G Model
ARTICLE IN PRESS
RESUS 5976 1–7
Resuscitation xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Resuscitation journal homepage: www.elsevier.com/locate/resuscitation
Clinical Paper
1
Feasibility of bispectral index monitoring to guide early post-resuscitation cardiac arrest triage夽,夽夽
2
3
4
Q1
5 6 7
Q2 Q3
8 9
David B. Seder a,b,∗ , John Dziodzio a , Kahsi A. Smith a , Paige Hickey c , Brittany Bolduc d , Philip Stone d , Teresa May a , Barbara McCrum a , Gilles L. Fraser a , Richard R. Riker a,b a
Maine Medical Center, Department of Critical Care Services, Portland, ME, United States Maine Medical Center, Neuroscience Institute, Portland, ME, United States c Furman University, Greenville, SC, United States d University of New England, Biddeford, ME, United States b
10
11 29
a r t i c l e
i n f o
a b s t r a c t
12 13 14 15 16
Article history: Received 13 December 2013 Received in revised form 13 March 2014 Accepted 14 April 2014
17
28
Keywords: Cardiac Arrest EEG Hypothermia Triage Severity BIS Bispectral index Resuscitation CPR
30
1. Introduction
18 19 Q4 20 21 22 23 24 25 26 27
31 32 33
Introduction: Triage after resuscitation from cardiac arrest is hindered by reliable early estimation of brain injury. We evaluated the performance of a triage model based on early bispectral index (BIS) findings and cardiac risk classes. Methods: Retrospective evaluation of serial patients resuscitated from cardiac arrest, unable to follow commands, and undergoing hypothermia. Patients were assigned to a cardiac risk group: STEMI, VT/VF shock, VT/VF no shock, or PEA/asystole, and to a neurological dysfunction group, based on the BIS score following first neuromuscular blockade (BISi), and classified as BISi > 20, BISi 10–20, or BISi < 10. Cause of death was described as neurological or circulatory. Results: BISi in 171 patients was measured at 267(±177) min after resuscitation and 35(±1.7)◦ C. BISi < 10 suffered 82% neurological-cause and 91% overall mortality, BISi 10–20 35% neurological and 55% overall mortality, and BISi > 20 12% neurological and 36% overall mortality. 33 patients presented with STEMI, 15 VT/VF-shock, 41 VT/VF-no shock, and 80 PEA/asystole. Among BISi > 20 patients, 75% with STEMI underwent urgent cardiac catheterization (cath) and 94% had good outcome. When BISi > 20 with VT/VF and shock, urgent cath was infrequent (33%), and 4 deaths (44%) were uniformly of circulatory etiology. Of 56 VT/VF patients without STEMI, 24 were BISi > 20 but did not undergo urgent cath – 5(20.8%) of these had circulatory-etiology death. Circulatory-etiology death also occurred in 26.5% BIS > 20 patients with PEA/asystole. When BISi < 10, a neurological etiology death dominated independent of cardiac risk group. Conclusions: Neurocardiac triage based on very early processed EEG (BIS) is feasible, and may identify patients appropriate for individualized post-resuscitation care. This and other triage models warrant further study. © 2014 Elsevier Ireland Ltd. All rights reserved.
Outcomes of patients surviving in- and out-of-hospital cardiac arrest have improved with routine application of therapeutic temperature management and evidence-based post-resuscitation care
夽 A Spanish translated version of the summary of this article appears as Appendix in the final online version at http://dx.doi.org/10.1016/j.resuscitation.2014.04.016. 夽夽 These data were presented in part at the 2013 American Heart Association Resuscitation Science Symposium. ∗ Corresponding author at: Neurocritical Care, Maine Medical Center, 22 Bramhall Street, Portland, ME 04102, United States. E-mail addresses: [email protected], [email protected] (D.B. Seder).
guidelines.1,2 These guidelines offer several concrete recommendations, including therapeutic hypothermia to 32–34 ◦ C for 12–24 h in survivors of VT/VF arrest, and urgent coronary angiography and PCI as appropriate for patients with STEMI at the time of presentation. Most of the other care is discretionary, however, and there is evidence of great variability in practices.3 One reason for this variability is the wide range of injuries suffered by hospitalized cardiac arrest survivors related to the severity and type of neurological and cardiac injuries at the time of hospital presentation, when any combination of mild, moderate, and severe brain injury, seizures, STEMI, acute coronary syndrome, pulmonary embolism, metabolic acidosis, shock, and cardiomyopathy may coexist. Despite this variation in patient condition and context, triage schemes remain rudimentary. For example,
http://dx.doi.org/10.1016/j.resuscitation.2014.04.016 0300-9572/© 2014 Elsevier Ireland Ltd. All rights reserved.
Please cite this article in press as: Seder DB, et al. Feasibility of bispectral index monitoring to guide early post-resuscitation cardiac arrest triage. Resuscitation (2014), http://dx.doi.org/10.1016/j.resuscitation.2014.04.016
34 35 36 37 38 39 40 41 42 43 44 45 46 47
G Model RESUS 5976 1–7
D.B. Seder et al. / Resuscitation xxx (2014) xxx–xxx
2 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71
72
73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110
ARTICLE IN PRESS
2010 AHA Guidelines suggest urgent coronary angiography for all OHCA survivors with STEMI, even those with severe brain injury.1 Urgent coronary angiography for OHCA survivors with VT/VF arrest remains controversial despite evidence of benefit, and many investigators and practitioners have demanded a more sophisticated assessment paradigm.4–9 Asphyxial rat models of brain injury following cardiac arrest suggest that processed EEG performed as early as 60 min after resuscitation predict outcome with a high degree of sensitivity and specificity.10–14 In humans, we have previously shown that simple and commercially available processed EEG (the bispectral index) under a standardized sedation and neuromuscular blockade algorithm accurately predicts outcome at a mean of 4.5 h after resuscitation.15,16 Believing that post-resuscitation CA care should be individualized to consider the relative risk/benefit profile of various treatment strategies for individual patients, we developed a possible risk stratification model, characterizing CA survivors based on commonly accepted cardiac risk groups, and an assessment of the severity of brain injury by processed EEG.15 We then populated the model with a large, retrospective cohort, to see if it reliably classified patients according to their ultimate risk of neurological or cardiac etiology death, and might therefore be used to triage patients prospectively to a treatment regimen tailored to their injuries.
2. Materials and methods This was a retrospective study involving serial patients admitted to the intensive care unit of a single United States center after in-hospital or out-of-hospital cardiopulmonary arrest between September, 2008 and May, 2012. Patients were included if over 18 years of age and a complete data set was available. Patients that died on hospital days 1–2 were excluded due to the difficulty of determining the cause of death in this subgroup, and the high frequency of psychosocial issues, such as the role of advanced age and DNR status in determining outcome.17 Deidentified demographic, clinical, treatment, and outcomes data are collected prospectively into an IRB-supported database built around the International Cardiac Arrest Registry (INTCAR) platform. Processed EEG data are routinely uploaded from the bedside bispectral index monitors (Covidien Medical, Boston, MA, models A2000 and VISTA). All patients are treated under a standardized post-resuscitation therapeutic hypothermia protocol, which has been previously described18,19 and involves rapid surface cooling to 33 ◦ C using a commercially available servo-regulated surface cooling device, 20 mcg/kg/min infusion of propofol, 25 mcg/h infusion of fentanyl, and intermittent bolus-dose vecuronium to initiate the cooling process and then in response to visible shivering.20,21 All patients are monitored with the bispectral index continuously during the hypothermia protocol to verify the adequacy of sedation,22 and with continuous EEG to detect seizure activity. STEMI was defined as ST elevations >1 mm in 2 or more contiguous leads. Shock was defined as requiring vasopressor support to maintain systolic blood pressure >90 mmHg during the first hour of hospitalization. Left ventricular function was based on the first measurement, typically within 12 h of presentation. Clinicians described the cause of death: When patients became brain dead or died after withdrawal of life support due to perceived neurological futility, the etiology of death was described as neurological etiology. If the patient died of shock, re-arrest, refractory arrhythmia, or progressive lactic acidosis and multiorgan system failure, the cause of death was described as circulatory. All inpatient deaths met these criteria. At our center, when patients do not awaken in the first days after rewarming, board certified neurologists are consulted to remark
on prognosis. This prognosis is typically delivered based on serial examinations off analgesia and sedation, continuous EEG findings, MRI typically performed between days 3 and 5 post-arrest, and serum neuron-specific enolase levels obtained at 24, 48, and 72 h. Although, in light of recent criticisms of the 2006 AAN Practice Parameters23,24 none of these modalities is considered infallible, we believe that in concert they frequently paint a picture of mild, moderate, or grave brain injury, with a corresponding high-tolow likelihood of functional recovery. Occasionally, somatosensory evoked potentials are then obtained, for confirmation of severe injury. Families use this information, in conference with attending neurology and critical care specialists, to determine whether discontinuation of life support measures is advisable, or prolonged supportive care is appropriate. Such conferences are tailored to individual family needs, and a dogmatic approach to interpreting prognostic data is discouraged. The timing of these meetings is not uniform, but delay ≥72 h after rewarming is recommended. The BIS monitor is removed after 48 h of temperature management, and although not blinded early on, is not available to consulting neurologists and is not part of prognostication. Indeed, the authors have strongly discouraged this experimental modality be used to influence bedside discussions of prognosis in out unit, and its primary clinical utility is to titrate sedation during neuromuscular blockade.19 Post-anoxic seizures and frequent periodic discharges are routinely treated with antiepileptic and suppressive agents, but they are also thought to reflect a more significant injury – especially if arising from a suppressed or burst-suppressed background,24,25 and may therefore influence prognostication. The BISi is defined as the mean bispectral index plateau value recorded during 10 min following the administration of the first dose of neuromuscular blockade.15 Neuromuscular blockade is necessary to eliminate muscle artifact that confounds the BIS reading, and the measurement is typically performed after arrival in the ICU, at which point cooling pads are placed, the BIS monitor applied, sedation and analgesia infused, and a bolus dose of 0.1 mg/kg vecuronium administered to augment the cooling process. The practice standard in our institution includes urgent cardiac catheterization for most CA survivors found to have ST-elevation myocardial infarction, but the most appropriate therapy is ultimately determined by the attending interventional cardiologist, and varies somewhat. Other cardiac arrest survivors, including those with a presumed cardiac cause of the arrest, rarely undergo urgent coronary angiography and PCI (defined as occurring prior to neurological assessment or awakening), but among patients who demonstrate neurological recovery, most undergo angiography on a delayed basis. We believe this approximates standard care in our region, and most areas of the United States, though some hospitals have adopted a more aggressive approach.26,27 Based on commonly employed cardiac risk strata and previously observed associations between processed EEG findings and neurological outcomes,15 a simple grid for risk stratification was developed (see Table 2 below). Patients in the database were classified according to the grid, and within each category the following treatment and outcomes considered: survival, death due to neurological etiology, death due to circulatory failure, early coronary angiography, and early percutaneous coronary intervention.
3. Results One hundred and seventy one patients met inclusion criteria. Demographics, number of comorbid medical conditions, characteristics of the arrest, presence of factors such as STEMI, cardiomyopathy, and shock, severity of neurological dysfunction as described
Please cite this article in press as: Seder DB, et al. Feasibility of bispectral index monitoring to guide early post-resuscitation cardiac arrest triage. Resuscitation (2014), http://dx.doi.org/10.1016/j.resuscitation.2014.04.016
111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167
168
169 170 171 172
G Model
ARTICLE IN PRESS
RESUS 5976 1–7
D.B. Seder et al. / Resuscitation xxx (2014) xxx–xxx
3
Table 1
Q6 Demographics, clinical factors, and differences in cause of death. P value contrasts patients with circulatory to cerebral death. Demographics
All patients (n = 171)
Circulatory death
Cerebral death
P
Age, years Gender female Number of comorb Witnessed Bystander CPR Rhythm VT/VF TIT (min) STEMI Shock BISi < 10 10 < BISi < 20 BISi > 20 LVEF < 30% LVEF < 50% EEG seizures Myoclonic activity
60.1 ± 16.0 51/171 (29.8%) 2.2 ± 1.5 136/167 (81.4%) 76/167 (45.5%) 82/169 (48.5%) 23.7 ± 15.6 33/171 (19.3%) 48/170 (28.2%) 66/171 (38.6%) 19/171 (11.1%) 86/171 (50.3%) 31/120 (25.8%) 75/120 (62.5%) 10/171 (5.8%) 40/171 (23.4%)
64.4 ± 16.5 14/27 (51.9%) 3.0 ± 1.6 25/27 (92.6%) 8/27 (29.6%) 12/26 (46.2%) 21.1 ± 14.1 2/27 (7.4%) 9/27 (33.3%) 6/27 (22.2%) 3/27 (11.1%) 18/27 (66.6%) 10/20 (50.0%) 18/20 (90.0%) 0/27 (0%) 7/27 (25.9%)
58.9 ± 17.8 23/73 (31.5%) 2.3 ± 1.6 45/70 (64.3%) 30/70 (42.9%) 20/72 (27.8%) 27.7 ± 16.1 8/73 (11.0%) 14/72 (19.4%) 54/73 (74.0%) 6/73 (8.2%) 13/73 (17.8%) 6/45 (13.3%) 19/45 (42.2%) 8/73 (11.0%) 26/73 (35.6%)
0.17 0.064 0.049 0.006 0.2 0.09 0.09 0.6 0.15 20 Class I # Patients/survival
Moderate brain injury BISi 10–20 Class II # Patients/survival
Severe brain injury BISi < 10 Class III # Patients/survival
Overall STEMI VT-VF with shock VT/VF without shock PEA or Asystole
81/68% 16/94% 9/56% 22/82% 34/50%
22/45% 8/75% 1/100% 3/67% 10/10%
66/9% 9/22% 5/0% 16/25% 36/0%
Please cite this article in press as: Seder DB, et al. Feasibility of bispectral index monitoring to guide early post-resuscitation cardiac arrest triage. Resuscitation (2014), http://dx.doi.org/10.1016/j.resuscitation.2014.04.016
216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235
G Model RESUS 5976 1–7 4
ARTICLE IN PRESS D.B. Seder et al. / Resuscitation xxx (2014) xxx–xxx
Fig. 1. (a) Absolute mortality and risk of circulatory-etiology etiology death by cardiac risk-class stratification. (b) Absolute mortality and risk of neurological-etiology etiology death by BISi stratification. (c) Relative risk of cerebral vs. circulatory etiology death by BISi stratification.
236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262
will create the framework for individualized post-resuscitation cardiac arrest care, with the potential to save lives, effectively allocate scarce resources, and improve the cost-effectiveness of care. Because of its apparent simplicity, widespread use, proprietary algorithm, and variable performance in anesthesia sedation monitoring trials,28–30 the Bispectral Index is a controversial processed EEG modality. Unilateral surface EEG is obtained from a frontotemporal adhesive sensor and undergoes algorithmic analysis – a weighted sum of electroencephalographic parameters including the frequency below which 95% of the power spectrum resides, the relative beta ratio, synchrony of fast and slow waves, and the burst-suppression ratio.31 The mathematical transformation of these signals yields a number from 0 to 100, which is purported to describe the level of awareness. BIS is also simple to apply and interpret, offering a gross estimation of brain electrical activity without interobserver variability, which could be used for triage purposes by non-neurologists. Like conventional EEG, the use of BIS monitoring to assess the severity of brain injury is potentially confounded by medications, temperature, tissue perfusion, and metabolic issues. Unlike conventional EEG, the BIS assesses only a frontotemporal location, so that patients with a heterogeneous brain injury might be misclassified. Despite these limitations, the accuracy of our early BIS based prognostication technique was previously shown in a similar cohort of patients,15 and very low BIS scores in an uncontrolled setting after cardiac arrest have been shown by other groups to correlate closely with poor outcome.32–35 These findings should not
be a surprise – reports of the correlation between EEG background activity and severity of brain injury after cardiac arrest date back to the 1960s.36 In 1965, Hockaday proposed an EEG based grading system based on evaluation of their own cases,37 and a subsequent review of multiple case series revealed no survivors among 408 patients meeting Hockaday criteria for Category IV or V injury – burst suppression or electrocerebral silence (ECS).38 More recently, several groups have characterized the relationship between EEG background and neurological outcome in patients undergoing therapeutic hypothemia,39–42 though none has carefully described the timing of this assessment, or suggested specifically how the information might be used to alter the course of therapy. Very early processed EEG is shown in multiple animal models to correlate with severity of brain injury after cardiac arrest, and based on these experiments, we suggest that it may be possible to move back the timing of brain injury severity assessment much closer to the time of resuscitation without losing accuracy.10–14 Clinicians considering the prospective application of this technique should be mindful of the unknown effects of sedation, narcotic analgesics, and temperature on our assessment model, and confirmation of severe brain injury with other (non-EEG based) modalities prior to consideration of de-escalating or withdrawing life support should be considered mandatory. Our patients were treated under a standard protocol involving lower doses of propofol and fentanyl than some investigators have described,43 and we do not routinely use benzodiazepines, barbiturates, or high-dose propofol. Use of such medications, especially at higher doses prior
Please cite this article in press as: Seder DB, et al. Feasibility of bispectral index monitoring to guide early post-resuscitation cardiac arrest triage. Resuscitation (2014), http://dx.doi.org/10.1016/j.resuscitation.2014.04.016
263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289
G Model
ARTICLE IN PRESS
RESUS 5976 1–7
D.B. Seder et al. / Resuscitation xxx (2014) xxx–xxx
5
Table 3
Q7 Neurocardiac triage grid: serial patients retrospectively assigned to neurological and circulatory risk category based on the initial BIS score after neuromuscular blockade, and standard cardiac risk categories. PCI = percutaneous coronary angiography. Neurological GRADE
Cause of death A
STEMI
Angiography PCI
Cause of death B
VT/VF shock
Angiography PCI
Cardiac Class Cause of death C
VT/VF no shock
Angiography PCI
Cause of death D
PEA/asystole
Angiography PCI
290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324
to neurological assessment, or overdose of such agents actually precipitating the arrest, might confound the assessment of neurological dysfunction, overestimating its association with severe brain injury. Conversely, if in the wake of recent work by Nielsen and colleagues,2 clinicians begin to target a core body temperature of 36 ◦ C after cardiac arrest, BISi scores may increase due to warmer temperature,44,45 and established cutoffs could prove unreliable at identifying mild brain injury. The effect of seizures on BIS scores is poorly characterized, and seizures at the time of the BISi assessment must be excluded if treatment decisions are to be based on the number. Muscle artifact due to shivering is the most common confounder of BIS, and prognostic measurements should not be made unless full neuromuscular blockade is employed – ideally verified by train-of-four or similar means. Finally, we suggest that BISi measurements be made 4–8 h after ROSC, since we have not evaluated the accuracy of earlier or later BISi measurements, and it is not known how the evolution of brain electrical activity affects the relationship between EEG background and outcome. Ultra-early estimation of the severity of brain injury after cardiac arrest may identify patients that benefit from intensified or individualized therapy in prospective trials. Although it was recently demonstrated that high-quality post-resuscitation care administered at 33 ◦ C was not superior to 36 ◦ C in an unselected group of unresponsive cardiac arrest survivors,2 hypothermia is shown in many laboratory models to be powerfully neuroprotective. It is reasonable to consider that intensified (longer or deeper hypothermia, blood pressure augmentation, or the addition of neuroprotective agents) therapy applied to a severely brain injured group of patients might improve the dismal outcomes seen in our cohort (9% survival when BISi < 10). Other prognostication schemes have been proposed,46–48 but do not clearly distinguish the source of risk as neurological or cardiac, and therefore cannot effectively guide therapeutic decisions. Many patients in this series (20/169) died of circulatory collapse despite BISi > 10 (mild or moderate brain injury), and we believe
Count Cerebral Circulatory Early Delayed Early Delayed Count Cerebral Circulatory Early Delayed Early Delayed Count Cerebral Circulatory Early Delayed Early Delayed Count Cerebral Circulatory Early Delayed Early Delayed
I BISi>20
II 10
ARTICLE IN PRESS
RESUS 5976 1–7
Resuscitation xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Resuscitation journal homepage: www.elsevier.com/locate/resuscitation
Clinical Paper
1
Feasibility of bispectral index monitoring to guide early post-resuscitation cardiac arrest triage夽,夽夽
2
3
4
Q1
5 6 7
Q2 Q3
8 9
David B. Seder a,b,∗ , John Dziodzio a , Kahsi A. Smith a , Paige Hickey c , Brittany Bolduc d , Philip Stone d , Teresa May a , Barbara McCrum a , Gilles L. Fraser a , Richard R. Riker a,b a
Maine Medical Center, Department of Critical Care Services, Portland, ME, United States Maine Medical Center, Neuroscience Institute, Portland, ME, United States c Furman University, Greenville, SC, United States d University of New England, Biddeford, ME, United States b
10
11 29
a r t i c l e
i n f o
a b s t r a c t
12 13 14 15 16
Article history: Received 13 December 2013 Received in revised form 13 March 2014 Accepted 14 April 2014
17
28
Keywords: Cardiac Arrest EEG Hypothermia Triage Severity BIS Bispectral index Resuscitation CPR
30
1. Introduction
18 19 Q4 20 21 22 23 24 25 26 27
31 32 33
Introduction: Triage after resuscitation from cardiac arrest is hindered by reliable early estimation of brain injury. We evaluated the performance of a triage model based on early bispectral index (BIS) findings and cardiac risk classes. Methods: Retrospective evaluation of serial patients resuscitated from cardiac arrest, unable to follow commands, and undergoing hypothermia. Patients were assigned to a cardiac risk group: STEMI, VT/VF shock, VT/VF no shock, or PEA/asystole, and to a neurological dysfunction group, based on the BIS score following first neuromuscular blockade (BISi), and classified as BISi > 20, BISi 10–20, or BISi < 10. Cause of death was described as neurological or circulatory. Results: BISi in 171 patients was measured at 267(±177) min after resuscitation and 35(±1.7)◦ C. BISi < 10 suffered 82% neurological-cause and 91% overall mortality, BISi 10–20 35% neurological and 55% overall mortality, and BISi > 20 12% neurological and 36% overall mortality. 33 patients presented with STEMI, 15 VT/VF-shock, 41 VT/VF-no shock, and 80 PEA/asystole. Among BISi > 20 patients, 75% with STEMI underwent urgent cardiac catheterization (cath) and 94% had good outcome. When BISi > 20 with VT/VF and shock, urgent cath was infrequent (33%), and 4 deaths (44%) were uniformly of circulatory etiology. Of 56 VT/VF patients without STEMI, 24 were BISi > 20 but did not undergo urgent cath – 5(20.8%) of these had circulatory-etiology death. Circulatory-etiology death also occurred in 26.5% BIS > 20 patients with PEA/asystole. When BISi < 10, a neurological etiology death dominated independent of cardiac risk group. Conclusions: Neurocardiac triage based on very early processed EEG (BIS) is feasible, and may identify patients appropriate for individualized post-resuscitation care. This and other triage models warrant further study. © 2014 Elsevier Ireland Ltd. All rights reserved.
Outcomes of patients surviving in- and out-of-hospital cardiac arrest have improved with routine application of therapeutic temperature management and evidence-based post-resuscitation care
夽 A Spanish translated version of the summary of this article appears as Appendix in the final online version at http://dx.doi.org/10.1016/j.resuscitation.2014.04.016. 夽夽 These data were presented in part at the 2013 American Heart Association Resuscitation Science Symposium. ∗ Corresponding author at: Neurocritical Care, Maine Medical Center, 22 Bramhall Street, Portland, ME 04102, United States. E-mail addresses: [email protected], [email protected] (D.B. Seder).
guidelines.1,2 These guidelines offer several concrete recommendations, including therapeutic hypothermia to 32–34 ◦ C for 12–24 h in survivors of VT/VF arrest, and urgent coronary angiography and PCI as appropriate for patients with STEMI at the time of presentation. Most of the other care is discretionary, however, and there is evidence of great variability in practices.3 One reason for this variability is the wide range of injuries suffered by hospitalized cardiac arrest survivors related to the severity and type of neurological and cardiac injuries at the time of hospital presentation, when any combination of mild, moderate, and severe brain injury, seizures, STEMI, acute coronary syndrome, pulmonary embolism, metabolic acidosis, shock, and cardiomyopathy may coexist. Despite this variation in patient condition and context, triage schemes remain rudimentary. For example,
http://dx.doi.org/10.1016/j.resuscitation.2014.04.016 0300-9572/© 2014 Elsevier Ireland Ltd. All rights reserved.
Please cite this article in press as: Seder DB, et al. Feasibility of bispectral index monitoring to guide early post-resuscitation cardiac arrest triage. Resuscitation (2014), http://dx.doi.org/10.1016/j.resuscitation.2014.04.016
34 35 36 37 38 39 40 41 42 43 44 45 46 47
G Model RESUS 5976 1–7
D.B. Seder et al. / Resuscitation xxx (2014) xxx–xxx
2 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71
72
73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110
ARTICLE IN PRESS
2010 AHA Guidelines suggest urgent coronary angiography for all OHCA survivors with STEMI, even those with severe brain injury.1 Urgent coronary angiography for OHCA survivors with VT/VF arrest remains controversial despite evidence of benefit, and many investigators and practitioners have demanded a more sophisticated assessment paradigm.4–9 Asphyxial rat models of brain injury following cardiac arrest suggest that processed EEG performed as early as 60 min after resuscitation predict outcome with a high degree of sensitivity and specificity.10–14 In humans, we have previously shown that simple and commercially available processed EEG (the bispectral index) under a standardized sedation and neuromuscular blockade algorithm accurately predicts outcome at a mean of 4.5 h after resuscitation.15,16 Believing that post-resuscitation CA care should be individualized to consider the relative risk/benefit profile of various treatment strategies for individual patients, we developed a possible risk stratification model, characterizing CA survivors based on commonly accepted cardiac risk groups, and an assessment of the severity of brain injury by processed EEG.15 We then populated the model with a large, retrospective cohort, to see if it reliably classified patients according to their ultimate risk of neurological or cardiac etiology death, and might therefore be used to triage patients prospectively to a treatment regimen tailored to their injuries.
2. Materials and methods This was a retrospective study involving serial patients admitted to the intensive care unit of a single United States center after in-hospital or out-of-hospital cardiopulmonary arrest between September, 2008 and May, 2012. Patients were included if over 18 years of age and a complete data set was available. Patients that died on hospital days 1–2 were excluded due to the difficulty of determining the cause of death in this subgroup, and the high frequency of psychosocial issues, such as the role of advanced age and DNR status in determining outcome.17 Deidentified demographic, clinical, treatment, and outcomes data are collected prospectively into an IRB-supported database built around the International Cardiac Arrest Registry (INTCAR) platform. Processed EEG data are routinely uploaded from the bedside bispectral index monitors (Covidien Medical, Boston, MA, models A2000 and VISTA). All patients are treated under a standardized post-resuscitation therapeutic hypothermia protocol, which has been previously described18,19 and involves rapid surface cooling to 33 ◦ C using a commercially available servo-regulated surface cooling device, 20 mcg/kg/min infusion of propofol, 25 mcg/h infusion of fentanyl, and intermittent bolus-dose vecuronium to initiate the cooling process and then in response to visible shivering.20,21 All patients are monitored with the bispectral index continuously during the hypothermia protocol to verify the adequacy of sedation,22 and with continuous EEG to detect seizure activity. STEMI was defined as ST elevations >1 mm in 2 or more contiguous leads. Shock was defined as requiring vasopressor support to maintain systolic blood pressure >90 mmHg during the first hour of hospitalization. Left ventricular function was based on the first measurement, typically within 12 h of presentation. Clinicians described the cause of death: When patients became brain dead or died after withdrawal of life support due to perceived neurological futility, the etiology of death was described as neurological etiology. If the patient died of shock, re-arrest, refractory arrhythmia, or progressive lactic acidosis and multiorgan system failure, the cause of death was described as circulatory. All inpatient deaths met these criteria. At our center, when patients do not awaken in the first days after rewarming, board certified neurologists are consulted to remark
on prognosis. This prognosis is typically delivered based on serial examinations off analgesia and sedation, continuous EEG findings, MRI typically performed between days 3 and 5 post-arrest, and serum neuron-specific enolase levels obtained at 24, 48, and 72 h. Although, in light of recent criticisms of the 2006 AAN Practice Parameters23,24 none of these modalities is considered infallible, we believe that in concert they frequently paint a picture of mild, moderate, or grave brain injury, with a corresponding high-tolow likelihood of functional recovery. Occasionally, somatosensory evoked potentials are then obtained, for confirmation of severe injury. Families use this information, in conference with attending neurology and critical care specialists, to determine whether discontinuation of life support measures is advisable, or prolonged supportive care is appropriate. Such conferences are tailored to individual family needs, and a dogmatic approach to interpreting prognostic data is discouraged. The timing of these meetings is not uniform, but delay ≥72 h after rewarming is recommended. The BIS monitor is removed after 48 h of temperature management, and although not blinded early on, is not available to consulting neurologists and is not part of prognostication. Indeed, the authors have strongly discouraged this experimental modality be used to influence bedside discussions of prognosis in out unit, and its primary clinical utility is to titrate sedation during neuromuscular blockade.19 Post-anoxic seizures and frequent periodic discharges are routinely treated with antiepileptic and suppressive agents, but they are also thought to reflect a more significant injury – especially if arising from a suppressed or burst-suppressed background,24,25 and may therefore influence prognostication. The BISi is defined as the mean bispectral index plateau value recorded during 10 min following the administration of the first dose of neuromuscular blockade.15 Neuromuscular blockade is necessary to eliminate muscle artifact that confounds the BIS reading, and the measurement is typically performed after arrival in the ICU, at which point cooling pads are placed, the BIS monitor applied, sedation and analgesia infused, and a bolus dose of 0.1 mg/kg vecuronium administered to augment the cooling process. The practice standard in our institution includes urgent cardiac catheterization for most CA survivors found to have ST-elevation myocardial infarction, but the most appropriate therapy is ultimately determined by the attending interventional cardiologist, and varies somewhat. Other cardiac arrest survivors, including those with a presumed cardiac cause of the arrest, rarely undergo urgent coronary angiography and PCI (defined as occurring prior to neurological assessment or awakening), but among patients who demonstrate neurological recovery, most undergo angiography on a delayed basis. We believe this approximates standard care in our region, and most areas of the United States, though some hospitals have adopted a more aggressive approach.26,27 Based on commonly employed cardiac risk strata and previously observed associations between processed EEG findings and neurological outcomes,15 a simple grid for risk stratification was developed (see Table 2 below). Patients in the database were classified according to the grid, and within each category the following treatment and outcomes considered: survival, death due to neurological etiology, death due to circulatory failure, early coronary angiography, and early percutaneous coronary intervention.
3. Results One hundred and seventy one patients met inclusion criteria. Demographics, number of comorbid medical conditions, characteristics of the arrest, presence of factors such as STEMI, cardiomyopathy, and shock, severity of neurological dysfunction as described
Please cite this article in press as: Seder DB, et al. Feasibility of bispectral index monitoring to guide early post-resuscitation cardiac arrest triage. Resuscitation (2014), http://dx.doi.org/10.1016/j.resuscitation.2014.04.016
111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167
168
169 170 171 172
G Model
ARTICLE IN PRESS
RESUS 5976 1–7
D.B. Seder et al. / Resuscitation xxx (2014) xxx–xxx
3
Table 1
Q6 Demographics, clinical factors, and differences in cause of death. P value contrasts patients with circulatory to cerebral death. Demographics
All patients (n = 171)
Circulatory death
Cerebral death
P
Age, years Gender female Number of comorb Witnessed Bystander CPR Rhythm VT/VF TIT (min) STEMI Shock BISi < 10 10 < BISi < 20 BISi > 20 LVEF < 30% LVEF < 50% EEG seizures Myoclonic activity
60.1 ± 16.0 51/171 (29.8%) 2.2 ± 1.5 136/167 (81.4%) 76/167 (45.5%) 82/169 (48.5%) 23.7 ± 15.6 33/171 (19.3%) 48/170 (28.2%) 66/171 (38.6%) 19/171 (11.1%) 86/171 (50.3%) 31/120 (25.8%) 75/120 (62.5%) 10/171 (5.8%) 40/171 (23.4%)
64.4 ± 16.5 14/27 (51.9%) 3.0 ± 1.6 25/27 (92.6%) 8/27 (29.6%) 12/26 (46.2%) 21.1 ± 14.1 2/27 (7.4%) 9/27 (33.3%) 6/27 (22.2%) 3/27 (11.1%) 18/27 (66.6%) 10/20 (50.0%) 18/20 (90.0%) 0/27 (0%) 7/27 (25.9%)
58.9 ± 17.8 23/73 (31.5%) 2.3 ± 1.6 45/70 (64.3%) 30/70 (42.9%) 20/72 (27.8%) 27.7 ± 16.1 8/73 (11.0%) 14/72 (19.4%) 54/73 (74.0%) 6/73 (8.2%) 13/73 (17.8%) 6/45 (13.3%) 19/45 (42.2%) 8/73 (11.0%) 26/73 (35.6%)
0.17 0.064 0.049 0.006 0.2 0.09 0.09 0.6 0.15 20 Class I # Patients/survival
Moderate brain injury BISi 10–20 Class II # Patients/survival
Severe brain injury BISi < 10 Class III # Patients/survival
Overall STEMI VT-VF with shock VT/VF without shock PEA or Asystole
81/68% 16/94% 9/56% 22/82% 34/50%
22/45% 8/75% 1/100% 3/67% 10/10%
66/9% 9/22% 5/0% 16/25% 36/0%
Please cite this article in press as: Seder DB, et al. Feasibility of bispectral index monitoring to guide early post-resuscitation cardiac arrest triage. Resuscitation (2014), http://dx.doi.org/10.1016/j.resuscitation.2014.04.016
216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235
G Model RESUS 5976 1–7 4
ARTICLE IN PRESS D.B. Seder et al. / Resuscitation xxx (2014) xxx–xxx
Fig. 1. (a) Absolute mortality and risk of circulatory-etiology etiology death by cardiac risk-class stratification. (b) Absolute mortality and risk of neurological-etiology etiology death by BISi stratification. (c) Relative risk of cerebral vs. circulatory etiology death by BISi stratification.
236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262
will create the framework for individualized post-resuscitation cardiac arrest care, with the potential to save lives, effectively allocate scarce resources, and improve the cost-effectiveness of care. Because of its apparent simplicity, widespread use, proprietary algorithm, and variable performance in anesthesia sedation monitoring trials,28–30 the Bispectral Index is a controversial processed EEG modality. Unilateral surface EEG is obtained from a frontotemporal adhesive sensor and undergoes algorithmic analysis – a weighted sum of electroencephalographic parameters including the frequency below which 95% of the power spectrum resides, the relative beta ratio, synchrony of fast and slow waves, and the burst-suppression ratio.31 The mathematical transformation of these signals yields a number from 0 to 100, which is purported to describe the level of awareness. BIS is also simple to apply and interpret, offering a gross estimation of brain electrical activity without interobserver variability, which could be used for triage purposes by non-neurologists. Like conventional EEG, the use of BIS monitoring to assess the severity of brain injury is potentially confounded by medications, temperature, tissue perfusion, and metabolic issues. Unlike conventional EEG, the BIS assesses only a frontotemporal location, so that patients with a heterogeneous brain injury might be misclassified. Despite these limitations, the accuracy of our early BIS based prognostication technique was previously shown in a similar cohort of patients,15 and very low BIS scores in an uncontrolled setting after cardiac arrest have been shown by other groups to correlate closely with poor outcome.32–35 These findings should not
be a surprise – reports of the correlation between EEG background activity and severity of brain injury after cardiac arrest date back to the 1960s.36 In 1965, Hockaday proposed an EEG based grading system based on evaluation of their own cases,37 and a subsequent review of multiple case series revealed no survivors among 408 patients meeting Hockaday criteria for Category IV or V injury – burst suppression or electrocerebral silence (ECS).38 More recently, several groups have characterized the relationship between EEG background and neurological outcome in patients undergoing therapeutic hypothemia,39–42 though none has carefully described the timing of this assessment, or suggested specifically how the information might be used to alter the course of therapy. Very early processed EEG is shown in multiple animal models to correlate with severity of brain injury after cardiac arrest, and based on these experiments, we suggest that it may be possible to move back the timing of brain injury severity assessment much closer to the time of resuscitation without losing accuracy.10–14 Clinicians considering the prospective application of this technique should be mindful of the unknown effects of sedation, narcotic analgesics, and temperature on our assessment model, and confirmation of severe brain injury with other (non-EEG based) modalities prior to consideration of de-escalating or withdrawing life support should be considered mandatory. Our patients were treated under a standard protocol involving lower doses of propofol and fentanyl than some investigators have described,43 and we do not routinely use benzodiazepines, barbiturates, or high-dose propofol. Use of such medications, especially at higher doses prior
Please cite this article in press as: Seder DB, et al. Feasibility of bispectral index monitoring to guide early post-resuscitation cardiac arrest triage. Resuscitation (2014), http://dx.doi.org/10.1016/j.resuscitation.2014.04.016
263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289
G Model
ARTICLE IN PRESS
RESUS 5976 1–7
D.B. Seder et al. / Resuscitation xxx (2014) xxx–xxx
5
Table 3
Q7 Neurocardiac triage grid: serial patients retrospectively assigned to neurological and circulatory risk category based on the initial BIS score after neuromuscular blockade, and standard cardiac risk categories. PCI = percutaneous coronary angiography. Neurological GRADE
Cause of death A
STEMI
Angiography PCI
Cause of death B
VT/VF shock
Angiography PCI
Cardiac Class Cause of death C
VT/VF no shock
Angiography PCI
Cause of death D
PEA/asystole
Angiography PCI
290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324
to neurological assessment, or overdose of such agents actually precipitating the arrest, might confound the assessment of neurological dysfunction, overestimating its association with severe brain injury. Conversely, if in the wake of recent work by Nielsen and colleagues,2 clinicians begin to target a core body temperature of 36 ◦ C after cardiac arrest, BISi scores may increase due to warmer temperature,44,45 and established cutoffs could prove unreliable at identifying mild brain injury. The effect of seizures on BIS scores is poorly characterized, and seizures at the time of the BISi assessment must be excluded if treatment decisions are to be based on the number. Muscle artifact due to shivering is the most common confounder of BIS, and prognostic measurements should not be made unless full neuromuscular blockade is employed – ideally verified by train-of-four or similar means. Finally, we suggest that BISi measurements be made 4–8 h after ROSC, since we have not evaluated the accuracy of earlier or later BISi measurements, and it is not known how the evolution of brain electrical activity affects the relationship between EEG background and outcome. Ultra-early estimation of the severity of brain injury after cardiac arrest may identify patients that benefit from intensified or individualized therapy in prospective trials. Although it was recently demonstrated that high-quality post-resuscitation care administered at 33 ◦ C was not superior to 36 ◦ C in an unselected group of unresponsive cardiac arrest survivors,2 hypothermia is shown in many laboratory models to be powerfully neuroprotective. It is reasonable to consider that intensified (longer or deeper hypothermia, blood pressure augmentation, or the addition of neuroprotective agents) therapy applied to a severely brain injured group of patients might improve the dismal outcomes seen in our cohort (9% survival when BISi < 10). Other prognostication schemes have been proposed,46–48 but do not clearly distinguish the source of risk as neurological or cardiac, and therefore cannot effectively guide therapeutic decisions. Many patients in this series (20/169) died of circulatory collapse despite BISi > 10 (mild or moderate brain injury), and we believe
Count Cerebral Circulatory Early Delayed Early Delayed Count Cerebral Circulatory Early Delayed Early Delayed Count Cerebral Circulatory Early Delayed Early Delayed Count Cerebral Circulatory Early Delayed Early Delayed
I BISi>20
II 10
