- Email: [email protected]
Highly malignant routine EEG predicts poor prognosis after cardiac arrest in the Target Temperature Management trial
Accepted Manuscript Title: Highly malignant routine EEG predicts poor prognosis after cardiac arrest in the Target Temperature Management trial Authors: S. Backman, T. Cronberg, H. Friberg, S. Ull´en, J. Horn, J. Kjaergaard, C. Hassager, M. Wanscher, N. Nielsen, E. Westhall PII: DOI: Reference:
S0300-9572(18)30367-8 https://doi.org/10.1016/j.resuscitation.2018.07.024 RESUS 7693
To appear in:
Resuscitation
Received date: Revised date: Accepted date:
31-5-2018 12-7-2018 24-7-2018
Please cite this article as: Backman S, Cronberg T, Friberg H, Ull´en S, Horn J, Kjaergaard J, Hassager C, Wanscher M, Nielsen N, Westhall E, Highly malignant routine EEG predicts poor prognosis after cardiac arrest in the Target Temperature Management trial, Resuscitation (2018), https://doi.org/10.1016/j.resuscitation.2018.07.024 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Resuscitation Backman Page 1
Highly malignant routine EEG predicts poor prognosis after cardiac arrest in the Target Temperature Management trial
Backman S, Cronberg C, Friberg H, Ullén S, Horn J, Kjaergaard J, Hassager C, Wanscher M,
IP T
Nielsen N, Westhall E.
SC R
Sofia Backman, MD, Lund University, Skane University Hospital, Department of Clinical Sciences Lund, Clinical Neurophysiology, Lund, Sweden, [email protected] (corresponding author)
U
Tobias Cronberg, MD, PhD, Lund University, Skane University Hospital, Department of Clinical Sciences Lund, Neurology, Lund, Sweden, [email protected]
N
Hans Friberg, MD, PhD, Lund University, Skane University Hospital, Department of Clinical Sciences Lund, Intensive and Perioperative Care, Lund, Sweden, [email protected]
M
A
Susann Ullén, PhD, Clinical Studies Sweden – Forum South, Skane University Hospital, Lund, Sweden, [email protected]
ED
Janneke Horn, MD, PhD, Department of Intensive Care Medicine, Academic Medical Center, University of Amsterdam, Amsterdam, The Netherlands, [email protected]
PT
Jesper Kjaergaard, MD, DMSci, Department of Cardiology, Rigshospitalet and Department of Clinical Medicine, University of Copenhagen, Copenhagen, Denmark, [email protected]
CC E
Christian Hassager, MD, DMSci, Department of Cardiology, Rigshospitalet and Department of Clinical Medicine, University of Copenhagen, Copenhagen, Denmark, [email protected] Michael C. Jaeger Wanscher, MD, PhD, Department of Cardiothoracic Anaesthesia, Rigshospitalet and Department of Clinical Medicine, University of Copenhagen, Copenhagen, Denmark, [email protected]
A
Niklas Nielsen, MD, PhD, Lund University, Helsingborg Hospital, Department of Clinical Sciences Lund, Intensive and Perioperative Care, Lund, Sweden, [email protected] Erik Westhall, MD, PhD, Lund University, Skane University Hospital, Department of Clinical Sciences Lund, Clinical Neurophysiology, Lund, Sweden, [email protected]
Resuscitation Backman Page 2
Contents Title: 116 characters Abstract: 243 Main text: 1481 Tables and figures: 3, Supplementary tables and figures: 5 References: 20
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Corresponding author Sofia Backman, MD Department of Clinical Neurophysiology Skane University Hospital, S-221 85 Lund, Sweden Email: [email protected] Phone: +46 46 173380 Fax: +46 46 146528
Resuscitation Backman Page 3
Abstract
Introduction Routine EEG is widely used and accessible for post arrest neuroprognostication. Recent studies, using standardised EEG terminology, have proposed highly malignant EEG patterns with
IP T
promising predictive ability.
SC R
Objectives
To validate the performance of standardised routine EEG patterns to predict neurological
U
outcome after cardiac arrest.
N
Methods
A
In the prospective multicenter Target Temperature Management trial, comatose cardiac arrest
M
patients were randomised to different temperature levels (950 patients, 36 sites). According to
ED
the prospective protocol a routine EEG was performed in patients who remained comatose after the 36 hours temperature control intervention. EEGs were retrospectively reviewed blinded to
PT
outcome using the standardised American Clinical Neurophysiology Society terminology. Highly malignant, malignant and benign EEG patterns were correlated to poor and good
CC E
outcome, defined by best achieved Cerebral Performance Category up to 180 days.
A
Results
At 20 sites 207 patients had a routine EEG performed at median 76 hours after cardiac arrest. Highly malignant patterns (suppression or burst-suppression with or without discharges) had a high specificity for poor outcome (98%, CI 92-100), but with limited sensitivity (31%, CI 2439). Our false positive patient had a burst-suppression pattern during ongoing sedation. A
Resuscitation Backman Page 4
benign EEG, i.e. continuous normal-voltage background without malignant features, identified patients with good outcome with 77% (CI 66-86) sensitivity and 80% (CI 73-86) specificity.
Conclusion Highly malignant routine EEG after targeted temperature management is a strong predictor of
IP T
poor outcome. A benign EEG is an important indicator of a good outcome for patients
SC R
remaining in coma.
Keywords ACNS nomenclature
U
EEG
N
Cardiac arrest
A
CC E
PT
ED
Outcome prediction
M
Hypoxic-ischaemic encephalopathy
A
Coma
Resuscitation Backman Page 5
List of abbreviations CA=Cardiac Arrest CI=95% Confidence Intervals CPC=Cerebral Performance Category scale EEG=Electroencephalography
IP T
IQR=Interquartile Range NSE=Neuron-specific enolase
SC R
TTM=Target Temperature Management
A
CC E
PT
ED
M
A
N
U
WLST=Withdrawal of Life-Sustaining Therapy
Resuscitation Backman Page 6
Introduction Routine EEG is commonly used for neurological prognostication of comatose patients after cardiac arrest (CA) [1]. Due to temporal evolution of EEG patterns during the first days after a hypoxic-ischaemic event and possible interference of sedative agents, it is crucial to define robust EEG patterns and time-windows for prognostication. In the multicenter Target
IP T
Temperature Management trial (TTM-trial) comatose survivors were randomised to two target temperature levels. An analysis plan of the routine EEGs in the TTM-trial was published [2],
SC R
including studies on interrater agreement, prognostic ability and the present validation study. In the previous studies on eight selected TTM-sites, we found substantial interrater agreement
[3] and high prognostic ability [4] for prespecified highly malignant EEG patterns. The aim of
N
U
the present study was to validate these patterns in the remaining TTM-cohort.
A
Methods
M
Adult comatose survivors after out-of-hospital CA of a presumed cardiac cause were
ED
randomised to 33ºC versus 36ºC temperature management at 36 sites in Europe and Australia
PT
as previously described [5]. Ethical permissions were obtained in the participating countries.
Sedation was tapered after rewarming, approximately 36 hours after CA, if not indicated for
CC E
intensive care reasons. A routine EEG was included in the trial protocol 48-72 hours after CA, in comatose patients. A detailed protocol for the design of the present study was published [2].
A
Routines for neurological prognostication and practise of withdrawal of life-sustaining therapy (WLST) has been reported [6, 7].
Resuscitation Backman Page 7
Neurological function was evaluated using the Cerebral Performance Category scale (CPC), daily in the intensive care unit, at hospital discharge and at a face-to-face follow-up at six months. A best CPC of 1-2 at any time-point was considered a good outcome.
The patient´s first recorded EEG between 36 hours and 14 days after CA were included.
IP T
Recordings with less than ten minutes duration were excluded. The EEGs were retrospectively analysed by senior clinical neurophysiologists (SB or EW), blinded to clinical parameters and
SC R
outcome, using the standardised terminology according to the American Clinical
Neurophysiology Society [8]. EEGs were classified into three predefined categories [2]: highly malignant EEG, EEG with any malignant feature or benign EEG (Table S1). Since 73% of the
U
EEGs in the present study did not include notations of standardised reactivity testing, mainly
A
N
due to data format reasons, we excluded reactivity from the definitions.
M
We used SPSS Statistics 25.0 for Mann-Whitney U-test, Chi-square or Fisher´s exact test where
ED
applicable. Data is presented as median and interquartile range (IQR). 95% confidence intervals
Results
PT
for percentages were calculated using Wilson´s method.
CC E
All 36 sites performed routine EEGs. We previously published data from eight selected sites [4]. Additional eight sites were excluded due to pruned EEG recordings or lack of technical
A
possibility to export registrations (Fig. 1). EEGs recorded from 207 patients at the remaining 20 sites were included in the present study. Patients with EEG had longer duration from CA to return of spontaneous circulation, higher neuron-specific enolase (NSE) levels, longer hospital stay and worse outcome (Table 1). The median time from CA to EEG was 76 hours (IQR 62-
Resuscitation Backman Page 8
104, range 38-309 hours). 95% of the EEGs were recorded with at least 12 electrodes and 87% with at least 16 electrodes. The median duration of the recordings was 21 minutes (IQR 15-30).
Poor outcome was seen in 141 patients (68%), of whom 44 patients had a highly malignant EEG pattern. One patient with highly malignant EEG had a good outcome, resulting in 98%
IP T
specificity and 31% sensitivity to predict poor outcome (Table 2). The EEG of the false positive
patient showed burst-suppression without discharges or highly epileptiform bursts, at 75 hours
SC R
after CA, with ongoing sedation during the EEG registration (Suppl. Fig. S1). The prognostic performance of highly malignant patterns were similar when CPC at six months was used as end-point instead of best CPC (Suppl. Table S2). Using a more conservative definition of poor
U
outcome (best CPC 4-5), there were two more false positives with highly malignant EEG
N
patterns, one with suppressed background and continuous periodic discharges and one with
M
A
burst-suppression without discharges.
ED
The presence of malignant rhythmic or periodic features had 96% specificity and 43% sensitivity for poor outcome (Table 2). A malignant EEG background predicted poor outcome
PT
with 80% specificity and 68% sensitivity.
CC E
A benign EEG was present in 66 patients (32%) and identified patients with good outcome with
A
77% (CI 66-86) sensitivity and 80% (CI 73-86) specificity (Suppl Table S3).
EEG was performed early after the intervention in 44% of the patients (up to 72 hours after CA), with a 23% prevalence for highly malignant patterns and 64% for any malignant feature. We found no significant differences for prevalence, specificity or sensitivity for highly malignant or malignant patterns between the early and later group (Suppl Table S4).
Resuscitation Backman Page 9
During the EEG registration, 34% of the patients had ongoing sedation, 34% had antiepileptic treatment, 18% had ongoing myoclonus and one patient had tonic-clonic seizures. Among the sedated patients, 19% had highly a malignant pattern and 55% had at least one malignant feature. There were no significant differences between the sedated and the unsedated group
IP T
regarding prevalence or prognostic ability of highly malignant or malignant patterns to predict
SC R
poor outcome (Suppl Table S4).
Discussion
U
Guidelines suggest EEG for prognostication after CA, but knowledge gaps exist and there is a
N
need for large prospective studies using standardised definitions [9, 10]. In the present study
A
we show that standardised highly malignant as well as benign routine EEG patterns in comatose
ED
M
post arrest patients have strong prognostic implications after the 36 hours intervention period.
The results validates our previous results [4] in this separate and larger patient cohort of the
PT
TTM-trial. Highly malignant EEG patterns had high specificity (98%) for poor outcome with relatively narrow confidence intervals (92-100%). Two recent studies, using the same
CC E
standardised definitions for a highly malignant EEG, reported similar results for specificity
A
when EEGs were performed after rewarming [11, 12].
Burst-suppression in patients with good outcome has previously been reported early after CA [13-15], but burst-suppression with identical bursts is a predictor of poor prognosis [13, 16]. Identical bursts are transient and disappear around rewarming and were therefore not
Resuscitation Backman Page 10
systematically evaluated in the present study. Our false positive patient with burst-suppression had neither identical bursts nor epileptiform discharges.
The sensitivity for poor outcome of highly malignant patterns was low (31%) and range from 40-62% in recent studies [4, 11, 12]. Although these patterns have shown substantial interrater
IP T
agreement [3], the differences regarding sensitivity may reflect a more conservative
interpretation of the standardised patterns in the present study or be due to different patient
SC R
cohorts.
It is well known that an early continuous normal-voltage background identified by continuous
U
EEG-monitoring is a strong predictor of favorable prognosis [16, 19]. An EEG that is reactive
N
to external stimuli has previously been shown to be an important predictor [17] but with limited
A
interrater agreement [3, 18]. In this study, benign EEG was a strong predictor of a good
ED
M
outcome, although reactivity testing was not part of the analyses.
Our false positive patient had ongoing sedation during the EEG, which should be taken into
PT
consideration as a risk of reducing the reliability of EEG for the individual patient. The potiental risk for false positive predictions directly after rewarming, due to effects of residual sedation,
CC E
was however not confirmed in our comparison between EEGs that were performed early after rewarming with those performed beyond 72 hours after CA. Notably, 29% of the patients in the
A
late group still had ongoing sedation. Further, we could not find any significant differences for EEG as a predictor of poor outcome, comparing the unsedated and the sedated group, but larger prospective studies with systematical assessment of sedatives are needed.
Resuscitation Backman Page 11
A limitation of our study is the large proportion of patients excluded due to lack of EEG. In the TTM-trial, 48% of the included patients woke up and 15% died before the time-point of prognostication and the majority of these had no EEG performed [7]. The group of patients with EEG is therefore likely to have a worse brain injury, supported by the findings of higher NSElevels and longer time to return of circulation. The results of this study are nevertheless
IP T
important as they represent the comatose patients in clinical practise undergoing EEG for
SC R
prognostication.
We tried to limit the risk of the self-fulfilling prophecy by using a strict and predefined protocol for prognostication in the TTM-study [20]. The highly malignant EEG patterns were not
U
included in the criteria for predicting poor outcome in the trial. The EEG pattern that was
N
considered as part of the WLST decision was a treatment-refractory status epilepticus combined
M
A
with deep coma three days after rewarming [6], but the decision was never based on EEG alone.
ED
Conclusion
We confirm that strictly defined highly malignant EEG patterns on a routine-EEG performed
PT
after the temperature management intervention reliably predict poor neurological outcome. A
CC E
benign EEG is useful to identify patients with a better prognosis.
A
Conflicts of interest: None of the authors have potential conflicts of interest to be disclosed.
Acknowledgement We would like to acknowledge the site investigators and the clinical staff at the EEG laboratories for contributing the EEG data:
Resuscitation Backman Page 12
Liverpool Hospital, St George Hospital, North Shore Hospital, Australia; General University Hospital Prague, Czech Republic; Copenhagen University Hospital Rigshospitalet, Denmark; Santa Maria degli Angeli Hospital Pordenone, Cattinara Hospital Trieste, San Martino Hospital Genoa, Italy; Centre Hospitalier de Luxembourg, Luxembourg; Leeuwarden Medisch Centrum, Rijnstate Hospital Arnhem, Onze Lieuwe Vrouve Gasthuis, The Netherlands; Linköping
IP T
University Hospital, Norrköping Hospital, Sweden; Geneva University Hospital, Switzerland; University Hospital of Wales Cardiff, Royal Bournemouth Hospital, Royal Berkshire Hospital,
SC R
Guys and St Thomas London, St Georges Hospital London, United Kingdom.
Funding source: Supported by The Swedish Heart and Lung Association; the Skåne University
U
Hospital Foundations; the Gyllenstierna-Krapperup Foundation; the Segerfalk foundation; the
N
Swedish National Health System (ALF); the County Council of Skåne; the Swedish Society of
A
Medicine; the Koch Foundation, The Swedish Heart-Lung Foundation, Skane University
M
Hospital foundation AFA Insurance, The Swedish Research Council and Hans-Gabriel and
ED
Alice Trolle-Wachtmeister Foundation; all in Sweden. The Tryg Foundation; Denmark. EU
PT
programme Interreg IV A.
CC E
Financial support: The authors have no financial interests in the study.
A
Legends
Legend Table 1: Patient characteristics. ROSC, return of spontaneous circulation; ICU, intensive care unit; NSE, neuron-specific enolase maximum value at 48 and 72 hours after cardiac arrest; CPC, Cerebral Performance Category
Legend Table 2: Prediction of poor outcome defined as CPC 3-5 as best CPC at any assessed time-point during the first six months after cardiac arrest. Definitions of EEG patterns based on the American Society of Clinical Neurophysiology terminology.
Resuscitation Backman Page 13
A
CC E
PT
ED
M
A
N
U
SC R
IP T
TP, true positive; FP, false positive, TN, true negative; FN, false negative.
Resuscitation Backman Page 14
Table 1 Patient characteristics Patients with EEG 20 sites, n = 207 median (IQR) or n (%)
Patients without EEG 20 sites, n = 403 median (IQR) or n (%)
324 (80.4%) 64 (55-72)
109/401 (27.2%) 28/402 (7.0%)
2/206 (1.0%) 29/205 (14.1%)
9/402 (2.2%) 61/401 (15.2%)
23 (15-35) (n=402) 102 (65-172) (n=402)
A
Time from cardiac arrest - ROSC (min) 27 (18-40) (n=207) - ICU discharge 198 (132-288) (n=207) (hours) - Hospital discharge 460 (264-975) (n=111) (hours)
SC R
60/207 (29.0%) 14/205 (6.8%)
41.9 (18.1-123.3) (n = 167)
Best CPC
1 2 3 4 5
M
NSE (ng/mL)
PT
ED
51 (24.6%) 15 (7.2%) 22 (10.6%) 119 (57.5%) 0
0.001 <0.001
363 (207-539) (n= 298)
0.025
17.7 (12.4-32.0) (n = 259)
<0.001
236 (58.6%) 28 (6.9%) 27 (6.7%) 112 (27.8%) 0
<0.001
N
History of ischemic heart disease - transient ischemic attack or stroke - epilepsy - diabetes
209 (51.9%)
IP T
Temperature target 36ºC 94 (45.4%) (others: 33ºC) Male gender 175 (84.5%) Age (year) 63 (56-70)
p value
U
Variable
A
CC E
Legend Table 1: ROSC, return of spontaneous circulation; ICU, intensive care unit; NSE, neuron-specific enolase maximum value at 48 and 72 hours after cardiac arrest; CPC, Cerebral Performance Category.
Resuscitation Backman Page 15
Table 2 Prediction of poor outcome defined as best CPC 3-5 during the first six months after cardiac arrest Sensitivity (95% CI)
Specificity (95% CI)
TP n
FP n
TN n
FN n
45 (22)
31 (24-39)
98 (92-100)
44
1
65
97
20 (10)
14 (9-21)
100 (94-100)
20
0
66
121
4 (2) 21 (10)
3 (1-7) 14 (6-16)
100 (94-100) 98 (92-100)
4 20
0 1
63 (30)
43 (35-51)
96 (87-98)
60
3
63
81
54 (26)
38 (30-46)
98 (92-100)
53
1
65
88
9 (4)
5 (2-10)
97 (90-99)
7
2
64
134
6 (3)
4 (2-8)
98 (92-100)
5
1
65
136
109 (53)
68 (60-75)
80 (69-88)
96
13
53
45
88 (43) 76 (37)
57 (49-65) 48 (40-56)
89 (80-95) 88 (78-94)
81 68
7 8
59 58
60 73
6 (3-12)
98 (92-100)
9
1
65
132
At least one malignant 128 (62) feature
80 (73-86)
77 (66-86)
113
15
51
28
Both malignant background and rhythmic/ 44 (21) periodic features
30 (24-39)
98 (92-100)
43
1
65
98
10 (5)
CC E
PT
ED
66 65
SC R
U
N
Malignant background Discontinuous (>10% suppression) Low-voltage (<20µV) Reversed anterioposterior gradient
A
Malignant rhythmic or periodic features Periodic discharges (≥50%) Rhythmic spike-andwave (≥50%) Unequivocal seizures or status epilepticus
M
Highly malignant pattern Suppressed background without discharges Suppressed background with continuous periodic discharges Burst-suppression
IP T
Prevalence n (%)
137 121
A
Legend Table 2: Prediction of poor outcome defined as CPC 3-5 as best CPC at any assessed time-point during the first six months after cardiac arrest. 207 included patients. Definitions of EEG patterns based on the American Society of Clinical Neurophysiology terminology. TP, true positive; FP, false positive, TN, true negative; FN, false negative.
Resuscitation Backman Page 16
Fig. 1
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Flow chart of study inclusion
Resuscitation Backman Page 17
References
A
CC E
PT
ED
M
A
N
U
SC R
IP T
[1] Friberg H, Cronberg T, Dunser MW, Duranteau J, Horn J, Oddo M. Survey on current practices for neurological prognostication after cardiac arrest. Resuscitation 2015;90:158-62. [2] Westhall E, Rosen I, Rossetti AO, van Rootselaar AF, Kjaer TW, Horn J, et al. Electroencephalography (EEG) for neurological prognostication after cardiac arrest and targeted temperature management; rationale and study design. BMC Neurol 2014;14:159. [3] Westhall E, Rosen I, Rossetti AO, van Rootselaar AF, Wesenberg Kjaer T, Friberg H, et al. Interrater variability of EEG interpretation in comatose cardiac arrest patients. Clin Neurophysiol 2015;126:2397-404. [4] Westhall E, Rossetti AO, van Rootselaar AF, Wesenberg Kjaer T, Horn J, Ullen S, et al. Standardized EEG interpretation accurately predicts prognosis after cardiac arrest. Neurology 2016;86:1482-90. [5] Nielsen N, Wetterslev J, Cronberg T, Erlinge D, Gasche Y, Hassager C, et al. Targeted temperature management at 33 degrees C versus 36 degrees C after cardiac arrest. N Engl J Med 2013;369:2197-206. [6] Nielsen N, Wetterslev J, al-Subaie N, Andersson B, Bro-Jeppesen J, Bishop G, et al. Target Temperature Management after out-of-hospital cardiac arrest--a randomized, parallel-group, assessor-blinded clinical trial--rationale and design. Am Heart J 2012;163:541-8. [7] Dragancea I, Wise MP, Al-Subaie N, Cranshaw J, Friberg H, Glover G, et al. Protocoldriven neurological prognostication and withdrawal of life-sustaining therapy after cardiac arrest and targeted temperature management. Resuscitation 2017;117:50-7. [8] Hirsch LJ, LaRoche SM, Gaspard N, Gerard E, Svoronos A, Herman ST, et al. American Clinical Neurophysiology Society's Standardized Critical Care EEG Terminology: 2012 version. J Clin Neurophysiol 2013;30:1-27. [9] Callaway CW, Soar J, Aibiki M, Bottiger BW, Brooks SC, Deakin CD, et al. Part 4: Advanced Life Support: 2015 International Consensus on Cardiopulmonary Resuscitation and Emergency Cardiovascular Care Science With Treatment Recommendations. Circulation 2015;132:S84-145. [10] Sandroni C, Cariou A, Cavallaro F, Cronberg T, Friberg H, Hoedemaekers C, et al. Prognostication in comatose survivors of cardiac arrest: an advisory statement from the European Resuscitation Council and the European Society of Intensive Care Medicine. Intensive Care Med 2014;40:1816-31. [11] Beuchat I, Solari D, Novy J, Oddo M, Rossetti AO. Standardized EEG interpretation in patients after cardiac arrest: Correlation with other prognostic predictors. Resuscitation 2018;126:143-6. [12] Rossetti AO, Tovar Quiroga DF, Juan E, Novy J, White RD, Ben-Hamouda N, et al. Electroencephalography Predicts Poor and Good Outcomes After Cardiac Arrest: A TwoCenter Study. Crit Care Med 2017;45:e674-e82. [13] Tjepkema-Cloostermans MC, Hofmeijer J, Trof RJ, Blans MJ, Beishuizen A, van Putten MJ. Electroencephalogram predicts outcome in patients with postanoxic coma during mild therapeutic hypothermia. Crit Care Med 2015;43:159-67. [14] Crepeau AZ, Rabinstein AA, Fugate JE, Mandrekar J, Wijdicks EF, White RD, et al. Continuous EEG in therapeutic hypothermia after cardiac arrest: prognostic and clinical value. Neurology 2013;80:339-44. [15] Amorim E, Rittenberger JC, Baldwin ME, Callaway CW, Popescu A, Post Cardiac Arrest S. Malignant EEG patterns in cardiac arrest patients treated with targeted temperature management who survive to hospital discharge. Resuscitation 2015;90:127-32.
Resuscitation Backman Page 18
A
CC E
PT
ED
M
A
N
U
SC R
IP T
[16] Sivaraju A, Gilmore EJ, Wira CR, Stevens A, Rampal N, Moeller JJ, et al. Prognostication of post-cardiac arrest coma: early clinical and electroencephalographic predictors of outcome. Intensive Care Med 2015;41:1264-72. [17] Rossetti AO, Carrera E, Oddo M. Early EEG correlates of neuronal injury after brain anoxia. Neurology 2012;78:796-802. [18] Hermans MC, Westover MB, van Putten M, Hirsch LJ, Gaspard N. Quantification of EEG reactivity in comatose patients. Clin Neurophysiol 2016;127:571-80. [19] Hofmeijer J, Beernink TM, Bosch FH, Beishuizen A, Tjepkema-Cloostermans MC, van Putten MJ. Early EEG contributes to multimodal outcome prediction of postanoxic coma. Neurology 2015;85:137-43. [20] Cronberg T, Horn J, Kuiper MA, Friberg H, Nielsen N. A structured approach to neurologic prognostication in clinical cardiac arrest trials. Scand J Trauma Resusc Emerg Med 2013;21:45.
Resuscitation Backman Page 19
Fig. 1
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Flow chart of study inclusion
Resuscitation Backman Page 20
Table 1 Patient characteristics
175 (84.5%) 63 (56-70)
324 (80.4%) 64 (55-72)
60/207 (29.0%) 14/205 (6.8%)
109/401 (27.2%) 28/402 (7.0%)
2/206 (1.0%) 29/205 (14.1%)
9/402 (2.2%) 61/401 (15.2%)
27 (18-40) (n=207) 198 (132-288) (n=207)
23 (15-35) (n=402) 102 (65-172) (n=402)
460 (264-975) (n=111)
41.9 (18.1-123.3) (n = 167)
Best CPC
1 2 3 4 5
M
NSE (ng/mL)
PT
ED
51 (24.6%) 15 (7.2%) 22 (10.6%) 119 (57.5%) 0
IP T
209 (51.9%)
SC R
Time from cardiac arrest - ROSC (min) - ICU discharge (hours) - Hospital discharge (hours)
94 (45.4%)
p value
U
History of ischemic heart disease - transient ischemic attack or stroke - epilepsy - diabetes
Patients without EEG 20 sites, n = 403 median (IQR) or n (%)
0.001 <0.001
363 (207-539) (n= 298)
0.025
17.7 (12.4-32.0) (n = 259)
<0.001
236 (58.6%) 28 (6.9%) 27 (6.7%) 112 (27.8%) 0
<0.001
N
Temperature target 36ºC (others: 33ºC) Male gender Age (year)
Patients with EEG 20 sites, n = 207 median (IQR) or n (%)
A
Variable
A
CC E
Legend Table 1: ROSC, return of spontaneous circulation; ICU, intensive care unit; NSE, neuron-specific enolase maximum value at 48 and 72 hours after cardiac arrest; CPC, Cerebral Performance Category.
Resuscitation Backman Page 21
Table 2 Prediction of poor outcome defined as best CPC 3-5 during the first six months after cardiac arrest
At least one malignant feature
TN n
98 (92-100)
44
1
65
97
20 (10)
14 (9-21)
100 (94-100)
20
0
66
121
4 (2) 21 (10)
3 (1-7) 14 (6-16)
100 (94-100) 98 (92-100)
4 20
0 1
63 (30)
43 (35-51)
96 (87-98)
60
3
63
81
54 (26)
38 (30-46)
98 (92-100)
53
1
65
88
9 (4)
5 (2-10)
97 (90-99)
7
2
64
134
6 (3)
4 (2-8)
98 (92-100)
5
1
65
136
109 (53)
68 (60-75)
80 (69-88)
96
13
53
45
88 (43) 76 (37)
57 (49-65) 48 (40-56)
89 (80-95) 88 (78-94)
81 68
7 8
59 58
60 73
6 (3-12)
98 (92-100)
9
1
65
132
80 (73-86)
77 (66-86)
113
15
51
28
30 (24-39)
98 (92-100)
43
1
65
98
10 (5)
128 (62)
CC E
44 (21)
66 65
N
U
SC R
31 (24-39)
FN n
IP T
FP n
PT
Both malignant background and rhythmic/ periodic features
45 (22)
Specificity (95% CI)
A
Malignant background Discontinuous (>10% suppression) Low-voltage (<20µV) Reversed anterioposterior gradient
TP n
M
Malignant rhythmic or periodic features Periodic discharges (≥50%) Rhythmic spike-andwave (≥50%) Unequivocal seizures or status epilepticus
Sensitivity (95% CI)
ED
Highly malignant pattern Suppressed background without discharges Suppressed background with continuous periodic discharges Burst-suppression
Prevalence n (%)
137 121
A
Legend Table 2: Prediction of poor outcome defined as CPC 3-5 as best CPC at any assessed time-point during the first six months after cardiac arrest. 207 included patients. Definitions of EEG patterns based on the American Society of Clinical Neurophysiology terminology. TP, true positive; FP, false positive, TN, true negative; FN, false negative.
S0300-9572(18)30367-8 https://doi.org/10.1016/j.resuscitation.2018.07.024 RESUS 7693
To appear in:
Resuscitation
Received date: Revised date: Accepted date:
31-5-2018 12-7-2018 24-7-2018
Please cite this article as: Backman S, Cronberg T, Friberg H, Ull´en S, Horn J, Kjaergaard J, Hassager C, Wanscher M, Nielsen N, Westhall E, Highly malignant routine EEG predicts poor prognosis after cardiac arrest in the Target Temperature Management trial, Resuscitation (2018), https://doi.org/10.1016/j.resuscitation.2018.07.024 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Resuscitation Backman Page 1
Highly malignant routine EEG predicts poor prognosis after cardiac arrest in the Target Temperature Management trial
Backman S, Cronberg C, Friberg H, Ullén S, Horn J, Kjaergaard J, Hassager C, Wanscher M,
IP T
Nielsen N, Westhall E.
SC R
Sofia Backman, MD, Lund University, Skane University Hospital, Department of Clinical Sciences Lund, Clinical Neurophysiology, Lund, Sweden, [email protected] (corresponding author)
U
Tobias Cronberg, MD, PhD, Lund University, Skane University Hospital, Department of Clinical Sciences Lund, Neurology, Lund, Sweden, [email protected]
N
Hans Friberg, MD, PhD, Lund University, Skane University Hospital, Department of Clinical Sciences Lund, Intensive and Perioperative Care, Lund, Sweden, [email protected]
M
A
Susann Ullén, PhD, Clinical Studies Sweden – Forum South, Skane University Hospital, Lund, Sweden, [email protected]
ED
Janneke Horn, MD, PhD, Department of Intensive Care Medicine, Academic Medical Center, University of Amsterdam, Amsterdam, The Netherlands, [email protected]
PT
Jesper Kjaergaard, MD, DMSci, Department of Cardiology, Rigshospitalet and Department of Clinical Medicine, University of Copenhagen, Copenhagen, Denmark, [email protected]
CC E
Christian Hassager, MD, DMSci, Department of Cardiology, Rigshospitalet and Department of Clinical Medicine, University of Copenhagen, Copenhagen, Denmark, [email protected] Michael C. Jaeger Wanscher, MD, PhD, Department of Cardiothoracic Anaesthesia, Rigshospitalet and Department of Clinical Medicine, University of Copenhagen, Copenhagen, Denmark, [email protected]
A
Niklas Nielsen, MD, PhD, Lund University, Helsingborg Hospital, Department of Clinical Sciences Lund, Intensive and Perioperative Care, Lund, Sweden, [email protected] Erik Westhall, MD, PhD, Lund University, Skane University Hospital, Department of Clinical Sciences Lund, Clinical Neurophysiology, Lund, Sweden, [email protected]
Resuscitation Backman Page 2
Contents Title: 116 characters Abstract: 243 Main text: 1481 Tables and figures: 3, Supplementary tables and figures: 5 References: 20
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Corresponding author Sofia Backman, MD Department of Clinical Neurophysiology Skane University Hospital, S-221 85 Lund, Sweden Email: [email protected] Phone: +46 46 173380 Fax: +46 46 146528
Resuscitation Backman Page 3
Abstract
Introduction Routine EEG is widely used and accessible for post arrest neuroprognostication. Recent studies, using standardised EEG terminology, have proposed highly malignant EEG patterns with
IP T
promising predictive ability.
SC R
Objectives
To validate the performance of standardised routine EEG patterns to predict neurological
U
outcome after cardiac arrest.
N
Methods
A
In the prospective multicenter Target Temperature Management trial, comatose cardiac arrest
M
patients were randomised to different temperature levels (950 patients, 36 sites). According to
ED
the prospective protocol a routine EEG was performed in patients who remained comatose after the 36 hours temperature control intervention. EEGs were retrospectively reviewed blinded to
PT
outcome using the standardised American Clinical Neurophysiology Society terminology. Highly malignant, malignant and benign EEG patterns were correlated to poor and good
CC E
outcome, defined by best achieved Cerebral Performance Category up to 180 days.
A
Results
At 20 sites 207 patients had a routine EEG performed at median 76 hours after cardiac arrest. Highly malignant patterns (suppression or burst-suppression with or without discharges) had a high specificity for poor outcome (98%, CI 92-100), but with limited sensitivity (31%, CI 2439). Our false positive patient had a burst-suppression pattern during ongoing sedation. A
Resuscitation Backman Page 4
benign EEG, i.e. continuous normal-voltage background without malignant features, identified patients with good outcome with 77% (CI 66-86) sensitivity and 80% (CI 73-86) specificity.
Conclusion Highly malignant routine EEG after targeted temperature management is a strong predictor of
IP T
poor outcome. A benign EEG is an important indicator of a good outcome for patients
SC R
remaining in coma.
Keywords ACNS nomenclature
U
EEG
N
Cardiac arrest
A
CC E
PT
ED
Outcome prediction
M
Hypoxic-ischaemic encephalopathy
A
Coma
Resuscitation Backman Page 5
List of abbreviations CA=Cardiac Arrest CI=95% Confidence Intervals CPC=Cerebral Performance Category scale EEG=Electroencephalography
IP T
IQR=Interquartile Range NSE=Neuron-specific enolase
SC R
TTM=Target Temperature Management
A
CC E
PT
ED
M
A
N
U
WLST=Withdrawal of Life-Sustaining Therapy
Resuscitation Backman Page 6
Introduction Routine EEG is commonly used for neurological prognostication of comatose patients after cardiac arrest (CA) [1]. Due to temporal evolution of EEG patterns during the first days after a hypoxic-ischaemic event and possible interference of sedative agents, it is crucial to define robust EEG patterns and time-windows for prognostication. In the multicenter Target
IP T
Temperature Management trial (TTM-trial) comatose survivors were randomised to two target temperature levels. An analysis plan of the routine EEGs in the TTM-trial was published [2],
SC R
including studies on interrater agreement, prognostic ability and the present validation study. In the previous studies on eight selected TTM-sites, we found substantial interrater agreement
[3] and high prognostic ability [4] for prespecified highly malignant EEG patterns. The aim of
N
U
the present study was to validate these patterns in the remaining TTM-cohort.
A
Methods
M
Adult comatose survivors after out-of-hospital CA of a presumed cardiac cause were
ED
randomised to 33ºC versus 36ºC temperature management at 36 sites in Europe and Australia
PT
as previously described [5]. Ethical permissions were obtained in the participating countries.
Sedation was tapered after rewarming, approximately 36 hours after CA, if not indicated for
CC E
intensive care reasons. A routine EEG was included in the trial protocol 48-72 hours after CA, in comatose patients. A detailed protocol for the design of the present study was published [2].
A
Routines for neurological prognostication and practise of withdrawal of life-sustaining therapy (WLST) has been reported [6, 7].
Resuscitation Backman Page 7
Neurological function was evaluated using the Cerebral Performance Category scale (CPC), daily in the intensive care unit, at hospital discharge and at a face-to-face follow-up at six months. A best CPC of 1-2 at any time-point was considered a good outcome.
The patient´s first recorded EEG between 36 hours and 14 days after CA were included.
IP T
Recordings with less than ten minutes duration were excluded. The EEGs were retrospectively analysed by senior clinical neurophysiologists (SB or EW), blinded to clinical parameters and
SC R
outcome, using the standardised terminology according to the American Clinical
Neurophysiology Society [8]. EEGs were classified into three predefined categories [2]: highly malignant EEG, EEG with any malignant feature or benign EEG (Table S1). Since 73% of the
U
EEGs in the present study did not include notations of standardised reactivity testing, mainly
A
N
due to data format reasons, we excluded reactivity from the definitions.
M
We used SPSS Statistics 25.0 for Mann-Whitney U-test, Chi-square or Fisher´s exact test where
ED
applicable. Data is presented as median and interquartile range (IQR). 95% confidence intervals
Results
PT
for percentages were calculated using Wilson´s method.
CC E
All 36 sites performed routine EEGs. We previously published data from eight selected sites [4]. Additional eight sites were excluded due to pruned EEG recordings or lack of technical
A
possibility to export registrations (Fig. 1). EEGs recorded from 207 patients at the remaining 20 sites were included in the present study. Patients with EEG had longer duration from CA to return of spontaneous circulation, higher neuron-specific enolase (NSE) levels, longer hospital stay and worse outcome (Table 1). The median time from CA to EEG was 76 hours (IQR 62-
Resuscitation Backman Page 8
104, range 38-309 hours). 95% of the EEGs were recorded with at least 12 electrodes and 87% with at least 16 electrodes. The median duration of the recordings was 21 minutes (IQR 15-30).
Poor outcome was seen in 141 patients (68%), of whom 44 patients had a highly malignant EEG pattern. One patient with highly malignant EEG had a good outcome, resulting in 98%
IP T
specificity and 31% sensitivity to predict poor outcome (Table 2). The EEG of the false positive
patient showed burst-suppression without discharges or highly epileptiform bursts, at 75 hours
SC R
after CA, with ongoing sedation during the EEG registration (Suppl. Fig. S1). The prognostic performance of highly malignant patterns were similar when CPC at six months was used as end-point instead of best CPC (Suppl. Table S2). Using a more conservative definition of poor
U
outcome (best CPC 4-5), there were two more false positives with highly malignant EEG
N
patterns, one with suppressed background and continuous periodic discharges and one with
M
A
burst-suppression without discharges.
ED
The presence of malignant rhythmic or periodic features had 96% specificity and 43% sensitivity for poor outcome (Table 2). A malignant EEG background predicted poor outcome
PT
with 80% specificity and 68% sensitivity.
CC E
A benign EEG was present in 66 patients (32%) and identified patients with good outcome with
A
77% (CI 66-86) sensitivity and 80% (CI 73-86) specificity (Suppl Table S3).
EEG was performed early after the intervention in 44% of the patients (up to 72 hours after CA), with a 23% prevalence for highly malignant patterns and 64% for any malignant feature. We found no significant differences for prevalence, specificity or sensitivity for highly malignant or malignant patterns between the early and later group (Suppl Table S4).
Resuscitation Backman Page 9
During the EEG registration, 34% of the patients had ongoing sedation, 34% had antiepileptic treatment, 18% had ongoing myoclonus and one patient had tonic-clonic seizures. Among the sedated patients, 19% had highly a malignant pattern and 55% had at least one malignant feature. There were no significant differences between the sedated and the unsedated group
IP T
regarding prevalence or prognostic ability of highly malignant or malignant patterns to predict
SC R
poor outcome (Suppl Table S4).
Discussion
U
Guidelines suggest EEG for prognostication after CA, but knowledge gaps exist and there is a
N
need for large prospective studies using standardised definitions [9, 10]. In the present study
A
we show that standardised highly malignant as well as benign routine EEG patterns in comatose
ED
M
post arrest patients have strong prognostic implications after the 36 hours intervention period.
The results validates our previous results [4] in this separate and larger patient cohort of the
PT
TTM-trial. Highly malignant EEG patterns had high specificity (98%) for poor outcome with relatively narrow confidence intervals (92-100%). Two recent studies, using the same
CC E
standardised definitions for a highly malignant EEG, reported similar results for specificity
A
when EEGs were performed after rewarming [11, 12].
Burst-suppression in patients with good outcome has previously been reported early after CA [13-15], but burst-suppression with identical bursts is a predictor of poor prognosis [13, 16]. Identical bursts are transient and disappear around rewarming and were therefore not
Resuscitation Backman Page 10
systematically evaluated in the present study. Our false positive patient with burst-suppression had neither identical bursts nor epileptiform discharges.
The sensitivity for poor outcome of highly malignant patterns was low (31%) and range from 40-62% in recent studies [4, 11, 12]. Although these patterns have shown substantial interrater
IP T
agreement [3], the differences regarding sensitivity may reflect a more conservative
interpretation of the standardised patterns in the present study or be due to different patient
SC R
cohorts.
It is well known that an early continuous normal-voltage background identified by continuous
U
EEG-monitoring is a strong predictor of favorable prognosis [16, 19]. An EEG that is reactive
N
to external stimuli has previously been shown to be an important predictor [17] but with limited
A
interrater agreement [3, 18]. In this study, benign EEG was a strong predictor of a good
ED
M
outcome, although reactivity testing was not part of the analyses.
Our false positive patient had ongoing sedation during the EEG, which should be taken into
PT
consideration as a risk of reducing the reliability of EEG for the individual patient. The potiental risk for false positive predictions directly after rewarming, due to effects of residual sedation,
CC E
was however not confirmed in our comparison between EEGs that were performed early after rewarming with those performed beyond 72 hours after CA. Notably, 29% of the patients in the
A
late group still had ongoing sedation. Further, we could not find any significant differences for EEG as a predictor of poor outcome, comparing the unsedated and the sedated group, but larger prospective studies with systematical assessment of sedatives are needed.
Resuscitation Backman Page 11
A limitation of our study is the large proportion of patients excluded due to lack of EEG. In the TTM-trial, 48% of the included patients woke up and 15% died before the time-point of prognostication and the majority of these had no EEG performed [7]. The group of patients with EEG is therefore likely to have a worse brain injury, supported by the findings of higher NSElevels and longer time to return of circulation. The results of this study are nevertheless
IP T
important as they represent the comatose patients in clinical practise undergoing EEG for
SC R
prognostication.
We tried to limit the risk of the self-fulfilling prophecy by using a strict and predefined protocol for prognostication in the TTM-study [20]. The highly malignant EEG patterns were not
U
included in the criteria for predicting poor outcome in the trial. The EEG pattern that was
N
considered as part of the WLST decision was a treatment-refractory status epilepticus combined
M
A
with deep coma three days after rewarming [6], but the decision was never based on EEG alone.
ED
Conclusion
We confirm that strictly defined highly malignant EEG patterns on a routine-EEG performed
PT
after the temperature management intervention reliably predict poor neurological outcome. A
CC E
benign EEG is useful to identify patients with a better prognosis.
A
Conflicts of interest: None of the authors have potential conflicts of interest to be disclosed.
Acknowledgement We would like to acknowledge the site investigators and the clinical staff at the EEG laboratories for contributing the EEG data:
Resuscitation Backman Page 12
Liverpool Hospital, St George Hospital, North Shore Hospital, Australia; General University Hospital Prague, Czech Republic; Copenhagen University Hospital Rigshospitalet, Denmark; Santa Maria degli Angeli Hospital Pordenone, Cattinara Hospital Trieste, San Martino Hospital Genoa, Italy; Centre Hospitalier de Luxembourg, Luxembourg; Leeuwarden Medisch Centrum, Rijnstate Hospital Arnhem, Onze Lieuwe Vrouve Gasthuis, The Netherlands; Linköping
IP T
University Hospital, Norrköping Hospital, Sweden; Geneva University Hospital, Switzerland; University Hospital of Wales Cardiff, Royal Bournemouth Hospital, Royal Berkshire Hospital,
SC R
Guys and St Thomas London, St Georges Hospital London, United Kingdom.
Funding source: Supported by The Swedish Heart and Lung Association; the Skåne University
U
Hospital Foundations; the Gyllenstierna-Krapperup Foundation; the Segerfalk foundation; the
N
Swedish National Health System (ALF); the County Council of Skåne; the Swedish Society of
A
Medicine; the Koch Foundation, The Swedish Heart-Lung Foundation, Skane University
M
Hospital foundation AFA Insurance, The Swedish Research Council and Hans-Gabriel and
ED
Alice Trolle-Wachtmeister Foundation; all in Sweden. The Tryg Foundation; Denmark. EU
PT
programme Interreg IV A.
CC E
Financial support: The authors have no financial interests in the study.
A
Legends
Legend Table 1: Patient characteristics. ROSC, return of spontaneous circulation; ICU, intensive care unit; NSE, neuron-specific enolase maximum value at 48 and 72 hours after cardiac arrest; CPC, Cerebral Performance Category
Legend Table 2: Prediction of poor outcome defined as CPC 3-5 as best CPC at any assessed time-point during the first six months after cardiac arrest. Definitions of EEG patterns based on the American Society of Clinical Neurophysiology terminology.
Resuscitation Backman Page 13
A
CC E
PT
ED
M
A
N
U
SC R
IP T
TP, true positive; FP, false positive, TN, true negative; FN, false negative.
Resuscitation Backman Page 14
Table 1 Patient characteristics Patients with EEG 20 sites, n = 207 median (IQR) or n (%)
Patients without EEG 20 sites, n = 403 median (IQR) or n (%)
324 (80.4%) 64 (55-72)
109/401 (27.2%) 28/402 (7.0%)
2/206 (1.0%) 29/205 (14.1%)
9/402 (2.2%) 61/401 (15.2%)
23 (15-35) (n=402) 102 (65-172) (n=402)
A
Time from cardiac arrest - ROSC (min) 27 (18-40) (n=207) - ICU discharge 198 (132-288) (n=207) (hours) - Hospital discharge 460 (264-975) (n=111) (hours)
SC R
60/207 (29.0%) 14/205 (6.8%)
41.9 (18.1-123.3) (n = 167)
Best CPC
1 2 3 4 5
M
NSE (ng/mL)
PT
ED
51 (24.6%) 15 (7.2%) 22 (10.6%) 119 (57.5%) 0
0.001 <0.001
363 (207-539) (n= 298)
0.025
17.7 (12.4-32.0) (n = 259)
<0.001
236 (58.6%) 28 (6.9%) 27 (6.7%) 112 (27.8%) 0
<0.001
N
History of ischemic heart disease - transient ischemic attack or stroke - epilepsy - diabetes
209 (51.9%)
IP T
Temperature target 36ºC 94 (45.4%) (others: 33ºC) Male gender 175 (84.5%) Age (year) 63 (56-70)
p value
U
Variable
A
CC E
Legend Table 1: ROSC, return of spontaneous circulation; ICU, intensive care unit; NSE, neuron-specific enolase maximum value at 48 and 72 hours after cardiac arrest; CPC, Cerebral Performance Category.
Resuscitation Backman Page 15
Table 2 Prediction of poor outcome defined as best CPC 3-5 during the first six months after cardiac arrest Sensitivity (95% CI)
Specificity (95% CI)
TP n
FP n
TN n
FN n
45 (22)
31 (24-39)
98 (92-100)
44
1
65
97
20 (10)
14 (9-21)
100 (94-100)
20
0
66
121
4 (2) 21 (10)
3 (1-7) 14 (6-16)
100 (94-100) 98 (92-100)
4 20
0 1
63 (30)
43 (35-51)
96 (87-98)
60
3
63
81
54 (26)
38 (30-46)
98 (92-100)
53
1
65
88
9 (4)
5 (2-10)
97 (90-99)
7
2
64
134
6 (3)
4 (2-8)
98 (92-100)
5
1
65
136
109 (53)
68 (60-75)
80 (69-88)
96
13
53
45
88 (43) 76 (37)
57 (49-65) 48 (40-56)
89 (80-95) 88 (78-94)
81 68
7 8
59 58
60 73
6 (3-12)
98 (92-100)
9
1
65
132
At least one malignant 128 (62) feature
80 (73-86)
77 (66-86)
113
15
51
28
Both malignant background and rhythmic/ 44 (21) periodic features
30 (24-39)
98 (92-100)
43
1
65
98
10 (5)
CC E
PT
ED
66 65
SC R
U
N
Malignant background Discontinuous (>10% suppression) Low-voltage (<20µV) Reversed anterioposterior gradient
A
Malignant rhythmic or periodic features Periodic discharges (≥50%) Rhythmic spike-andwave (≥50%) Unequivocal seizures or status epilepticus
M
Highly malignant pattern Suppressed background without discharges Suppressed background with continuous periodic discharges Burst-suppression
IP T
Prevalence n (%)
137 121
A
Legend Table 2: Prediction of poor outcome defined as CPC 3-5 as best CPC at any assessed time-point during the first six months after cardiac arrest. 207 included patients. Definitions of EEG patterns based on the American Society of Clinical Neurophysiology terminology. TP, true positive; FP, false positive, TN, true negative; FN, false negative.
Resuscitation Backman Page 16
Fig. 1
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Flow chart of study inclusion
Resuscitation Backman Page 17
References
A
CC E
PT
ED
M
A
N
U
SC R
IP T
[1] Friberg H, Cronberg T, Dunser MW, Duranteau J, Horn J, Oddo M. Survey on current practices for neurological prognostication after cardiac arrest. Resuscitation 2015;90:158-62. [2] Westhall E, Rosen I, Rossetti AO, van Rootselaar AF, Kjaer TW, Horn J, et al. Electroencephalography (EEG) for neurological prognostication after cardiac arrest and targeted temperature management; rationale and study design. BMC Neurol 2014;14:159. [3] Westhall E, Rosen I, Rossetti AO, van Rootselaar AF, Wesenberg Kjaer T, Friberg H, et al. Interrater variability of EEG interpretation in comatose cardiac arrest patients. Clin Neurophysiol 2015;126:2397-404. [4] Westhall E, Rossetti AO, van Rootselaar AF, Wesenberg Kjaer T, Horn J, Ullen S, et al. Standardized EEG interpretation accurately predicts prognosis after cardiac arrest. Neurology 2016;86:1482-90. [5] Nielsen N, Wetterslev J, Cronberg T, Erlinge D, Gasche Y, Hassager C, et al. Targeted temperature management at 33 degrees C versus 36 degrees C after cardiac arrest. N Engl J Med 2013;369:2197-206. [6] Nielsen N, Wetterslev J, al-Subaie N, Andersson B, Bro-Jeppesen J, Bishop G, et al. Target Temperature Management after out-of-hospital cardiac arrest--a randomized, parallel-group, assessor-blinded clinical trial--rationale and design. Am Heart J 2012;163:541-8. [7] Dragancea I, Wise MP, Al-Subaie N, Cranshaw J, Friberg H, Glover G, et al. Protocoldriven neurological prognostication and withdrawal of life-sustaining therapy after cardiac arrest and targeted temperature management. Resuscitation 2017;117:50-7. [8] Hirsch LJ, LaRoche SM, Gaspard N, Gerard E, Svoronos A, Herman ST, et al. American Clinical Neurophysiology Society's Standardized Critical Care EEG Terminology: 2012 version. J Clin Neurophysiol 2013;30:1-27. [9] Callaway CW, Soar J, Aibiki M, Bottiger BW, Brooks SC, Deakin CD, et al. Part 4: Advanced Life Support: 2015 International Consensus on Cardiopulmonary Resuscitation and Emergency Cardiovascular Care Science With Treatment Recommendations. Circulation 2015;132:S84-145. [10] Sandroni C, Cariou A, Cavallaro F, Cronberg T, Friberg H, Hoedemaekers C, et al. Prognostication in comatose survivors of cardiac arrest: an advisory statement from the European Resuscitation Council and the European Society of Intensive Care Medicine. Intensive Care Med 2014;40:1816-31. [11] Beuchat I, Solari D, Novy J, Oddo M, Rossetti AO. Standardized EEG interpretation in patients after cardiac arrest: Correlation with other prognostic predictors. Resuscitation 2018;126:143-6. [12] Rossetti AO, Tovar Quiroga DF, Juan E, Novy J, White RD, Ben-Hamouda N, et al. Electroencephalography Predicts Poor and Good Outcomes After Cardiac Arrest: A TwoCenter Study. Crit Care Med 2017;45:e674-e82. [13] Tjepkema-Cloostermans MC, Hofmeijer J, Trof RJ, Blans MJ, Beishuizen A, van Putten MJ. Electroencephalogram predicts outcome in patients with postanoxic coma during mild therapeutic hypothermia. Crit Care Med 2015;43:159-67. [14] Crepeau AZ, Rabinstein AA, Fugate JE, Mandrekar J, Wijdicks EF, White RD, et al. Continuous EEG in therapeutic hypothermia after cardiac arrest: prognostic and clinical value. Neurology 2013;80:339-44. [15] Amorim E, Rittenberger JC, Baldwin ME, Callaway CW, Popescu A, Post Cardiac Arrest S. Malignant EEG patterns in cardiac arrest patients treated with targeted temperature management who survive to hospital discharge. Resuscitation 2015;90:127-32.
Resuscitation Backman Page 18
A
CC E
PT
ED
M
A
N
U
SC R
IP T
[16] Sivaraju A, Gilmore EJ, Wira CR, Stevens A, Rampal N, Moeller JJ, et al. Prognostication of post-cardiac arrest coma: early clinical and electroencephalographic predictors of outcome. Intensive Care Med 2015;41:1264-72. [17] Rossetti AO, Carrera E, Oddo M. Early EEG correlates of neuronal injury after brain anoxia. Neurology 2012;78:796-802. [18] Hermans MC, Westover MB, van Putten M, Hirsch LJ, Gaspard N. Quantification of EEG reactivity in comatose patients. Clin Neurophysiol 2016;127:571-80. [19] Hofmeijer J, Beernink TM, Bosch FH, Beishuizen A, Tjepkema-Cloostermans MC, van Putten MJ. Early EEG contributes to multimodal outcome prediction of postanoxic coma. Neurology 2015;85:137-43. [20] Cronberg T, Horn J, Kuiper MA, Friberg H, Nielsen N. A structured approach to neurologic prognostication in clinical cardiac arrest trials. Scand J Trauma Resusc Emerg Med 2013;21:45.
Resuscitation Backman Page 19
Fig. 1
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Flow chart of study inclusion
Resuscitation Backman Page 20
Table 1 Patient characteristics
175 (84.5%) 63 (56-70)
324 (80.4%) 64 (55-72)
60/207 (29.0%) 14/205 (6.8%)
109/401 (27.2%) 28/402 (7.0%)
2/206 (1.0%) 29/205 (14.1%)
9/402 (2.2%) 61/401 (15.2%)
27 (18-40) (n=207) 198 (132-288) (n=207)
23 (15-35) (n=402) 102 (65-172) (n=402)
460 (264-975) (n=111)
41.9 (18.1-123.3) (n = 167)
Best CPC
1 2 3 4 5
M
NSE (ng/mL)
PT
ED
51 (24.6%) 15 (7.2%) 22 (10.6%) 119 (57.5%) 0
IP T
209 (51.9%)
SC R
Time from cardiac arrest - ROSC (min) - ICU discharge (hours) - Hospital discharge (hours)
94 (45.4%)
p value
U
History of ischemic heart disease - transient ischemic attack or stroke - epilepsy - diabetes
Patients without EEG 20 sites, n = 403 median (IQR) or n (%)
0.001 <0.001
363 (207-539) (n= 298)
0.025
17.7 (12.4-32.0) (n = 259)
<0.001
236 (58.6%) 28 (6.9%) 27 (6.7%) 112 (27.8%) 0
<0.001
N
Temperature target 36ºC (others: 33ºC) Male gender Age (year)
Patients with EEG 20 sites, n = 207 median (IQR) or n (%)
A
Variable
A
CC E
Legend Table 1: ROSC, return of spontaneous circulation; ICU, intensive care unit; NSE, neuron-specific enolase maximum value at 48 and 72 hours after cardiac arrest; CPC, Cerebral Performance Category.
Resuscitation Backman Page 21
Table 2 Prediction of poor outcome defined as best CPC 3-5 during the first six months after cardiac arrest
At least one malignant feature
TN n
98 (92-100)
44
1
65
97
20 (10)
14 (9-21)
100 (94-100)
20
0
66
121
4 (2) 21 (10)
3 (1-7) 14 (6-16)
100 (94-100) 98 (92-100)
4 20
0 1
63 (30)
43 (35-51)
96 (87-98)
60
3
63
81
54 (26)
38 (30-46)
98 (92-100)
53
1
65
88
9 (4)
5 (2-10)
97 (90-99)
7
2
64
134
6 (3)
4 (2-8)
98 (92-100)
5
1
65
136
109 (53)
68 (60-75)
80 (69-88)
96
13
53
45
88 (43) 76 (37)
57 (49-65) 48 (40-56)
89 (80-95) 88 (78-94)
81 68
7 8
59 58
60 73
6 (3-12)
98 (92-100)
9
1
65
132
80 (73-86)
77 (66-86)
113
15
51
28
30 (24-39)
98 (92-100)
43
1
65
98
10 (5)
128 (62)
CC E
44 (21)
66 65
N
U
SC R
31 (24-39)
FN n
IP T
FP n
PT
Both malignant background and rhythmic/ periodic features
45 (22)
Specificity (95% CI)
A
Malignant background Discontinuous (>10% suppression) Low-voltage (<20µV) Reversed anterioposterior gradient
TP n
M
Malignant rhythmic or periodic features Periodic discharges (≥50%) Rhythmic spike-andwave (≥50%) Unequivocal seizures or status epilepticus
Sensitivity (95% CI)
ED
Highly malignant pattern Suppressed background without discharges Suppressed background with continuous periodic discharges Burst-suppression
Prevalence n (%)
137 121
A
Legend Table 2: Prediction of poor outcome defined as CPC 3-5 as best CPC at any assessed time-point during the first six months after cardiac arrest. 207 included patients. Definitions of EEG patterns based on the American Society of Clinical Neurophysiology terminology. TP, true positive; FP, false positive, TN, true negative; FN, false negative.