Propofol and Midazolam in the Treatment of Refractory Status Epilepticus


Address correspondence and reprint requests to Dr. T. P. Bleck at Department of Neurology, Box 394, University of Virginia, Charlottesville, VA 22908, U.S.A. E-mail:


Summary:  Purpose: To explore outcome differences between propofol and midazolam (MDL) therapy for refractory status epilepticus (RSE).

Methods: Retrospective chart review of consecutive patients treated for RSE between 1995 and 1999.

Results: We found 14 patients treated primarily with propofol and six with MDL. Propofol and MDL therapy achieved 64 and 67% complete clinical seizure suppression, and 78 and 67% electrographic seizure suppression, respectively. Overall mortality, although not statistically significant, was higher with propofol (57%) than with MDL (17%) (p = 0.16). Subgroup mortality data in propofol and MDL patients based on APACHE II (Acute Physiology and Chronic Health Evaluation) score did not show statistically significant differences except for propofol-treated patients with APACHE II score ≥20, who had a higher mortality (p = 0.05). Reclassifying the one patient treated with both agents to the MDL group eliminated this statistically significant difference (p = 0.22).

Conclusions: In our small sample of RSE patients, propofol and MDL did not differ in clinical and electrographic seizure control. Seizure control and overall survival rates, with the goal of electrographic seizure elimination or burst suppression rather than latter alone, were similar to previous reports. In RSE patients with APACHE II score ≥20, survival with MDL may be better than with propofol. A large multicenter, prospective, randomized comparison is needed to clarify these data. If comparable efficacy of these agents in seizure control is borne out, tolerance with regard to hemodynamic compromise, complications, and mortality may dictate the choice of RSE agents.

The mortality rate associated with status epilepticus (SE) varies between 3 and 35% depending on the age of the patient, the etiology of seizures, and the duration of SE (1). Available reports of treated refractory SE (RSE) patients consist of small series (2–6). Depending on the definition of RSE used in these studies, mortality rates vary between 32 and 77%. An incidence of 6,000–20,000 RSE cases/year in the United States indicates the magnitude of the problem (7). The high prevalence of concurrent systemic illnesses adds to its burden (2).

Significant advances in the management of RSE have occurred during the last two decades. Three agents [barbiturates, propofol, and midazolam (MDL)] have emerged as treatment for RSE, but success rates vary (2–5,8–12). The total number of patients treated with these agents remains low, and therefore a consensus about RSE management has yet to emerge. The use of high-dose barbiturates to treat RSE is associated with a high morbidity and mortality (13). Among the newer agents to treat RSE, propofol is efficacious and has favorable pharmacokinetics (2,13). Several case reports have described the efficacy of propofol therapy in controlling SE (8–10,14–25). One small, open-labeled, nonrandomized study with a prospective component compared the outcome of RSE patients treated with propofol to those treated with high-dose barbiturates (2). Open-labeled, uncontrolled studies of continuous intravenous (i.v.) MDL use in RSE demonstrated moderate efficacy (11,12,26–28) and favorable pharmacokinetics (29). The aim of this study was to examine our experience with propofol and MDL therapy in adult patients with RSE and to explore possible differences in efficacy and complications.


Patient identification

The medical records of consecutive RSE patients (age 17 or older) admitted and treated at University of Virginia Medical Center between 1995 through 1999 were reviewed retrospectively; before this time, propofol was not used to treat RSE patients at our institution. Patients were identified from the SE cases in neuroscience intensive care unit (ICU) admission logbook and from EEG reports that described the use of propofol or MDL. Patients were categorized based on which of the two agents they received. If patients received both agents, the agent used for the greater time was designated.

We followed the RSE clinical criteria defined by Stecker et al. (2) for inclusion. These included acute seizures persisting >2 h, or seizures recurring at a rate of at least two per hour without recovery to baseline between seizures, despite treatment with conventional anticonvulsants (AEDs) including phenytoin (PHT) and lorazepam (LZP) or phenobarbital (PB). The presence of altered mental status was necessary, thus excluding simple partial SE. Patients satisfying these criteria were included in the study only if EEG before or during RSE therapy showed electrographic seizures. EEGs from referring hospitals were acceptable if available for review and met the criteria.

Pediatric patients (age 17 years or younger) were excluded because their prognosis is much better than that of adult patients, and a propofol treatment group was not available for comparison. Patients with a history of witnessed motor manifestations in the absence of electrographic seizures were also excluded. EEG findings of burst suppression, suppression, or diffuse slowing alone did not constitute criteria for inclusion.

Clinical assessment

To assess clinical characteristics, we reviewed the inpatient and outpatient charts and EEG of study patients. The suspected etiology of SE was determined from all available clinical, radiologic, or laboratory data. However, contributions from other medical illnesses could not be entirely excluded in some patients. The duration of SE before treatment was determined from the time the first motor manifestation occurred. If the duration of RSE after seizure onset was not well defined, it was rounded off to the nearest 12 h. An APACHE II score was calculated for each patient at the time of RSE diagnosis to serve as a severity-of-illness index (30). A cut-off score of ≥20 was chosen to indicate moderate to severe illness, based on the observation that all patients with scores ≤20 survived.

We noted the occurrences of hemodynamic compromise, infectious complications, number of days on a ventilator, and mortality. Hemodynamic compromise was defined as persistent or recurrent hypotension despite pressor support, necessitating change or withdrawal of the RSE agent. Infectious complications included sepsis, chest radiographs showing pneumonia, and pyuria. Mortality included death during or within 2 weeks after the discontinuation of RSE therapy.

Outcome assessment

Clinical seizure outcome was studied independently from electrographic seizure outcome. Data on clinical seizures, both overt and subtle, were entirely dependent on the records. A seizure was considered subtle if it consisted of subtle twitches of the extremities, face, or nystagmoid eye movements in an unconscious patient (31). Cessation of seizures was defined as complete suppression of motor manifestations. Clinical–electrographic correlation was not available for some presumed abnormal motor activities that fitted in the description of subtle seizures either because the EEG was not archived, or because the subtle movements occurred when intermittent EEG was performed. These documented abnormal movements, therefore, were considered seizures. Clinical seizures during planned tapering were not included in the outcome. However, clinical seizures during maintenance phase, when drug dosage could not be increased further due to complications, and after 24 h after withdrawal of the RSE agent, were included as treatment failure.

Electrographic seizure outcome was based on EEG reports and review of available archived EEG. EEG reports were available in the charts on a daily basis for the duration of monitoring. Board-certified electroencephalographers read all studies. Important segments of archived EEGs on microfilm or CD-ROM were available on all patients. These were reviewed to corroborate the EEG reports. Electrographic seizure suppression was defined as complete elimination of electrographic seizure regardless of burst suppression or complete suppression of EEG activity. Periodic complexes associated with ictal motor manifestations were considered clinical seizures. Periodic complexes without ictal motor manifestations were also considered seizures except in a terminal patient, when they were considered to represent brain damage, rather than ongoing SE (32). Periodic complexes seen briefly after ictal motor manifestations were considered to be due to the postictal state.

Statistical analysis

All statistical tests were performed with SigmaStat 2.0 software (SPSS Corporation, Chicago, IL, U.S.A.). Student's t test was used to analyze age and duration of SE before treatment. The APACHE II score was analyzed by the rank-sum test. Fisher's exact two-tailed test was used to analyze dichotomized variables. Duration of RSE agent infusion and days on ventilator were analyzed by the Mann–Whitney rank sum test. Alpha was set at ≤0.05. Primary analyses were performed, including the one patient who received both agents in the propofol group. A sensitivity analysis was undertaken to determine if reclassification of this patient to the MDL group changed the results.



The clinical characteristics of the patients treated with propofol and MDL are given in Tables 1 and 2, and a comparison of their demographic data is given in Table 3. Fourteen patients (eight women, six men) received primarily propofol, and six patients (five women, one man) received MDL. There was no clinically significant demographic difference between propofol and MDL patients.

Table 1.  Clinical profile of refractory status epilepticus patients treated with propofol
Cause of
refractory status
Other known
of SE before
II score at
RSE diagnosis
Other drug
  • ARF, acute renal failure; CBZ, carbamazepine; CRF, chronic renal failure; CVA, cerebrovascular accident; DM, diabetes mellitus, HTN, hypertension; ICA, internal carotid artery; L, left; MCA, middle cerebral artery; PCA, posterior cerebral artery; PB, phenobarbital; PHT, phenytoin; PLEDs, periodic lateralized epileptiform discharges; R, right; RSE, refractory status epilepticus; S/P, status post; VPA, valproic acid.

  • a

     12-h midazolam infusion preceded propofol therapy.

1/17/MEncephalitisNoneDiscrete L/R temporal ictal activity24 h20PHT 23, VPA 75
2/18/M? EncephalitisNoneGeneralized periodic complexes L > R48 h18PHT 18, PB 24
3/21/MEncephalitisNoneContinuous ictal activity L > R12 h20PHT 22, PB 34, VPA 40
4/28/MRemote symptomatic epilepsyStatic encephalopathyL hemisphere continuous ictal activity12 h17PHT 19, PB 24
5/28/FCerebral edemaMultiorgan failure, S/P liver transplant rejectionL temporal PLEDs6 h24PHT 11
6/47/FHyponatremiaMalabsorption syndrome, head traumaR frontal PLEDs2 h10PHT 18
7/52/FMetastases brainCa breast, DM, HTNPeriodic complexes R > L frontal24 h26PHT 22
8/58/FCryptococcal meningitisHepatic encephalopathy ARF, HTNIndependent L/R continuous ictal activity12 h29PHT 17, PB 24
9/60/FHypoglycemiaDMBifrontal periodic complexes24 h21PHT 26
10/62/FRemote symptomatic epilepsyDM, HTN, old CVA L MCAL/R hemisphere continuous ictal activity24 h28PHT 20
11/66/Ma? EncephalitisR ICA complete occlusionDiscrete L/R temporal ictal activity48 h20PB 46, CBZ 5
12/66/MS/P Basilar art. aneurysm surgery/brain edemaNoneR frontal ictal activity with generalization24 h15PHT 20
13/76/FIntracranial hemorrhageS/P meningioma resection, CRF HTNDiscrete L/R frontal ictal activity16 h20PHT 25, PB 15
14/81/FRemote symptomatic epilepsyDM, HTN, old CVA L PCADiscrete L occipital ictal activity24 h19PHT 14, VPA 25
Table 2.  Clinical profile of refractory status epilepticus patients treated with midazolam
Cause of
refractory status
Other known
EEG before
of SE
II score at
RSE diagnosis
drug levels
  1. CBZ, carbamazepine; CVA, cerebrovascular accident; DM, diabetes mellitus; ESRD, end-stage renal disease; HTN, hypertension; ICA, internal carotid artery; L, left; MCA, middle cerebral artery; PB, phenobarbital; PHT, phenytoin; R, right; RSE, refractory status epilepticus; SDH, subdural hematoma; VPA, valproic acid.

1/19/FRemote symptomatic epilepsyStatic encephalopathy, cardiomyopathyDiscrete R > L temporal ictal activity24 h7PHT 22, PB 23
2/33/FUremiaDM, HTN, CVA R MCA, ESRDL hemisphere continuous ictal activity24 h26PHT 18, CBZ 8
3/38/FRemote symptomatic epilepsyStatic encephalopathy R ICA complete occlusionDiscrete R frontal ictal activity8 h11PHT 26
4/47/MIntracerebral hemorrhageCirrhosis liverL temporal ictal activity with generalization18 h20PHT 20, PB 9
5/74/FIschemia–hypoxiaMultiple traumaDiscrete L/R temporal ictal activity2 h25PHT 22, PB 26
6/79/FL acute on chronic SDHESRD, HTNDiscrete L/R temporal ictal activity16 h20PHT 16, VPA 75
Table 3.  Summary of demographics and outcomes of refractory status epilepticus patients treated with propofol and midazolam
Demographics and outcomesPropofol
(n = 14)
(n = 6)
  • a

     p value of 0.16 is a comparison of overall mortality between propofol and midazolam therapies.

Median age in yr (range)48.6 (17–81)48.3 (19–79)
Median APACHE II score (range)20 (10–29)20 (7–26)
Median duration of SE before
 therapy in h (range)
24 (2–48)17 (2–24)
Median duration of therapy in h
31 (7–408)43 (40–118)
Clinical seizure suppression (%) (n)64 (n = 9)67 (n = 4)
Electrographic seizure suppression
 (%) (n)
78 (n = 11)67 (n = 4)
Elimination of all epileptiform
 discharges (%) (n)
29 (n = 4)17 (n = 1)
Infectious complications (%) (n)29 (n = 4)50 (n = 3)
Hemodynamic compromise (%) (n)14 (n = 2)0 (n = 0)
Median number of days on ventilator
4 (1–32)4 (0–11)
Overall mortality (%) (n)a57 (n = 8)17 (n = 1)

Initial treatment

All patients were admitted to the neuroscience ICU and managed by neurointensivists. Three patients were intubated before admission to the ICU. All except one of the remaining patients were intubated before initiation of propofol or MDL. Central venous or pulmonary arterial catheters and arterial lines were placed in all patients.

Treatment of patients naïve to conventional AEDs consisted of standard loading doses of PHT, 20 mg/kg, and LZP, 0.05–0.1 mg/kg, i.v. Patients treated with PHT and PB before the episode of RSE received additional loading doses if these levels were subtherapeutic. Patients receiving other maintenance AEDs such as valproic acid (VPA) and carbamazepine (CBZ) continued to receive these agents. Every attempt was made to keep AED levels above therapeutic levels throughout the episode of SE. No differences were observed in the serum PHT and PB levels between the two groups before initiation of RSE agents (p = 0.78 and 0.28, respectively) (Tables 1 and 2).

Propofol/midazolam treatment

Propofol and MDL dose (Tables 4 and 5) depended on individual treating physician preference and hemodynamic status. A propofol bolus dose of 1–3 mg/kg was administered in only five patients. Maintenance infusion dose varied typically between 1 and 10 mg/kg/h (range, 0.1–24 mg/kg/h). All MDL-treated patients except one received an MDL bolus (dose, 2–12 mg). Maintenance infusion dose varied between 0.05 and 0.8 mg/kg/h. The end point of titration of the RSE agent was elimination of electrographic seizure or achievement of electrographic burst suppression/suppression. After ∼12 or 24 h of attaining the goal, an attempt was made to reduce the RSE agent by approximately half.

Table 4.  Complications, therapy, and outcome of refractory status epilepticus patients treated with propofol
bolus (mg/kg)/
infusion (mg/kg/h)
Duration of
Days of
  • Pre-SE, prestatus epilepticus.

  • a

     Subtle seizures.

12/2.0–12.046 hYesYes5Alive/mild deficit
2Pneumonia2/5.0–9.017 dNoaYes32Alive/mild deficit
3Hemodynamic compromise1/5.0–10.548 hNoaNo2Died
4Pneumonia2/1.0–3.096 hYesYes4Alive/pre-SE
50/2.0–24.082 hNoaNo4Died
60/2.5–10.014 hYesYes2Alive/normal
7Pneumonia3/2.5–5.018 hNoaYes4Died
8Sepsis0/2.0–5.022 hNoaNo12Died
9Hemodynamic compromise0/5.0–9.040 hYesYes9Died
100/3.0–5.07 hYesYes1Died
110/2.0–3.048 hYesYes15Died
120/1.0–4.072 hYesYes4Alive/mild deficit
130/2.0–20.014 hYesYes4Died
140/0.1–0.39 hYesYes5Alive pre-SE
Table 5.  Complications, therapy, and outcome of refractory status epilepticus patients treated with midazolam
of infusion
  • Pre-Se, pre–status epilepticus.

  • a

     Subtle seizures.

2Pneumonia12/0.1–0.545YesYes3Alive, pre-SE
3Pneumonia5/0.1–0.882YesYes8Alive pre-SE
4None16/0.05–0.1118NoaYes11Alive pre-SE
6None0/0.5–0.140YesYes7Alive/moderate deficit

Management while receiving RSE therapy

Volume expansion with saline, and pressure support with phenylephrine, dopamine, and norepinephrine were used to maintain the patient's blood pressure (BP) within an acceptable range, as necessary. If pressure-support measures did not rapidly control BP, infusion of propofol or MDL was withheld until the patient's BP reached an acceptable value.

EEG was performed before initiating/continuing RSE agent or as soon as possible thereafter. Digital EEG monitoring continued throughout the period of RSE treatment. Once overt and initial subtle seizure control was achieved, intermittent EEG monitoring was continued for a variable period depending on patient's clinical condition.


The outcomes of patients treated with propofol and MDL are given in Tables 4 and 5, and they are compared in Table 3. The propofol group had more patients (64%) with acute central nervous system (CNS) injury compared with the MDL group (33%) (Tables 1 and 2); and also had a greater median duration of SE before therapy (24 vs. 17 h; Table 3). However, median duration of therapy was greater with MDL (43 h) compared with propofol (31 h; Table 3).

Burst suppression or complete suppression was not attempted or achieved in eight of 20 patients (three in the propofol group and five in the MDL group). Nevertheless, propofol and MDL therapy achieved 64 and 67% complete clinical seizure suppression, 78 and 67% electrographic seizure suppression, and 29 and 17% elimination of all epileptiform discharges, respectively, and did not show any difference (all p values ≥0.61; Table 3). Subtle motor manifestations noted in seven patients were evaluated as subtle seizures due to lack of clinical–electrographic correlation.

Table 3 presents the rates of infectious complications, hemodynamic compromise, and number of days on ventilator. Hemodynamic compromise was seen in two of 14 patients treated with propofol but in none of six in the MDL group.

In addition to the overall mortality rates, subgroup mortality rates for moderate to severe disease (APACHE II score) were assessed. A comparison of overall mortality data did not show statistically significant difference (all p values ≥0.16). Patients with APACHE II score ≥20 treated with propofol had a higher mortality than did those treated with MDL (p = 0.05). Sensitivity analysis with reclassification of the patient treated with both agents to the MDL group resulted in loss of significance (p = 0.22).


Our small series of adult RSE patients treated primarily with propofol (n = 14) and MDL (n = 6) demonstrated no significant difference in the clinical or electrographic seizure control, infectious complications, hemodynamic compromise, and number of days on ventilator. These findings do not suggest differences in efficacy between the two RSE agents for seizure control and reduction in number of days on ventilator. A larger study is needed, particularly to compare the infectious complications and hemodynamic compromise. If comparable efficacy in seizure control were borne out in a large prospective randomized control study, tolerability with regard to hemodynamic compromise, complications, and mortality with these therapies might dictate the choice of RSE agents.

A higher percentage of patients treated with propofol showed electrographic seizure suppression than complete clinical seizure suppression. This could be due to lack of continuous EEG data available for review, including those of some motor manifestations that we included as clinical seizures. The majority of motor manifestations resembled those described by Treiman (31); nonetheless, they could also represent side effects or withdrawals from AEDs.

Our end point of RSE agent titration differed somewhat from other studies in that either elimination of electrographic/clinical seizure or achievement of electrographic burst suppression/complete EEG suppression was acceptable. Complete clinical seizure suppression seen in 65% of patients was comparable to seizure outcome (74%) achieved by Stecker et al. (2) using burst suppression/suppression as the end point of RSE titration. It is possible that lower doses of an RSE agent can eliminate the electrographic/clinical seizure without necessarily producing persistent burst suppression/suppression. Lower doses of RSE agents may decrease the frequency of complications including hemodynamic compromise and will still protect the brain from seizure-induced injury (4). Experimental work supports this theory, showing that high-frequency spike activity on the EEG is associated with neuronal injury, whereas occasional spikes or intermittent bursts of ictal activity do not cause apparent injury (33,34).

Overall mortality and subgroup mortality findings in patients with APACHE II score ≥20 were interesting. A higher mortality rate was found in the propofol-treated patients with APACHE II score ≥20 than in MDL-treated, but this finding was found after multiple comparisons and was lost in sensitivity analysis. The overall mortality rate was nonsignificantly higher with propofol (57%) than with MDL (17%) therapy, perhaps owing to the small sample size. Interestingly, a higher proportion of the propofol group had acute CNS injury and also had longer median duration of SE before therapy compared with the MDL group. Stecker et al. (2) also reported a nonsignificantly higher mortality rate with propofol (88%) compared with high-dose barbiturate therapy (50%). In both studies, small sample size limited the statistical power to detect a difference in overall mortality, although in both cases, it was close to 40%. A large prospective randomized series of RSE patients treated by these agents is needed to answer whether this finding is real.

In a small study of 16 RSE patients treated with high-dose barbiturates and propofol, overall mortality was 69% compared with 45% in our series (2). The patient populations may be different in these studies; therefore, a direct comparison is not possible. The lower mortality in our series may be partly attributed to lower median APACHE II score of 20 (range, 7–29) compared with 27.5 (range, 19–38) in the previous study. This finding suggests that the APACHE II score may be a potent predictor of outcome in RSE therapy.

Our study has several limitations. Nonrandom assignment to treatment groups may have resulted in imbalances in the groups. Specific patient characteristics, such as hemodynamic status at presentation, may have influenced treatment selection. The retrospective data collection limited the ability to characterize the patients fully. The small sample size renders the study underpowered, perhaps resulting in type II errors. Because of the exploratory nature of this study, we considered many potentially important predictors of outcomes. Because of the number of comparisons, the difference between the groups found may be due to chance.

A multiinstitutional prospective randomized study of RSE patients could examine efficacy, complication rates, and mortality. Design questions to be addressed include (a) which RSE agents should be compared; (b) what should be the titration goal, electrographic seizure elimination or burst suppression/suppression; and (c) which prespecified subgroup analyses must be considered when calculating the sample size. Because patient age, underlying etiology (acute CNS, acute non-CNS, and remote symptomatic), and disease severity all affect survival, any planned study would need to consider if one agent is superior in any particular subgroup. In considering a prospective trial with alpha set at 0.05 and beta at 0.2 (80% power) to detect a clinically meaningful difference in mortality rate of ∼40%, 56 patients, 28 in each group, would be needed for the primary outcome, and additional patients would be needed for planned subgroup analyses. Although both propofol and midazolam cost ∼$500 a day, other costs such as number of ICU days and number of days on ventilator may be important to consider in assessing the cost–benefit analysis of RSE therapy.