The principles of care described for patients with epilepsy admitted electively to the intensive care unit (ICU) also apply to emergent ICU admissions. Efforts should be made to continue the patient's outpatient antiepileptic drug (AED) regimen. Some of the newer agents may be administered via nasogastric tubes or other alimentary tubes (Table: Antiepileptic drug preparations that can be delivered via enteric tubes with reliable kinetics). The choices of IV substitutes for newer AEDs are limited, and it may be necessary to deliver periodic doses of benzodiazepines in the short-term interval while loading the patient with an acceptable IV medication preparation. Targets for serum levels should be in the therapeutic range (e.g., 10–20 mg/dL for phenytoin, 10–20 mg/dL for phenobarbital, and 50–100 mg/dL for valproic acid).

Acutely, it is usually more important to establish adequate levels of a therapeutic AED rapidly than to make an optimal choice for long-term therapy. Considerations about interactions with other medications are more important during chronic treatment than in the ICU. Nevertheless, several pharmacologic principles must be taken into account when treating individuals with epilepsy in the ICU:

  • Lowering of seizure threshold. The point is not absolutely to avoid medications that can lower seizure threshold, but rather to be aware of this association during their use. Several drugs and drug classes commonly encountered in the ICU were listed in the discussion of treating seizures in electively admitted patients.
  • Protein binding. Many AEDs (e.g., phenytoin, phenobarbital, valproic acid) are highly protein bound, and most agents have some protein binding. Free levels of these agents are increased in the setting of low albumin, which frequently occurs with malnutrition in the ICU. In addition, medications or conditions that compete with these medications for binding to albumin increase free levels. The most common of these conditions is uremia, in which blood urea nitrogen displaces other agents bound to albumin. In these settings, monitor free serum levels of AEDs to adjust medication doses. Notable exceptions to this rule are gabapentin and levetiracetam, which are not metabolized hepatically or significantly protein bound.
  • Increased metabolism, autoinduction, and tolerance. These phenomena usually occur only after a number of days or, more commonly, weeks of AED therapy. In the ICU, they generally are a major concern only for individuals with tremendous hepatic induction or tolerance induced before ICU admission. Drugs such as phenytoin, carbamazepine, and phenobarbital cause autoinduction of hepatic enzymes, and individuals with a recent history of taking these medications may require higher-than-typical doses of these medications. Subjects with a history of significant chronic use of alcohol or benzodiazepines often require much higher doses of benzodiazepines than usual to achieve an antiepileptic effect. Finally, many patients are rapid acetylators, in whom it can be difficult to achieve therapeutic AED levels, despite large doses.37 This condition is usually apparent early after initiation of therapy and is best addressed by changing to another AED that is not dependent on the affected enzyme system for metabolism. Large increases in medication dose help in this situation, but doses may be difficult to titrate, with shifts from subtherapeutic to toxic levels. In our experience, this often results in wasted time and prolonged periods in which the patient has low serum levels of AED.

Difficulties associated with specific AEDs

Vigilant monitoring for particular side effects or difficulties associated with the use of specific AEDs is important in the ICU. All AEDs can cause allergic reactions, hepatic dysfunction, and encephalopathy. The ICU must also be aware of specific side effects or medication-associated challenges presented by specific AEDs:

  • Valproic acid can cause thrombocytopenia and increased risk of bleeding. This agent is often avoided when active bleeding or increased risk of bleeding is a problem, although the magnitude of these risks is difficult to assess.37 Valproate also carries with it an increased risk of hyperammonemia and encephalopathy in the elderly, which can be difficult to recognize in ICU patients unless one is specifically aware of this difficulty.38,39
  • Carbamazepine and oxcarbazepine can cause leukopenia and hyponatremia. These side effects may be dose-related or idiosyncratic. Both types of side effects may be severe, but usually are not. Hyponatremia, when due to carbamazepine, is usually due to inappropriate antidiuretic hormone section and may respond, to some degree, to fluid restriction.
  • Phenytoin can cause zero-order kinetics after a level of more than 10–12 ìg/mL. The metabolizing enzyme becomes saturated, and levels may rise dramatically in response to a small increase in dose. Phenytoin is highly protein bound. Monitor free levels in uremia, hypoalbuminemia, and pregnancy.
  • Topiramate can cause acute psychosis, depression, decreased ability to speak clearly, and nephrolithiasis.
  • Lamotrigine may cause a rash leading to erythema multiforme if rapidly titrated. This problem is exacerbated by simultaneous administration of valproic acid. Risks of severe allergic reaction leading to erythema multiforme are increased in those under 16 years of age.
  • Phenobarbital can cause prolonged sedation at high levels owing to its long half-life. Large amounts of protein binding may result in elevated free levels in uremia and hypoalbuminemia. The half-life may be prolonged in hepatic failure.
  • Benzodiazepines may cause prolonged sedation with extended infusion, owing to storage and slow release from adipose tissue.

Adapted from: Kolb SJ and Litt B. Management of epilepsy and comorbid disorders in the emergency room and intensive care unit. In: Ettinger AB and Devinsky O, eds. Managing epilepsy and co-existing disorders. Boston: Butterworth-Heinemann; 2002;515–535. With permission from Elsevier (

Authored By: 
Steven J. Kolb Md PhD
Brian Litt MD
Reviewed By: 
Steven C. Schachter MD
Friday, April 30, 2004