You are on page 1of 7


A new clinical score for the prognosis of status epilepticus in adults

!lez-Cuevasa, E. Santamarinaa, M. Toledoa, M. Quintanab, J. Salaa, M. Sueirasc, L. Guzmanc and
M. Gonza
J. Salas-Puiga

Epilepsy Unit, Neurology Department, Vall dHebron University Hospital, Universitat Autonoma de Barcelona, Barcelona; bNeurology
Department, Vall dHebron University Hospital, Universitat Autonoma de Barcelona, Barcelona; and cNeurophysiology Department, Vall
dHebron University Hospital, Universitat Autonoma de Barcelona, Barcelona, Spain


Keywords: Background and purpose: The severity of status epilepticus (SE) has an impor-
age, modified Rankin tant impact in clinical outcomes. The Status Epilepticus Severity Score
Scale, prognosis, score, (STESS) is a score for predicting mortality in SE at admission. The baseline
status epilepticus modified Rankin Scale (mRS) might be a prognostic factor for assessing the
short-tem outcomes of SE. Therefore, our aim was to evaluate the eectiveness
Received 25 January 2016 of mRS and whether its addition to the STESS improves the prediction of
Accepted 8 June 2016 mortality.
Methods: Consecutive patients with SE and aged >16 years were recruited
European Journal of
during 3 years. Receiver operating characteristic curves and a logistic regres-
Neurology 2016, 0: 17
sion model were developed to estimate the scores of the new score, designated
doi:10.1111/ene.13073 as modified STESS (mSTESS), and it was subsequently compared with the
Results: In all, 136 patients were included. Mean age was 62.01 ! 17.62 (19
95) years, and 54% were male. The capacity of the STESS to predict mortality
was 74.3% (95% confidence interval 63.8%81.8%), whilst the capacity of the
mRS to predict mortality was 65.2% (95% confidence interval 54.2%76.2%).
The logistic regression model and receiver operating characteristic curves
enabled the classification of mRS as follows: 0, mRS = 0; 1, mRS = 13; and
2, mRS > 3. These values, when added to the other items of the STESS,
resulted in the mSTESS with scores between 0 and 8 points. The capacity of
the mSTESS to predict mortality was 80.1%. An mSTESS > 4 established an
overall accuracy of 81.8% for predicting mortality, which was considerably
higher than the overall accuracy of STESS 3 (59.6%).
Conclusions: The baseline mRS was associated with high mortality risk. It is
proposed to use mSTESS to improve the prediction of mortality risk in SE.

Identifying the clinical factors that predict the out-

come of patients with SE is important because these
Status epilepticus (SE) is a serious medical emergency indicators may also be useful for clinical decision-
with an incidence of 10"41 per 100 000 people [1,2] making. Previous studies have focused on short-term
and results in short-term mortality in 7%39% of prognostic factors of SE and have demonstrated that
patients [35]. older age and acute symptomatic etiology are related
to a poorer outcome [3,5,6]. Results are less consistent
for other variables such as altered level of conscious-
Correspondence: M. Gonz! alez-Cuevas, Epilepsy Unit, Neurology ness [4], total duration of SE [7,8], time to treatment
Department, Vall dHebron University Hospital, Universitat or seizure duration before initiation of therapy [9]. A
Autonoma de Barcelona, Passeig Vall dHebron 119-121, Barcelona
recent study of serum concentrations of acute-phase
08035, Spain (tel./fax: +34 93 489 4257; e-mail: proteins measured early in SE revealed an indepen-
This study is not industry sponsored. dent association between albumin serum levels and

2016 EAN 1
2 !

mortality [10]. Another recent study found that the SE [19]. Patients with post-anoxic SE or incomplete
serum concentrations of procalcitonin levels measured clinical data were excluded.
at SE onset are independently associated with unfa- In addition to demographic characteristics, the type
vorable outcome [11]. of status (convulsive or non-convulsive), a previous
A simple clinical tool for predicting individual out- history of seizures, seizure semiology, level of con-
comes before obtaining the results of diagnostic tests sciousness before treatment, SE etiology and baseline
was proposed by Rossetti et al. [12]. These researchers mRS collected at arrival at the emergency department
developed a simple clinical score for the prognosis of were identified by interview of the patients family
SE in adults (Status Epilepticus Severity Score, or and by the patients history.
STESS) to use prior to initiating treatment. The score Seizure semiology was classified according to the
consists of four variables that are available at presen- worst manifestation prior to treatment (in descending
tation: history of seizures, age, seizure type and con- order of gravity: non-convulsive SE in coma, general-
sciousness impairment. It was concluded that the ized convulsive, complex partial, myoclonic or absence
STESS could identify patients with a high probability or simple partial). Level of consciousness prior to
of survival after SE. Later, a prospective study by the treatment was categorized as alert, somnolent or con-
same group confirmed that the STESS is able to reli- fused, stuporous and comatose. SE etiology was clas-
ably identify SE patients who are likely to survive and sified according to the last report of the International
proposed that, in these cases, an early aggressive treat- League Against Epilepsy [19] as being acute
ment could not be routinely justified, to reduce the symptomatic, remote symptomatic, progressive symp-
inherent risks that are related to an aggressive treat- tomatic or idiopathic/cryptogenic. Clinical outcome
ment plan. However, the STESS has low predictive was assessed at hospital discharge (dead or alive).
value for bad outcomes and has a ceiling eect espe- The mRS was scored as follows: 0, no symptoms; 1,
cially in patients older than 65 years without pre-exist- no significant disability despite symptoms (i.e. able to
ing epilepsy [12,13]. carry out all usual duties and activities); 2, slight dis-
The modified Rankin Scale (mRS) is an objective, ability (i.e. unable to carry out all previous activities
reliable measure of disability or dependence in daily but able to look after own aairs without assistance);
activities of a patient. It is commonly used in the eval- 3, moderate disability (i.e. requiring some help but able
uation of patients with neurological emergencies, such to walk without assistance); 4, moderately severe dis-
as stroke or SE, as a tool to measure their functional ability (i.e. unable to walk without assistance and
state [14,15]. This state may on many occasions be unable to attend to own bodily needs without assis-
influenced by the presence of dierent comorbidities, tance); 5, severe disability (i.e. bedridden, incontinent
and some recent studies have suggested that coexisting and requiring constant nursing care and attention) [17].
comorbidities might be associated with poor short- The STESS relies on the assessment of age, where
term prognosis in SE [1618]. under 65 years of age is 0 points and 65 years of age
Therefore, it is hypothesized that the patients base- or older is 2 points; previous history of seizures is 0
line mRS may be a prognostic factor that determines and no previous seizures 1; severity of seizure type
the short-term outcome of SE and that the inclusion (simple partial, complex partial, absence, myoclonic as
of the baseline functional condition of the patient (i.e. complicating idiopathic generalized epilepsy 0; gener-
the mRS) in the STESS can improve the prediction of alized convulsive 1; non-convulsive SE in coma 2);
mortality. and level of consciousness (alert or somnolent/con-
fused 0; stuporous or comatose 1). A score of 02 is
defined as favorable, which indicates a low risk of
death, and a score of 3 points or higher is defined as
A retrospective registry of all adult patients (16 years unfavorable [12].
of age and older) who experienced SE and were A prognosis was made according to the likelihood
admitted to our hospital between March 2011 and of patient death before hospital discharge.
March 2014 was analyzed. The local ethics committee
authorized the study without obtaining informed con-
Statistical analysis
sent from all patients.
Status epilepticus was defined as a condition result- Statistical analysis was undertaken using SPSS Statis-
ing either from the failure of the mechanisms respon- tics for Windows, version 17.0 (SPSS Inc, Chicago,
sible for seizure termination or from the initiation of USA).
mechanisms which lead to abnormally prolonged sei- Prognosis was assessed according to the clinical out-
zures after 5 min for tonic-clonic SE and 10 min focal come (alive or dead) at discharge. To determine

2016 EAN

variables associated with mortality, statistical signifi- The demographics and the most relevant clinical
cance was assessed by Pearsons chi-squared or Fish- variables of the cohort were classified by outcome and
ers exact test for categorical variables and the are presented in Table 1. In the univariate analysis, it
Students t or MannWhitney U test for numerical was found that age, non-convulsive SE in coma,
variables. The cuto age with best sensitivity and speci- marked consciousness impairment, etiology, the
ficity to predict mortality was determined through a STESS and mRS were associated with higher mortal-
receiver operating characteristic (ROC) curve. ROC ity.
curves were also obtained to assess the capacity of the The use of anesthetic treatment was also included
mRS, the STESS and the modified STESS (mSTESS) in the univariate analysis, and even though mortality
to predict death. The predictive capacity was deter- was higher in patients requiring anesthetics this dier-
mined as the area under the ROC curve with a confi- ence was not statistically significant. Regarding age,
dence interval (CI) of 95%. A logistic regression model the best cuto point to predict mortality in our sam-
was performed to assess the mRS and STESS as inde- ple (ROC curve) was 70 years old (P = 0.003).
pendent predictors of death using the Hosmer and The results of the logistic regression analysis
Lemeshow statistic to assess the goodness of fit. The showed that the STESS (OR 1.863, 95% CI 1.345
score added to the STESS that was used to obtain the 2.577; P < 0.001) and mRS (OR 1.459, 95% CI
final mSTESS scale was based upon the coecient val- 1.0042.19; P = 0.047) were the only independent pre-
ues (b coecients) of the logistic regression after con- dictors of mortality. The data model had an excellent
sidering the results of the ROC curves for each cuto goodness of fit (Hosmer and Lemeshow test,
point of mRS used to predict mortality. Sensitivities, P = 0.240) with a regression coecient of the STESS
specificities, as well as positive and negative predictive [b 0.622, SE(b) 0.166] which is almost twice that of
values (PPVs and NPVs) were calculated to evaluate the mRS [b 0.378, SE(b) 0.190].
the prediction accuracy for the best cuto points of the Considering both the values of the coecients of
STESS and the mSTESS. The overall accuracy for the the logistic regression, where a higher weight in the
cuto point of each scale was obtained using the pro- STESS was observed, and the results of the ROC
portion of patients that were classified for the predic- curve of mRS that was used to predict mortality
tion of mortality. An ROC curve was also performed [where two good cutos were obtained, one very sen-
to establish cuto points of the mSTESS to classify sitive >0 (sensitivity 96.4%, specificity 23.1%) and the
patients into dierent risks of mortality. MantelHaen- other very specific >3 (sensitivity 21.4%, specificity
szel v2 tests were used to test the homogeneity of the 94.4%)], it was possible to divide the mRS into three
odds ratio (OR) across strata. A P value <0.05 was con- groups, to which the following scores were assigned:
sidered to be statistically significant. 0, mRS = 0; 1, mRS = 13; and 2, mRS > 3. These
scores were added to the STESS of each patient. The
cuto point of age to 70 years old was also modified
following our ROC curve results regarding age.
Of the 162 patients who were registered as having SE, Finally a new score that was variable between 0 and 8
136 were included and 26 were excluded due to insu- points was obtained, the modified STESS (mSTESS;
cient clinical data (seven cases) or to hypoxic etiology Table 2). The STESS, mRS and mSTESS predictive
(19 cases). capacities are detailed in Fig. 1, where it is noted that
The mean age was 62.01 ! 17.62 (1995) years, the mSTESS has a greater capacity to predict mortal-
and 54.4% of patients were males. Also, 60.2% of ity (80.1%, 95% CI 70.6%89.6%) compared to
patients had a de novo SE episode. The total mortality STESS (74.3%, 95% CI 63.8%81.8%).
was 22.1%. The most common etiologies of SE were The best cuto point to predict mortality in the
acute symptomatic (47.1%) and remote symptomatic mSTESS was a score of 4. When the established cuto
(32.4%), which was mainly secondary to vascular of STESS 3 was compared, an improvement in the
(30.9%) and tumor (19.9%) lesions. Of the patients capacity to predict death was found with a higher over-
who died, 14% were due to status itself (homeostatic all accuracy (81.8% vs. 59.6%), as shown in Table 3.
failure), 34% were from complications during hospi- This new scale provides a better PPV than the
talization, mainly respiratory infections, and 52% died STESS, thereby allowing those patients with a higher
due to the etiology causing the status (17% acute risk of mortality to be better identified. Furthermore,
stroke, 6% intracranial hemorrhage, 10% subarach- in the ROC curve another cuto point was observed
noid hemorrhage, 10% meningitis, 3% encephalitis that allowed the patients with a very low risk of death
and 6% due to traumatic brain injury leading to brain (mSTESS < 02) to be qualified successfully, enabling
edema). three types of patients to be dierentiated according

2016 EAN
4 !

Table 1 Variables of interest to predict risk of mortality

Total 136 (100%) Alive 106 (77.9%) Dead 30 (22.1%) P

Age (mean, SD) 62.01 (!17.6) 60.1 ! 17.2 68.5 ! 17.6 0.022
Male 74 (54.4%) 56 (52.8%) 18 (60%) 0.486
History of previous seizures 54 (39.7%) 46 (43.4%) 8 (26.7%) 0.098
Seizure type
SP or CP 80 (58.8%) 68 (64.1%) 12 (40%) 0.018
GC 42 (30.9%) 33 (31.1%) 9 (30%) 0.906
NCSEC 14 (10.3%) 5 (4.7%) 9 (30%) <0.001
Alert or somnolent/confused 92 (67.6%) 82 (77.3%) 10 (33.3%) <0.001
Stuporous or comatose 44 (32.4%) 24 (22.6%) 20 (66.6%)
mRS (median, IQR) 1 (12) 1 (12) 2 (13) 0.002
Acute symptomatic 64 (47.1%) 43 (40.6%) 21 (70%) 0.004
Remote symptomatic 44 (32.4%) 41 (38.7%) 3 (10%) 0.003
Progressive symptomatic 11 (8.1%) 10 (9.4%) 1 (3.3%) 0.455
Cryptogenic/idiopathic 17 (12.5%) 12 (11.3%) 5 (16.7%) 0.531
STESS (median, IQR) 3 (13) 2 (13) 3.5 (25) <0.001
Anesthetic treatment 41 (30.1%) 28 (26.4%) 13 (43.3%) 0.075

CP, complex partial; GC, generalized convulsive; IQR, interquartile range; mRS, modified Rankin Scale; NCSEC, non-convulsive status epilep-
ticus with coma; SP, simple partial; STESS, Status Epilepticus Severity Score.

Table 2 Modified STESS (mSTESS) 1.0

0.8 Reference line
Alert or somnolent/confused 0
Stuporous or comatose 1
Seizure type
SP or CP 0 0.6

GC 1
Age (years)
<70 0 0.4
70 2
History of seizures
Area under
SE 95% CI
Yes 0 ROC curve
STESS (06) 0.743 0.054 0.6380.818
No 1 0.2 mRS (05) 0.652 0.056 0.5420.762
mRS mSTESS (08) 0.801 0.049 0.7060.896
0 0
13 1
4 2 0.0
Total 08 0.0 0.2 0.4 0.6 0.8 1.0
1 - specificity
CP, complex partial; GC, generalized convulsive; mRS, modified
Rankin Scale; mSTESS, modified Status Epilepticus Severity Score; Figure 1 Capacity of the new scale versus the STESS to predict
NCSEC, non-convulsive status epilepticus with coma; SP, simple mortality.
partial; STESS, Status Epilepticus Severity Score.

patients in the group with an mSTESS score of 34

to risk of death: mSTESS 02, low risk; 34, interme- (intermediate risk), 21.2% of patients died in the
diate risk; >4, high risk (Fig. 2). exposed to anesthetic drugs group versus 25% of
Whether the decision to start sedation was associ- patients who were not. Amongst patients with an
ated with mortality for dierent scores in the mSTESS mSTESS score >4 (high risk), 50% of patients died in
was also analyzed using a stratified analysis. Amongst the unexposed group versus 70% of patients in the
patients with an mSTESS score of 02 (low risk), exposed group. The dierences were not significant
2.3% died following exposure to anesthetic drugs ver- amongst the dierent risk groups after obtaining an
sus 9.1% of patients who were not exposed. Amongst estimate of a common OR score of 1.74 (95% CI

2016 EAN

Table 3 Comparison of test performances between STESS 3 and mSTESS > 4

Sensitivity (%) Specificity (%) PPV (%) NPV (%) accuracy (%) OR (CI 95%) P

STESS 3 73.3 55.7 31.9 88.1 59.6 3.452 (1.4108.451) 0.007

mSTESS > 4 50 90.4 58.3 87.0 81.8 9.400 (3.50425.217) <0.001

CI, confidence interval; mSTESS, modified Status Epilepticus Severity Score; NPV, negative predictive value; OR, odds ratio; PPV, positive
predictive value; STESS, Status Epilepticus Severity Score.

predicting mortality in SE by classifying age into

seven groups by assigning specific scores for each dec-
ade of life (from age 21 to >80). Nevertheless, it was
observed that mortality rates might change over time
in relation to age and examples were cited of out-
comes in a population-based study that was con-
ducted between 1982 and 1986 and included patients
that were substantially worse than patients in a recent
study [20,21]. This was particularly true for patients
over 60 years of age. Therefore, the age score in the
EMSE and STESS should be adopted, especially
because there has been a significant improvement in
the survival rate of certain age groups. Conversely,
there were patients who, despite being young, may
have a higher risk of mortality in relation to their
Figure 2 Prediction of mortality based upon the mSTESS value medical history and basal condition. In our series the
according to risk group.
best cuto point for predicting mortality was
70 years.
0.664.57) with an elevated degree of homogeneity It has been demonstrated that comorbidities may
between the groups (P = 0.264). Therefore, mortality influence the prognosis of SE. Thus, it has been sug-
in our series did not depend on sedation, regardless of gested in previous studies that patients with a higher
the dierent mSTESS scores. number of comorbid conditions have a worse out-
come [16]. In another prospective study, the role of
pre-existing comorbidities on the outcome of patients
with SE (assessed with the Charlson Comorbidity
From a therapeutic perspective, it is important to Index) was examined. It was found that this measure
have tools that enable an accurate prognosis of was not independently related to outcome and
patients with SE in an early stage. The STESS is a instead a model that included etiology, the STESS
useful tool in this regard and has already been vali- and each variable of the Charlson Comorbidity
dated in previous studies; however, this tool has a Index was proposed. This produced an excellent pre-
ceiling eect, particularly in patients older than dictive model for mortality [17]. Data provided by
65 years of age who do not have pre-existing epilepsy the EMSE suggest an important role for comorbidi-
[13,14,20]. ties [18].
The age of patients had been previously determined Dierent comorbidities may impact on the func-
by other studies to be an important risk factor for tional state of patients; therefore it is hypothesized
mortality after SE [4,5,13,16]. Given that older that the use of an easy and fast score such as the
patients are less resistant to complications of SE and mRS to evaluate the functional state at emergency
its treatment, such as pneumonia, this could be one of arrival could be useful for predicting the prognosis of
the reasons why these patients result in a higher inci- patients with SE. This score has been demonstrated to
dence of unfavorable outcomes. be a useful tool in other neurological emergencies,
To determine mortality in relation to age, Towne such as acute stroke [14].
et al. used a range of 30 years [7]. Rossetti et al. cate- In a very recent study performed to identify factors
gorized into age younger than 65 years or 65 and influencing long-term outcome in refractory SE the
older [4,13]. In a recent study by Leitinger et al. functional outcome was quantified using the mRS.
another score (the EMSE) was proposed for For every refractory SE episode two outcomes were

2016 EAN
6 !

evaluated, one at discharge and one long-term out- Other functional scales, such as the Barthel scale,
come, and compared to baseline mRS. In the results it may be more reliable and less subjective in assessing
can be observed that patients who had a poor long- disability; however, these scales are longer and less
term outcome had a slightly higher median in mRS suitable for rapid patient assessment.
compared with patients with a good outcome [3 (14) One limitation of this study is that it is a retrospec-
vs. 2 (04)], but this dierence was not statistically sig- tive cohort series from a single center, and it has not
nificant (P = 0.233). However, this study focused only been validated. Another limitation is that mortality
on refractory SE [22]. was considered as the primary outcome and only clini-
In our series, it was observed that the baseline func- cal factors were considered. Electroencephalogram pat-
tional condition that was also defined by the mRS terns and other variables that might influence the
was independently associated with mortality in results such as seizure duration before initiation of
patients with SE. By adding this variable to the therapy and total duration of SE were ignored because
STESS (increasing the cuto point of age), a new the aim was to focus only on the factors that can be
score was obtained: the modified STESS (4 mSTESS, evaluated very early in the first assessment of the
NPV 87.0%, PPV 58.3%, overall accuracy 81.8%). patient. Also, like many clinical studies, only mortality
This new score not only allowed the prediction of on discharge was investigated and a longer period of
mortality to be improved but also allowed patients evaluation would be very interesting for future studies.
who had a very low risk of death (mSTESS 2) to be
With this new scale, it was observed that the ceiling
eect of the STESS decreases; e.g. a 70-year-old The STESS is a useful tool in predicting mortality in
patient without pre-existing epilepsy and a good base- patients with SE at admission; these predictions can
line situation (mRS = 0) would score 3 points on our be improved by adding baseline disability, as assessed
scale that would correspond to intermediate risk of by the mRS, and increasing the cuto point of age to
mortality and not as an unfavorable prognosis as in 70 years. A new proposal, the modified STESS, can
the STESS. In addition the higher value of PPV in be more accurate in the short-term prognosis of
our score allows us to better identify those patients patients with SE.
with a really higher risk of mortality.
On the other hand, a recent study performed for
intensive care unit patients with SE has shown that
the use of intravenous anesthetic drugs in SE was The authors thank the ER neurology and neurophysi-
associated with an increased relative risk of death ologist consultants of the Vall dHebron University
independent of possible confounders and without sig- Hospital, Spain, for their help with data acquisition.
nificant changes in risk by dierent grades of SE Furthermore, we want to thank Dr Andrea O. Ros-
severity and dierent SE etiologies [23]. The analyzed setti for reviewing the manuscript and for advice.
cohort of these studies was exclusively composed of
patients seen in intensive care units, thus potentially
Disclosure of conflicts of interest
limiting the generalizability. In our series, globally, a
higher mortality was observed in patients who The authors declare no financial or other conflicts of
required the use of anesthetics as part of treatment; interest.
however, this result was not statistically significant
(P = 0.075). Moreover the mortality in our series did
not depend on coma induction, regardless of the dif-
ferent STESS scores. This result is similar to that pre- 1. DeLorenzo RJ, Hauser WA, Towne AR. A prospective,
population-based epidemiologic study of status epilepti-
viously reported by Rossetti et al. [13]. It has to be
cus in Richmond, Virginia. Neurology 1996; 46: 1029
taken into account that our analyses had a retrospec- 1035.
tive design, and findings were not adjusted for all 2. Knake S, Rosenow F, Vescovi M, et al; Status Epilepti-
known outcome predictors. cus Study Group Hessen (SESGH). Incidence of status
Following the methodology of Rossetti et al. [12,13] epilepticus in adults in Germany: a prospective, popula-
tion-based study. Epilepsia 2001; 40: 759762.
etiology was not included in our new score because
3. Logroscino G, Hesdorer DC, Cascino G, Annegers JF,
the cause can only be determined after several exami- Hauser WA. Short-term mortality after a first episode
nations that require a delay of up to several hours or of status epilepticus. Epilepsia 1997; 38: 13441349.
days, which can be crucial for therapeutic manage- 4. Rossetti AO, Hurwitz S, Logroscino G, Bromfield EB.
ment. Prognosis of status epilepticus: role of aetiology, age,

2016 EAN

and consciousness impairment at presentation. J Neurol 14. Bonita R, Beaglehole R. Modification of Rankin Scale:
Neurosurg Psychiatry 2006; 77: 611615. recovery of motor function after stroke. Stroke 1998; 19:
5. Chin RFM, Neville BGR, Scott RC. A systematic 14971500.
review of the epidemiology of status epilepticus. Eur J 15. Rathakrishnan R, Sidik NP, Huak CY, Wilder-Smith
Neurol 2004; 11: 800810. EP. Generalized convulsive status epilepticus in Singa-
6. Hui A, Joyn GM, Li H, Wong KS. Status epilepticus pore: clinical outcomes and potential prognostic mark-
in Hong Kong Chinese: aetiology, outcome and pre- ers. Seizure 2009; 18: 202205.
dictors of death and morbidity. Seizure 2003; 12: 478 16. Koubeissi M, Alshekhlee A. In-hospital mortality of
482. generalized convulsive status epilepticus: a large US
7. Towne AR, Pellock JM, Ko D, DeLorenzo RJ. Deter- sample. Neurology 2007; 69: 886893.
minants of mortality in status epilepticus. Epilepsia 17. Alvarez V, Januel JM, Burnand B, Rossetti AO. Role of
1994; 35: 2734. comorbidities in outcome prediction after status epilepti-
8. Lowenstein DH, Alldredge BK. Status epilepticus at an cus. Epilepsia 2012; 53: e89e92.
urban public hospital in the 1980s. Neurology 1993; 43: 18. Leitinger M, H oller Y, Kalss G, et al. Epidemiology-
483488. based mortality score in status epilepticus (EMSE). Neu-
9. Sagduyu A, Tarlaci S, Sirin H. Generalized tonic-clinic rocrit Care 2015; 22: 273282.
status epilepticus: causes, treatment, complications and 19. Trinka E, Cock H, Hesdorer D, et al. A definition and
predictor of fatality. J Neurol 1998; 245: 640646. classification of status epilepticus Report of the ILAE
10. Sutter R, Grize L, Fuhr P, R uegg S, Marsch S. Acute Task Force on Classification of Status Epilepticus.
phase proteins and mortality in status epilepticus: a five- Epilepsia 2015; 56: 15151523.
year observational cohort study. Crit Care Med 2013; 20. Sutter R, Kaplan PW, R uegg S. Independent external
41: 15261533. validation of the status epilepticus severity score. Crit
11. Sutter R, Valenca M, Tschudin-Sutter S, R uegg S, Care Med 2013; 41: e475e479.
Marsch S. Procalcitonin and mortality in status epilepti- 21. Neligan A, Shorvon SD. Prognostic factors, morbidity
cus: an observational cohort study. Crit Care 2015; 9: and mortality in tonic-clonic status epilepticus: a review.
361. Epilepsy Res 2011; 93: 110.
12. Rossetti AO, Logroscino G, Bromfield EB. A clinical 22. Mad#zar D, Geyer A, Knappe RU, Gollwitzer S, Kura-
score for prognosis of status epilepticus in adults. Neu- matsu JB, Gerner ST. Association of seizure duration
rology 2006; 66: 17361738. and outcome in refractory status epilepticus. J Neurol
13. Rossetti AO, Logroscino G, Milligan TA, Michaelides 2016; 263: 485491.
C, Rueux C, Bromfield EB. Status Epilepticus Severity 23. Sutter R, Marsch S, Fuhr P, Kaplan PW, R uegg S.
Score (STESS): a tool to orient early treatment strategy. Anesthetic drugs in status epilepticus: risk or rescue? A
J Neurol 2008; 255: 15611566. 6-year cohort study. Neurology 2014; 82: 656664.

2016 EAN