Predictive Factors for Refractory Convulsive Status Epilepticus: a 7-year Retrospective Study

Marc Laurence L. Fernandez, Leah L. Shiong Shu, Natasha L. Fabiana, 
Paul Matthew D. Pasco, Leonor I. Cabral-Lim, and Josephine C. Gutierrez

Department of Neurosciences, College of Medicine,
Philippine General Hospital, University of the Philippines Manila


Status epilepticus (SE) is a clinical condition warranting emergent neurologic intensive care  management.  It  is characterized as onset of epileptic seizure or series of seizure attacks lasting for 30 minutes  without regaining consciousness.Generalized convulsive SE (GCSE) is described as continuous, convulsive epileptic episodes lasting more than 5 minutes, or two or more seizures, with impairment of consciousness in between seizures.2,3 On the other hand, non-convulsive SE (NCSE) is defined as change in mental status from baseline for at least 30 minutes, with associated ictal discharges on electroencephalographic (EEG) studies.2

Etiologies of SE are variable. Studies reveal that at least one-third of the SE cases are due to an exacerbation of a previously existing seizure disorder, while another one-third of the cases are  first onset  or de novo  seizure episodes. Previous or new onset cerebrovascular disease, subarachnoid hemorrhage, CNS infections, hypoxic injury, isoniazid toxicity, and alcohol withdrawal are known precipitating factors for SE.2

Worldwide, the incidence of SE ranges from 10 to 50 per 100,000 persons.2 SE-related mortalities range from 7 to 20%.

About 31%  to 43% of patients with SE proceed to become refractory SE (RSE), further classified into convulsive RSE (RCSE)  and the non-convulsive type (RNCSE).3,7 RCSE is defined as convulsive seizure activity lasting more than 60 minutes that fails to respond after firstand second-line anti-epileptic drugs (AEDs) are instituted.The development of RCSE may be due to delays in recognition and management, and by the severity of other underlying conditions. In the Veteran Administrative (VA) cooperative study, about 56% of the SE cases responded to first-line treatment regimen, while about 7% resolve after subsequent administration of second-line regimen, and 3% after the third-line agent.2,5 The pathophysiology behind the failure of epileptic  seizures  to respond to the treatment regimens in RCSE is still being studied.

In general, RSE is associated with a poor prognosis, as only one-third of the cases would return to their pre-morbid states. Patients  with RSE usually develop systemic complications such as rhabdomyolysis, pulmonary edema, cerebral edema, hyperglycemia, and multi-organ failure resulting in a longer hospital stay.It is also associated with higher mortality rates (68%) on 30-day outcome  studies, especially if associated with severe pre-morbid conditions.2,6

This study aims to determine the incidence of RCSE among patients with SE within a 7-year period in a tertiary hospital.  Specifically, this  study would  also identify risk factors associated with the progression of SE into RCSE.


This is a retrospective, descriptive study of adult patients admitted for SE. All patients, with a diagnosis of status epilepticus managed by the adult neurology service of the Philippine General Hospital  from January 2003  to September 2010 were included in the study. Patients were identified using manual screening of patient records, annual census, and computerized search. A total of 108 patients with SE were identified;  however,  only 66  records were available. Of these, patients who  had  uncontrolled generalized tonic clonic seizures lasting more than 60 minutes and had been given two or more intravenous antiepileptic loading doses, drips, or anesthetic agents were classified as RCSE. 

The following variables  were obtained: age, sex, etiology of SE, presence of co-morbidities, time of first-onset of seizure, compliance to SE protocol upon admission, AEDs, EEG, cranial imaging, blood chemistry (serum Na), CSF studies, length of hospital stay, and outcomes. Retrievable records of EEG tracings and digital records were reviewed using the Neurofax µ EEG-9100 software (Nihon Kohden, Japan).

Univariate analysis using Pearson’s  correlation coefficient (chi  square test)  was used to determine the association of the different variables with patient outcomes. Significant variables on univariate analysis (p-value<0.05) were included in the multivariate analysis using logistic regression analysis. Significant correlation is defined by a p-value of <0.05. Stepwise logistic regression was performed on the variables to further limit the number of significantly correlated variables. Sensitivity and specificity of the variables most significantly associated with the dependent variable would be computed. Data were analyzed using Stata 9.2 (Stata Corp, LP) software.


Sixty six (66) patients clinically diagnosed as SE were included  in the study. Of these, 11 patients (16%) were identified to have RCSE. The mean age of patients with RCSE is 33.18±20.1  years.  Eight of the RCSE cases (73%) developed from first-onset seizures. About 55% of the RCSE cases (n=6) were caused by acute viral meningoencephalitis. The diagnosis of viral CNS infection was based on clinical symptomatology, CSF pleocytosis, electroencephalographic and neuro-imaging findings. However, only 1 case of viral encephalitis was confirmed using herpes simplex virological test.  Other notable etiologies for RCSE are previously uncontrolled seizure disorder (27%, n=3) and brain tumors (18%, n=2). 

EEG was performed in 34 out of the 66 SE patients. Ten of these EEGs had normal results but these tests  were performed after seizure control. Of the 11 RCSE patients, only 9 patients had EEG studies (Table 1). The presence of generalized slowing, intermittent rhythmic delta activity, periodic lateralized epileptiform discharges, and generalized spike and wave complexes were observed in all of the RCSE EEG studies.

Table  1. Factors associated with the development of RCSE among patients with SE

Magnetic resonance imaging (MRI) and computed tomography (CT) scans were performed in 52 out of 66 SE patients (Table 1). Nine out of the 11 RCSE neuroimaging studies revealed abnormalities such as cerebral edema, white matter changes, and focal post-contrast enhancements (tumoral lesions).

Seizure control in 45%  of  the  RCSE  cases (n=5)  was achieved using three or more IV AEDs. Midazolam drip (0.1 mg/kg/hr) was used as a second-line IV AED in six cases (60%) of RCSE, while Phenobarbital (20mg/kg) and Valproic Acid  (20mg/kg)  were used in  four  cases (40%). Levetiracetam (1 g IV loading dose) was used as an adjunct IV therapy in four cases (40%). Seizure control in a patient with previously diagnosed epilepsy was achieved using Propofol (6 mg/kg/hr) in addition to four other IV AEDs.

Table 1 shows the risk factors associated with the development of RCSE among patients with SE. Using univariate analysis, only three variables were significantly correlated with RCSE among patients with SE. These risk factors are: first-onset seizures (p=0.036), abnormal cranial
imaging (p=0.008), and serum sodium levels <135 (p=0.034).

However, using multivariate analysis (Table 2), only two variables were significantly associated with RCSE: first onset seizures (p=0.0214) and abnormal cranial imaging (p=0.0131). Sensitivity of first onset seizures and abnormal cranial imaging as predictive factors for developing RCSE is
81.82%. On the other hand, the  specificity of  the two variables is 52.7 and 63.4%, respectively.

Table 2. Association of Significant Risk factors for RCSE by Multivariate Analysis

If both variables are taken together, the probability of predicting progression to RCSE using the area under the receiver-operator characteristic (ROC) curve, which compares the true positive rate and false negative rates, is 0.8891. This translates to an 88.91% probability of an RCSE event occurring if the two variables, first-onset seizures and abnormal cranial imaging, are  present at the time of the convulsive seizure attack.

The average length of hospital stay was 31.5±97.30 hours. The range of hospital stay was from 4 to 115 days. The mortality rate for RCSE was 3 in 11 cases (27%).


RCSE is one of the most common neurologic emergencies with high rates of mortality and morbidity. There is paucity of comprehensive studies that analyze its underlying factors, therapeutic response, and management outcomes.1,2 However, it is important to identify patients who are at risk of developing RCSE to prevent long-term deleterious complications.1

In this study, the prevalence of RCSE among patients with SE was 16%. This is lower than a previous German study where 43% of patients with SE developed into RSE.The difference  may be due to the heterogeneity of the populations under study.Management of RSE also varied across different neurocritical care facilities.The actual rate of RCSE in this study may be higher if patients had been monitored in a neurological intensive care unit, or if they had been referred immediately at the onset of the symptoms.

A recent related retrospective study revealed that de novo RCSEs generally have poorer prognosis, causing a mortality rate of 55%, and development of long-term neurologic sequelae such as post-SE symptomatic seizures in 29% of the patients.7

Abnormal cranial imaging was also strongly correlated with RCSE. Of the 24  patients with abnormal CT/MRI findings, a total of 10 (42%) had RCSE. Most of these abnormalities were consistent with CNS infections, brain metastasis, or encephalomalacia. These underlying structural pathologies cause variedly distributed epileptogenic foci which may predispose the patient to SE, and eventually RCSE.A retrospective study on the predictors of RCSE in a German neurologic facility revealed 88.9% of the patients with RSE were secondarily caused by focal CNS diseases.5

All patients (n=6)  with viral CNS infection initially presenting with uncontrolled first-onset seizures proceeded into RCSE. In a study by Misra et al. (2008), about 37 out of 93 patients (40%) with CNS infection  develop  into SE.9 However, the proportion of patients progressing from SE to RSE was only 10.9%. This is comparable to our data (9.1%, n=6). Encephalitis is cited in several studies as an important etiologic predictor for developing SE and RCSE. A recent study in India also demonstrated that a large proportion of RCSE patients have CNS infections (44.4%), particularly viral encephalitis.10 Furthermore, a subset of RCSE would further progress to a protracted course. The super refractory SE (SRSE) is defined as SE continuing for greater than 24 hours after initiation of anesthetic therapy, including cases of SE recurring upon withdrawal or reduction of the anesthesia.11 A recent Chinese study  indicated that  a majority of SRSE is caused by viral encephalitis (67.7%).12 Encephalitis usually has multiple focal epileptogenic  foci which are cortically distributed. Other cases such as herpes virus encephalitis may also cause intractable seizures, as the virus has predilection for the temporal lobe, a region with a low seizure threshold.In summary, early determination and identification of risk factors for SE is important to prevent progression to RSE. However, this study did not  determine whether discharged patients returned to their pre-morbid functional baseline or if these patients developed long-term neurologic complications. In addition,  patients’ Glasgow Outcome Scores (GOS) and APACHE II scores should be obtained on admission and compared to pre-discharge scores but were not in this study.

This study identified two predictive factors for developing RCSE: abnormal cranial imaging and first-onset seizures.  The presence  of both variables can predict occurrence of RCSE with a relatively high sensitivity rate. The study is limited by the small number of patients identified with RCSE, as well as limited information regarding the variables leading to RCSE.


  1. Mayer SA, Claassen J, Lokin J, Mendelsohn F, Dennis LJ, Fitzsimmons BF. Refractory status epilepticus: frequency, risk factors, and impact on outcome. Arch Neurol. 2002; 59(2):205-10.
  2. Shorvon S, et al. The management of status epilepticus. J Neurol Neurosurg Psychiatry. 2001; 70(suppl II):ii22–ii27.
  3. Bleck T, et al. Refractory status epilepticus. Curr Opin Crit Care. 2005; 11:117-20.
  4. Holtkamp M, Othman J, Buchheim K, et al. Predictors and prognosis of refractory status epilepticus treated in a neurological intensive care unit. J Neurol Neurosurg Psychiatry. 2005; 76:534-9.
  5. Lowenstein DH, Bleck T, Macdonald RL. It’s time to revise the definition of status epilepticus. Epilepsia. 1999; 40:120-2.
  6. Misra UK, Kalita J, Nair PP. Status epilepticus in central nervous system infections: an experience from a developing country. Am J Med. 2008; 618-23.
  7. Treiman DM, Meyers PD, Walton NY, et al. for the Veterans Affairs Status Epilepticus Cooperative Study Group. A comparison of four treatments for generalized convulsive status epilepticus. N Engl J Med. 1998; 339:792-8.
  8. Tsai MH, Chuang YC, Chang HW, et al. Factors predictive of outcome in patients with de novo status epilepticus. QJM. 2009; 102(1):57-62.
  9. Kalita J, Nair PP, Misra UK. A clinical, radiological and outcome study of status epilepticus from India. J Neurol. 2009; 257(2):224-9
  10. Vootori S,  et al. Prognosis and predictors of outcome of refractory generalized convulsive status epilepticus in adults treated in neurointensive care unit. Clin Neurol Neurosurg. 2014; 126C:7-10.
  11. Shorvon S,  Ferlisi M.  The treatment of super-refractory status epilepticus: a critical review of available therapies and a clinical treatment protocol. Brain. 2011; 134:2802-18.
  12. Tian L, Li Y, Xue X, et al. Super-refractory status epilepticus in West China. Acta Neurol Scand. 2014. doi:10.1111/ane.12336