- Email: [email protected]
External validation of STESS and EMSE as outcome prediction scores in an Egyptian cohort with status epilepticus
Epilepsy & Behavior 102 (2020) 106686
Contents lists available at ScienceDirect
Epilepsy & Behavior journal homepage: www.elsevier.com/locate/yebeh
External validation of STESS and EMSE as outcome prediction scores in an Egyptian cohort with status epilepticus Mohamed S. El-Tamawy a, Hanan Amer a, Nirmeen A. Kishk a, Amani M. Nawito b, Mye A. Basheer b, Nelly Alieldin c, Rehab Magdy a,⁎, Alshimaa S. Othman a a b c
Department of Neurology, Kasr Al-Ainy Faculty of Medicine, Cairo University, Cairo, Egypt Department of Neurophysiology, Kasr Al-Ainy Faculty of Medicine, Cairo University, Cairo, Egypt Department of Cancer Epidemiology, National Cancer Institute, Cairo University, Cairo, Egypt
a r t i c l e
i n f o
Article history: Received 14 October 2019 Revised 4 November 2019 Accepted 4 November 2019 Available online xxxx Keywords: Status epilepticus STESS EMSE Mortality outcome
a b s t r a c t Purpose: There is a lack of data concerning the performance of the outcome prediction scores in patients with status epilepticus (SE) in developing countries. The aim of this study was to compare the predictive performances of the status epilepticus severity score (STESS) and the epidemiology-based mortality score in status epilepticus (EMSE) and adaptation of such scoring system to be compatible with the nature of society. Method: This is a prospective study, conducted in Egypt from the period of January 2017 to June 2018. The main outcome measure was survival versus death, on hospital discharge. The cutoff point with the best sensitivity and specificity to predict mortality was determined through a receiver operating characteristic (ROC) curve. Results: Among the 144 patients with SE with a mean age of 39.3 ± 19.5 years recruited into the study, 38 patients (26.3%) died in the hospital with the survival of 99 patients while 7 patients (4.9%) were referred to other centers with an unknown outcome. Although EMSE had a bit larger area under the curve (AUC) (0.846) than STESS-3 (AUC 0.824), STESS-3 had the best performance as in-hospital death prediction score as it has a higher negative predictive value (94.6%) than that of EMSE (90.9%) in order not to miss high-risk patients. Conclusion: In the Egyptian population, STESS and EMSE are useful tools in predicting mortality outcome of SE. The STESS performed significantly better than EMSEE combinations as a mortality prediction score. © 2019 Elsevier Inc. All rights reserved.
1. Introduction Status epilepticus (SE) is a major neurological emergency associated with significant mortality outcomes. Such an outcome may influence treatment strategy choice [1]. For that purpose, two clinical scoring tools have been developed for in-hospital mortality prediction after SE, based on different sets of prognostic parameters. The Status Epilepticus Severity Score (STESS) was originally validated by Rossetti, Logroscino [2] followed by Epidemiology-Based Mortality Score in
Abbreviations: SE, status epilepticus; STESS, status epilepticus severity score; EMSE, epidemiology-based mortality score in status epilepticus; EACLDE, etiology (E), age (A), comorbidity (C), level of consciousness before treatment (L), duration (D), EEG (E); ER, emergency room; NICU, neurological intensive care unit; CSE, convulsive status epilepticus; NCSE, nonconvulsive status epilepticus; ILAE, the International League Against Epilepsy; CCI, Charlson Comorbidity Index; EEG, electroencephalogram; SD, standard deviation; ROC curve, receiver operator characteristic curve; AUC, area under the curve; PPV, positive predictive values; NPV, negative predictive values; CNS, central nervous system; HIV, human immunodeficiency virus. ⁎ Corresponding author. E-mail address: [email protected] (R. Magdy).
https://doi.org/10.1016/j.yebeh.2019.106686 1525-5050/© 2019 Elsevier Inc. All rights reserved.
Status Epilepticus (EMSE) by Leitinger, Holler [3], then received external validation in several developed countries [4,5,6]. However, there is no available data in that concern in developing countries. Apparent differences between the developed and developing countries exist in terms of economics and healthcare facilities, and these differences might influence the etiologies, clinical profiles of patients with SE, and then outcome [7]. The aim of this study was to compare the performance of the STESS and EMSE and adaptation of such scoring system in light of unique demographic and clinical characteristics of patients in Egypt as the largest country in terms of population, in the Eastern Mediterranean region. 2. Subjects and methods This is a prospective study, conducted over a total working duration of 18 months (January 2017 to June 2018). Patients were recruited from the emergency room (ER), neurological intensive care unit (NICU), or other medical intensive care units of Cairo University Hospitals. The study was approved by the ethical committee of the neurology department, Cairo University. Informed consents were waived.
2
M.S. El-Tamawy et al. / Epilepsy & Behavior 102 (2020) 106686
2.1. Subjects
2.3. Outcome
The eligibility criteria for this study were that all patients had been with diagnosed SE according to the new the International League Against Epilepsy (ILAE) Report on definition and classification of SE [8], including those with both convulsive SE (CSE) and nonconvulsive SE (NCSE) semiologies. Patients' ≥12 years old of both genders were included.
The main outcome measure was survival versus death, on hospital discharge.
2.2. Methods A set of clinical data was recorded to build up the two clinical scoring tools, including age, history of previous seizures, seizure semiology, seizure duration, level of consciousness before treatment, comorbidity by Charlson Comorbidity Index (CCI) [9], and onset electroencephalogram (EEG), if available. The etiology of SE was determined after a set of laboratory and radiological imaging according to the clinical situation. 2.2.1. Status epilepticus severity score (STESS) [2] This scale was used to predict the mortality risk in patients with nonanoxic SE. A score of three points or higher indicates the risk of death. The STESS score included the following four parameters: level of consciousness (alert or somnolent/confused = 0 point, stuporous or comatose = 1 point), “worst” seizure type (simple-partial, complexpartial, absence, myoclonic as complicating idiopathic generalized epilepsy = 0 point; generalized–convulsive = 1 point; nonconvulsive SE in coma = 2 points), age (under 65 years = 0 point, 65 or older = 2 points), and history of previous seizures, as a surrogate for acute etiology, (yes = 0 point, no or unknown = 1 point). 2.2.2. Epidemiology-based mortality score in status epilepticus (EMSE) [3] This scale was used to predict the mortality risk in a population with SE on more comprehensive scoring system. The long version of the score has six-item categories: age (A), etiology (E), comorbidity (C), level of consciousness before treatment (L), duration (D), and EEG (E). The evaluation of the EMSE scoring system is summarized in Table 1. In this study, the following combinations were applied: EAC, EACL, EACLD, and EACLDE. The score gives a clear weight for epidemiological data. However, some limitations exist but had been resolved by communicating with the corresponding author of the score via email (Supplementary material 1 & 2).
2.4. Statistical analysis The SPSS (statistical package for social sciences) version 23.0 was used for data management and data analysis. To describe the data, frequency (percent) and mean ± standard deviation (SD) were used. P-values less than 0.05 were considered statistically significant. The receiver operator characteristic (ROC) curve was used to select the best cutoff for scores that best predict surviving after an episode of SE. After selecting the best cutoff, different accuracy measures were calculated including sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV), in addition to total accuracy. 3. Results 3.1. Demographic and clinical characteristics of the patients One hundred forty-four patients with SE were enrolled in this study. Their ages ranged from 13 to 80 years, with a mean of 39.3 ± 19.5 years, 79 males (54.9%) versus 65 females (45.1%). The clinical characteristics of the study population are summarized in Table 2. Only five patients were post anoxic, so they were excluded from STESS analyses as this scale was intended to be used only in patients with nonanoxic SE. Electroencephalogram was performed for only 54 patients (37.5% of the total cohort) because of limited resources availability. 3.2. SE outcome on hospital discharge Apart from 7 patients (4.9%) who were referred to other centers because of the unavailability of intensive care unit (ICU) beds, the outcome was favorable for the majority of the study population, with the survival of 99 patients (68.8%). On the other hand, 38 patients had a fatal outcome. Relation of outcome prediction scales to the outcome: The STESS and different EMSE combinations were significantly higher among patients with fatal outcome (Table 3). This included EMSEEACLDE, which was available for only 54 patients, of whom 16 died.
Table 1 Scoring system of epidemiology-based mortality score in status epilepticus (EMSE). Etiology (score one stratum) CNS anomalies 2 Hydrocephalus 8 Drug abuse 11 Brain tumor 16 CNS infection 33
Drug resistant\withdrawal, poor compliance 2 Remote cerebrovascular disease, brain injury 7 Head trauma 12 Metabolic; sodium imbalance 17 Acute cerebrovascular disease 26
Age (score one stratum) 21–30 y 1
31–40 y 2
41–50 y 3
51–60 y 5
Multiple sclerosis 5 Alcohol abuse 10 Cryptogenic 12 Metabolic disorders 22 Anoxia 65
61–70 y 7
Comorbidity (score each disease) Hemiplegia, Moderate to severe renal Myocardial infarction, Congestive heart failure, Peripheral vascular disease, cerebrovascular disease, dementia, chronic pulmonary disease, Connective disease, diabetes with end organ damage, any tumor including tissue disease, Peptic ulcer disease, mild liver disease 10 Leukemia\Lymphoma 20 LOC (score one stratum) Awake 0
Stupor 14
Somnolence 5
Duration of seizure (score one stratum) b1 h 3 EEG (score the worst stratum) Burst suppression 60
ASIDs 40
71–80 y 8
N80 y 10
Moderate to severe liver disease 30 Metastatic Solid tumor 60
Coma 23
≥1 h 33 LPDs 40
GPDs 40
No ASIDs, LPDs or GPDs 0
CNS: central nervous system, EEG: electroencephalography, ASID: after status ictal discharge, GPD: generalized periodic discharge, LOC: level of consciousness, LPD: lateralized periodic discharge.
M.S. El-Tamawy et al. / Epilepsy & Behavior 102 (2020) 106686 Table 2 Clinical characteristics of the patients. n = 144 Semiology of SE episode according to the new (2015) ILAE classification system SE with prominent motor features 139 (96.5%) SE without prominent motor features (NCSE) 5 (3.5%) History of previous epilepsy 76 (52.8%) Worst seizure type Simple partial, CPS, absence, myoclonic 22 (15.27%) Generalized convulsive 121 (84.02%) Non-convulsive SE in coma 1 (0.69%) LOC pretreatment Alert 8 (5.6%) Confused\somnolent 48 (33.3%) Stuporous 41 (28.5%) Comatose 47 (32.6%) Charlson Comorbidity Index 0–70 (5.9 ± 11.6) Etiology of SE Acute symptomatic 93 (64.6%) Remote symptomatic 39 (27.1%) Progressive symptomatic 9 (6.3%) SE in defined electroclinical syndromes 2 (1.4%) Unknown (Cryptogenic) 1 (0.7%) EEGa Burst suppression 3 (5.6%) ASIDs 3 (5.6%) PDs 5 (9.3%) No ASIDs or PDs 43 (79.6%) ILAE: international league against epilepsy, SE: status epilepticus, NCSE: nonconvulsive status epilepticus, CPS: complex partial seizures, LOC: level of consciousness, CSE: convulsive status epilepticus, EEG: electroencephalogram, ASID: after status ictal discharge, PDs: periodic discharges. a EEG was performed for only 54 patients.
3.3. Performance of the STESS and EMSE combinations in predicting inhospital death According to the ROC curve for predicting in-hospital death, the best sensitivity/specificity trade-off was set at cutoff points of 3, 28, 35, 67, and 67 for STESS, EMSE-EAC, EMSE-EACL, EMSE-EACLD, and EMSEEACLDE, respectively (Figs. 1, 2, 3, & 4).
3.4. Comparison of the STESS and EMSE combinations as in-hospital death prediction scores The EMSE-EACLDE had the largest area under the curve (AUC) followed by STESS-3. Although EMSE-EACLDE had a bit larger AUC (0.846) than STESS-3 (AUC 0.824), STESS-3 had the best performance as in-hospital death prediction score as it has a higher NPV (94.6%) than that of EMSE-EACLDE (90.9%) (Figs. 1, 2, 3, & 4).
4. Discussion This was the first independent, external validation study of the predictive accuracy of two clinical scoring systems (STESS and EMSE) used
Table 3 Relation of outcome prediction scores to outcome among the study population. Outcome prediction scores (Mean ± SD)
Survivors (n = 99)
Non survivors (n = 38)
P-value
STESSa EMSE-EAC EMSE-EACL EMSE-EACLD
2.0 ± 1.1 19.5 ± 16.9 30.4 ± 19.7 48.9 ± 45.3
3.4 ± 1.0 40.2 ± 29.8 59.5 ± 31.7 87.8 ± 33.8
b0.001 b0.001 b0.001 b0.001
STESS: status epilepticus severity score, EMSE: epidemiology-based mortality score in status epilepticus, E: etiology, A: age, C: comorbidity, L: level of consciousness, D: duration. a Five SE episodes due to hypoxic brain insult were excluded from STESS scoring (4 of them died and only 1 survived).
3
for in-hospital death prediction among patients with SE, in developing countries. The scoring system must be adapted to different regions in the world according to their distinctive demographic characteristics. The differences in the demographic factors, etiologies, facilities, e.g., nonavailability of continuous EEG monitoring, and treatment variables are likely to be responsible for such heterogeneity in SE outcome noted between developing and developed countries [7]. The most common etiology of SE in developing countries is antiepileptic drug (AED)-related factors. However, vascular causes represent the majority in the developed world. Furthermore, central nervous system (CNS) infections and autoimmune entities predominate in developing countries [10]. In this study, the optimal cutoff point of the STESS for the best discrimination between survivors and nonsurvivors (≥ 3) was similar to the one calculated by the original investigators with high NPV (94.6%). Some studies identified the optimal cutoff point of the STESS at (≥ 4) [4,11]. On the other hand, our results revealed the optimal cutoff point of the EMSE-EACLDE for the best discrimination between survivors and nonsurvivors (≥67) was larger than the one calculated by the original investigators [3] with high NPV (90.9%). All these results can be explained by different epidemiological characteristics of the studied populations. In fact, the primary aim of the development of such scores was to predict the in-hospital short-term outcome of patients with SE in a trail to design treatment plans, and this is the value of a high NPV with accepted accuracy in order not to miss high-risk patients [2,3]. For that purpose, STESS-3 had the best performance as in-hospital death prediction score as it has a higher NPV (94.6%) than that of EMSE-EACLDE (90.9%) at the expense of the small difference between the measured AUC of both scores. The best combination of EMSE score variables was given when EACLDE combinations were used (AUC 0.846). However, predictive accuracy was not too lowered on excluding EEG as a variable (AUC 0.816). This EMSE combination (EACLD) was not previously analyzed in previous studies, but it was intentionally analyzed in our study to suit our center's capabilities where the portable EEG is not available all the time. Furthermore, the superiority of the performance of STESS over EMSE was in favor of this limitation in epilepsy centers in developing countries, as EEG is not including in STESS scoring system. A point worthy of discussion, strengths, and weaknesses of these two clinical scoring systems can be accentuated or embellished according to different demographic characteristics and etiology of SE around the world. It is fast to apply STESS, whereas, EMSE may take some time until the tests reveal the etiology. For the etiology of SE, EMSE allowed detailed etiology-risk stratification; however, this may need adapting to some new emerging etiologies such as autoimmune entities. Nevertheless, it is unclear how the test weighs nonlesional drug-resistant epilepsy with proper compliance. To some extent, this obstacle was overcome by STESS as it was satisfied with asking only about the history of previous seizures, as a surrogate for acute etiology. For comorbidities, they were put into consideration in the case of the EMSE while omitted in the case of STESS. However, some comorbidities were omitted by CCI, i.e., human immunodeficiency virus (HIV), and others may need scoring modification after significant improvement in their five-year survival rate, i.e., leukemia and lymphoma, with world regional variation in risk stratification according to available treatment facilities. Finally, the clinical ingredient based on seizure semiology was inadvertently omitted in EMSE as it relies primarily on the epidemiological measure but was admitted through STESS scoring as the worst seizure type. It was noted that the use of “worst seizure type” as a variable, does not allow the evaluation of patients with justice. Such scoring parameter weighs equally for a patient who had an episode of NCSE and the one who had an episode of generalized-convulsive SE evolving to NCSE. Therefore, change in SE semiology has to be studied as a reliable predictor of SE mortality outcome than “worst seizure type.”
4
M.S. El-Tamawy et al. / Epilepsy & Behavior 102 (2020) 106686
Fig. 1. Receiver-operating characteristic curve for STESS for predicting SE outcome. STESS: status epilepticus severity score, AUC: area under curve, CI: confidence interval, NPV: negative predictive value, PPV: positive predictive value.
Fig. 2. Receiver-operating characteristic curve for EMSE-EAC for predicting SE outcome. EMSE: Epidemiology-based mortality score in status epilepticus, EAC: etiology-age-comorbidity, AUC: area under curve, CI: confidence interval, NPV: negative predictive value, PPV: positive predictive value.
M.S. El-Tamawy et al. / Epilepsy & Behavior 102 (2020) 106686
5
Fig. 3. Receiver-operating characteristic curve for EMSE-EACL and EMSE-EACLD for predicting SE outcome. EMSE: Epidemiology-based mortality score in status epilepticus, EACL: etiologyage-comorbidity-level of consciousness, EACLD: etiology-age-comorbidity-level of consciousness-duration, AUC: area under curve, CI: confidence interval, NPV: negative predictive value, PPV: positive predictive value.
Fig. 4. Receiver-operating characteristic curve for EMSE-EACLDE for predicting SE outcome. EMSE: Epidemiology-based mortality score in status epilepticus, EACLDE: etiology-agecomorbidity-level of consciousness-duration-EEG, AUC: area under curve, CI: confidence interval, NPV: negative predictive value, PPV: positive predictive value.
6
M.S. El-Tamawy et al. / Epilepsy & Behavior 102 (2020) 106686
There is an urgent need for future multicenter studies for adaptation of SE outcome prediction scores taking into consideration the previously mentioned strengths and weaknesses of STESS and EMSE. The strengths of this study are the large cohort and its prospective observational design from the largest and oldest tertiary referring center in the region. One limitation of this study was being a single center-study. Another limitation was that the functional outcome was not considered as an outcome parameter.
5. Conclusion In the Egyptian population, STESS and EMSE scores are useful tools to predict SE mortality. However, STESS is preferred than EMSE in order not to miss high-risk patients. This is an urgent call for a multicenter study for adaptation of outcome prediction scores in SE to advances in medicine, as new data emerge.
Funding sources This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Declaration of competing interest The authors declare that they have no conflicts of interest. Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.yebeh.2019.106686.
References [1] Leitinger M, Kalss G, Rohracher A, Pilz G, Novak H, Höfler J, et al. Predicting outcome of status epilepticus. Epilepsy Behav 2015;49:126–30. https://doi.org/10.1016/j. yebeh.2015.04.066. [2] Rossetti AO, Logroscino G, Milligan TA, Michaelides C, Ruffieux C, Bromfield EB. Status Epilepticus Severity Score (STESS): a tool to orient early treatment strategy. J Neurol 2008 Oct;255(10):1561–6. https://doi.org/10.1007/s00415-008-0989-1 [PubMed PMID: 18769858; eng.]. [3] Leitinger M, Holler Y, Kalss G, Rohracher A, Novak HF, Höfler J, et al. Epidemiologybased mortality score in status epilepticus (EMSE). Neurocritical Care 2015 Apr;22 (2):273–82. https://doi.org/10.1007/s12028-014-0080-y [PubMed PMID: 25412806; eng.]. [4] Sutter R, Kaplan PW, Ruegg S. Independent external validation of the status epilepticus severity score. Critical Care Med 2013 Dec;41(12):e475–9. https://doi.org/10. 1097/CCM.0b013e31829eca06 [PubMed PMID: 23982031; eng.]. [5] Kang BS, Kim DW, Kim KK, Moon Hye Jin, Kim Young-Soo, Kim Hyun Kyung, et al. Prediction of mortality and functional outcome from status epilepticus and independent external validation of STESS and EMSE scores [journal article]. Crit Care 2016 January 27;20(1):25. https://doi.org/10.1186/s13054-016-1190-z. [6] Yechoor N, Adeli A, Hafeez S. External validation of the epidemiology-based mortality score in status epilepticus in an American intensive care population. Epilepsy Res 2018 Dec;148:32–6. https://doi.org/10.1016/j.eplepsyres.2018.10.001 [PubMed PMID: 30342324; eng.]. [7] Hassan H, Rajiv KR, Menon R, Menon D, Nair M, Radhakrishnan A. An audit of the predictors of outcome in status epilepticus from a resource-poor country: a comparison with developed countries. Epileptic Disorders 2016 Jun 1;18(2):163–72. https://doi.org/10.1684/epd.2016.0832 [PubMed PMID: 27248976; eng.]. [8] 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. Epilepsia 2015 Oct;56(10):1515–23. https://doi.org/10. 1111/epi.13121 [PubMed PMID: 26336950; eng.]. [9] Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chron Dis 1987;40(5):373–83 [PubMed PMID: 3558716; eng.]. [10] Phabphal K, Geater A, Limapichart K, Sathirapanya P, Setthawatcharawanich S. Adult tonic–clonic convulsive status epilepticus over the last 11 years in a resource-poor country: a tertiary referral centre study from southern Thailand. Epileptic Disord 2013 Sep;15(3):255–61. https://doi.org/10.1684/epd.2013.0604 [PubMed PMID: 24007808; eng.]. [11] Zhang Y, Chen D, Xu D, Tan G, Liu L. Clinical utility of EMSE and STESS in predicting hospital mortality for status epilepticus. Seizure 2018 Aug;60:23–8. https://doi.org/ 10.1016/j.seizure.2018.05.017 [PubMed PMID: 29906706; eng.].
Contents lists available at ScienceDirect
Epilepsy & Behavior journal homepage: www.elsevier.com/locate/yebeh
External validation of STESS and EMSE as outcome prediction scores in an Egyptian cohort with status epilepticus Mohamed S. El-Tamawy a, Hanan Amer a, Nirmeen A. Kishk a, Amani M. Nawito b, Mye A. Basheer b, Nelly Alieldin c, Rehab Magdy a,⁎, Alshimaa S. Othman a a b c
Department of Neurology, Kasr Al-Ainy Faculty of Medicine, Cairo University, Cairo, Egypt Department of Neurophysiology, Kasr Al-Ainy Faculty of Medicine, Cairo University, Cairo, Egypt Department of Cancer Epidemiology, National Cancer Institute, Cairo University, Cairo, Egypt
a r t i c l e
i n f o
Article history: Received 14 October 2019 Revised 4 November 2019 Accepted 4 November 2019 Available online xxxx Keywords: Status epilepticus STESS EMSE Mortality outcome
a b s t r a c t Purpose: There is a lack of data concerning the performance of the outcome prediction scores in patients with status epilepticus (SE) in developing countries. The aim of this study was to compare the predictive performances of the status epilepticus severity score (STESS) and the epidemiology-based mortality score in status epilepticus (EMSE) and adaptation of such scoring system to be compatible with the nature of society. Method: This is a prospective study, conducted in Egypt from the period of January 2017 to June 2018. The main outcome measure was survival versus death, on hospital discharge. The cutoff point with the best sensitivity and specificity to predict mortality was determined through a receiver operating characteristic (ROC) curve. Results: Among the 144 patients with SE with a mean age of 39.3 ± 19.5 years recruited into the study, 38 patients (26.3%) died in the hospital with the survival of 99 patients while 7 patients (4.9%) were referred to other centers with an unknown outcome. Although EMSE had a bit larger area under the curve (AUC) (0.846) than STESS-3 (AUC 0.824), STESS-3 had the best performance as in-hospital death prediction score as it has a higher negative predictive value (94.6%) than that of EMSE (90.9%) in order not to miss high-risk patients. Conclusion: In the Egyptian population, STESS and EMSE are useful tools in predicting mortality outcome of SE. The STESS performed significantly better than EMSEE combinations as a mortality prediction score. © 2019 Elsevier Inc. All rights reserved.
1. Introduction Status epilepticus (SE) is a major neurological emergency associated with significant mortality outcomes. Such an outcome may influence treatment strategy choice [1]. For that purpose, two clinical scoring tools have been developed for in-hospital mortality prediction after SE, based on different sets of prognostic parameters. The Status Epilepticus Severity Score (STESS) was originally validated by Rossetti, Logroscino [2] followed by Epidemiology-Based Mortality Score in
Abbreviations: SE, status epilepticus; STESS, status epilepticus severity score; EMSE, epidemiology-based mortality score in status epilepticus; EACLDE, etiology (E), age (A), comorbidity (C), level of consciousness before treatment (L), duration (D), EEG (E); ER, emergency room; NICU, neurological intensive care unit; CSE, convulsive status epilepticus; NCSE, nonconvulsive status epilepticus; ILAE, the International League Against Epilepsy; CCI, Charlson Comorbidity Index; EEG, electroencephalogram; SD, standard deviation; ROC curve, receiver operator characteristic curve; AUC, area under the curve; PPV, positive predictive values; NPV, negative predictive values; CNS, central nervous system; HIV, human immunodeficiency virus. ⁎ Corresponding author. E-mail address: [email protected] (R. Magdy).
https://doi.org/10.1016/j.yebeh.2019.106686 1525-5050/© 2019 Elsevier Inc. All rights reserved.
Status Epilepticus (EMSE) by Leitinger, Holler [3], then received external validation in several developed countries [4,5,6]. However, there is no available data in that concern in developing countries. Apparent differences between the developed and developing countries exist in terms of economics and healthcare facilities, and these differences might influence the etiologies, clinical profiles of patients with SE, and then outcome [7]. The aim of this study was to compare the performance of the STESS and EMSE and adaptation of such scoring system in light of unique demographic and clinical characteristics of patients in Egypt as the largest country in terms of population, in the Eastern Mediterranean region. 2. Subjects and methods This is a prospective study, conducted over a total working duration of 18 months (January 2017 to June 2018). Patients were recruited from the emergency room (ER), neurological intensive care unit (NICU), or other medical intensive care units of Cairo University Hospitals. The study was approved by the ethical committee of the neurology department, Cairo University. Informed consents were waived.
2
M.S. El-Tamawy et al. / Epilepsy & Behavior 102 (2020) 106686
2.1. Subjects
2.3. Outcome
The eligibility criteria for this study were that all patients had been with diagnosed SE according to the new the International League Against Epilepsy (ILAE) Report on definition and classification of SE [8], including those with both convulsive SE (CSE) and nonconvulsive SE (NCSE) semiologies. Patients' ≥12 years old of both genders were included.
The main outcome measure was survival versus death, on hospital discharge.
2.2. Methods A set of clinical data was recorded to build up the two clinical scoring tools, including age, history of previous seizures, seizure semiology, seizure duration, level of consciousness before treatment, comorbidity by Charlson Comorbidity Index (CCI) [9], and onset electroencephalogram (EEG), if available. The etiology of SE was determined after a set of laboratory and radiological imaging according to the clinical situation. 2.2.1. Status epilepticus severity score (STESS) [2] This scale was used to predict the mortality risk in patients with nonanoxic SE. A score of three points or higher indicates the risk of death. The STESS score included the following four parameters: level of consciousness (alert or somnolent/confused = 0 point, stuporous or comatose = 1 point), “worst” seizure type (simple-partial, complexpartial, absence, myoclonic as complicating idiopathic generalized epilepsy = 0 point; generalized–convulsive = 1 point; nonconvulsive SE in coma = 2 points), age (under 65 years = 0 point, 65 or older = 2 points), and history of previous seizures, as a surrogate for acute etiology, (yes = 0 point, no or unknown = 1 point). 2.2.2. Epidemiology-based mortality score in status epilepticus (EMSE) [3] This scale was used to predict the mortality risk in a population with SE on more comprehensive scoring system. The long version of the score has six-item categories: age (A), etiology (E), comorbidity (C), level of consciousness before treatment (L), duration (D), and EEG (E). The evaluation of the EMSE scoring system is summarized in Table 1. In this study, the following combinations were applied: EAC, EACL, EACLD, and EACLDE. The score gives a clear weight for epidemiological data. However, some limitations exist but had been resolved by communicating with the corresponding author of the score via email (Supplementary material 1 & 2).
2.4. Statistical analysis The SPSS (statistical package for social sciences) version 23.0 was used for data management and data analysis. To describe the data, frequency (percent) and mean ± standard deviation (SD) were used. P-values less than 0.05 were considered statistically significant. The receiver operator characteristic (ROC) curve was used to select the best cutoff for scores that best predict surviving after an episode of SE. After selecting the best cutoff, different accuracy measures were calculated including sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV), in addition to total accuracy. 3. Results 3.1. Demographic and clinical characteristics of the patients One hundred forty-four patients with SE were enrolled in this study. Their ages ranged from 13 to 80 years, with a mean of 39.3 ± 19.5 years, 79 males (54.9%) versus 65 females (45.1%). The clinical characteristics of the study population are summarized in Table 2. Only five patients were post anoxic, so they were excluded from STESS analyses as this scale was intended to be used only in patients with nonanoxic SE. Electroencephalogram was performed for only 54 patients (37.5% of the total cohort) because of limited resources availability. 3.2. SE outcome on hospital discharge Apart from 7 patients (4.9%) who were referred to other centers because of the unavailability of intensive care unit (ICU) beds, the outcome was favorable for the majority of the study population, with the survival of 99 patients (68.8%). On the other hand, 38 patients had a fatal outcome. Relation of outcome prediction scales to the outcome: The STESS and different EMSE combinations were significantly higher among patients with fatal outcome (Table 3). This included EMSEEACLDE, which was available for only 54 patients, of whom 16 died.
Table 1 Scoring system of epidemiology-based mortality score in status epilepticus (EMSE). Etiology (score one stratum) CNS anomalies 2 Hydrocephalus 8 Drug abuse 11 Brain tumor 16 CNS infection 33
Drug resistant\withdrawal, poor compliance 2 Remote cerebrovascular disease, brain injury 7 Head trauma 12 Metabolic; sodium imbalance 17 Acute cerebrovascular disease 26
Age (score one stratum) 21–30 y 1
31–40 y 2
41–50 y 3
51–60 y 5
Multiple sclerosis 5 Alcohol abuse 10 Cryptogenic 12 Metabolic disorders 22 Anoxia 65
61–70 y 7
Comorbidity (score each disease) Hemiplegia, Moderate to severe renal Myocardial infarction, Congestive heart failure, Peripheral vascular disease, cerebrovascular disease, dementia, chronic pulmonary disease, Connective disease, diabetes with end organ damage, any tumor including tissue disease, Peptic ulcer disease, mild liver disease 10 Leukemia\Lymphoma 20 LOC (score one stratum) Awake 0
Stupor 14
Somnolence 5
Duration of seizure (score one stratum) b1 h 3 EEG (score the worst stratum) Burst suppression 60
ASIDs 40
71–80 y 8
N80 y 10
Moderate to severe liver disease 30 Metastatic Solid tumor 60
Coma 23
≥1 h 33 LPDs 40
GPDs 40
No ASIDs, LPDs or GPDs 0
CNS: central nervous system, EEG: electroencephalography, ASID: after status ictal discharge, GPD: generalized periodic discharge, LOC: level of consciousness, LPD: lateralized periodic discharge.
M.S. El-Tamawy et al. / Epilepsy & Behavior 102 (2020) 106686 Table 2 Clinical characteristics of the patients. n = 144 Semiology of SE episode according to the new (2015) ILAE classification system SE with prominent motor features 139 (96.5%) SE without prominent motor features (NCSE) 5 (3.5%) History of previous epilepsy 76 (52.8%) Worst seizure type Simple partial, CPS, absence, myoclonic 22 (15.27%) Generalized convulsive 121 (84.02%) Non-convulsive SE in coma 1 (0.69%) LOC pretreatment Alert 8 (5.6%) Confused\somnolent 48 (33.3%) Stuporous 41 (28.5%) Comatose 47 (32.6%) Charlson Comorbidity Index 0–70 (5.9 ± 11.6) Etiology of SE Acute symptomatic 93 (64.6%) Remote symptomatic 39 (27.1%) Progressive symptomatic 9 (6.3%) SE in defined electroclinical syndromes 2 (1.4%) Unknown (Cryptogenic) 1 (0.7%) EEGa Burst suppression 3 (5.6%) ASIDs 3 (5.6%) PDs 5 (9.3%) No ASIDs or PDs 43 (79.6%) ILAE: international league against epilepsy, SE: status epilepticus, NCSE: nonconvulsive status epilepticus, CPS: complex partial seizures, LOC: level of consciousness, CSE: convulsive status epilepticus, EEG: electroencephalogram, ASID: after status ictal discharge, PDs: periodic discharges. a EEG was performed for only 54 patients.
3.3. Performance of the STESS and EMSE combinations in predicting inhospital death According to the ROC curve for predicting in-hospital death, the best sensitivity/specificity trade-off was set at cutoff points of 3, 28, 35, 67, and 67 for STESS, EMSE-EAC, EMSE-EACL, EMSE-EACLD, and EMSEEACLDE, respectively (Figs. 1, 2, 3, & 4).
3.4. Comparison of the STESS and EMSE combinations as in-hospital death prediction scores The EMSE-EACLDE had the largest area under the curve (AUC) followed by STESS-3. Although EMSE-EACLDE had a bit larger AUC (0.846) than STESS-3 (AUC 0.824), STESS-3 had the best performance as in-hospital death prediction score as it has a higher NPV (94.6%) than that of EMSE-EACLDE (90.9%) (Figs. 1, 2, 3, & 4).
4. Discussion This was the first independent, external validation study of the predictive accuracy of two clinical scoring systems (STESS and EMSE) used
Table 3 Relation of outcome prediction scores to outcome among the study population. Outcome prediction scores (Mean ± SD)
Survivors (n = 99)
Non survivors (n = 38)
P-value
STESSa EMSE-EAC EMSE-EACL EMSE-EACLD
2.0 ± 1.1 19.5 ± 16.9 30.4 ± 19.7 48.9 ± 45.3
3.4 ± 1.0 40.2 ± 29.8 59.5 ± 31.7 87.8 ± 33.8
b0.001 b0.001 b0.001 b0.001
STESS: status epilepticus severity score, EMSE: epidemiology-based mortality score in status epilepticus, E: etiology, A: age, C: comorbidity, L: level of consciousness, D: duration. a Five SE episodes due to hypoxic brain insult were excluded from STESS scoring (4 of them died and only 1 survived).
3
for in-hospital death prediction among patients with SE, in developing countries. The scoring system must be adapted to different regions in the world according to their distinctive demographic characteristics. The differences in the demographic factors, etiologies, facilities, e.g., nonavailability of continuous EEG monitoring, and treatment variables are likely to be responsible for such heterogeneity in SE outcome noted between developing and developed countries [7]. The most common etiology of SE in developing countries is antiepileptic drug (AED)-related factors. However, vascular causes represent the majority in the developed world. Furthermore, central nervous system (CNS) infections and autoimmune entities predominate in developing countries [10]. In this study, the optimal cutoff point of the STESS for the best discrimination between survivors and nonsurvivors (≥ 3) was similar to the one calculated by the original investigators with high NPV (94.6%). Some studies identified the optimal cutoff point of the STESS at (≥ 4) [4,11]. On the other hand, our results revealed the optimal cutoff point of the EMSE-EACLDE for the best discrimination between survivors and nonsurvivors (≥67) was larger than the one calculated by the original investigators [3] with high NPV (90.9%). All these results can be explained by different epidemiological characteristics of the studied populations. In fact, the primary aim of the development of such scores was to predict the in-hospital short-term outcome of patients with SE in a trail to design treatment plans, and this is the value of a high NPV with accepted accuracy in order not to miss high-risk patients [2,3]. For that purpose, STESS-3 had the best performance as in-hospital death prediction score as it has a higher NPV (94.6%) than that of EMSE-EACLDE (90.9%) at the expense of the small difference between the measured AUC of both scores. The best combination of EMSE score variables was given when EACLDE combinations were used (AUC 0.846). However, predictive accuracy was not too lowered on excluding EEG as a variable (AUC 0.816). This EMSE combination (EACLD) was not previously analyzed in previous studies, but it was intentionally analyzed in our study to suit our center's capabilities where the portable EEG is not available all the time. Furthermore, the superiority of the performance of STESS over EMSE was in favor of this limitation in epilepsy centers in developing countries, as EEG is not including in STESS scoring system. A point worthy of discussion, strengths, and weaknesses of these two clinical scoring systems can be accentuated or embellished according to different demographic characteristics and etiology of SE around the world. It is fast to apply STESS, whereas, EMSE may take some time until the tests reveal the etiology. For the etiology of SE, EMSE allowed detailed etiology-risk stratification; however, this may need adapting to some new emerging etiologies such as autoimmune entities. Nevertheless, it is unclear how the test weighs nonlesional drug-resistant epilepsy with proper compliance. To some extent, this obstacle was overcome by STESS as it was satisfied with asking only about the history of previous seizures, as a surrogate for acute etiology. For comorbidities, they were put into consideration in the case of the EMSE while omitted in the case of STESS. However, some comorbidities were omitted by CCI, i.e., human immunodeficiency virus (HIV), and others may need scoring modification after significant improvement in their five-year survival rate, i.e., leukemia and lymphoma, with world regional variation in risk stratification according to available treatment facilities. Finally, the clinical ingredient based on seizure semiology was inadvertently omitted in EMSE as it relies primarily on the epidemiological measure but was admitted through STESS scoring as the worst seizure type. It was noted that the use of “worst seizure type” as a variable, does not allow the evaluation of patients with justice. Such scoring parameter weighs equally for a patient who had an episode of NCSE and the one who had an episode of generalized-convulsive SE evolving to NCSE. Therefore, change in SE semiology has to be studied as a reliable predictor of SE mortality outcome than “worst seizure type.”
4
M.S. El-Tamawy et al. / Epilepsy & Behavior 102 (2020) 106686
Fig. 1. Receiver-operating characteristic curve for STESS for predicting SE outcome. STESS: status epilepticus severity score, AUC: area under curve, CI: confidence interval, NPV: negative predictive value, PPV: positive predictive value.
Fig. 2. Receiver-operating characteristic curve for EMSE-EAC for predicting SE outcome. EMSE: Epidemiology-based mortality score in status epilepticus, EAC: etiology-age-comorbidity, AUC: area under curve, CI: confidence interval, NPV: negative predictive value, PPV: positive predictive value.
M.S. El-Tamawy et al. / Epilepsy & Behavior 102 (2020) 106686
5
Fig. 3. Receiver-operating characteristic curve for EMSE-EACL and EMSE-EACLD for predicting SE outcome. EMSE: Epidemiology-based mortality score in status epilepticus, EACL: etiologyage-comorbidity-level of consciousness, EACLD: etiology-age-comorbidity-level of consciousness-duration, AUC: area under curve, CI: confidence interval, NPV: negative predictive value, PPV: positive predictive value.
Fig. 4. Receiver-operating characteristic curve for EMSE-EACLDE for predicting SE outcome. EMSE: Epidemiology-based mortality score in status epilepticus, EACLDE: etiology-agecomorbidity-level of consciousness-duration-EEG, AUC: area under curve, CI: confidence interval, NPV: negative predictive value, PPV: positive predictive value.
6
M.S. El-Tamawy et al. / Epilepsy & Behavior 102 (2020) 106686
There is an urgent need for future multicenter studies for adaptation of SE outcome prediction scores taking into consideration the previously mentioned strengths and weaknesses of STESS and EMSE. The strengths of this study are the large cohort and its prospective observational design from the largest and oldest tertiary referring center in the region. One limitation of this study was being a single center-study. Another limitation was that the functional outcome was not considered as an outcome parameter.
5. Conclusion In the Egyptian population, STESS and EMSE scores are useful tools to predict SE mortality. However, STESS is preferred than EMSE in order not to miss high-risk patients. This is an urgent call for a multicenter study for adaptation of outcome prediction scores in SE to advances in medicine, as new data emerge.
Funding sources This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Declaration of competing interest The authors declare that they have no conflicts of interest. Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.yebeh.2019.106686.
References [1] Leitinger M, Kalss G, Rohracher A, Pilz G, Novak H, Höfler J, et al. Predicting outcome of status epilepticus. Epilepsy Behav 2015;49:126–30. https://doi.org/10.1016/j. yebeh.2015.04.066. [2] Rossetti AO, Logroscino G, Milligan TA, Michaelides C, Ruffieux C, Bromfield EB. Status Epilepticus Severity Score (STESS): a tool to orient early treatment strategy. J Neurol 2008 Oct;255(10):1561–6. https://doi.org/10.1007/s00415-008-0989-1 [PubMed PMID: 18769858; eng.]. [3] Leitinger M, Holler Y, Kalss G, Rohracher A, Novak HF, Höfler J, et al. Epidemiologybased mortality score in status epilepticus (EMSE). Neurocritical Care 2015 Apr;22 (2):273–82. https://doi.org/10.1007/s12028-014-0080-y [PubMed PMID: 25412806; eng.]. [4] Sutter R, Kaplan PW, Ruegg S. Independent external validation of the status epilepticus severity score. Critical Care Med 2013 Dec;41(12):e475–9. https://doi.org/10. 1097/CCM.0b013e31829eca06 [PubMed PMID: 23982031; eng.]. [5] Kang BS, Kim DW, Kim KK, Moon Hye Jin, Kim Young-Soo, Kim Hyun Kyung, et al. Prediction of mortality and functional outcome from status epilepticus and independent external validation of STESS and EMSE scores [journal article]. Crit Care 2016 January 27;20(1):25. https://doi.org/10.1186/s13054-016-1190-z. [6] Yechoor N, Adeli A, Hafeez S. External validation of the epidemiology-based mortality score in status epilepticus in an American intensive care population. Epilepsy Res 2018 Dec;148:32–6. https://doi.org/10.1016/j.eplepsyres.2018.10.001 [PubMed PMID: 30342324; eng.]. [7] Hassan H, Rajiv KR, Menon R, Menon D, Nair M, Radhakrishnan A. An audit of the predictors of outcome in status epilepticus from a resource-poor country: a comparison with developed countries. Epileptic Disorders 2016 Jun 1;18(2):163–72. https://doi.org/10.1684/epd.2016.0832 [PubMed PMID: 27248976; eng.]. [8] 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. Epilepsia 2015 Oct;56(10):1515–23. https://doi.org/10. 1111/epi.13121 [PubMed PMID: 26336950; eng.]. [9] Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chron Dis 1987;40(5):373–83 [PubMed PMID: 3558716; eng.]. [10] Phabphal K, Geater A, Limapichart K, Sathirapanya P, Setthawatcharawanich S. Adult tonic–clonic convulsive status epilepticus over the last 11 years in a resource-poor country: a tertiary referral centre study from southern Thailand. Epileptic Disord 2013 Sep;15(3):255–61. https://doi.org/10.1684/epd.2013.0604 [PubMed PMID: 24007808; eng.]. [11] Zhang Y, Chen D, Xu D, Tan G, Liu L. Clinical utility of EMSE and STESS in predicting hospital mortality for status epilepticus. Seizure 2018 Aug;60:23–8. https://doi.org/ 10.1016/j.seizure.2018.05.017 [PubMed PMID: 29906706; eng.].