Chapter 64. Seizures in the Intensive Care Unit

• Seizures are a relatively common occurrence in the intensive care unit, but may be difficult to recognize.
• Seizures that persist longer than 5 to 7 minutes should be treated to prevent progression to status epilepticus.
• Three major factors determine outcome in status epilepticus: type of seizure, cause, and duration.
• Electroencephalographic (EEG) monitoring to titrate therapy should be implemented in seizing patients who do not awaken promptly after institution of antiepileptics, even if tonic-clonic motor activity resolves.
• Lorazepam is a preferred agent for initial treatment, followed by consideration of additional agents for long-term management or to “break” status epilepticus.
• Patients with refractory status epilepticus require intubation, mechanical ventilation, and aggressive treatment with antiepileptics titrated to the EEG.
• The underlying cause of the seizure disorder must be sought in tandem with treatment of the seizure disorder itself.

Seizures are a relatively common occurrence in the ICU, complicating the course of about 3% of adult intensive care unit patients admitted for nonneurologic conditions.1 Status epilepticus (SE) may be the primary indication for admission, or it may occur in any ICU patient during a critical illness. A seizure may be the first indication of a central nervous system (CNS) complication or the result of overwhelming systemic disease. Seizures in the setting of critical illness are often difficult to recognize and require a complex diagnostic and management strategy. Delay in recognition and treatment of seizures is associated with increased mortality;2 thus the rapid diagnosis of this disorder is mandatory.

Status epilepticus refers to a protracted seizure episode or multiple frequent seizures lasting 30 minutes or longer. Although conventional definitions of SE have used this time window, clinicians should recognize that most seizures will terminate spontaneously within a few minutes.3 Recent data suggest that only half of seizure episodes lasting 10 to 29 minutes will stop spontaneously.4 Therefore seizures that persist longer than 5 to 7 minutes should be treated to prevent progression to SE.5

Limited data are available on the epidemiology of seizures in the ICU. A 10-year retrospective study of all ICU patients with seizures at the Mayo Clinic revealed that 7 patients had seizures per 1000 ICU admissions.5 Our 2-year prospective study of medical ICU patients identified 35 with seizures per 1000 admissions.1 The incidence of generalized convulsive SE (GCSE) in the United States is estimated to be up to 195,000 episodes per year,6 but it is unknown how many of these patients require care in an ICU. The incidence of SE in the elderly is almost twice that of the general population.7 Eight percent of hospitalized comatose patients in a recent series were found to be in electrographic status epilepticus.8 Seizures are probably even more frequent in the pediatric ICU, as children in the first year of life have the highest incidence of SE of any age group studied.5

Table 64-1 summarizes the most common causes of SE in adults and children in the community. An analysis of 204 cases of SE in Virginia revealed that the primary etiology in children was infection with fever, followed by remote symptomatic epilepsy, and subtherapeutic levels of anticonvulsant drugs. In adults, cerebrovascular disease and low antiepileptic drug levels were the most prevalent causes.5 A recent study from Brazil found anticonvulsant noncompliance to be the main cause of SE in patients with a prior history of epilepsy, and CNS infection, stroke, and metabolic disturbances predominated in the group without previous seizures.9 A prospective study of neurologic complications in medical ICU patients determined that two thirds of patients had a vascular, infectious, or neoplastic explanation for their seizures;1 metabolic and toxic etiologies are common in the ICU as well. A review of 100 cases of nonconvulsive SE (NCSE) demonstrated that 14% were due to acute neurologic events, 28% due to acute systemic causes, and 31% due to epilepsy, with the remainder due to multiple causes or a cryptogenic etiology.10

A prospective study of neurologic complications in medical ICU patients showed that having one seizure in the ICU doubled mortality.1 Up to 61% of patients developing SE during hospitalization die.11 SE in and of itself confers a mortality rate of 26% to adults older than 16 years and 38% to those 60 years and older.5 Multiple reports corroborate an especially poor outcome in the elderly.8,12 The mortality rate of SE in children is 3% in the general population and 6% in the ICU.13

Three major factors determine outcome in SE: the type of SE, the cause, and the duration. Myoclonic SE following an anoxic episode carries a very poor prognosis for survival. Complex partial SE (CPSE) can produce limbic system damage, usually manifested as a memory disturbance. The mortality of patients with NCSE has been reported between 18%9 and 57%,14 and correlates with the underlying etiology, severity of impairment of mental status, and the development of acute complications (especially respiratory failure and infection). Causes associated with increased mortality included anoxia, intracranial hemorrhage, tumor, infection, and trauma. SE in the setting of acute ischemic stroke has a very high mortality, approaching 35%.15 Prolonged seizure duration is a negative prognostic factor.16 A study of 253 adult SE patients showed a greater than tenfold increase in mortality rate associated with seizures lasting ≥60 minutes compared with those lasting 30 to 59 minutes.17

The International League Against Epilepsy's (ILAE) classification of seizures is generally accepted. The system allows classification on the basis of clinical criteria without inferring cause. Knowledge of interictal or ictal electroencephalographic (EEG) findings is not necessary to classify seizures except for absence seizures, which are not likely to be a problem in the ICU. The classification system divides seizures into two types: partial, which have a focal or localized onset, and generalized, in which the cortex of both cerebral hemispheres is involved simultaneously at onset. Partial seizures can further be categorized as simple, in which consciousness remains intact throughout the event, or complex, in which consciousness is disrupted or altered (but not lost), often resulting in amnesia for the event. Seizures that start locally and then spread to involve the entire cortex are termed secondary generalized. Generalized seizures are of two types: convulsive, in which tonic, clonic, or myoclonic movements are prominent, and nonconvulsive, in which a patient has an altered level of consciousness with or without very subtle motor manifestations.

The clinical manifestation of partial seizures varies with the location of their onset. Motor seizures are usually due to a lesion in the contralateral frontal lobe. Deviation of eyes and head toward the irritative focus is often seen at the onset of seizure activity and is termed versive movement. Careful observation of the direction of this initial movement provides important diagnostic information regarding the location of brain pathology. Muscle contractions may be localized to a small region, such as the face or fingers, or be more extensive, involving the entire hemibody. Movements are usually tonic or clonic, but dystonic posturing is also common. Sensory seizures can be primarily auditory, somatosensory, visual, or consist of vague visceral sensations. Patients with complex partial seizures may demonstrate any combination of the above symptoms and have associated motor automatisms, such as lip smacking or swallowing.

Generalized convulsive seizures are usually of the tonic-clonic type. During the tonic phase, initial extension of the trunk is followed by extension of the arms, legs, neck, and back. The respiratory muscles may be involved in the tonic spasm, resulting in cyanosis and decreased oxygen saturation if the tonic phase is long enough, although this is rare. The clonic phase follows and is manifest by repetitive muscle contractions. Fixed and dilated pupils, tachycardia, and hypertension are well described during tonic-clonic seizures. Incontinence usually follows termination of the seizure. The frequency of the clonus eventually wanes and respiration commences when the seizure stops. Patients may initially be deeply comatose but should begin to regain consciousness within 15 to 20 minutes.

Status epilepticus refers to prolonged or serial seizures without interictal resumption of baseline mental status. Refractory SE refers to SE that is resistant to treatment with first-line measures and requires more aggressive therapy. Further description of specific treatment modalities will be reviewed below. Epilepsia partialis continua is a special type of focal motor epilepsy that consists of near constant muscle contractions of a specific muscle group. These movements can last for months or years without generalizing.

There are theoretically as many different types of SE as there are seizures, since SE is a prolonged seizure. However, SE cannot be classified in exactly the same manner as individual seizures, because seizures are discrete time-limited events with symptomatology restricted to the brief duration of their occurrence. SE, on the other hand, can evolve over time and therefore can have a symptomatology that may encompass more than one seizure type. Furthermore, NCSE can have similar signs and symptoms with different EEG signatures and etiologies. The simplest classification divides SE into generalized convulsive SE and nonconvulsive SE, depending on whether convulsive movements are present. Since NCSE includes everything that is not convulsive, it describes a wide variety of clinical entities and scenarios.

The conventional method of subcategorizing NCSE is to divide it into absence SE and complex partial SE. This works well for patients with a previous history of epilepsy. In this context, absence SE denotes confusion, typically mild, in a patient with generalized, approximately 3-Hz spike-wave discharges on EEG and a history of generalized epilepsy. Complex partial SE denotes confusion, typically waxing and waning, or recurrent complex partial seizures associated with focal seizures in a patient with focal epilepsy. As defined herein, both types of NCSE imply that the encephalopathy is due to seizure activity. Historically, NCSE was labeled “absence” type if generalized EEG changes were found and “complex partial” if focal EEG changes were found, regardless of whether a history of epilepsy was present.

Many patients with NCSE do not have a history of epilepsy and do not fit into the conventional categorization elaborated above. For example, in a retrospective study of NCSE, we did not find any association between EEG findings and mortality,9 emphasizing that this categorization is not very useful. This is particularly a problem in ICU patients in whom there are typically numerous factors contributing to encephalopathy. This nosologic uncertainty has given rise to several terms to describe NCSE arising in the ICU, including ICU status, subtle generalized convulsive status epilepticus, EEG status, and status in the critically ill. An important aspect of ICU status is that encephalopathy often has other causes in addition to the seizure activity.

NCSE is of particular importance to the intensivist when it occurs as a sequela of inadequately treated GCSE. After prolonged generalized convulsions, visible motor activity may stop, but the electrochemical seizure continues. Patients who do not start to awaken after 20 minutes should be assumed to have entered NCSE. NCSE following GCSE is a dangerous problem because the destructive effects of SE continue even without obvious motor activity. NCSE in this setting demands emergent treatment guided by electroencephalographic monitoring to prevent further cerebral damage since there are no clear clinical criteria to indicate whether therapy is effective.

NCSE can occur as a late stage of convulsive SE from any etiology, or as an initial form of SE from another cause. Failure to recognize NCSE is common in patients presenting with nonspecific neurobehavioral abnormalities, such as delirium, lethargy, bizarre behavior, cataplexy or mutism.18 A high level of suspicion for this disorder should be maintained in patients with unexplained alteration in level of consciousness or cognition who are admitted to the ICU.

Two special circumstances with which the intensivist should be familiar are myoclonus and febrile seizures. Brief, shock-like, involuntary muscle contractions constitute myoclonus. Myoclonic jerks are arrhythmic, of variable amplitude, and involve both small and large muscles. In patients with postanoxic coma, myoclonus may be continuous or evoked by stimuli such a noise or touch. While this disorder has been associated with epileptiform discharges in the EEG,19 not all episodes of myoclonus are epileptic; an EEG can clarify whether it is epileptic in individual cases. Postanoxic myoclonus also occurs in patients who have regained consciousness (the Lance-Adams syndrome); in this setting the myoclonus is probably of cerebellar origin and is not a seizure. Febrile seizures are specific to young children and are usually generalized motor convulsions that occur in association with fever, typically as the temperature is rising. These seizures should not be confused with those that transpire in the setting of fever secondary to infection of the nervous system. Febrile seizures are usually brief, but can be prolonged and recurrent, prompting admission to an ICU.

Clinical judgment is required to classify seizures in the ICU. Patients in whom consciousness has already been altered by drugs, hypotension, sepsis, or intracranial pathology may be difficult to classify using only the ILAE classification because it depends heavily on whether the seizure activity has altered consciousness. However, focal seizure activity on EEG or focal neurologic deficits often helps determine whether the seizure is focal or generalized in onset. The ILAE continues to work towards revising and updating the current classification system. The goal is a multi-axis diagnostic scheme that incorporates anatomic, etiologic, therapeutic, and prognostic implications. For the most recent information regarding this ongoing project, refer to http://www.epilepsy.org.20

The systemic and cerebral pathophysiology of GCSE can be divided into early and late phases.21 The early phase of systemic manifestations results from an adrenergic surge and excessive muscle activity.22 The adrenergic surge causes tachycardia, hypertension, and hyperglycemia. These are augmented by extreme muscle activity that causes hyperthermia and acidosis and can lead to muscle breakdown, rhabdomyolysis, and secondary acute renal failure. This stage is generally well compensated by homeostatic mechanisms so that the excessive demands are met with increased supply or other compensatory mechanisms.

Most facets of GCSE begin to slow down late in GCSE, so only a rare patient continues to have continuous convulsive motor activity for more than an hour. Cessation of continuous motor activity would seem to be a beneficial turn of events, but this is actually coincident with a sharp increase in mortality and in complications. Although systemic factors such as heart rate and blood pressure normalize, they may be inadequate to meet increased demands of intermittent convulsions or electrographic seizure activity, even in the absence of convulsions. Thus mortality increases dramatically for SE lasting longer than an hour.16 Death may result from a number of causes, but in a prospective study of cardiovascular changes during GCSE 58% of patients had potentially fatal arrhythmias.23 Patients with atherosclerotic cardiovascular risk factors may have a gradual deterioration in hemodynamic parameters as their cardiovascular reserve is expended, while other patients decline acutely, presumably from arrhythmias.24

SE may cause neuronal injury in surviving patients. Some neuronal injury is caused by systemic factors; for example, hyperthermia causes cerebellar neuronal injury. However, neuronal injury continues during electrographic SE, even without motor manifestations or when physiologic parameters are held in the normal range. This is illustrated most clearly in experimental GCSE. Neuronal injury is prominent in the hippocampus and temporal lobe in primates with experimental GCSE. The injury persisted even when muscle activity was eliminated by paralysis, and pulse, blood pressure, temperature, and oxygenation were kept normal.

Neuronal injury during SE is due in part to the excitotoxic effects of glutamate-mediated neuronal seizure activity.20 Glutamate is the most common excitatory neurotransmitter in the brain. It mediates transfer of information between neurons under normal conditions via several receptors. However, glutamate excessively activates the N-methyl-D-aspartate (NMDA) subtype of receptor in the robust conditions of SE. NMDA receptors have a limited normal function, but during SE they cause very prolonged depolarization of neurons. This results in intracellular accumulation of calcium and other cellular changes that result in both immediate and delayed cell death.20

There are two important clinical implications of the pathophysiology of SE. First, neuronal injury continues during electrical SE even after control of motor manifestations. Therefore it is imperative to exclude ongoing seizure activity if patients are pharmacologically paralyzed after GCSE or do not awaken soon after motor activity stops. These circumstances require EEG monitoring to exclude ongoing seizure activity. Second, pharmacologic treatment is aimed at augmenting inhibition, via drugs that act on γ-aminobutyric acid (GABA), such as barbiturates and benzodiazepines. There will probably also be a role for NMDA antagonists. Ketamine is the only currently available NMDA antagonist, but others are likely to be helpful in the future.

Three problems complicate seizure recognition in the ICU: (1) occurrence of complex partial or nonconvulsive seizures in the setting of depressed consciousness, (2) masking of seizures by pharmacologically-induced paralysis or sedation, and (3) misinterpretation of other abnormal movements as seizures. ICU patients often have decreased levels of consciousness in the absence of seizures that are ascribable to the underlying disease and its complications.25 An encephalopathic patient may be unable to appreciate or report symptoms of seizure. Fluctuations in mental status are frequently subtle and may go unrecognized by staff. A decline in baseline alertness may reflect a seizure; an EEG may be required to confirm that one has occurred.

Patients receiving neuromuscular junction blocking agents do not manifest the motor signs of seizures. Patients with refractory intracranial hypertension, severe pulmonary disease, or other critical illnesses may be both paralyzed and sedated, making identification of seizures particularly challenging. Tachycardia and hypertension are signs of seizure that can be misinterpreted as evidence of inadequate sedation. Continuous EEG monitoring is warranted in this population if seizures are suspected.

Patients with metabolic disturbances, anoxia, and other types of nervous system injury may demonstrate abnormal movements that can be confused with seizure. Asterixis, or flapping tremor, is a brief arrhythmic loss of tone that can appear in the setting of hepatic encephalopathy, hypercarbia, drug intoxication, or CNS pathology.26 Myoclonus in postanoxic coma has been reported in the presence18 and absence27 of epileptiform discharges. Therefore EEG is absolutely indicated in this setting to evaluate for ongoing seizures. Action myoclonus in a patient recovering from hypoxic encephalopathy is evoked during movements directed at a target, such as an examiner's finger. It is frequently associated with cerebellar ataxia and postural lapses, which when combined with myoclonus can severely impair ambulation. Myoclonus associated with etomidate is described,28 but whether it is cortically mediated remains unclear. Brain-injured patients may suffer from so-called “hypothalamic seizures.” Tetanus patients do not lose consciousness during their spasms, and describe excruciating pain associated with the sustained whole-body contractions. Psychiatric disturbances in the ICU occasionally resemble complex partial seizures. If doubt about the nature of abnormal movements persists, an EEG should be performed.

The initial approach to seizure management is the same as that for any other acute medical problem: airway, breathing, and circulation. As described above, generalized convulsive status epilepticus often causes apnea and/or poor oxygen saturation. Hypertension and tachycardia may be marked. However, respiratory and hemodynamic dysfunction is transient, and with seizure termination rapidly returns to normal. Padded tongue blades or similar items should not be placed inside the mouth; they are more likely to obstruct the airway than to preserve it. Medication to treat tachycardia and hypertension before the seizure activity stops is not warranted.

When a patient has a seizure, one has a natural tendency to try to stop the event. This leads to both diagnostic confusion and iatrogenic complications. Beyond protecting the patient from harm, very little can be done rapidly to influence the course of the seizure. The seizures of most patients stop before any medication can reach the brain in an effective concentration. Observation is the most important activity to perform when a patient has a single seizure. This is the time to collect evidence of a partial onset in order to implicate structural brain disease. The postictal examination is similarly valuable; language, motor, sensory, or reflex abnormalities after an apparently generalized seizure are evidence of focal pathology.

Seizures in ICU patients have many potential causes that must be investigated. Drugs are a major cause of seizures in critically ill patients, especially in the setting of renal or hepatic dysfunction. Theophylline can provoke seizures or SE if it has been rapidly loaded or if high concentrations of the drug occur; however, these complications can also arise with normal serum drug levels.10 Imipenem-cilastatin29 and fluoroquinolones30 have the potential to lower the seizure threshold, particularly in patients with impaired renal function. Accumulation of a metabolite of meperidine, normeperidine, causes seizures, even in patients with normal renal function. Sevoflurane, a volatile anesthetic agent, also causes electrographic and clinical seizures without a history of epilepsy or CNS pathology.31

Recreational drugs are frequently-overlooked offenders in patients presenting to the ICU. Acute cocaine or methamphetamine intoxication is characterized by a state of hypersympathetic activity followed by seizures.32Ethanol withdrawal is a common cause of seizures between 6 and 96 hours after the patient's last drink, but concomitant causes must not be overlooked. Narcotic withdrawal may produce seizures in the critically ill5 and in newborns of opioid-dependent mothers.33 Both bupropion hydrochloride34 and tricyclic antidepressants are associated with seizure in overdose and occasionally at therapeutic doses. In the absence of other clear causes for seizure, a complete toxicology screen should be performed upon admission.

Serum glucose, electrolyte concentrations, and serum osmolality should also be measured. Nonketotic hyperglycemia can precipitate both focal and generalized seizures;35,36 epilepsia partialis continua was the most common type seen in a recent series.37 Seizure activity may infrequently be the first presenting sign of diabetes mellitus. Both severe, rapidly developing hyponatremia and hypoglycemia can cause seizures. The patient's blood glucose concentration should be measured immediately upon presentation, and dextrose and thiamine administered if hypoglycemia is present. Hypocalcemia rarely causes seizures beyond the neonatal period; identifying even moderate hypocalcemia must not signal the end of the diagnostic work-up. Hypomagnesemia has an equally unwarranted reputation as the cause of seizures in malnourished alcoholic patients.

The physical examination should emphasize assessment for both global and focal abnormalities of the CNS. Evidence of cardiovascular disease or systemic infection should be sought and the skin and fundi examined closely. Particular attention should be given to the funduscopic examination of infants presenting from the community with seizures, as retinal hemorrhages may be the only evidence of brain trauma induced by child abuse (the “shaken baby syndrome”).

New-onset seizures almost always warrant brain imaging. Considering the large number of critically ill patients with neurologic pathology as a primary or contributing cause for seizures, acute brain processes must be ruled out. Computed tomography (CT) scanning is a rapid modality with which the trained clinician can detect acute blood, swelling, large tumors or abscesses, and subacute or remote ischemic strokes. With current technology, there are exceptionally few patients who cannot undergo CT scanning. Magnetic resonance imaging (MRI) is particularly helpful in detecting evidence of acute ischemic stroke, encephalitis, small tumors, subdural empyemas, and cerebral edema. Most cardiac pacemakers are a contraindication to MRI, but many other medical devices, such as inferior vena cava filters, intracranial pressure monitors, and cerebral aneurysm clips, are now manufactured using MRI-compatible material. Patients with altered mental status who need cerebrospinal fluid analysis require imaging of the brain first, to rule out a mass, swelling, or other cause of impending brain herniation. When CNS infection is suspected, empiric antibiotic treatment should be started while imaging studies are being obtained.

In contrast to the patient with a single or a few seizures, the SE patient requires simultaneous diagnostic and therapeutic efforts. Although 30 minutes of continuous or recurrent seizure activity usually defines SE, one should not stand by waiting for this period to pass to start treatment. Since most seizures in critically ill patients stop within 2 to 3 minutes, it is reasonable to start treatment after 5 minutes of continuous seizure activity or after the second or third seizure occurs without recovery between the spells.

The Electroencephalogram

Treatment for recognized SE should not be delayed to obtain an EEG, but such recognition is not always straightforward. A prospective evaluation of 164 patients demonstrated that nearly half manifested persistent electrographic seizures in the 24 hours after clinical control of convulsive SE, and 14% went into electrographic status epilepticus.38 Subclinical seizures have been observed during aggressive treatment for SE, even in patients treated with high-dose barbiturates to produce a burst-suppression pattern on EEG. These data suggest that EEG monitoring after control of convulsive SE can be essential in directing the course of treatment. Emergent EEG is necessary to exclude NCSE in those patients who do not begin to awaken soon after visible seizure activity has stopped. Patients who develop refractory SE or receive neuromuscular junction blockade require continuous EEG monitoring, since ongoing seizure activity can cause neuronal injury via excitotoxic mechanisms as outlined above.

A variety of findings may be present in the EEG, depending on the seizure type, duration, and level of pharmacologic intervention. Prospective data indicate that EEG patterns may also be helpful in determining prognosis. One study found that the presence of burst-suppression, post-SE ictal discharges, and periodic lateralized epileptiform discharges during the initial 24 hours after control of SE were statistically significantly correlated with mortality and poor outcome.39 Burst-suppression secondary to pharmacologic coma for treatment of SE must be differentiated from burst-suppression due to widespread cortical injury, or that seen as the last stage of the EEG evolution of SE. The availability of continuous paperless electroencephalographic monitoring allows for detection of seizure activity over a long period.

Isolated Seizures

Not all patients who have seizures require anticonvulsant therapy. Making the decision to administer anticonvulsants to a hospitalized patient who experiences one or a few seizures mandates consideration of a provisional cause, estimation of the likelihood of recurrence, and recognition of the utility and limitations of anticonvulsants. For example, seizures due to ethanol or other hypnosedative withdrawal do not need chronic treatment, but short-term therapy with benzodiazepines for repeated or prolonged seizures may be warranted (Table 64-2). Seizures caused by metabolic disturbances such as hyponatremia are often refractory to conventional anticonvulsant medications such as phenytoin, and are best treated with correction of the underlying disorder (benzodiazepines may be useful for seizure suppression if needed while the metabolic problem is being corrected). Seizures related to nonketotic hyperglycemia respond best to correction of hyperglycemia with insulin and rehydration.36

A patient with CNS disease who has even one seizure should receive anticonvulsant therapy because the risk of seizure recurrence is very high. However, this treatment should be reviewed before discharge. Initiating this treatment after the first unprovoked seizure may help delay the appearance of subsequent seizures,40 but probably does not influence whether epilepsy subsequently develops.41 Prophylactic therapy in patients at high risk for seizure, especially if his or her condition would be seriously complicated by a convulsion, is not unreasonable. Patients with traumatic brain injury, intracerebral hemorrhages, and subarachnoid hemorrhages are frequently placed on anticonvulsants immediately upon admission, although no prospective randomized trials have proven a positive effect on outcome.

In the ICU setting, phenytoin is often the first drug selected due to ease of administration and rapid assessment of blood levels. While the efficacy of phenytoin in the control of seizures is well established, several inherent properties of the drug limit its tolerability. In order to improve aqueous solubility, phenytoin is suspended in a highly alkaline solution that is comprised of 40% propylene glycol.42 The propylene glycol vehicle has been linked to hypotension and cardiac arrhythmias during phenytoin infusion; however, phenytoin itself may be partly responsible for hemodynamic instability. The caustic pH of the parenteral formulation can cause injection site reactions that can range from burning at the IV site to necrosis in the event of extravasation.

The phenytoin prodrug, fosphenytoin, is water soluble; therefore the parenteral formulation is more neutral than that of phenytoin and contains no organic solvents. Cardiovascular side effects were initially thought to be less common with fosphenytoin, but subsequent experience suggests that hypotension and arrhythmias may follow its infusion. Pain at the infusion site is significantly less common with fosphenytoin than with phenytoin.43 In patients without IV access, fosphenytoin can be safely administered intramuscularly. IM doses of fosphenytoin are well tolerated, require no cardiac monitoring, and are completely absorbed. Fosphenytoin is rapidly converted to phenytoin in vivo and free phenytoin levels after fosphenytoin administration are not markedly different compared to phenytoin, although the time to reach the peak level after IM administration is several hours.

A 20 mg/kg loading dose of phenytoin brings most patients to the desired concentration of 20 μg/mL (corresponding to an unbound or free concentration of 2 μg/mL). Fosphenytoin is dosed by phenytoin-equivalent units (PE); therefore no dosage adjustments are needed when converting patients from phenytoin to fosphenytoin. Fosphenytoin can be administered via intravenous infusion at rates of up to 150 mg PE/min, compared with a maximum rate of 50 mg/min for phenytoin. Both of these drug infusions should be started at a lower rate and increased as tolerated. When loading doses of fosphenytoin are given IM, two divided doses of 10 mg/kg each are recommended. After fosphenytoin administration, phenytoin concentrations should not be measured until the biologic conversion to phenytoin is complete and the drug has equilibrated throughout the body, about 2 hours after an intravenous infusion or 4 hours after an intramuscular injection of fosphenytoin. Phenytoin is approximately 90% protein bound in normal hosts, but the unbound fraction is the active component. Patients with renal or hepatic dysfunction or those taking drugs that compete for protein binding may benefit from measuring the free (unbound) serum phenytoin concentration before increasing phenytoin doses due to apparently subtherapeutic total phenytoin concentrations.

The maintenance dose for phenytoin is typically in the range of 5 to 7 mg/kg per day, but is highly variable because of individual differences in metabolism and interactions with other drugs metabolized via the cytochrome P450 system. Maintenance doses can be given either enterally or parenterally. Maintenance doses of IV or enteral liquid suspension phenytoin must be given in twice-daily divided doses since their half-life is less than 24 hours. Extended-release capsules can be given once a day. However, patients often do not tolerate more than 300 or 400 mg of phenytoin enterally in any one dose secondary to nausea. Therefore patients requiring more than this amount in capsules should usually receive divided doses.

Hypersensitivity is the major adverse effect of concern to the intensivist. This may manifest itself solely as fever, but commonly includes rash, eosinophilia, and elevated liver enzymes. Adverse reactions to phenytoin and other anticonvulsants have been reviewed elsewhere.44

Phenobarbital remains a useful anticonvulsant for those intolerant to phenytoin or those who have persistent seizures after adequate phenytoin administration. The loading IV dose is 15 to 20 mg/kg, and the target serum concentration is 20 to 40 μg/mL. The serum concentration may be altered by hepatic and renal dysfunction. Furthermore, phenobarbital can also induce P450-related metabolism, thereby affecting the metabolism of other drugs that undergo hepatic clearance. Since the usual clearance half-life of phenobarbital is about 96 hours, maintenance doses of this agent should be given once a day. A steady-state level takes about 3 weeks to become established. Sedation is the major adverse effect; allergy to the drug occurs rarely.

Carbamazepine is rarely initiated in the ICU because it is not available in parenteral form and absorption from the gastrointestinal tract is relatively slow. Carbamazepine has significant interactions with many drugs that are used in hospitalized patients, such as corticosteroids, theophylline, warfarin, and cimetidine. Adjusting blood levels of carbamazepine in the setting of polypharmacy can be unpredictable. Carbamazepine and the newer anticonvulsant oxcarbazepine can both cause hyponatremia with chronic use, probably due to a combination of the syndrome of inappropriate secretion of antidiuretic hormone (SIADH) and salt-wasting nephropathy.

Status Epilepticus

Status epilepticus is a medical emergency. While proper diagnosis of the cause is critical, the most important initial goal is to expeditiously stop the clinical and electrographic seizures. The likelihood of successfully treating SE is inversely related to the duration of seizures; the longer seizures last, the more difficult they are to terminate. Aggressive and rapid management of status epilepticus is essential to limit further neurologic and systemic complications.

The conventional agents used for first-line treatment of SE are the benzodiazepines (especially lorazepam, diazepam, and midazolam), phenytoin, and phenobarbital. The Veterans Affairs Status Epilepticus Cooperative Study Group trial compared four regimens for the initial treatment of GCSE and demonstrated that lorazepam was more efficacious than phenytoin, and easier to use than phenobarbital or phenytoin plus diazepam.45 Lorazepam has been our agent of first choice for terminating SE for many years and remains so with support from this study.

The major advantage of lorazepam over diazepam is its longer duration of action, thereby limiting seizure recurrence. Lorazepam has traditionally been given in 2-mg doses repeated at 5-minute intervals if seizures do not terminate. Since this is often an inadequate dose and valuable time passes before definitive treatment is instituted, we recommend instead a single IV dose of 0.1 mg/kg of lorazepam. If lorazepam is not available, a single IV dose of 0.15 mg/kg of diazepam is an alternative. However, another agent such as phenytoin or phenobarbital should be started immediately, as the duration of action of diazepam against SE is only about 20 minutes. In the event that IV access is unattainable, 0.2 mg/kg of midazolam administered IM will be rapidly and reliably absorbed. The use of midazolam in refractory SE will be discussed below. All benzodiazepines carry a risk of hypotension and respiratory depression. However, these are also sequelae of prolonged or inadequately treated SE. The intensivist should be prepared to intubate or use vasopressors if necessary.

Phenytoin is an effective anti-SE agent; however, the constraint on the rate of intravenous administration is of concern when treating SE. Fosphenytoin may be a better drug for use in SE since it can be loaded up to three times faster, although its 7-minute conversion half-life means that the serum phenytoin level does not reach its target much faster. Phenytoin has a long duration of action when an adequate dose is given (a 20-mg/kg dose produces a serum level above 20 mg/mL for 24 hours). Adding an additional 5 mg/kg if the initial load fails to stop SE may be useful. Intramuscular injection of fosphenytoin in SE patients may be supported by the known pharmacokinetics of this route, but it should not be considered to be acceptable therapy for SE and should be reserved for only those rare circumstances in which IV access cannot be obtained.

Phenobarbital in the management of acute SE is not routinely recommended, except when phenytoin is contraindicated. However, the Veterans Affairs study showed no difference in efficacy between lorazepam and phenobarbital as first-line agents in SE, but phenobarbital took longer to administer.45 Furthermore, in the patients that did not respond to lorazepam or phenytoin, the response rate to phenobarbital was only 2.1% (unpublished data). We therefore recommend pursuing a more definitive treatment strategy for patients who have entered refractory SE (RSE).

Refractory Status Epilepticus

Patients in RSE require doses of medication that are highly likely to cause significant respiratory suppression and hypotension. Therefore mechanical ventilation is necessary, and invasive hemodynamic monitoring is frequently required. Concomitant continuous EEG monitoring is also mandatory to confirm treatment success and monitor depth of sedation. The traditional goal of therapy is burst-suppression pattern on EEG for 12 to 24 hours prior to any attempts to wean medication. Since the available data suggest that successful treatment and improved outcome probably required seizure suppression regardless of background EEG activity,46 we recommend cessation of electrographic seizures as the goal instead.

High-dose barbiturates, most commonly pentobarbital, are extremely useful in RSE, but side effects can be severe and may limit use (Table 64-3). Hypotension can be refractory to initial resuscitative efforts, and the patient may benefit from pulmonary artery catheterization to plan fluid and vasopressor management. Pulmonary infection is common due to prolonged intubation and impaired function of both respiratory cilia and leukocytes. The intensivist must be vigilant in monitoring for infection since barbiturate-induced poikilothermia may mask fever. Despite these side effects, barbiturate anesthesia should not be rapidly discontinued if it is successful in terminating refractory SE. Continuing therapy for at least 48 hours, gradual tapering of the infusion dose, and the administration of phenobarbital during the drug taper are recommended.47 Pentobarbital is loaded at 5 to 12 mg/kg followed by an infusion of 1 to 10 mg/kg per hour. As an alternative, thiopental sodium may be given in 75- to 125-mg IV boluses followed by infusion rates of 1 to 5 mg/kg per hour. Both medications rapidly redistribute into adipose tissue; recovery of consciousness usually takes much longer after thiopental infusions than after pentobarbital. Elimination times can be greatly increased in obese patients after prolonged infusions.

Midazolam is a water-soluble benzodiazepine that has demonstrated high efficacy in refractory SE in adults and children.48–50 Midazolam is loaded at 0.2 mg/kg followed by continuous infusion of 0.05 to 2.0 mg/kg per hour. Respiratory depression may be encountered less frequently than with other hypnosedatives, but should be anticipated. Since most patients with RSE are already intubated, concern for respiratory effects should not limit use. Clinically significant hypotension is rare even at the very high doses that are often required to address tachyphylaxis.51 Sedation is quickly reversed after short-term infusions are discontinued. However, terminal half-lives of three to eight times normal have been reported with extended administration.52 In addition, prolonged elimination times have been associated with critical illness and hepatorenal dysfunction.

Propofol is an intravenous anesthetic agent that acts primarily on the GABAA receptor. Case reports documenting its efficacy in RSE are abundant, but studies examining direct comparisons with other agents have had mixed results.53–55 An initial bolus of 1 to 2 mg/kg should be followed with a maintenance infusion at 1 to 15 mg/kg per hour. Propofol is fast acting, highly lipid soluble, and has little propensity to accumulate even with prolonged infusions.47 Because of its rapid clearance, propofol should not be abruptly discontinued, but instead tapered gradually. Respiratory depression and hypotension are extremely common, especially after the initial bolus. Nutritional support must be adjusted in the setting of propofol infusion due to the high lipid and calorie content of the solution. Acidosis and rhabdomyolysis have been reported in both adults56 and children.57 Careful monitoring of creatine kinase and blood pH are prudent.

Alternative regimens with reported success in the termination of RSE include isoflurane, intravenous valproate, ketamine, and topiramate. Further studies are indicated to elucidate the role of these agents in the treatment of seizure emergencies.

Once SE is addressed, one must manage the major systemic complications of SE. Patients with GCSE should be screened for rhabdomyolysis with urine myoglobin and serum creatinine kinase (CK) determination. If myoglobinuria is present or if the CK concentration is more than 10 times the upper limit of normal, rehydration and urinary alkalinization should be instituted.58 Prolonged or severe hyperthermia should be aggressively treated with cooling blankets, ice packs, or other cooling modalities.

Special Considerations for Children

Treatment of seizures or SE in critically ill children generally parallels that for adults. Intravenous access is often more difficult to achieve in children. Lorazepam and diazepam can both be administered by the rectal route (usually 0.5 mg/kg per rectum for both agents) and midazolam (0.2 mg/kg) via the IM, nasal, or buccal routes. Lorazepam is probably the first-line drug of choice for terminating SE in children as for adults. One study of 86 children presenting with seizure found that those who received lorazepam had a higher incidence of termination of seizure activity and less frequent respiratory depression than those treated with diazepam.59 Midazolam administered by continuous infusion appears effective in RSE in children.60–62 Although all eight patients in one study were mechanically ventilated, none demonstrated cardiovascular instability despite midazolam doses resulting in burst-suppression.59

As with adults, rapid control of SE in children achieved with benzodiazepines should be followed by administration of a longer-acting agent such as phenytoin (20 mg/kg IV), fosphenytoin (20 mg PE/kg IV), or phenobarbital (10 to 20 mg/kg IV).63 The rate of conversion of fosphenytoin to phenytoin is probably the same in children as in adults. Intramuscular injection of fosphenytoin may be particularly advantageous for prevention of recurrent seizures in children without IV access. The use of IV fosphenytoin over IV phenytoin is prudent in infants and neonates, whose small limbs are at especially high risk of extensive necrosis and amputation in the event of a phenytoin extravasation.

Intravenous valproate appears to be safe and effective in children;64 however, more data are needed to fully evaluate its use in pediatric SE.