described for the “anesthetic model’(Gloor et al., 1968; tion, failure of clearance of hydrogen ions, and lactate Steriade et al., 1990). In most of these diseases, in which and free radical production on reperfusion (Dawson et there is death or extensive loss of neocortical neurons, al., 1988, 1993; Xia et al., 1996). the EEG changes deteriorate to various degrees of sup- The cortical N20 response to median nerve stimulation pression and there is no clinical reversibility. approaches the ideal prognostic test (Bassetti et al., 1996; The EEG shows evolutionary changes related to the Goldie et al., 1981; Rothstein et al., 1991; Sandroni et al., depth of anesthesia and the dose of drugs (Stockard and Bickford, 1975): 1. Desynchronization or “fast activity” 2. Increase in rhythmicity and voltage, especially ␦-activity (Fig. 2) 3. Mixtures of slower frequencies with the faster frequencies; increased ␦-activity with deeper levels 4. Burst–suppression, with duration of suppression becoming longer with deeper levels (Fig. 3) 5. Suppression; isoelectric EEG (Fig. 4) Many metabolic disorders follow this basic model but show some special features or departures as well. Table 1 (Young et al., 1992) lists various metabolic and toxic diseases as they relate to the anesthesia model.

EEG IN GENERALIZED ISCHEMIC ENCEPHALOPATHY FIG. 4. Generalized suppression followed cardiac arrest and severe From experimental studies, it is clear that ischemia is ischemic encephalopathy. The patient died without recovery of con- the essential component in producing neuronal death in sciousness. (Reprinted with permission from Young GB, Kreeft JH, McLachlan RS, Demelo J. EEG and clinical associations with mortality cardiac arrest. Hypoxia alone, even with arterial oxygen in comatose patients in a general intensive care unit. J Clin Neuro- concentrations of less than 25 mm Hg, does not produce physiol 1999;16:354–460).

J Clin Neurophysiol, Vol. 17, No. 5, 2000 EEG IN COMA 479

TABLE 1. Metabolic encephalopathies related to the anesthetic model

Reversibility of burst–suppression or Etiology Additional features suppression

Drug intoxication (Haider et al., 1971) Some may produce epileptiform activity; a Completely reversible mixture of ␤ ϩ ␪/␦ suggestive of barbiturate or benzodiazepine overdose Hepatic failure (Bickford and Butt, 1955; Triphasic waves* (Fig. 5) Not reversible (older literature); potentially Karnaze and Bickford, 1984; Kennedy et reversible if liver function improves or with al., 1973; Parsons–Smith et al., 1957) detoxification and if severe cerebral edema is prevented Reyes syndrome (Aoki and Lombroso, 1973; 14 and 6 positive spikes in 50% Probably not reversible (dependent on damage Yamada et al., 1977) (controversial significance) from brain swelling and/or hypoglycemia Uremic encephalopathy (Bolton and Young, Triphasic waves* or epileptiform activity in Probably reversible 1990) some; photic sensitivity Septic encephalopathy (Young et al., 1990a) Triphasic waves* in some Potentially reversible if patient survives multiple organ failure Dialytic encephalopathy (Hughes and Epileptiform activity (generalized spike and Not reversible in advanced cases (aluminum Schreeder, 1980) wave) poisoning) Porphyria (Dow, 1961) Focal features (e.g., spikes, suppression Largely reversible; EEG may be slow to seen occasionally, usually transient) improve Hypothyroidism (Nieman, 1959) Low amplitude Should resolve, except in cretinism Wernicke’s encephalopathy (Fournet and May show sharp and slow wave complexes, EEG may improve many patients with severe Lanternier, 1956) periods of suppression in very advanced amnesia and ataxia cases Addison’s disease (Dreyfus–Bresac and Majority are normal; others show diffuse Resolves, especially with corticosteroids Mises, 1953) slowing Hypercalcemia (Karnaze and Bickford, 1984; May have triphasic waves or FIRDA Rare, resolves Moure, 1967) Hypoglycemia (Paschen et al., 1991; Regan May show epileptiform activity, periodic Partly reversible, but neuronal death can occur and Brown–Mayers, 1956; Shagass and complexes if severe Rowsell, 1954) Hypothermia (Hughes, 1978; Sadove et al., Temperature dependent: no change until Burst–suppression at 20 to 22°C, suppression 1967) Ͻ 30°C with temperature Ͻ 18°C completely reversible

* Triphasic waves were originally thought to be specific for hepatic encephalopathy (Bickford and Butt, 1955). However, they can occur in a variety of metabolic encephalopathies, septic encephalopathies, and neurodegenerative conditions (Karnaze and Bickford, 1984; Sundaram and Blume, 1987). In the context of an acute alteration in alertness, however, they are highly suggestive of a metabolic encephalopathy. In our experience, hepatic or renal failure or septic encephalopathy are the most common underlying etiologies in this situation (Young et al., 1990a). FIRDA, frontal intermittent rhythmic delta activity. (Reprinted with permission from the American Journal of EEG Technology. Source: Young et al., 1992.)

1995; Zandbergen et al., 1998). The absence of the N20 response from median nerve stimulation is specific but is not especially sensitive for a hopeless prognosis. The lack of this response shows nearly 100% specificity for an outcome no better than a permanent vegetative state, but many patients with a preserved N20 response die without recovery of consciousness. The EEG must be compared with this as a prognostic test. From a theoretical point of view, the EEG should be a suitable test for cortical damage or integrity. The neurons that are the most sensitive to generalized ischemic damage are those of the cerebral cortex that generate postsyn- aptic potentials that are recorded from scalp EEG. The EEG is flat during resuscitation; this may persist for several hours after circulation is restored (Jorgensen and FIG. 5. Triphasic waves are present in this patient with septic enceph- Holm, 1998; Pampiglioni and Harden, 1968; Prior, alopathy who later died of multiorgan failure. (Reprinted with permission from Young GB, Kreeft JH, McLachlan RS, Demelo J. EEG and 1973a, 1973b). EEG rhythms in the very young (i.e., clinical associations with mortality in comatose patients in a general neonates and prematures) may take longer to recover intensive care unit. J Clin Neurophysiol 1999;16:354–460.)

J Clin Neurophysiol, Vol. 17, No. 5, 2000 480 G. B. YOUNG than with older patients. Thus, as a general rule, it is better and normal blood pressure are a reliable indication of an to wait 24 hours or more from the event until the first EEG, outcome no better than permanent vegetative state. unless seizures are suspected. There is a caveat with respect to the suppressed EEG. EEGs performed after the first day of arrest may show Matsuo (1985) has noted that in patients with permanent deteriorating patterns associated with a fatal outcome. This vegetative state after cardiac arrest, the EEG can vary in is supported by experimental models, in which it has been amplitude. Some patients appear to have a desynchro- shown that the phenomenon of “delayed neuronal death” nized EEG that gives the appearance of electrocerebral may take more than 24 hours to develop (Kirino, 1982). silence, yet with polysomnography they may show One must be on guard that such later suppression is not higher voltage ␦-activity as a fragment of slow-wave related to drugs, shock, or sepsis, however. sleep, alternating with rapid eye movement sleep. There There is general agreement that the following EEG was not, however, the full development of all four stages patterns found after cardiac arrest are strongly associated of slow-wave sleep. with an poor neurologic outcome: generalized suppression; generalized burst–suppression; generalized periodic patterns, especially with epileptiform activity; and Generalized Burst–-Suppression ␣-or␣-/⌰-pattern coma (Binnie and Prior, 1994; Binnie et al, 1970; Hockaday et al., 1965; Mo¨ller et al., 1978; Generalized burst–suppression after cardiac arrest has Pampiglioni and Harden, 1968; Prior, 1973b, pp 244– usually, but not invariably, been associated with an 254; Scollo–Lavizzari and Bassetti, 1987; Yamashita et outcome of no better than persistent vegetative state. The al., 1995; Zandbergen et al., 1998; Zaret, 1985). These term burst–suppression (see Fig. 3) is meant to indicate patterns are addressed in the following subsections. episodes of generalized attenuation in voltage with the burst containing various frequencies (International Fed- eration of Clinical Neurophysiology, 1983). If the bursts Generalized Suppression contain only epileptiform discharges or bisynchronous Generalized suppression is not well defined in the short-duration complexes and are roughly periodic, the older EEG literature on postcardiac arrest patients. pattern could be called generalized periodic complexes, Terms such as nearly flat, no EEG at all (presumably but some may still refer to the pattern as a burst– complete or isoelectric suppression) are not sufficiently suppression pattern. Burst–suppression again has not precise, and technical criteria are not given in sufficient been well defined in published studies, and in many detail to know the voltage threshold described for these cases burst–suppression is lumped with other patterns. A terms. In some studies, suppression is combined with fatal outcome occurred in all 14 patients with various other patterns that are thought to indicate an unfavorable generalized burst–suppression patterns in the series of Bren- prognosis (burst–suppression or ␣-pattern coma). Sup- ner et al. (1975), and in all 27 children with unspecified pression sometimes takes time to evolve. It may not be burst–suppression in the series of Pampligioni and Harden present even at 24 hours, but only by 2 or 3 days from the (1968). However, occasional survivors are found, especially time of the cardiac arrest. In Prior’s (1973b) study, 43 of in those without epileptiform discharges in the bursts (Chen 43 patients (100%) with cardiac arrest with Hockaday et et al., 1996; Cloche et al., 1968). The presence or absence al.’s (1965) grade V category (incomplete or complete of simple or even fairly complex movements (e.g., tongue suppression), either initially or on subsequent EEGs, thrusting, grimacing) does not alter the usually unfavorable died without recovery of consciousness. The original outcome (Reeves et al., 1997). Several series have found report by Hockaday et al. (1965) contained 14 patients that when generalized epileptiform activity occurs within with cardiac or respiratory arrest, and all 14 of patients in the bursts, patients rarely recover consciousness (Pampigli- the group V category died. We have had at least one one, 1964; Synek, 1988; Young et al., 1999). Similarly, patient with incomplete suppression—more than 10 ␮V generalized epileptiform discharges, either as periodic, ste- but less than 20 ␮V—who recovered conscious aware- reotyped phenomena or as a burst of discharges, as the only ness, but none with less than 10 ␮V recovered conscious- EEG activity on an otherwise totally suppressed back- ness. There are no reports of recovery of conscious ground, are almost always associated with lack of recovery awareness in a child or an adult with an isoelectric EEG of awareness. that was recorded more than 24 hours after restoration of The electroencephalographer should be aware that circulation. Autopsy reports show severe cortical damage drugs, especially midazolam and propofol, commonly (Brierley et al., 1971). Thus, an isoelectric EEG, re- used in intensive care units, may produce a burst–sup- corded more than 24 hours from cardiac resuscitation, pression pattern. Opiates may produce a burst–suppres-

J Clin Neurophysiol, Vol. 17, No. 5, 2000 EEG IN COMA 481

FIG. 7. Generalized pseudoperiodic complexes appear against a suppressed background in a comatose patient after cardiac arrest. No myoclonus was noted.

from other causes, as described by Trieman et al. (1990). (Treiman’s five stages include (1) discrete seizures: epileptiform activity occurs in discrete epochs separated by FIG. 6. The patient experienced a cardiac arrest 4 days before this recording, which shows continuous generalized epileptiform dis- postictal slow-frequency waves; (2) merging seizures: charges. These discharges were present after paralysis with skeletal epileptiform activity persists throughout, but the dis- muscle relaxants, establishing that they were not related to artifact from charges wax and wane in frequency and amplitude; (3) myoclonus. (Reprinted with permission from Young GB, Kreeft JH, McLachlan RS, Demelo J. EEG and clinical associations with mortality continuous seizures: spikes and spike–waves are rhyth- in comatose patients in a general intensive care unit. J Clin Neuro- mic and relatively constant throughout the seizure; (4) physiol 1999;16:354–460.) continuous with flat periods: ongoing epileptiform activity is interrupted by generalized attenuation of voltage from 0.5 to 8 seconds; and (5) periodic epileptiform sion pattern with increased epileptiform potentials in the discharges on a flat background: high-voltage, repetitive neonate (da Silva et al., 1999). epileptiform discharges occur regularly against a flat or suppressed background.) The mortality rate of patients Epileptiform and Generalized Periodic Discharges with such seizures without cardiac arrest ranges from 6% to 35% (Hauser, 1983). It is important to differentiate the A theoretical concern involves patients with convul- fatal status epilepticus that occurs after cardiac arrest sive and subtle status epilepticus after cardiac arrest. from the status epilepticus of seizure disorders. If this After cardiac arrest, the occurrence of generalized epi- cannot be done clinically or with EEG, other tests to leptiform activity, either as sequential spikes or as spike confirm the nonviability of the cortex, such as somato- and wave (Fig. 6), single or generalized epileptiform sensory testing, sequential EEGs, or evolutionary discharges as part of a burst–suppression pattern, or changes from continuous EEG monitoring, may be valu- generalized periodic or pseudoperiodic complexes (Fig. able (Goldie et al., 1981). 7) has been associated with a fatal outcome without A caveat is the occurrence of postcardiac arrest elec- recovery of consciousness (Celesia et al., 1988; Gaches, trographic epileptiform activity without a flat back- 1971; Kuroiwa and Celesia, 1980; McCarty and Mar- ground, but with continuous background rhythmic shall, 1981; Nilsson et al., 1972; Simon and Aminoff, waves. Such patients recover consciousness more com- 1986; Wijdicks et al., 1994). This is often associated with monly than those with a true burst–suppression pattern myoclonus or eye opening accompanying the bursts of with epileptiform activity, according to Synek (1988). epileptiform activity (Simon and Aminoff, 1986; Young We have had a single case without a marked deficit. In et al, 1990b). However, Mori and Tsuruta (1983) re- the extensive study by Jorgensen and Holm (1998), ported such a case with a favorable recovery. however, no patients with this pattern recovered con- There is a striking similarity of these patterns to stages sciousness. Thus, this pattern is usually but not invari- 3, 4, or 5 of the sequential changes of status epilepticus ably evidence of severe generalized ischemic damage.

J Clin Neurophysiol, Vol. 17, No. 5, 2000 482 G. B. YOUNG

more continuous rhythms and develop electrographic reactivity to stimulation (Kaplan et al., 1999; Young et al., 1996). Alternatively, somatosensory evoked potential testing may establish the prognosis sooner (Rothstein et al., 1991).

OPTIMAL EEG USE AFTER CARDIAC ARREST The literature on cardiac arrest victims indicates that somatosensory evoked potentials are a better test than a single EEG recording for determining a prognosis of no better than permanent vegetative state. EEG lacks the specificity of somatosensory evoked potentials in this regard. Furthermore, EEG is hampered by reversible factors, including drugs, sepsis, and metabolic conditions, that are prevalent in the intensive care unit, whereas the N20 response is not as affected by these factors. Apart from complete EEG suppression after 24 hours from cardiac arrest, no other single EEG pattern has a 100% association with an outcome no better than permanent vegetative state. The value of single recordings, taken in isolation, is limited to various probabilities, FIG. 8. Poor outcome. The top EEG, showing a diffuse ␣-/-⌰-pattern, but never to certainty of poor outcome. was recorded 1 day after cardiac arrest. The bottom EEG, recorded 5 The predictive value of EEG is greatly enhanced, days later, shows a burst–suppression pattern. The patient remained in coma until death. (Reprinted with permission from Young GB, Blume however, if serial or continuous EEGs are performed WT, Campbell VC, et al. Alpha, alpha–theta and theta coma: a clinical over several days to note trends or evolution, reactivity, outcome study utilizing serial recordings. Electroencephalogr Clin and variability (Holmes and Lombroso, 1993). Further- Neurophysiol 1994;91:93–99.) more, if other factors—including even partial absence of cranial nerve reflexes, absence of EEG reactivity, and ␣-, ⌰-, ␣-/⌰-Pattern Comas variability—are included, the odds ratios for mortality without recovery of consciousness are markedly in- ␣ After cardiac arrest, -pattern coma consists of diffuse creased (Young et al., 1999). More research into the ␣ or frontally predominant, continuous rhythms in the - prognostic value of continuous and serial EEG studies (8–13 Hz) frequency band (Westmoreland et al., 1975). and the use of combined variables will be valuable. A ␣ ␣ Unlike normal -activity, -pattern coma is not respon- carefully performed, prospective, multicenter study with sive to eye opening or to stimulation. There is no fun- sufficient numbers should provide information that will damental difference in the etiologies, clinical picture, better define the optimal prognostic role of EEG, evoked ␣ ⌰ and prognostic importance among -, -, and a mixture potentials, and various clinical factors individually and in ␣ ⌰ of -/ -patterns, and they can be combined for practical combination. purposes (Synek and Synek, 1984, 1987; Young et al., 1994). Patients usually remain vegetative or die without recovering conscious awareness. However, again, some References patients may make a satisfactory recovery. Thus, finding ␣ Aoki Y, Lombroso CT. Prognostic value of electroencephalography in an -pattern coma after cardiac arrest is not in itself Reye’s syndrome. Neurology 1973;23:333–43. definitive prognostically (Fig. 8, top), but it should be Bassetti C, Bomio F, Mathis J, Hess CW. Early prognosis in coma after regarded as a “transitional pattern” (Jorgensen and Holm, cardiac arrest: a prospective clinical electrophysiological, and biochemical study of 60 patients. J Neurol Neurosurg Psychiatry 1998). By 6 days after arrest, the pattern usually evolves 1996;61:610–5. into a prognostically definitive pattern in more than 90% Bergamasco B, Bergamini L, Doriguzzi T, Fabiani D. EEG sleep of patients (Young et al., 1997). Those who die without patterns as a prognostic criterion in post-traumatic coma. Electro- encephalogr Clin Neurophysiol 1968;24:374–7. recovering consciousness develop a burst–suppression Bickford RG, Butt HR. Hepatic coma: the electroencephalographic pattern without reactivity, and those who survive have pattern. J Clin Invest 1955;34:790–9.

J Clin Neurophysiol, Vol. 17, No. 5, 2000 EEG IN COMA 483

Binnie CD, Prior PF. Electroencephalography. J Neurol Neurosurg Ellis JM, Lee SI. Acute prolonged confusion in the elderly as an ictal Psychiatry 1994;57:1308–18. state. Epilepsia 1978;19:119–28. Binnie CD, Prior PF, Lloyd DSL, Scott DF, Margerison JH. Electro- Fagan KJ, Lee SI. Prolonged confusion following convulsions due to encephalographic prediction of fatal anoxic brain damage after generalized nonconvulsive status epilepticus. Neurology 1990;40: resuscitation from cardiac arrest. BMJ 1970;4:265–8. 1689–94. Bolton CF, Young GB. Neurological complications of renal disease. Fishgold H, Mathis P. Obnubilations comas et stupeurs: e´tudes e´lec- Boston: Butterworths, 1990:47, 58–69. troencephalographiques. Electroencephalogr Clin Neurophysiol Brenner RP, Schwartzman RJ, Richey ET. Prognostic significance of 1959;11(suppl 11):27–68. episodic low amplitude or relatively isoelectric EEG patterns. Dis Fournet A, Lanternier M. Constations e´lectroencephalographique dans Nerv Syst 1975;36:582–7. 17 cas d’encephalopathie Gayet–Wernicke. Rev Neurol (Paris) Brierley JB, Adams JH, Graham DI, Simpson JA. Neocortical death 1956;95:644–5. after cardiac arrest. Lancet 1971;2:560–5. Gaches J. Activite´s periodiques en EEG. Rev EEG Neurophysiol Cant BR, Shaw NA. Monitoring by compressed spectral array in 1971;1:9–33. prolonged coma. Neurology 1984;34:35–9. Glaser G. Metabolic, endocrine and toxic diseases. In: Remond A, ed. Celesia GG, Grigg MM, Ross E. Generalized status myoclonus in acute Handbook of electroencephalography and clinical neurophysiol- anoxic and toxic–metabolic encephalopathies. Arch Neurol 1988; ogy. Amsterdam: Elsevier, 1972–1977:15C-3–5. 45:781–4. Gloor P, Kalabay O, Giard N. The electroencephalogram in diffuse Chase TN, Moretti L, Prensky AL. Clinical and electroencephalo- encephalopathies. Brain 1968;91:779–802. graphic manifestations of vascular disease of the pons. Neurology Goldie WD, Chiappa KH, Young RR, Brooks EB. Brainstem auditory 1968;18:357–68. and somatosensory evoked responses in brain death. Neurology Chatrian GE, White LE, Daly D. Electroencephalographic patterns 1981;31:248–56. resembling those of sleep in certain comatose states after injuries Grigg–Damberger MM, Coker SB, Halsey CL, Anderson CL. Neonatal to the head. Electroencephalogr Clin Neurophysiol 1963;15:272– burst–suppression: its developmental significance. Pediatr Neurol 80. 1989;5:84–92. Chen R, Bolton CF, Young GB. Prediction of outcome in patients with Gupta PC, Seth P. Periodic complexes in herpes simplex encephalitis. anoxic coma: a clinical and electrophysiological study. Crit Care A clinical and experimental study. Electroencephalogr Clin Neu- Med 1996;24:672–8. rophysiol 1973;35:67–74. Chequer RS, Tharp BR, Dreimane D, et al. Prognostic value of EEG in Haider I, Matthew H, Oswald I. Electroencephalographic changes in neonatal meningitis. Pediatr Neurol 1992;8:417–22. acute drug poisoning. Electroencephalogr Clin Neurophysiol Chiappa KH, Hoch DB. Electrophysiologic monitoring. In: Ropper 1971;30:23–31. AH, ed. Neurological and neurosurgical intensive care. New Hansotia P, Gottschalk P, Green P, Zais D. Spindle coma: incidence, York: Raven Press, 1993:147–83. clinicopathological correlates and prognostic value. Neurology Chiappa KH, Ropper AH. Long-term electrophysiologic monitoring of 1981;31:83–7. patients in the neurology intensive care unit. Semin Neurol 1984; Harner R, Naquet R. Altered states of consciousness, coma, cerebral 4:469–79. death. In: Remond A, ed. Handbook of electroencephalography Cloche R, Desmonts JM, Hennetier G. Morphological and evolutive and clinical neurophysiology. Vol. 12. Amsterdam: Elsevier, aspects of electroencephalographic findings in acute cerebral an- 1975:37–46. oxia. (42 cases). Aneth Analg Re´anim 1968;25:579–80. Hauser WA. Status epilepticus. Etiology and neurological sequelae. Courjon J, Scherzer E. Traumatic disorders. In: Remond A, Magnus O, Adv Neurol 1983;34:3–14. Courjon J, eds. Handbook of electroencephalography and clinical Hockaday JM, Potts F, Epstein E, Bonazzi A, Schwabb RS. Electro- neurophysiology. Amsterdam: Elsevier, 1972:8–95. encephalographic changes in acute cerebral anoxia from cardiac or Creutzfeldt OD, Meisch JJ. Changes in cortical neuronal activity and respiratory arrest. Electroencephalogr Clin Neurophysiol 1965; EEG changes during hypoglycemia. In: Hernandez–Peon E, ed. 18:575–86. The physiological basis of mental activity. Electroencephalogr Holmes GL, Lombroso CT. Prognostic value of background patterns in Clin Neurophysiol 1963;15(suppl 24):158–71. neonatal EEG. J Clin Neurophysiol 1993;10:323–52. da Silva O, Alexandrou D, Knoppert D, Young GB. Seizure and Hughes JR. Limitations of the EEG in coma and brain death. AnnNY electrographic changes in the newborn period induced by opiates Acad Sci 1978;315:121–36. and corrected by naloxone infusion. J Perinatol 1999;19:120–3. Hughes JR, Boshes B, Leesma J. Electroclinical and pathological Dawson VL, Dawson TM, Bartley DA, Uhl GR, Snyder SH. Mecha- correlations in comatose patients. Clin EEG 1976;7:13–30. nisms of nitric oxide-mediated neurotoxicity in primary brain Hughes JR, Schreeder MT. EEG in dialysis encephalopathy. Neurology cultures. J Neurosci 1993;13:2651–61. 1980;30:1148–54. Dawson VL, Dawson TM, London ED, Bredt DS, Snyder SH. Nitric Illis LS, Taylor FM. The electroencephalogram in herpes simplex oxide mediates glutamate neurotoxicity in primary cortical cul- encephalitis. Lancet 1972;1:718–21. tures. Proc Natl Acad SciUSA1988;88:6368–71. International Federation of Clinical Neurophysiology. Recommenda- DeLorenzo RJ, Waterhouse EJ, Towne AR, et al. Persistent noncon- tions for the practice of clinical neurophysiology. Amsterdam: vulsive status epilepticus after the control of convulsive status Elsevier, 1983:13. epilepticus. Epilepsia 1998;39:833–40. Jordan K. Continuous EEG monitoring (CEEG) in the neurological Dow RS. The electroencephalographic findings in acute intermittent ICU. Neurology 1990;40(suppl 1):180. Abstract. porphyria. Electroencephalogr Clin Neurophysiol 1961;13:425– Jordan KG. Nonconvulsive seizures (NCS) and nonconvulsive status 37. epilepticus (NCSE) detected by continuous EEG monitoring in the Dow R, Ulett G, Raaf J. EEG studies immediately following head neuro ICU (NICU-CEEG). Neurology 1992;42(suppl 3):194–5. injury. Am J Psychiatry 1944;101:174–83. Jordan KG. Continuous EEG and evoked potential monitoring in the Dreyfus–Bresac C, Mises R. E´ tude EEG de 28 addisoniens. Electro- neuroscience intensive care unit. J Clin Neurophysiol 1993;10: encephalogr Clin Neurophysiol 1953;5:133–4. 445–75. Dusser A, Navelet Y, Devictor D, Landrieu P. Short- and long-term Jorgensen EO, Holm S. The natural course of neurological recovery prognostic value of the electroencephalogram in children with following cardiopulmonary resuscitation. Resuscitation 1998;36: severe head injury. Electroencephalogr Clin Neurophysiol 1989; 111–22. 73:85–93. Kaplan PW, Genoud D, Ho TW, Jallon P. Etiology, neurologic corre-

J Clin Neurophysiol, Vol. 17, No. 5, 2000 484 G. B. YOUNG

lations, and prognosis in alpha coma. Clin Neurophysiol 1999; the burst–suppression EEG pattern. J Clin Neurophysiol 1997;14: 110:205–13. 150–3. Karnaze DS, Bickford RG. Triphasic waves: a reassessment of their Regan PF, Browne–Mayers AN. Electroencephalography: frequency significance. Electroencephalogr Clin Neurophysiol 1984;57:193–8. analysis and consciousness. A correlation during insulin-induced Karnaze DS, Marshall LF, Bickford RG. EEG monitoring of clinical hypoglycemia. J Nerv Ment Dis 1956;124:142–7. coma: the compressed spectral array. Neurology 1982;32:289–92. Rothstein TL, Thomas EM, Sumi SM. Predicting coma in hypoxic– Kennedy J, Parbhoo SP, MacGillivray B, Sherlock S. Effect of extra- ischemic coma. Electroencephalogr Clin Neurophysiol 1991;79: corporeal liver perfusion on the electroencephalogram on coma 101–7. due to acute liver failure. Q J Med 1973;42:549–61. Rumpl E. Craniocerebral trauma. In: Neidermeyer E, Lopes da Silva F, Kirino T. Delayed neuronal death in the gerbil hippocampus following eds. Electroencephalography: basic principles, clinical applica- ischemia. Brain Res 1982;239:57–69. tions and related fields. Baltimore: Urban and Schwarzenberg, Kuroiwa Y, Celesia GG. Clinical significance of periodic EEG pat- 1987:347–67. terns. Arch Neurol 1980;37:15–9. Rumpl E, Lorenzi E, Hackl JM, Gerstenbrand F, Hengl W. The EEG at Lindenberg R. Anoxia does not produce brain damage. Jpn J Legal different stages of acute secondary traumatic midbrain and bulbar Med 1982;36:38–57. brain syndromes. Electroencephalogr Clin Neurophysiol 1979;46: Loeb C, Rosadini G, Poggio G. EEG during coma. Neurology 1959; 487–97. 9:610–8. Rumpl E, Prugger M, Bauer G, et al. Incidence and prognostic value of Matsuo F. EEG features of the apallic syndrome resulting from cerebral spindles in post-traumatic coma. Electroencephalogr Clin Neuro- anoxia. Electroencephalogr Clin Neurophysiol 1985;61:113–22. physiol 1983;56:420–9. McCarty GE, Marshall SW. Transient eye opening associated with Sadove MS, Becka D, Gibbs FA. Electroencephalography for anes- postanoxic EEG suppression–burst pattern. Arch Neurol 1981;38: thesiologists and surgeons. Philadelphia: Lippincott, 1967:17–8. 754–6. Sandroni C, Barelli A, Piazza O, et al. What is the best test to predict Mo¨ller M, Holm B, Sindrup E, Nielsen BL. Electroencephalographic outcome after prolonged cardiac arrest? Eur J Emerg Med 1995; prediction of anoxic brain damage after resuscitation from cardiac 2:33–7. arrest in patients with acute myocardial infarction. Acta Med Scollo–Lavizzari G, Bassetti C. Prognostic value of EEG in post- Scand 1978;203:31–7. anoxic coma after cardiac arrest. Eur Neurol 1987;26:161–70. Mori E, Tsuruta H. Transient eye opening with EEG burst–suppression Shagass C, Rowsell PN. Serial electroencephalographic and clinical pattern in postanoxic encephalopathy. Arch Neurol 1983;40:189– studies in a case of prolonged insulin coma. Arch Neurol Psychi- 90. atry 1954;72:705–11. Moure JMB. The electroencephalogram in hypercalcemia. Arch Neurol Sharbrough FW. EEG in different states of consciousness, purposeful 1967;17:34–51. responsiveness, including comatose and vegetative states. Pre- Niedermeyer E. Cerebrovascular disorders and the EEG. In: Niederm- sented at the American EEG Society. June 9, 1981. eyer E, Lopes da Silva F, eds. Electroencephalography: basic Silverman D. Retrospective study of the EEG in coma. EEG Clin principles, clinical applications and related fields. 2nd ed. Balti- Neurophysiol 1963;15:486–503. more: Urban and Schwarzenberg, 1987:275–99. Simon RP, Aminoff MJ. Electrographic status epilepticus in fatal Nieman EA. The electroencephalogram in myxoedema coma. Clinical anoxic coma. Ann Neurol 1986;20:351–5. and electroencephalographic study of three cases. BMJ 1959;1: Skoldenberg B, Forsgren M, Alestig K, et al. Acyclovir versus vidara- 1204–8. bine in herpes simplex encephalitis. Lancet 1984;2:707–11. Nillson BY, Olsson Y, Sourander P. Electroencephalographic and Steriade M, Gloor P, Llinas RR, Lopes da Silva FH, Mesulam M-M. histopathological changes resembling Jakob–Creutzfeldt disease Report of IFCN Committee on basic mechanisms: basic mecha- after transient cerebral ischemia due to cardiac arrest. Acta Neurol nisms of cerebral rhythmic activities. Electroencephalogr Clin Scand 1972;48:418–26. Neurophysiol 1990;76:481–508. Nuwer MR. Quantitative EEG: I. Techniques and problems of fre- Stockard JJ, Bickford RG. The neurophysiology of anesthesia. In: quency analysis and topographic mapping. J Clin Neurophysiol Gordon E, ed. A basis and practice of neuroanesthesia. Amster- 1988a;5:1–43. dam: Excerpta Medica, 1975:3–46. Nuwer MR. Quantitative EEG: II. Frequency analysis and topographic Stockard JJ, Bickford RG, Aung MH. The EEG in traumatic brain mapping in clinical settings. J Clin Neurophysiol 1988b;5:45–85. injury. In: Vinken PJ, Bruyn GW, eds. Handbook of clinical Pampiglione G. Some EEG observations during unconsciousness, soon neurology. Amsterdam: North Holland Publishing, 1975;23:317– after resuscitation. Acta Neurochir 1964;12:282–8. 67. Pampiglioni G, Harden A. Resuscitation after cardiopulmonary arrest. Sundaram MBM, Blume WT. Triphasic waves: clinical correlates and Lancet 1968;1:1261–4. morphology. Can J Neurol Sci 1987;14:136–40. Parsons–Smith BG, Summerskill WHJ, Dawson AM, Sherlock S. The Synek VM. Prognostically important coma patterns in diffuse anoxic electroencephalograph in liver disease. Lancet 1957;2:867–71. and traumatic encephalopathies in adults. J Clin Neurophysiol Paschen W, Bengtsson F, Rohn G, Bonnekoh P, et al. Cerebral 1988;5:161–74. polyamine metabolism in reversible hypoglycemia of rat: relation- Synek VM, Synek BJL. “Theta pattern coma”: a variant on alpha ship to energy metabolism and calcium. J Neurochem 1991;57: pattern coma. Clin EEG 1984;15:116–21. 204–15. Synek VM, Synek BJL. “Theta pattern coma” occurring in younger Pearigen P, Gwinn R, Simon RP. The effects of in vivo hypoxia on adults. Clin Electroencephalogr 1987;18:54–60. brain injury. Brain Res 1996;725:184–91. Treiman DM, Walton NY, Kendrick C. A prospective sequence of Prior PF. The EEG in cerebral anoxia. Amsterdam: Excerpta Medica, electroencephalographic changes during generalized convulsive 1973a:244–54. status epilepticus. Epilepsy Res 1990;5:49–60. Prior PF. The EEG in cerebral anoxia. Amsterdam: Excerpta Medica, Turrell RC, Roseman E. Electroencephalographic studies of the en- 1973b:129. cephalopathies. IV. Serial studies in meningococcal meningitis. Rae–Grant AD, Barbour PJ, Reed J. Development of a novel EEG Arch Neurol Psychiatry 1955;73:141–8. rating scale for head injury using dichotomous variables. Electro- Van der Drift JHA, Kok NKD. The EEG in cerebrovascular disorders encephalogr Clin Neurophysiol 1991;79:349–57. in relation to pathology. In: Gastaut H, Meyer JS, eds. Handbook Rath SA, Klein HJ. Current applications of the EEG in the comatose of electroencephalography and clinical neurophysiology. Amster- patient. Am J EEG Technol 1991;31:65–81. dam: Elsevier, 1972:180–96. Reeves AL, Westmoreland BF, Klass DW. Clinical accompaniments of Van Ness PC. Pentobarbital and EEG burst suppression in treatment of

J Clin Neurophysiol, Vol. 17, No. 5, 2000 EEG IN COMA 485

status epilepticus refractory to benzodiazepines and phenytoin. Young GB, Blume WT, Campbell VC, et al. Alpha, alpha–theta and Epilepsia 1990;31:61–7. theta coma: a clinical outcome study utilizing serial recordings. Vas GA, Cracco JB. Diffuse encephalopathies. In: Daly DD, Pedley Electroencephalogr Clin Neurophysiol 1994;91:93–9. TA, eds. Current practice of clinical electroencephalography. Young GB, Bolton CF, Austin TW, Archibald Y, Wells GA. The New York: Raven Press, 1990:371–99. encephalopathy associated with septic illness. Clin Invest Med Vespa PM, Nuwer MR, Nenov V, et al. Increased incidence and impact 1990a;13:297–304. of nonconvulsive and convulsive seizures after traumatic brain Young GB, Gilbert JJ, Zochodne DW. The significance of myoclonic injury as detected by continuous electroencephalographic EEG status epilepticus in post-anoxic coma. Neurology 1990b;40: monitoring. J Neurosurg 1999;91(electronic pages):1–19. 1843–8. Westmoreland B, Klass DW, Sharbrough FW, Reagan TJ. “Alpha Young GB, Jordan KG, Doig GS. An assessment of nonconvulsive coma.” Electroencephalographic, clinical, pathologic and etiolog- seizures in the intensive care unit using continuous EEG monitor- ical correlations. Arch Neurol 1975;32:713–8. ing: an investigation of variables associated with mortality. Neu- Wijdicks EFM, Parisi JE, Sharbrough FW. Prognostic value of myo- rology 1996;47:83–9. clonic status epilepticus in comatose survivors of cardiac arrest. Young GB, Kreeft JH, McLachlan RS, Demelo J. EEG and clinical Ann Neurol 1994;35:239–43. associations with mortality in comatose patients in a general Williams D. The EEG in acute head injuries. J Neurol Psychiatry intensive care unit. J Clin Neurophysiol 1999;16:354–60. 1941a;4:107–30. Young GB, Leung LS, Campbell V, DeMelo J, Schieven J, Tilsworth Williams D. The EEG in post-traumatic states. J Neurol Psychiatry R. The electroencephalogram in metabolic/toxic coma. Am J EEG 1941b;4:131–46. Technol 1992;32:243–59. Xia Y, Dawson VL, Dawson TM, Snyder SH, Zweier JL. Nitric oxide Young GB, McLachlan RS, Kreeft JH, DeMelo JD. An electroencepha- synthetase generates superoxide and nitric oxide in arginine- lographic classification system for coma. Can J Neurol Sci 1997; depleted cells leading to peroynitite-mediated cellular injury. Proc 24:320–5. Natl Acad SciUSA1996;93:6770–4. Zandbergen ED, de Haan RJ, Stoutenbeek CP, et al. Systematic review Yamada T, Young S, Kimura J. Significance of positive spike bursts in of early prediction of poor outcome in anoxic–ischaemic coma. Reye’s syndrome. Arch Neurol 1977;34:376–80. Lancet 1998;352:1808–12. Yamashita S, Morinaga T, Ohgo S, et al. Prognostic value of electro- Zaret BS. Prognostic and neurophysiological implications of concur- encephalogram (EEG) in anoxic encephalopathy after cardiopul- rent burst suppression and alpha patterns in the EEG of post- monary resuscitation: relationship among anoxic period, EEG anoxic coma. Electroencephalogr Clin Neurophysiol 1985;61: grading and outcome. Intern Med 1995;34:71–6. 199–209.

J Clin Neurophysiol, Vol. 17, No. 5, 2000