Intravenous ketamine for the treatment of refractory status epilepticus: A retrospective multicenter study


  • This study was presented in part at the 2012 Annual Meeting of the American Epilepsy Society in San Diego, CA.



To examine patterns of use, efficacy, and safety of intravenous ketamine for the treatment of refractory status epilepticus (RSE).


Multicenter retrospective review of medical records and electroencephalography (EEG) reports in 10 academic medical centers in North America and Europe, including 58 subjects, representing 60 episodes of RSE that were identified between 1999 and 2012. Seven episodes occurred after anoxic brain injury.

Key Findings

Permanent control of RSE was achieved in 57% (34 of 60) of episodes. Ketamine was felt to have contributed to permanent control (“possible” or “likely” responses) in 32% (19 of 60) including seven (12%) in which ketamine was the last drug added (likely responses). Four of the seven likely responses, but none of the 12 possible ones, occurred in patients with postanoxic brain injury. No likely responses were observed when infusion rates were lower than 0.9 mg/kg/h, when ketamine was introduced at least 8 days after SE onset, or after failure of seven or more drugs. Ketamine was discontinued due to possible adverse events in five patients. Complications were mostly attributed to concurrent drugs, especially other anesthetics. Mortality rate was 43% (26 of 60), but was lower when SE was controlled within 24 h of ketamine initiation (16% vs. 56%, p = 0.0047).


Ketamine appears to be a relatively effective and safe drug for the treatment of RSE. This retrospective series provides preliminary data on effective dose and appropriate time of intervention to aid in the design of a prospective trial to further define the role of ketamine in the treatment of RSE.

Refractory status epilepticus (RSE) is defined as failure to respond to appropriate doses of two antiepileptic drugs (Brophy et al., 2012). The treatment of RSE frequently requires anesthetic drugs, which can be associated with lethal complications. Prolonged seizures are accompanied by a decline in sensitivity to γ-aminobutyric acid (GABA) agonists (Goodkin & Kapur, 2009) but not to N-methyl-d-aspartate (NMDA) antagonists (Mazarati & Wasterlain, 1999). Because ketamine is an NMDA antagonist, and not associated with cardiorespiratory depression (Hijazi et al., 2003), it is an attractive option for the treatment of RSE. Animal data suggest that ketamine may be an option for the treatment of prolonged status epilepticus (Borris et al., 2000). Based on this and other animal studies, in the last several years ketamine has been used for the treatment of RSE in humans, but most published data come from case reports or small series (Walker et al., 1995; Kofke et al., 1997; Sheth & Gidal, 1998; Ubogu et al., 2003; Pruss & Holtkamp, 2008; Hsieh et al., 2010; Kramer, 2012; Rosati et al., 2012; Synowiec et al., 2013).

The purpose of this study was to describe the experience with intravenous (i.v.) ketamine in the treatment of RSE across centers in North America and Europe.


The study was approved by the institutional review board of each center. Cases of RSE treated with intravenous ketamine between 1999 and 2012 were retrospectively identified (Table S1). RSE was defined as failure of at least two anticonvulsant medications. Pharmacy records, electroencephalography (EEG) reports, and intensive care unit (ICU) databases were reviewed. Methods of identification were specific to each institution in order to minimize recollection bias.

Data were collected using a standardized datasheet, and all variables were available for all episodes except loading dose (available in 45 of 60), maximum infusion rate (54 of 60), and functional outcome (58 of 60). Outcome was measured at discharge using the modified Rankin Scale (mRS) for adults, and mortality and increase in dependency for children. Anesthetic drugs included propofol, midazolam, pentobarbital, and thiopental.

“Likely response” to ketamine was defined as permanent control of SE within 24 h of initiation if ketamine was the last drug added. “Possible response” was defined as permanent control of SE within 24 h of initiation when ketamine was not the last drug added. Permanent control was defined as no recurrence of SE during the same ICU stay. Control of SE was defined as cessation of clinical and electrographic ictal manifestations as ascertained by continuous video-EEG monitoring (59 cases) or clinical evaluation and intermittent EEG (one case of status epilepticus of infantile spasms). Adverse events were attributed to ketamine if they occurred after the initiation of ketamine and if they led to lowering of the dose or discontinuation.

We used Fisher exact, Mann-Whitney, Kruskal-Wallis, and Cochrane-Armitage tests, and multivariable stepwise logistic regression models, as appropriate.


We identified 60 episodes of RSE treated with intravenous ketamine in 46 adults and 12 children (Table 1). More than one half of the subjects had new-onset RSE of unknown etiology (sometimes referred to as NORSE), which was most often attributed to encephalitis with lack of a known causative organism or anti-neuronal antibody.

Table 1. Demographics and clinical data (N = 60 episodes of RSE)
  1. Data presented as N (row percentage) unless stated otherwise. h, hours; d, days; m, months; y, years; CEEG, continuous EEG monitoring; SE, status epilepticus.

  2. a

    Causes of nonanoxic brain injury and systemic etiologies included proven infectious (N = 4) or autoimmune (N = 2; both anti-NMDA) meningoencephalitis, subarachnoid hemorrhage (N = 2), ischemic stroke (N = 2), traumatic brain injury (N = 1), sepsis-associated encephalopathy (N = 1), and posterior reversible encephalopathy syndrome (N = 1).

Age; median (range)24 y (7 m–74 y)
Children (<18 y) (%)12 (20)
Female gender (%)30 (50)
Etiology (%) 
Unknown34 (57)
Acute symptomatic20 (33)
Nonanoxic brain injurya11 (18)
Postanoxic encephalopathy7 (12)
Systemic etiologya2 (3)
Remote symptomatic6 (10)
Prior history of epilepsy (%)9 (15)
Duration of SE (days); median (range)26.5 d (1 h–10 m)
CEEG (%)59 (98)
Time from onset of SE to CEEG (median, range)<24 h (0–17 d)
Classification of SE (%) 
Generalized convulsive14 (23)
Tonic–clonic5 (8)
Myoclonic6 (10)
Tonic3 (5)
Generalized nonconvulsive3 (5)
Focal convulsive4 (7)
Epilepsia partialis continua2 (3)
Hemiconvulsive2 (3)
Focal nonconvulsive38 (63)
Status epilepticus of infantile spasms1 (2)

Ketamine was introduced after a median of 9 days of SE (range 0–122). We found wide variability in how it was administered. Common features included use of a loading dose (median 1.5 mg/kg; maximum 5 mg/kg) followed by continuous infusion (median 2.75 mg/kg/h; maximum 10 mg/kg/h). Ketamine was always part of a multidrug regimen that ranged from 2 to 12 concurrent medications (Table S2). The duration of infusion varied from 6 h to 27 days.

Ketamine was likely responsible for control of SE in 12% of cases (7 of 60), and possibly responsible in an additional 20% (12/60; Table 2). SE was transiently controlled by ketamine in eight cases (13%). Likely response was not observed when infusion rates were lower than 0.9 mg/kg/h; ketamine was introduced at least 8 days after SE onset; or after failure of seven or more drugs. When given as the fourth-line drug, likely and possible responses were observed in 4 of 10 and 2 of 10 cases, respectively. There were no specific drug combinations that improved response to ketamine. Furthermore, we were not able to analyze the effect of ketamine on the used dose of concurrent benzodiazepines. Only the number of previously failed drugs was significantly associated with ketamine response (Table 2). Because four likely (but none of the 12 possible responses) occurred in postanoxic cases, we reanalyzed nonanoxic cases separately and found similar results. (Table S3).

Table 2. Determinants of ketamine efficacy (N = 60 episodes)
 Likely response (N = 7)Possible response (N = 12)Likely or possible response (N = 19)No response (N = 41)p-Value (univ.)$p-Value (multiv.)
  1. h, hours; d, days; m, months; univ., univariate analysis; multiv., multivariate analysis.

  2. $p-value refers to analysis using likely, possible, and no response as three separate categories.

  3. a

    Information available in 54 of 60 cases.

  4. b

    Information available in 46 of 60 cases.

  5. c

    Anesthetic drugs included pentobarbital, thiopental, midazolam, and propofol.

Latency to ketamine; median (range)12 h (6 h–7 d)5 d (18 h–30 d)4.5 d (6 h–30 d)10 d (12 h–122 d)0.0053NS
Number of previously failed drugs; median (range)4 (3–7)6 (3–11)6 (3–11)8 (3–16)0.0012<0.01
Unknown (N = 34)17826<0.001NS
Anoxic (N = 7)4043
Acute nonanoxic (N = 13)2249
Remote (N = 6)0333
SE classification      
Generalized convulsive (N = 14)2468NS
Generalized nonconvulsive (N = 3)0112
Focal convulsive (N = 4)0222
Focal nonconvulsive (N = 38)551028
Infantile spasms (N = 1)0001
Maximum infusion rate (mg/kg/h); median (range)a7 (0.9–10)1.8 (0.6–7)2 (0.6–10)3 (0.05–10)NS
Loading dose administeredb6/6 (100%)5/8 (63%)11/14 (79%)23/32 (72%)NS
Duration of administration1 (0–2)3 (0–10)2 (0–10)5 (0–27)<0.001NS
Number of concurrent drugs3 (1–5)5 (1–11)4 (1–11)6 (1–10)<0.001NS
Number of concurrent anesthetic drugsc1 (0–1)1 (1–3)1 (0–3)2 (1–3)<0.001NS

Ultimately, 34 episodes (57%) of status epilepticus were controlled. The overall mortality rate was 45% (26 of 60). Younger age and response to ketamine were associated with lower mortality in a stepwise multivariable logistic regression analysis (Table 3). Dose and duration of exposure to ketamine were not related to mortality. Among survivors, most had a poor functional status at discharge: only 2 of 46 (4%) adults had good outcome (mRS ≤ 2) and one of 12 children returned to their baseline status. There was no difference in functional outcome in survivors whether they responded to ketamine or not.

Table 3. Variables associated with mortality
(N = 60 cases)Alive at discharge (N = 34; 57%)Deceased (N = 26; 43%)p-value (univ.)p-value (multiv.)OR (multiv.)
  1. univ., univariate analysis; multiv., multivariate analysis.

  2. a

    Anesthetic drugs include pentobarbital, thiopental, midazolam, and propofol.

Duration of ketamine; days; median (range)4 (0–24)4 (0–27)NS
Dose of ketamine; mg/kg/h; median (range)1.7 (0.05–10)3.6 (0–10)NS
Response to ketamine (N = 19); N (row %)16 (84)3 (16)0.00140.0010.007–0.3
Age; years; median (range)26 (7 mo–85)51 (19–76)0.0025<0.0011.02–1.1/year
Unknown (N = 34); N (row %)18 (53)16 (47)0.0127NS
Anoxic (N = 7); N (row %)2 (29)5 (71)
Acute non anoxic (N = 13); N (row %)9 (69)4 (31)
Remote (N = 6); N (row %)5 (83)1 (17)
Status classification     
Generalized convulsive (N = 14); N (row %)8 (56)6 (44)NS
Generalized nonconvulsive (N = 3); (row %)1 (33)2 (67)
Focal convulsive (N = 4); N (row %)2 (50)2 (50)
Focal nonconvulsive (N = 38); N (row %)25 (66)13 (33)
Status epilepticus of infantile spasms; N (row %)1 (100)0 (0)
Any adverse event (N = 18); N (row %)8 (44)10 (66)NS
Concurrent anesthetic drugs; median (range)a2 (0–3)2 (1–3)NS

Most patients were mechanically ventilated (56 of 60) and treated with vasopressors (52 of 60) prior to ketamine. Vasopressors were decreased in 6 patients but increased in 21 after initiation of ketamine. One third of patients developed complications while receiving ketamine, most commonly sepsis, shock, organ failure, and pneumonia (Table S4). We found no relationship between these complications and the dose or duration of exposure to ketamine, but complications were higher when more than one other anesthetic drugs was used concurrently (Table 4).

Table 4. Variables associated with adverse events occurring during ketamine infusion
Adverse events occurring during ketamine infusion (N = 60 cases)Adverse events (N = 18; 30%)No adverse events (N = 42; 70%)p-value (univ.)
  1. h, hours; d, days; m, months; y, years; univ., univariate analysis; multiv., multivariate analysis.

  2. a

    Information available in 54 of 60 cases.

  3. b

    Anesthetic drugs include pentobarbital, thiopental, midazolam, and propofol.

Duration of ketamine; days; median (range)5 (1–17)3 (0–27)NS
Maximal infusion rate; mg/kg/h; median (range)a2.4 (0.6–7.5)2.75 (0–10)NS
Shock prior to ketamine (N = 8); N (row %)2 (25)6 (75)NS
Hepatic failure prior to ketamine (N = 6); N (row %)0 (0)6 (100)NS
Renal failure prior to ketamine (N = 10); N (row %)4 (40)6 (60)NS
Cardiac failure prior to ketamine (N = 7); N (row %)1 (14)6 (86)NS
Concurrent antiepileptic drugs; median (range)5 (4–11)5 (1–11)NS
Concurrent anesthetic drugs; median (range)b2 (1–3)1 (0–3)0.005

Ketamine was discontinued in four cases (7%) due to suspected treatment-related adverse events. One patient developed a syndrome similar to the propofol-related infusion syndrome after 4 days of high dose ketamine (4.5 mg/kg/h) and midazolam (2.5 mg/kg/h), but no recent propofol use, and he recovered. Two patients developed supraventricular tachycardia that resolved after ketamine discontinuation. One other required the administration of amiodarone for atrial fibrillation. In addition, ketamine was discontinued in one patient due to a probable adverse event that could not be reliably identified.

Intracranial pressure (ICP) monitoring was available during ketamine administration in seven cases. Two patients with cerebral edema from anoxic injury required osmotic therapy while on ketamine. Three patients were documented to have increased ICP prior to ketamine infusion but did not experience any worsening.


We provide retrospective, multicenter data on the use of intravenous ketamine for the management of RSE. Our case series suggests that ketamine may be efficacious and relatively safe in sedated and ventilated children and adults.

The study population was a highly selected group of patients with prolonged RSE, as evidenced by the median number of failed drug trials before ketamine was administered (six drugs) and the median latency from onset of SE to ketamine administration (9 days). Therefore, it is not surprising that 90% of cases were secondary to an acute brain injury or belonged to the category of new-onset RSE of unknown etiology (NORSE), which is usually attributed to encephalitis with lack of a known causative organism or anti-neuronal antibody and is associated with a relatively poor prognosis (Mayer et al., 2002; Holtkamp et al., 2005a,b; Costello et al., 2009).

The information regarding the therapeutic dosage or use of ketamine for the management of status epilepticus is scarce. Doses as high as 7.5 mg/kg/h used for up to 14 days have been described as safe (Kramer, 2012; Rosati et al., 2012). In our series, administration of doses up to 10 mg/kg/h for up to 27 days was not associated with increased complications or mortality compared to patients receiving lower doses and shorter duration of exposure. Most importantly, all likely responses (permanent cessation of SE within 24 h of ketamine initiation) occurred at doses of at least 0.9 mg/kg/h.

Permanent control of SE was likely or possibly attributed to ketamine in 32% (19 of 60) of episodes, and transient control in an additional 13% (8 of 60). Response rate was highest when ketamine was introduced “early.” When used as a third-line or fourth-line agent, ketamine was deemed responsible for control in 60% (6 of 10) of episodes of SE. Likely responses to ketamine were not reported when the drug was introduced more than 1 week after onset of SE or after the failure of seven or more drugs, although possible responses were seen. Available guidelines support the role of general anesthesia as the standard therapeutic approach to RSE, despite the fact that most available evidence comes from uncontrolled case series. A systematic review of these series reports the efficacy (ability to control RSE without breakthrough seizures) to be 42%, 66%, and 60%, respectively, for midazolam, propofol, and barbiturates (Claassen et al., 2002). A recent prospective trial comparing propofol to barbiturates was interrupted due to poor enrollment after 24 patients were enrolled (Rossetti et al., 2011). Efficacy of propofol and barbiturates was 43% and 22%, respectively. In the majority of published cases, anesthetics were used as a third-line or fourth-line agent. Therefore, the efficacy of ketamine in our series is similar to that of other anesthetic agents, although lower than previous ketamine case series that reported an efficacy (permanent control of SE) of 63% (Synowiec et al., 2013) and 72% (Rosati et al., 2012). In these two series, however, the majority of cases occurred in the setting of chronic epilepsy. This etiology carries a good prognostic factor compared to NORSE and acute brain injury, which accounted for the majority of cases in our series.

Consistent with previous reports, we identified few adverse events leading to the discontinuation of ketamine (Kramer, 2012; Rosati et al., 2012; Synowiec et al., 2013) and most of them were related to the use of concurrent anesthetics and not to dose or duration of ketamine. One incident of severe acidosis with coadministration of high doses of ketamine and midazolam was life-threatening, a complication that has been reported only once for each drug (Roervik & Stovner, 1974; Federman et al., 2009). Apart from two patients with cerebral edema secondary to anoxic injury, no other patients were documented to have changes in intracranial pressure. This confirms that ketamine does not significantly raise ICP in ventilated patients (Mayberg et al., 1995). Overall, the incidence of adverse events and mortality rate in our series is similar to other series of RSE (Novy et al., 2010; Hocker et al., 2012). We found that mortality was lower in subjects who responded to ketamine, but this may reflect lower severity of these episodes.

Data from animal models regarding the potential neurotoxicity of ketamine are controversial, as both neuronal apoptosis and neuroprotection have been reported (Green & Coté, 2009). Human data are limited to a single case of SE with a prolonged refractory course (Ubogu et al., 2003). Given the severity and duration of most cases in our series and the confounding presence of acute brain injury and comorbidities in many of our patients, we are unable to address this question, which should, however, be studied in any prospective trial.

This study is limited by its retrospective design, subject heterogeneity, variability in timing and dose, incomplete data in some patients, and other confounding factors, for example, further coadministration of other drugs. Any nonestablished treatment tends to be administered at the peak of severity of the disease. Therefore, spontaneous improvement, which can occur in SE, cannot be excluded. Given the high number of medications received by patients concomitantly to ketamine, efficacy may be attributed in part to the cumulative or delayed effect of these concurrent or previously administered drugs.

The higher efficacy that was seen with earlier administration may be due to the fact that only the most severe cases persisted long enough for ketamine to be administered. Despite these limitations, this series documents the potential utility and current patterns of use of ketamine for the treatment of RSE and provides useful preliminary data for planning a prospective trial.


Continuous ketamine infusion appeared to be safe and moderately efficient for the treatment of RSE, contributing to the control of SE in one third of cases, while being responsible for few significant adverse events. The role of ketamine for the treatment of RSE should be further investigated in prospective, randomized trials.


N.G. was the recipient of a Postdoctoral Research Fellowship from the Belgian American Education Foundation, and received research grant from the Commission for Educational Exchange between the United States and Belgium and the Belgian Neurological Society. B.F. received funding from Pfizer's Medical and Academic Partnerships Program.

L.M.J., J.N.B., B.R.N., B. M., A.A.-O., R.K., L.C.M., S.S., A.Z., G.P.K., H.P.G., and J.C. have nothing to disclose.

I.S.F. was funded by a grant for the study of Epileptic Encephalopathies from ‘Fundación Alfonso Martín Escudero.' B.L. has received funding for research from UCB Pharma; serves on the advisory board of UCB Pharma, GlaxoSmithKline (GSK), and Pfizer; and received speaker honoraria from UCB Pharma and GSK. J.P.S. has served on an advisory board for UCB Pharma. He serves on the editorial board of Epilepsy and Behavior, Epileptologia, and Restorative Neurology and Neuroscience. He has received honoraria from American Epilepsy Society and UCB Pharma, and he is on speakers' bureaus for UCB Pharma. He has received funding for research from the American Academy of Neurology, David-Phinney Foundation/Sunflower Revolution, UCB Pharma, NIH/NINDS, and the University of Cincinnati Research Council. He has served as an expert witness in legal proceedings for the government.

T.L. serves on the Laboratory Accreditation Board for Long Term (Epilepsy and Intensive Care Unit) Monitoring, on the Council of the American Clinical Neurophysiology Society, on the American Board of Clinical Neurophysiology, as an Associate Editor for Seizure, and performs video-electroencephalography long-term monitoring, electroencephalography, and other electrophysiologic studies at Boston Children's Hospital and bills for these procedures. T.L. receives research support from the National Institutes of Health/NINDS, a Career Development Fellowship Award from Harvard Medical School and Boston Children's Hospital, the Program for Quality and Safety at Boston Children's Hospital, the Payer Provider Quality Initiative, The Epilepsy Foundation of America, the Center for Integration of Medicine and Innovative Technology, the Epilepsy Therapy Project, the American Epilepsy Society, Cure, and from investigator initiated research grants from Lundbeck and Eisai.

C.D.H. is supported by research grants from the Canadian Institutes of Health Research, PSI Foundation and SickKids Foundation. L.J.H. has received research support for investigator-initiated studies from UCB-Pharma, Upsher-Smith, and Lundbeck; consultation fees for advising from Lundbeck, Upsher-Smith, UCB-Pharma, and GlaxoSmithKline; and royalties for authoring chapters for UpToDate-Neurology, and for co-authoring the book “Atlas of EEG in Critical Care,” by Hirsch and Brenner (2010). S.M.L. has received research support from the National Institutes of Health/SBIR, and for investigator-initiated studies from UCB Pharma. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.