Mortality,morbidityandrefractorinesspredictioninstatusepilepticus: Comparisonof

STESSandEMSEscores
Giada Giovanninia,b, Giulia Montia,b, Manuela Tondellia, Andrea Marudic,

Franco Valzaniab, Markus Leitingerd,e, Eugen Trinkad,e,f, Stefano Melettia,b,*

a DepartmentofBiomedical, Metabolic, andNeural

Science, CenterforNeuroscience andNeurotechnology,
UniversityofModenaandReggioEmilia,Modena, Italy
b UnitofNeurology,OCSAE Hospital, AOU
Modena, Italy c Intensive CareUnit,OCSAEHospital, AOU
Modena, Italy
d
DepartmentofNeurology,ChristianDoppler Klinik, Paracelsus MedicalUniversity,Salzburg, Austria e Centerfor
CognitiveNeuroscience, Salzburg, Austria
f
PublicHealth,Health Services ResearchandHTA,Universityfor Health Sciences,Medical InformaticsandTechnology,Halli.T., Austria

A R T I C LE I N F O Purpose: Status epilepticus (SE) is a neurological emergency, characterized by high short-term morbidity and mortality. We evaluated
and compared two scores that have beendeveloped to evaluate status epilepticus prognosis: STESS (Status Epilepticus Severity Score)
and EMSE(EpidemiologybasedMortality score inStatusEpilepticus).
Article history: Methods: A prospective observational study was performed on consecutive patients with SE admitted between September 2013 and
Received9November2016 August 2015. Demographics, clinical variables, STESS-3 and -4, and EMSE64 scores were calculated for each patient at baseline. SE
Receivedinrevisedform29January2017 drugresponse, 30-daymortalityand morbidity weretheoutcomes measure.
Accepted30January2017
Results: 162 episodes of SE were observed: 69% had a STESS 3; 34% had a STESS 4; 51% patients had an EMSE 64. The 30-days
mortalitywas31.5%:EMSE-64 showed greater negative predictive value (NPV) (97.5%), positivepredictive value(PPV)(59.8%) and
accuracy in the prediction of death than STESS-3 and STESS-4 (p < 0.001). At 30 days, the clinical condition had deteriorated in 59% of
Keywords: status
the cases:EMSE-64 showed greater NPV (71.3%), PPV (87.8%) andaccuracythanSTESS-3and STESS-4(p< 0.001) inthe prediction
epilepticus mortality
ofthisoutcome.In23%ofallcases, statusepilepticusproved refractoryto non-anaesthetic treatment. Allthreescalesshoweda highNPV
morbidity
(EMSE-64: 87.3%; STESS-4: 89.4%; STESS-3: 87.5%) but a low PPV (EMSE-64: 40.9%; STESS-4: 52.9%; STESS-3: 32%) for the
refractoriness
EMSE prediction of refractoriness to first and second line drugs. This means that accuracy for the prediction of refractoriness was equally poor
STESS forall scales.
AB S T R AC T Conclusions: EMSE-64 appears superior to STESS-3 and STESS-4 in the prediction of 30-days mortality and morbidity. All scales
showed poor accuracy in the prediction of response to first and second line antiepileptic drugs. At present, there are no reliable scores
capableofpredictingtreatment responsiveness.
©2017 BritishEpilepsyAssociation. PublishedbyElsevier Ltd. Allrightsreserved.

1. Introduction

Abbreviations: AUROC, area under the receiving operating curve; CSE, complex status
epilepticus; EMSE, Epidemiology based Mortality score in Status Epilepticus; ICU, Intensive Care
Unit; NCSE, non convulsive status epilepticus; NPV, negative predictive value; PPV, positive
predictive value; ROC, receiving operating curve; SE, status epilepticus; STESS, Status Epilepticus
Severity Score.
* Corresponding author at: Department of Biomedical, Metabolic, and Neural Sciences, Center for
Neuroscience and Neurotechnology, University of Modena and Reggio Emilia, NOCSE Hospital, via
Giardini1355,41126Modena, Italy.Fax: +39 0593961336.
E-mailaddress:stefano.meletti@unimore.it(S. Meletti).
http://dx.doi.org/10.1016/j.seizure.2017.01.004
1059-1311/©2017 BritishEpilepsy Association.Publishedby Elsevier Ltd. Allrights reserved. 32
etal./Seizure46(2017)31–37
treatment. Moreover, predicting refractoriness of SE would be important for the
correct management of SE itself (e.g. using continuous EEG monitoring, or rapid
referraltoan IntensiveCare Unit,ICU).
Status Epilepticus (SE) is a common neurological emergency with an annual
incidence that varies from 9.9 to 41 per 100.000 people [1–8]. SE ischaracterized by
significant short-term morbidity and mortality [9]. Case fatality ranges from 5% to
46% [2–8,10,11].
In this context, the assessment of the individual patient’s prognosis as early as
possibleappears tobecrucialin SE management, eithertoavoid over-treatment and
itspotentially harmfulconsequences[12]ortoavoidunder-detectionandunder-
G. Giovannini
To assess the short-term prognosis of SE, especially mortality, two different
scores have been developed: STESS (Status Epilepticus Severity Score) [13,14] and
EMSE(Epidemiologybased MortalityscoreinStatusEpilepticus)[15,16].
The STESS was the first score developed to assess in-hospital mortality after SE
[13].STESSisasimpleandquicklyassessable scorebasedonfourparameters:levelof
consciousnessbefore treatmentadministration,worstseizuretype,ageandhistoryof
previous seizures [13,14]. The original cut-off value was set at three points (STESS-
3), which means that scores 3 predict a high risk of death after a SE episode. STESS
showed a high negativepredictive value(NPV) for survival(100% and 97%), whileit
had a low positive predictive value (PPV) for death (37.5% and 39%) [13,14]. This
implies that it identifies patients who will survive a SE episode while it fails, in the
majorityof cases, to identifypatients who will die due to a SE episode. In a subsequent
study, Sutter et al. [18] set the cut-off at 4points (STESS-4)allowing an increase of the
correct classificationvalueofthescalefrom48.5%ofSTESS-3to73.1%of STESS-4.
TheEMSEwasintroducedin2015[15,16].Thisisthefirstscore basedonavailable
epidemiologicalorreal worlddata. Itusesfour parameters: etiology, age(stratifiedby
decade), comorbidity (based on the comorbidities reported in the Charlson’s
Comorbidity Index, [19]) and EEG, with four different patterns: lateralized periodic
discharges, generalized periodicdischarges, after status epilepticusictaldischarges,
and spontaneous burst suppression. The score was designed in a retrospective
explorative study and the best cut-off was found to be 64 points, i.e. values of 64 or
higher predict a higher likelihood of death. In that population EMSE had a NPV of
100%, PPV of 68.8% and correctly classified 89.1% of inhospital death [15,16]. This
meant it appeared to be useful in predicting bad outcomes as well as good outcomes
afteranepisode ofSE.
The aim of the present study was to evaluate and compare the predictive power of
STESS-3, STESS-4 and EMSE-64 scores to determine treatment response, 30-days
mortality, and 30-days disability in a group of adult patients followed up
prospectively.
2. Materialandmethods
2.1. Typeofstudy
This is a prospective observational study. We collected all SE episodes in young
adults and adults (14 years old) within a 2years period from September 1st 2013 to
August 31st 2015. Patients were observed at the Ospedale Civile Sant’Agostino
Estense of Modena (regional center for neurological diseases for Modena city and
district, Italy). For each included patient STESS and EMSE scores were calculated at
baseline.
2.2. Inclusion/exclusion criteriaandadopteddefinitions
Status epilepticus was defined as a continuous seizure or two or more discrete
seizuresbetween whichthereisnocomplete recoveryofconsciousnessthatlasts5min
for convulsive SE (CSE) [20,21]. In cases of non convulsive status epilepticus
(NCSE), which means a SE episode not accompanied by prominent motor
phenomena or with subtle motor phenomena, a 30-min cut-off time was adopted. The
most recent proposalforclassification ofSE used 10 min forNCSE[20],but at thetime
thedatawerecollected prospectively,30 mincut-offwasadopted.
Patients presenting two or more SE episodes during the study period were
consideredonlyforthefirstepisode.
2.3. Enrollmentstrategy
We created a specific “Status Epilepticus Form” as a data gathering form to
capture the information needed for each case. The form was filled in by the first
treating physician who took care of the patient (in the majority of cases a
neurologist, a neurointensivist, or an Accident and Emergencyphysician) or bythe
staff of the neurophysiology unit who performed the first EEG examination of a
suspected SE case. In our centre a neurologist is on duty 24 h/day for seven
days/week. The same neurophysiology staff recorded all EEGs and an
epileptologist reviewed and updated all the forms and EEGs. Any missing
information was completed using the hospital Informatics Database. As an
additional quality control of the study protocol, we also checked all the patients
discharged from the hospital in the two-year period being analysed with an
“epilepsy”or“seizure”ICD-9discharge codeattheendofthestudy.
2.4. Outcomeassessment
The outcomes were (a) response to treatment dichotomized in responsive or
refractory SE on admission; (b) mortality and (c) level of disability at 30 days from
SEonset.
Refractory Status Epilepticus (RSE) was defined as a SE episode that
continues, regardlessofthedelaysincetheonset ofthe seizure, afterfailureofa trial
ofat least intravenous benzodiazepines and at least one AED, appropriatelychosen
and at adequate dosage. Thus, it required admission to the Intensive Care Unit
(ICU) andtheapplicationofanestheticdrugtherapy.Finally, Super-RSE (SRSE)
was defined as SE that continues or recurs at least 24 h from the beginning of
anesthetictherapyorrecursduringthereduction orwithdrawalofanesthesia[22].
Functional outcome was assessed for each patient by calculating the modified
Rankin Scale (mRS). According to the scale parameters, no disability or slight
disability corresponded to a mRS 2; moderate disability was considered mRS 3–4
andsevere disabilitymRS5;mRS6wasdeath.Foreachpatient,thevalueof mRS at
the 30-days follow-up was compared to the level of disability before the hospital
admission(pre-mRS).
2.5. Statisticalanalysis
Descriptive statistics were used to analyse clinical and demographical variables
inthewholepopulation.
Diagnostic accuracy for STESS and EMSE-64 in classifying patients
according to each outcome was assessed by creating Receiver Operating
Characteristic (ROC) curves with different cutoff levels and calculating the
AUROC (Area under ROC). Sensitivity, specificity, PPV and NPV were
calculated for each scale. McNemar test was used to compare sensitivity and
specificitybetweenSTESS(withdifferentcutoffvalues)and EMSE-64.
The equivalencyofthe AUROCs was compared usingan algorithm suggested
by De Long et al. [23] in order to establish the best performance test for each
outcome.Allstatisticalanalyses wereperformedusingStata11Software.
3. Results
3.1. Patientpopulation
Duringthetwo-yearstudyweobserved175SEepisodes occurringin162patients.
Eight patients had two or more SE episodes during the years of observation, thus the
recurrencerate was4.6%.The162firstepisodesofSEreceivedfurtheranalysis. Table
1providesthedemographic,clinicaldataandEEGfeaturesof thepatients.
The detailed distribution of the STESS and EMSE parameters in our
population is reported in Table 2 and Table 3 respectively. The mean STESS score
was3points.Thisvaluewasobservedin57 patients(35%).Consideringthecut-off
of3 points112patients (69%)hada STESSscore3, whileincreasingthecut-offat4
points 55patients (34%)had aSTESSscore4. ConsideringtheEMSE scorewitha
cut-offvalueof64points,82patients(51%)presented anEMSE64points,while80
patients(49%)showedanEMSE<64 points.
3.2. STESSandEMSEinmortalityprediction
Weobserveda30-daysmortalityof31.5%(51deceased patients).Sensitivity,
specificity,PPVandNPVofeachscaleare showedinFig.1a.
Consideringthe30-days mortalityprediction, STESS-3 had a high sensitivity
(86.3%) and a low specificity (38.7%). Applying STESS-4 the specificity
increased(73%)whilethesensitivity decreased (49%).
A direct comparison of sensitivity between STESS and EMSE showed that
EMSE-64 had higher sensitivity compared to STESS-4 (p < 0.001) but not
compared to STESS-3 (p = 0.12). A direct comparison of specificity between
STESS and EMSE showed that EMSE-64 had higher specificity compared to
STESS-3(p<0.001) butnotcomparedtoSTESS-4(p=0.76).
Notably, EMSE-64 showed a high negative predictive value (NPV, 97.5%) and
positivepredictivevalue(PPV,59.8%).
 G.Giovannini etal./ Seizure46(2017)31–37 33

The equality test of AUROCS showed that EMSE-64 appeared to be superior to worsened functional conditions. Sensitivity, specificity, PPV and NPV of the three
STESS (considering both cutoff values of 3 and 4). Thus EMSE-64 has the highest scalesareshowedinFig.2a.
accuracy (0.83) in predicting 30days mortality after a SE episode, which was Again we found that STESS-3 had a high sensitivity (80%) and a low specificity
significantlysuperior toSTESS-scores(Fig.1b). (46.3%) while adopting STESS-4 the specificity increased to 79.1% while the
sensitivitydecreasedto43.2%.
3.3. STESSandEMSEinpredictionofclinicalworsening EMSE-64 showed higher sensitivity compared to STESS-4 (p < 0.001) but not
compared to STESS-3 (p = 0.57); specificity was higher for EMSE-64 in comparison
At baseline 103 patients (63.6%) had no disabilities or a slight disability (mRS 0- toSTESS-3(p<0.001)but notcomparedtoSTESS-4(p=0.76).EMSE-64showedthe
2). At 30 days post-SE only 55 patients (34%) had a mRS 0-2. Comparing the 30 days
highest
follow-up mRS values to the baseline values, we found that 95 patients (59%) had
Table 1 comparison of the accuracy of STESS-3 and STESS-4 did not show significant
Demographicandclinical variablesofthe patients.
differencesbetweenthem.ThecomparisonofAUROCofthethree scalesisshowedin
Clinicalcharacteristicsofpatientpopulation N(%)
Fig.2b.
Total 162
3.4. STESSandEMSEinrefractorinessprediction
Gender(Female) 98(60.5%)
108patients (66.7%)hada SE responsivetofirst orsecond line treatment, while37
Age(years) Range 14-94 patients (22.8%) had a refractory SE, thus needed admission to ICU and anesthetic
Average 70 therapy. Among them, 26 (16%) were defined as super-refractory. For 17 patients
(10.5%) it was not possible to define the treatment response, as though they did not
Median 74 respondtothesecondlinetreatment,therewasnotherapy escalationduetoan“end-of-
lifeapproach”andallofthemdied whilststillinSE.
Aetiology Cerebrovascular 34(21%) All three scales had a high NPV (87.3% for the EMSE-64, 87.5% for the STESS-3
Anoxia/hypoxia 24(14.8%) and 89.4% fortheSTESS-4)for prediction of refractoriness totreatment. Conversely
theyall showed a low PPV (40.9% for the EMSE, 32% for the STESS-3 and 52.9% for
Braintumors 23(14.2%) the STESS-4). The sensitivity was high for all three scales (without statistically
significant differences), while the highest specificity was showed by STESS-4
Others 81(50%) (STESS-4vsSTESS3,p=0.000;STESS-4 versusEMSE-64,p=0.006).
Thesensitivity,specificity,PPV,NPVofthethreescalesare showninFig.3a.
Etiologyclassification Acutesymptomatic 94(58%)
The comparison of the accuracy of the three scales in predicting refractoriness
Progressivesymptomatic 27(17%) showed that STESS-4 had the highest accuracy. The direct comparison of the
accuracyof STESS-4 and STESS-3 shows that STESS-4 was superior to STESS-3 (p
Remote symptomatic 18(11%) = 0.0001). When STESS-4 and STESS-3 were compared to EMSE-64 their
accuracies were not significantly different. The comparison of AUROC of the three
Multifactorial 15(9%)
scalesisshowninFig.3b.
Cryptogenic 7(4%)

Idiopathic 1(1%)

Semiology NCSE AllNCSE 115(71%)

NCSE only 73

GCSE !NCSE 23
15
FMSE!NCSE
4
MSE!NCSE 47(29%)
CSE GCSE,FMSE,MSE

30-days mortality 51(31.5%)

Patientsworsenedat30days 95(59%)

Responsetotreatment Responsive 108(67%)

Refractory 11(7%)
Super-Refractory 26(16%)
Unclassifieda 17(10%)
SE: Status epilepticus. NCSE: Non-convulsive status epilepticus. CSE: Convulsive status
epilepticus. GCSE: Generalized convulsive status epilepticus. FMSE: Focal motor status
epilepticus. MSE:Myoclonic statusepilepticus.
a
Unclassified: Patients for which it was not possible to define the treatment response, as even if they
did not respond to the second line treatment, there was no therapy escalation due to an “end-of-life
approach”andallof themdiedbeingstill in SE(seetext).
PPV(87.8%),andNPV(71.3%).
The comparison of the accuracy of prediction of clinical worsening among the
three scores showed that EMSE-64 had a significantly higher accuracy when
compared both to STESS-4 (p < 0.0001) and STESS-3 (p = 0.0001). The direct
34 G.Giovannini etal./ Seizure46(2017)31–37

Table 2
STESSparametersdistributioninthe observedpopulation.
Parameter Value N %

Levelof consciousness beforetreatment Alert/somnolent 108 66.7

Stupor/coma 54 33.3

Typesofseizures Simple partial,complexpartial,absence, myoclonic 84 51.9

Generalized-convulsive 41 25.3

Nonconvulsivestatus epilepticusincoma 37 22.8

Age <65years 51 31.5

65years 111 68.5

Historyofpreviousseizures Yes 46 28.4

NoorUnknown 116 71.6

4. Discussion the clinician that the patient has a high probability of surviving (98%) without
functional decline after the SE episode. Thus, this gives the possibility to avoid an
In thepresent studywefoundthat EMSEwas superiortoSTESS inpredictingboth unnecessary overtreatment and its related possible complications. In contrast an
short-term mortality and morbidity after SE, while both scores are less useful in EMSE64pointsinforms theclinicianthatthepatienthasahighprobability(88%)of
predicting refractoriness to treatment. In particular, an EMSE <64 points suggests to
Table 3 71–80 48 29.6
EMSE parametersdistributionintheobservedpopulation.
Parameter Value N % >80 46 28.4

Etiology CNS-anomalies 0 0
Comorbidity Myocardialinfarction 28 17.3
Drugsreduction/withdrawal/poorcompliance 12 7.4
Cerebrovasculardisease 30 18.5
Multiplesclerosis 3 1.9
Connectivetissuedisease 5 3.1
Remote braininjury 2 1.2
Diabeteswithoutendorganfailure 22 13.6
Remote cerebrovascular disease 21 13
Diabeteswithendorganfailure 1 0.6
Hydrocephalus 0 0
Metastaticsolidtumor 12 7.4
Alcoholabuse 2 1.2
Congestiveheartfailure 10 6.2
Drugoverdose 1 0.6
Dementia 33 20.4
Headtrauma 3 1.9
Ulcer 11 6.8
Cryptogenic 9 5.6
Hemiplegia 3 1.9
Braintumor 24 14.8
Anytumor 15 9.3
Metabolic:sodiumimbalance 4 2.5
AIDS 0 0
Metabolicdisorders 34 21
Peripheral vasculardisease 9 5.6
Acutecerebrovasculardisease 16 9.9
COPD 21 13
CNS-infection 7 4.3
Mildliverdisease 9 5.6
Anoxia 24 14.8
Moderatesevereliverdisease 3 1.9
Age <21 3 1.9
Moderatesevererenaldisease 8 4.9
21–30 4 2.5
EEG Spontaneousburstsuppression 12 7.4
31–40 4 2.5
ASIDs, LPDs,GPDs 52 32.1
41–50 7 4.3
NoASIDs,LPDs, GPDs 98 60.5
51–60 23 14.2
ASIDs: After Status epilepticus Ictal Discharges; LPDs: Lateralized Periodic Discharges; GPDs:
GeneralizedPeriodicDischarges.
61–70 27 16.7
worsening and death (60%) after the SE episode indicating that it is of paramount
importance to rapidly and intensively manage the patient (SE treatment, EEG
 G.Giovannini etal./ Seizure46(2017)31–37 35

Fig. 1. A: Sensitivity, Specificity, PPV, NPV comparison of EMSE, STESS-3, STESS-4 scores for mortality prediction. Sensitivity: STESS-3 versus STESS-4 comparison: *p < 0.0000; STESS-3 versus EMSE
comparison: p = 0.12; STESS-4 versus EMSE comparison: **p = 0.000. Specificity: STESS-3 versus STESS-4 comparison: *p < 0.0000; STESS-3 versus EMSE comparison: **p = 0.000; STESS-4 versus EMSE
comparison: p = 0.76. B: AUROC comparison of EMSE, STESS-3 and STESS-4 in mortality prediction. AUROC EMSE: 0.8317; AUROC STESS-3: 0.6251; AUROC STESS-4: 0.61. EMSE versus STESS-3
comparison:p<0.0001;EMSE versus STESS-4comparison:p<0.0001; STESS-3 versus STESS-4 comparison:p =0.7126.

Fig. 2. A: Sensitivity, Specificity, PPV, NPV comparison of EMSE, STESS-3, STESS-4 scores for clinical worsening prediction. Sensitivity: STESS-3 versus STESS-4 comparison: *p < 0.0000; STESS-3 versus
EMSE comparison: p = 0.570; STESS-4 versus EMSE comparison: **p = 0.000. Specificity: STESS-3 versus STESS-4 comparison: *p < 0.0000; STESS3 versus EMSE comparison: **p = 0.000; STESS-4 versus
EMSE comparison: p = 0.340. B: AUROC comparison of EMSE, STESS-3 and STESS-4 in clinical worsening prediction. AUROC EMSE: 0.8043; AUROC STESS-3: 0.6313; AUROC STESS-4: 0.6113. EMSE
versus STESS-3comparison:p =0.0001;EMSE versusSTESS-4comparison:p<0.0001; STESS-3
monitoring, etiology identification and treatment, complications avoidance) to 10.1016/j. seizure.2017.01.004, the results of the published studies of mortality
avoidapossiblenegative outcome. prediction with the STESS and EMSE-64 scores are summarized [13–
As far as mortality prediction EMSE-64 showed a high sensitivity (96%) but a 15,18,24,25].
relatively low specificity (70%) reflecting the low proportion of false negative With regardstotheprediction ofclinicalworsening, EMSE-64 again showed a
(2.5%) and the relatively high proportion of false positive (40%) cases. These significantly higher accuracy when compared to both STESS-3 and to STESS-4
patients mainly had a SE episode related to acute strokes, anoxic brain injury or with a sensitivityofnearly76% and specificityof85%. EMSE>64 points meansan
brain tumors: all etiologies known to be related to high mortality. Indeed, 88%probabilityof worseningtheclinicalconditionat30-daysfollow-up,whilean
considering the mortality at 6 months from SE onset 30% of these patients were EMSE score<64 points means a71% probabilityofregainingthe previous clinical
dead. Notably, since the EMSE uses four nonmodifiable parameters to assess status before the SE episode. Since the patient has a 29% risk of worsening as well,
mortality prediction it is nevertheless possible that medical interventions (e.g. this means that even if the test is negative the patient has a significant possibility of
rapid SE drug treatment, rapid identification and whenever possible treatment of experiencing an adverse outcome. This has to be kept in mind in the treatment/
the underlying conditions, management and prevention of complications) could managementoftheSEepisode.Inthiscaseonlytenpatientswere falsepositives.
improvetheoutcome. Tothebestofourknowledge,onlytwostudiescomparedSTESS andEMSEscores
Overall, these findings confirm previous results of the study of Leitinger et al. inpredictingfunctionaloutcomeathospital
[15,16], highlighting the good accuracy of EMSE-64 in predicting short-term versus STESS-4comparison:p =0.5994.
mortality after a SE episode and its superiority to STESS. In contrast, Kang et al.
[24]showedno differencesbetweenSTESSandEMSEinmortalityprediction,but discharge [17,24] (see also Supplementary Table S2 in the online version at DOI:
in that study the authors did not use the EMSE-64 score as originally proposed 10.1016/j.seizure.2017.01.004). Both studies [17,24] concluded that the EMSE
(different cut-offvaluesand different parameters), and thereforeit isnot possibleto had a higheraccuracyin predicting functional outcomewhen compared toSTESS
compare results. In Supplementary Table S1, in the online version at DOI: at hospital discharge and at 3 months post-discharge. They also found that the
 END-IT (Encephalitis,Non-convulsivestatusepilepticus,Diazepamresistance,
Image abnormality, Tracheal intubation) score was significantly superior to
STESS (with both cut-off levels 3 and 4) and without statistically significant
differences with the EMSE. With regards to the END-IT, however, the score was
developed to assess the 3 months prognosis in a convulsive SE population,
therefore we believed it was not suitable for use to determine 30 days morbidity in
ourpopulation(with115NCSEcases,71%ofourpopulation).
Finally, relating to refractoriness prediction, the three scales all showed a high
NPV thus a patient with a negative score has a high probability of treatment
responsiveness. On the other hand, they all had a low PPV (40.9% for the EMSE,
32% for the STESS-3 and 52.9% for the STESS-4) thus in a patient with positive
scoreit isnot possibletopredict theresponsetotreatment. Thelowaccuracyof the
three scales in SE refractoriness prediction implies that their clinical usefulness is
limited.
Whilst STESS wascreated toassessin-hospitalmortality, ithas sincebeen usedto
predict refractoriness to treatment [26,27]. In the study of Novy et al. [26], STESS 3
was found to be strongly associated with the presence of RSE (p = 0.001)but it was not
used topredictrefractoriness. InthestudyofSutteretal.[27],theonly study,tothebest
of our knowledge, that evaluated STESS-3 to predict SE refractoriness, the authors
concluded that STESS failed to reliably predict refractoriness (AUROC 0.58 in the
Swiss population and0.60intheUSpopulationrespectively).
36 G.Giovannini et al. /Seizure46 (2017)31–37
5. Conclusions
This studyhighlights the superiorityof EMSE on STESS in mortalityand clinical
worsening predictions after an episode of SE, thus EMSE could be adopted in
everydayclinicalpracticeasaquick bedsideevaluation. However, weconfirmthat,at
present,thereare noreliablescorestopredicttreatmentrefractoriness.Giventhat, we
think that it would be useful to create a more dynamic score, that consider modifiable
variables (as, for example, development of in-hospital complications) to be
calculated dailyin SE patients, as it is for NIHSS (National Institutes of Health Stroke
Scale) [28] in stroke patients, thus giving the clinician a more precise idea of the
clinicalevolutionand,perhaps,refractorinessdevelopment.
Studyfunding
Nofunding.
Disclosures
S. Meletti received Research grant support from the Emilia Romagna Region,
from the non-profit organization CarisMo Foundation; has received personal
compensationasscientific advisoryboardmemberforUCBandEISAI.

Fig. 3. A: Sensitivity, Specificity, PPV, NPV comparison of EMSE, STESS-3, STESS-4 scores for clinical refractoriness prediction. Sensitivity: STESS-3 versus STESS-4 comparison: p = 0.12; STESS-3 versus
EMSE comparison: p=0.210; STESS-4 versusEMSE comparison:p =1. Specificity: STESS-3 versusSTESS-4comparison:**p =0.000; STESS-3 versus EMSE comparison: #p =0.005; STESS-4 versus EMSE
comparison: ##p =0.006. B: AUROC comparison of EMSE, STESS-3 and STESS-4 in refractoriness prediction. AUROC EMSE-

64:0.6843;AUROC STESS-3:0.6134; AUROC STESS-4:0.7538.EMSE versus STESS-3comparison:p=0.1129;EMSE versus STESS-4comparison:p=0.1932; STESS-3 versus
STESS-4comparison:p=0.0001.
4.1. Studylimitations
Even if our population was prospectively and consecutively collected, this study
represents a ‘one center’ experience. This could represent a potential bias since the
accuracy of the evaluated scores could be different in relation to the severity of the
examined population. However, our hospital is a regional care clinic that seems to be
representativeofnon-selectedstatusepilepticus patients.Intheanalyzedpopulation,
only14patientshadaSE episoderelatingtoaprecipitatingfactorinapreviousepilepsy
history: an etiology well known to have a good prognosis and corresponding low
scores on STESS and EMSE. Our population showed a distribution of scores that
appeared to be similar to the ones previously reported in the Austrian population
(68.5% of patients had a STESS3)[15] but higher than the ones observed in the Swiss
population(51.9%ofpatientshadaSTESS3)[14]andin thestudyofGoyaletal.[25]in
whom only 27.3% of patients had a STESS 3. Therefore, larger multicenter
prospectivestudiesare stronglyadvisedtoevaluatetheaccuracyofthesescales.
E.Trinka received personal fees from Eisai, Everpharma, Medtronic, Bial,
Newbridge, GL Pharm, GlaxoSmithKline, Boehringer Viropharma, and Actavis;
grantsandpersonalfeesfrom Biogen Idec,UCBPharmaandEisai;andgrantsfrom
Red Bull, Merck, European Union, FWF Osterreichischer Fond zur
Wissenschaftsforderung, Bundesministerium fur Wissenschaft und Forschung,
and the Jubilaumsfond der Osterreichischen Nationalbank. GKa reports travel
grantsfromUCBPharmaandEisai.
M. Leitinger received travel grants from Medtronic and UCB Pharma and
personalfeesfromEverpharma. Theotherauthorsreportnodisclosures.
Acknowledgments
Wethank TheEEGtechnician stafffortheinvaluablehelp in the recruitment of
patients.WethankHollyNesslingforrevisingand editingtheEnglishlanguage.
References
[1] Logroscino G, Hesdorffer DC, Cascino G, Annegers JF, Hauser WA, et al. Time trends in
incidence, mortality, and case-fatality after first episode of status epilepticus. Epilepsia
2001;42(8):1031–5.
 G.Giovannini etal./ Seizure46(2017)31–37 37
[2] DeLorenzoRJ,HauserWA,Towne AR,BoggsJG, PellockJM,Penberthy L,et al. A prospective,
population-based epidemiologic study of status epilepticus in Richmond, Virginia. Neurology
1996;46:1029–35.
[3] Vignatelli L, Rinaldi R, Galeotti M, de Carolis P, D'Alessandro R, et al. Epidemiology of status
epilepticus in a rural area of northern Italy: a 2-year population-based study. Eur J Neurol
2005;12:897–902.
[4] Vignatelli L, Tonon C, D’Alessandro R, on behalf of the Bologna Group for the Study of Status
Epilepticus. Incidence and short-term prognosis of status epilepticus in adults in Bologna, Italy.
Epilepsia2003;44(7):964–8.
[5] Knake S, Rosenow F, Vescovi M, Oertel WH, Mueller HH, Wirbatz A, for the Status Epilepticus
Study Group Hessen (SESGH), et al. Incidence of status epilepticus in adults in Germany: a
prospective, population-basedstudy. Epilepsia2001;42(6):714–8.
[6] Jallon P, Coeytaux A, Galobardes B, Morabia A, et al. Incidence and case-fatality rate of status
epilepticusintheCantonofGeneva.The Lancet 1999;353:1496.
[7] Coeytaux A, Jallon P, Galobardes B, Morabia A, et al. Incidence of status epilepticus in French-
speaking Switzerland(EPISTAR). Neurology 2000;55:693–7.
[8] Govoni V, Fallica E, Monetti VC, Guerzoni F, Faggioli R, Casetta I, et al. Incidence of status
epilepticus in Southern Europe: a population study in the Health District of Ferrara, Italy. Eur
Neurol2008;59:120–6.
[9] Logroscino G, Hesdorffer DC, Cascino G, Hauser WA, Coeytaux A, Galobardes B, et al.
Mortality after a first episode of status epilepticus in the United States and Europe. Epilepsia
2005;46(Suppl. 11):46–8.
[10] GiovanniniG,Monti G,Polisi MM,Mirandola L, Marudi A,Pinelli G,etal. Aoneyearprospective
studyofrefractorystatusepilepticusin Modena, Italy. EpilepsyBehav2015;49:141–5.
[11] Logroscino G, Hesdorffer DC, Cascino G, Annegers JF, Hauser WA, et al. Shortterm mortality
afterafirst episodeofstatusepilepticus.Epilepsia 1997;38 (12):1344–9.
[12] Marchi NA, Novy J, Faouzi M, Stähli C, Burnand B, Rossetti AO, et al. Status epilepticus: impact
oftherapeuticcomaonoutcome.CritCareMed 2015;43:1003–9.
[13] Rossetti AO, Logroscino G, Bromfield EB. A clinical score for prognosis of status epilepticus in
adults.Neurology2006;66:1736–8.
[14] Rossetti AO, Logroscino G, Milligan TA, Michaelides C, Ruffieux C, Bromfield EB, et al. Status
Epilepticus Severity Score (STESS): a tool to orient early treatment strategy. J Neurol
2008;255:1561–6.
[15] Leitinger M, Holler Y, Kalss G, Rohracher A, Novak HF, Hofler J, et al. Epidemiology-based
mortalityscoreinstatusepilepticus(EMSE). Neurocrit Care2015;22:273–82.
[16] Leitinger M, Kalss G, Rohracher A, Pilz G, Novak H, Höfler J, et al. Predicting outcome of status
epilepticus. EpilepsyBehav2015;49:126–30.
[17] Gao Q, Ou-Yang TP, Sun XL, Yang F, Wu C, Kang T, et al. Prediction of functional outcome in
patientswithconvulsivestatusepilepticus:theEND-ITscore.Crit Care2016;20:46.
[18] Sutter R,Kaplan PW, StephanRueg̈ gS. Independentexternal validationof the statusepilepticus
severity score.CritCareMed2013;41:e475–9.
[19] Charlson ME, Pompei P, Ales KL, MacKenzie R, et al. A new method of classifying prognostic
comorbidity in longitudinal studies: development and validation. J Chron Dis 1987;40(5):373–
83.
[20] Trinka E, Cock H, Hesdorffer D, Rossetti AO, Scheffer IE, Shinnar S, et al. A definition and
classification of status epilepticus—report of the ILAE task force on classification of status
epilepticus. Epilepsia2015;56(10):1515–23.
[21] Lowenstein DH, Bleck T, Macdonald RL. It’s time to revise the definition of status epilepticus.
Epilepsia1999;40(1):120–2.
[22] Shorvon S, Ferlisi M. The treatment of super-refractory status epilepticus: a critical review of
available therapiesandaclinicaltreatment protocol. Brain 2011;134:2802–18.
[23] DeLong ER, DeLong DM, Clarke-Pearson DK. Comparing the areas under two or more
correlated receiver operating characteristic curves: a nonparametric approach. Biometrics
1988;44:837–45.
[24] Kang BS, Kim DW, Kim KK, Moon HJ, Kim YS, Kim HK, et al. Prediction of mortality and
functional outcome from status epilepticus and independent external validation of STESS and
EMSE scores.CritCare2016;20:25.
[25] Goyal MK, Chakravarthi S, Modi M, Bhalla A, Lal V, et al. Status epilepticus severity score
(STESS): a useful tool to predict outcome of status epilepticus. Clin Neurol Neurosurg
2015;139:96–9.
[26] Novy J, Logroscino G, Rossetti AO. Refractory status epilepticus: a prospective observational
study. Epilepsia2010;51(2):251–6.
[27] Sutter R, Kapland PW, Marscha S, Hammela EM, Ruegg S, Ziai WC, et al. Early predictors of
refractorystatusepilepticus:aninternationaltwo-centerstudy. EurJNeurol2015;22:79–85.
[28] Spilker J, Kongable G, Barch C, Braimah J, Brattina P, Daley S, et al. Usingthe NIH stroke scale to
assess stroke patients. The NINDS rt-PA Stroke Study Group. J Neurosci Nurs
1997;29(December(6)):384–92.