Protocol-driven neurological prognostication and withdrawal of life-sustaining therapy after cardiac arrest and targeted temperature management

Accepted Manuscript Title: Protocol-driven neurological prognostication and withdrawal of life-sustaining therapy after cardiac arrest and targeted temperature management Authors: Irina Dragancea MD, PhD Matthew P Wise MD, DPhil Nawaf Al-Subaie MD Julius Cranshaw MD PhD Hans Friberg MD, PhD Guy Glover MD Tommaso Pellis MD Rebecca Rylance Andrew Walden MD, PhD Niklas Nielsen MD, PhD Tobias Cronberg MD, PhD, the TTM trial investigators PII: DOI: Reference:

S0300-9572(17)30214-9 http://dx.doi.org/doi:10.1016/j.resuscitation.2017.05.014 RESUS 7186

To appear in:

Resuscitation

Received date: Revised date: Accepted date:

21-11-2016 25-4-2017 11-5-2017

Please cite this article as: Dragancea I, Wise MP, Al-Subaie N, Cranshaw J, Friberg H, Glover G, Pellis T, Rylance R, Walden A, Nielsen N, Cronberg T, Protocol-driven neurological prognostication and withdrawal of life-sustaining therapy after cardiac arrest and targeted temperature management, Resuscitation (2017), http://dx.doi.org/10.1016/j.resuscitation.2017.05.014 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.

1

Protocol-driven neurological prognostication and withdrawal of life-

2

sustaining therapy after cardiac arrest and targeted temperature

3

management

4 5

Irina Dragancea MD, PhD1, Matthew P Wise, MD, DPhil2, Nawaf Al-Subaie, MD3, Julius

7

Cranshaw, MD PhD4 , Hans Friberg, MD, PhD5 , Guy Glover, MD6, Tommaso Pellis, MD7,

8

Rebecca Rylance8, Andrew Walden, MD, PhD9, Niklas Nielsen, MD, PhD10, Tobias

9

Cronberg, MD, PhD1 and the TTM trial investigators*

us

an

10

cr

ip t

6

11 12

1

13

Neurology, Lund, Sweden

14

2

Adult Critical Care, University of Wales, Cardiff

15

3

Department of Intensive Care, St George’s University Hospital, London, UK

16

4

Department of Intensive Care, Royal Bournemouth Hospital, Bournemouth, UK

17

5

Lund University, Skane University Hospital, Department of Clinical Sciences Lund,

18

Anesthesia & Intensive care, Lund, Sweden

19

6

20

London, UK

21

7

Intensive Care Unit, Santa Maria degli Angeli, Pordenone, Italy

22

8

R&D Centre Skåne, Skåne University Hospital, Lund, Sweden

23

9

Department of Intensive Care, Royal Berkshire Hospital, Reading, UK

24

10

25

& Intensive care, Lund, Sweden

Ac ce pt e

d

M

Lund University, Skane University Hospital, Department of Clinical Sciences Lund,

Department of Intensive Care, Guy’s and St Thomas NHS Trust, King’s College Hospital,

Lund University, Helsingborg Hospital, Department of Clinical Sciences Lund, Anesthesia

26

1 Page 1 of 29

27

* A complete list of investigators participating in the TTM trial is available in the

28

Supplemental Data file.

29

Correspondence to: Dr Irina Dragancea, Department of Clinical Sciences, Division

31

of Neurology, Skåne University Hospital, 22185 Lund, Sweden. Tel.:+46 46 171428.

32

E-mail address: [email protected]

ip t

30

35

Number of words in the body of manuscript: 2941

36

Number of references: 32

37

Number of figure: 3

38

Number of tables: 3 + 2 (in the Supplementary appendix)

43 44 45 46 47 48 49 50 51

d

42

Ac ce pt e

41

M

39 40

us

Number of words in the abstract: 249

an

34

cr

33

52 53 54 55 56 57 58

2 Page 2 of 29

59

ABSTRACT

60

Background: Brain injury is reportedly the main cause of death for patients resuscitated after

62

out-of-hospital cardiac arrest (OHCA). However, the majority may actually die following

63

withdrawal of life-sustaining therapy (WLST) with a presumption of poor neurological

64

recovery. We investigated how the protocol for neurological prognostication was used and

65

how related treatment recommendations might have affected WLST decision-making and

66

outcome after OHCA in the Targeted Temperature Management (TTM) trial.

67

Methods: Analyses of prospectively recorded data: details of neurological prognostication;

68

recommended level-of-care; WLST decisions; presumed cause of death; and Cerebral

69

Performance Category (CPC) 6 months following randomisation.

70

Results: Of 939 patients, 452 (48%) woke and 139 (15%) died, mostly for non-neurological

71

reasons, before a scheduled time point for neurological prognostication (72 hours after the end

72

of TTM). Three hundred and thirteen (33%) unconscious patients underwent prognostication

73

at a median 117(IQR 93-137) hours after arrest. Thirty-three (3%) unconscious patients were

74

not neurologically prognosticated and for 2 patients (1%) data were missing. Related care

75

recommendations were: continue in 117 (37%); not escalate in 55 (18%); and withdraw in

76

141 (45%). WLST eventually occurred in 196 (63%) at median day 6 (IQR 5-8). At 6 months,

77

only 2 patients with WLST were alive and 248 (79%) of prognosticated patients had died.

78

There were significant differences in time to WLST and death after the different

79

recommendations (log rank <0.001).

80

Conclusion: Delayed prognostication was relevant for a minority of patients and related to

81

subsequent decisions on level-of-care with effects on ICU length-of-stay, survival time and

82

outcome.

Ac ce pt e

d

M

an

us

cr

ip t

61

83 84

3 Page 3 of 29

85

1. Introduction

86

Out-of-hospital cardiac arrest (OHCA) is a common cause of death worldwide with reported

88

survival less than 11%.1,2 Up to 50% of patients admitted to an intensive care unit (ICU) after

89

successful resuscitation survive,3,4 often with few sequelae or only mild cognitive disability.5

90

For those who die during hospitalisation, the most frequent presumed cause of death is brain

91

injury but the majority of these patients actually die following withdrawal of life-sustaining

92

therapy (WLST).6-12 Decisions about care, including WLST, often involve testing a patient’s

93

potential for a ‘meaningful’ neurological recovery.13 For neurological prognostication, the

94

reliability and optimal timing of clinical, radiological and electrophysiological tests is

95

debatable, especially if patients are treated with targeted temperature management (TTM).14

96

Current guidelines recommend a multimodal approach starting 72 hours after cardiac

97

arrest.15,16 Recent reports have raised concerns about ‘early’ WLST7,9,17,18 but the decision-

98

making processes leading to WLST are not well described or explained. The aim of the

99

present study is to describe the practice of protocol-driven neurological prognostication in the

101 102 103 104

cr

us

an

M

d

Ac ce pt e

100

ip t

87

TTM trial, related treatment recommendations, and outcomes for patients.

2. Methods

105

Post-hoc analysis of data from the TTM trial approved by the steering group before the

106

analyses commenced.4 The study protocol, design and primary outcome are published.19-21

107

Between November 2010 and January 2013, 939 adult comatose survivors of OHCA of

108

primary cardiac origin were randomized to two target temperatures (33 °C or 36 °C) after

109

return of spontaneous circulation (ROSC). To reduce risks of consecutive bias and ‘self-

110

fulfilling prophecies’, a standardized protocol for neurological prognostication and pre4 Page 4 of 29

specified justifications for WLST were included in the design.22 In unconscious patients, a

112

conventional EEG would be performed between 12 and 36 hours, and median nerve

113

somatosensory evoked potentials (SSEP) if available, would be obtained between 48 and 72

114

hours, after the end of TTM.23 Neurological evaluation and a related treatment

115

recommendation should then, in general, occur at least 72 hours after the end of TTM by an

116

assessor blinded to temperature assignment but not to other clinical records. If a patient, for

117

reasons mentioned below, was eligible for earlier prognostication, investigators were still

118

urged to have a blinded evaluation performed. The assessor’s prognostic evaluation would

119

comprise clinical neurological examination, considered unaffected by sedation, and

120

electrophysiological results. Although neuroimaging results were not part of the

121

prognostication protocol, they too were accessible, if available. On the basis of their

122

evaluation, the assessor would issue one of the following recommendations: “continue active

123

intensive care”; “do not escalate intensive care” (DNE); or “withdraw active intensive care.”

124

However, the final decision regarding the level of care was left to the responsible physician.

125

Pre-specified criteria of a poor neurological prognosis allowing for discontinuation of life-

126

support were: 21, 22

128 129 130 131 132

cr

us

an

M

d

1.

Ac ce pt e

127

ip t

111

Persisting coma (Glasgow Coma Score Motor component (GCS-M) ≤2) at least 72

hours after commencement of TTM and: a. absent N20 peaks on SSEP or b. treatment refractory status epilepticus

2. Myoclonic status within 24 hours of arrest and absent N20 peaks on SSEP after rewarming

133

3. Brain death at any time point

134

4. Ethical reasons at any time point

5 Page 5 of 29

For WLST occurring during the first week, the responsible physician reported the reasons

136

using pre-specified categories. More than one reason could be recorded per patient.

137

Neurological outcome was measured after 6 months using the Cerebral Performance Category

138

scale (CPC). A poor neurological outcome was defined as severe neurological dysfunction,

139

vegetative state or death (CPC 3-5). For patients who died before neurological

140

prognostication, the presumed cause of death, any limitations in care and their reasons were

141

recorded. All electronic case records were systematically examined for conflicting data.

cr

ip t

135

us

142

2.1 Statistical analysis

144

Continuous variables are expressed as median and range. Categorical variables are expressed

145

as number of patients and percentages. Mann-Whitney U tests were used for median

146

comparisons. Chi-square tests were used for proportional comparisons of categorical

147

variables. Log-rank tests were used for survival differences between recommendation groups.

148

Kaplan-Meier graphs were produced. Two-tailed p values were used and p < 0.05 was

149

considered significant. All calculations were performed using Stata 13®. Since the date, but

150

not the hour of WLST was recorded, the exact time between OHCA and WLST could not be

151

calculated.

153 154 155

M

d

Ac ce pt e

152

an

143

3. Results

156

Of the 939 patients included in the TTM study, 452 (48%) woke and 139 died (15%) before

157

neurological prognostication. Three hundred and thirteen patients (33%) underwent

158

neurological prognostication. Prognostication was not performed in 33 (3%) patients. Data

6 Page 6 of 29

159

were missing for 2 (1%) patients (Fig. 1). Characteristics of these groups are presented in

160

Table 1.

161

3.1 Patients awakening before neurological prognostication at 6-month follow-up

163

Of those who regained consciousness before neurological prognostication, 387 (86%) had

164

survived with good neurological function (CPC 1-2). Twenty-two (5%) had developed severe

165

neurological disability (CPC 3); 38 (8%) had died; and for 5 (1%) data were missing (Table

166

1).

us

cr

ip t

162

167

3.2 Patients dying before protocolised neurological prognostication could be performed

169

Of 139 patients, a decision of WLST was taken in 105 (76%) on median ICU day 3 (IQR 2-4)

170

with no difference between the two temperature groups (p = 0.316). Supplementary data 1

171

shows their characteristics stratified by TTM allocation. Reasons for WLST were: failing

172

circulation (57(54%)); MOF (39(37%)); ethical reasons (19(18%)); medical co-morbidity

173

(18(17%)); brain death (17(16%)); myoclonic status and absent SSEP-N20 peaks (5(5%));

174

persisting coma and absent SSEP-N20 peaks (2(2%)); and subarachnoid haemorrhage with

175

clinical signs of herniation (1(1%)).

M

d

Ac ce pt e

176

an

168

177

3.3 Unconscious patients who underwent protocolised neurological prognostication

178

Neurological prognostication was performed in 313 patients a median of 117 (IQR 93-137)

179

hours after arrest. Details are presented in Table 2. The assessor’s recommendation was to

180

“continue active intensive care” in 117 (37%), “DNE” in 55 (18%) and “withdraw active

181

intensive care” in 141 (45%). Multiple modalities were reported as supporting

182

recommendations in 241 (77%). In 55 (18%) clinical findings, in 4 (1%) SSEP, in 11 (4%)

183

EEG and in 2 (1%) brain CT were the single modalities reported (Supplementary data 2).

7 Page 7 of 29

Prognostication was reported before 72 hours after the end of TTM in 117 patients (Table 3).

185

The assessor’s recommendation was to “continue active intensive care” in 41 (35%), “DNE”

186

in 14 (12%) and “withdraw active intensive care” in 62 (53%).

187

Of those 117 patients, 7 had early status myoclonus and absent SSEP-N20 potentials and 6

188

were diagnosed brain dead. Fifty-four patients had one or several other pathological findings:

189

bilateral absent pupillary and corneal reflexes (n=20), absent SSEP-N20 potentials (n=25),

190

early status myoclonus (n=9), electrographic status epilepticus (n=20), and generalized

191

oedema on brain CT (n=14).

us

cr

ip t

184

192

3.3.1 Recommendation and WLST decision

194

Of those patients receiving a recommendation, 196 (63%) eventually had WLST at median

195

ICU day 6 (IQR 5-8) with no difference between the two temperature groups (p = 0.901). One

196

hundred and ninety-four had died before 6 month follow-up.

197

One hundred and thirty patients (92%) with a recommendation to “withdraw active intensive

198

care” had WLST (Table 2). In the 55 with “DNE” recommendations, WLST was eventually

199

performed in 34 (62%). In the 117 patients with a recommendation to “continue active

200

intensive care”, 32 (27%) had WLST. Kaplan-Meier curves of the three recommendation

201

groups separated during time from recommendation to WLST (log rank <0.001, Fig. 2a).

202

Among patients assessed before 72 hours after TTM ended, 84 (72%) had WLST, 44 (72%)

203

randomized to 33°C and 40 (71%) randomized to 36°C. Median time from cardiac arrest to

204

WLST was ICU day 5 (IQR 4-6) with no differences between the two temperature groups

205

(p=0.635). Reasons for WLST are presented in Table 3.

Ac ce pt e

d

M

an

193

206 207 208

8 Page 8 of 29

3.3.2. Recommendation and outcome at 6-month follow-up.

210

At 6-month follow-up, 139 (98%) patients with recommendations to “withdraw active

211

intensive care”, 51 (92%) with “DNE” recommendations and 58 (50%) with “continue active

212

intensive care” recommendations had died (Table 2). Kaplan-Meier curves of the three

213

recommendation groups separated during time from recommendation to death (log rank

214

<0.001, Fig. 2b). All but two patients with recommendations to “withdraw active intensive

215

care” had died before follow-up; one was discharged from ICU in CPC 4 and was in CPC 2 at

216

6 months; one, who did not have WLST, was in CPC 3 at 6 months. The presumed cause of

217

death among all assessed patients was classified as cerebral in 199 (80%), cardiovascular in

218

25 (10%), MOF in 15 (6%) and other in 9 (4%) (Fig. 3a).

an

us

cr

ip t

209

219

3.4 Outcome of patients who were not neurologically prognosticated

221

Thirty-three (3%) patients were not neurologically prognosticated. At 6 months, 22 (68%) had

222

died and WLST occurred in 10 (30%). One of the two patients with missing data about

223

prognostication had WLST.

d

Ac ce pt e

224

M

220

225

3.5 Time to death, cause of death and WLST among all patients in the trial

226

At 6 months, 449 patients had died, 411 (92%) before hospital discharge. The median time to

227

death was five days (IQR 3-10) after arrest. A cardiovascular cause of death was more

228

common during the first 2 days after arrest. A cerebral cause predominated from 3 days to six

229

months (Fig. 3b). WLST preceded in-hospital death in 307 (75%). There was no significant

230

difference in the proportion of patients who had WLST (p = 0.554), either in their time to

231

WLST (p = 0.967) or time to death (p = 0.476) between the two temperature groups.

232

WLST during days 1-3 in the ICU, occurred in 87 (21% of in-hospital deaths) patients.

233

Neurological reasons, including brain death, were given in 28 patients.

9 Page 9 of 29

4. Discussion

235

Outcome studies of OHCA survivors may be prone to bias because pessimistic and premature

236

determinations of neurological outcome might increase WLST and mortality. To address this

237

issue, the TTM trial protocol incorporated commonly used neurological prognostication tests

238

for unconscious patients after OHCA and linked criteria for WLST.21,22 These elements of the

239

trial design might represent a standard of clinical practice that could reduce the risk of

240

unfounded WLST.

241

However, we observed that prior to the specified neurological prognostic time point, 72 hours

242

after the end of TTM, nearly half the trial patients had regained consciousness and 15% had

243

died. In the third who were still unconscious, and for whom prognostic evaluation was

244

performed, less than half, had a recommendation to “withdraw active intensive care”. Some

245

patients with this recommendation did not have WLST and two patients were alive at 6

246

months, one in CPC 2 and one in CPC 3.

247

We consider these observations to be novel and informative. First, they indicate that, despite a

248

clinical trial setting, neurological prognostic evaluation might only be required in a minority

249

of survivors of OHCA. Second, although results may suggest further intensive care treatment

250

is futile, this need not result in rapid WLST. While a treating physician might concur with a

251

specialist recommendation, WLST may not be inevitable. Further research should analyse this

252

discordance.

253

Current guidelines in which combinations of clinical findings and test results are prerequisite

254

to consider a poor neurological prognosis “likely” or “very likely” after cardiac arrest,

255

recommend a multimodal approach.15,16 In these guidelines, most unconscious patients are

256

evaluated later than 72 hours after cardiac arrest, postponing the process of prognostication by

257

comparison to older algorithms.24 The TTM trial preceded these guidelines but the trial

Ac ce pt e

d

M

an

us

cr

ip t

234

10 Page 10 of 29

protocol has important similarities. Lacking other comparative data, our results may help

259

clinicians using up-to-date guidance estimate possible effects of guideline implementation.

260

For the majority of patients in the TTM trial, neurological prognosis was formulated using a

261

multimodal approach. Our results show that clinical findings, EEG interpretations and SSEP

262

reports were frequently involved in supporting recommendations. Despite omission from the

263

trial protocol, brain CT results were also often considered during prognostic evaluation. This

264

may reflect the clinical opinion that certain CT appearances of global ischaemic injury have

265

high specificity to predict poor outcome after OHCA.25,26

266

In 18 patients, a recommendation to “withdraw active intensive care” appeared solely based

267

on clinical findings. However, WLST decisions were taken in 16 patients at median ICU day

268

5 (IQR 4-6) and the reported reasons for WLST were not neurological in most. The trial

269

WLST criteria did not include the findings from ocular reflexes although their absence is

270

considered a reliable marker of adverse prognosis.15,27,28 GCS-M scores, which are

271

inconsistent predictors, were also excluded, except in combination with absent N20 peaks on

272

SSEP or treatment-refractory status epilepticus. We consider that under certain circumstances

273

and with specific results, a “withdraw active intensive care” recommendation might be

274

acceptable on clinical grounds alone. Whether the responsible clinician accepted such a

275

recommendation per se is, however, uncertain and simultaneous ethical and other unreported

276

aspects of a case may have influenced their decision-making.

277

The majority of our patients were assessed 72 hours after the end of TTM. Nevertheless we

278

found that 20 percent of assessed patients received a recommendation to “withdraw active

279

intensive care” before this time point. The protocol did provide opportunities for earlier tests

280

and assessments if brain death was suspected, status myoclonus occurred or if there were

281

ethical reasons not to extend intensive care further. We speculate that these exceptions added

282

variability to the conduct of the trial protocol. Established routines and guidelines may have

Ac ce pt e

d

M

an

us

cr

ip t

258

11 Page 11 of 29

also influenced the way the protocol was implemented locally and nationally. The utility of

284

multiple prognostication time points, as proposed in the 2015 ERC/ESICM guidelines, in our

285

opinion, requires further investigation.16

286

As a cohort, the patients undergoing neurological prognostication had poor outcomes. By 6

287

months, 79% had died and only 15% had achieved a good neurological outcome. More than

288

half did not survive beyond their ICU stay, and for ICU survivors the median time to ICU

289

discharge was 9 days after arrest. Nevertheless, we demonstrate that within this cohort, the

290

ICU course diverged in association with prognostic recommendation. Over 80% of patients

291

receiving a “withdraw active intensive care” recommendation had WLST within one day and

292

their median survival from OHCA was only five days. By contrast, for the majority, the

293

recommendations “DNE” or “continue active intensive care” were associated with several

294

more days in the ICU before WLST, when this occurred. WLST ultimately transpired in

295

almost two thirds of patients with a “DNE” recommendation and one third with a

296

recommendation to “continue active intensive care”. Median survival times were 8 and 14

297

days after OHCA respectively. While the recommendation to “continue active intensive care”

298

was associated with 6 month survival of 50% and good neurological outcome in 37%, “DNE”

299

and “withdraw active intensive care” recommendations were associated with a similar high 6

300

month mortality greater than 90% and good neurological outcomes in less than 4%. The latter

301

recommendations may be perceived as end-of-life decisions with no ethical or legal

302

distinction between them. Our observations support a practical similarity.29

303

However, we propose that observed contrasting ICU lengths of stay, survival times and

304

outcomes for patients undergoing neurological prognostication are potentially consequences

305

of the three distinct trial recommendations. These differences may be clinically important to

306

patients, their families and their carers. As the recommendations may be in current clinical

Ac ce pt e

d

M

an

us

cr

ip t

283

12 Page 12 of 29

use, the ethical implications are broad and should be considered in the context of a patient’s

308

wishes, preferences, feelings, beliefs and values as well as their society and healthcare system.

309

Overall, WLST was associated with three quarters of in-hospital deaths, more than reported in

310

similar studies.7,18,30 Clearly cultural and classification differences make between-study

311

comparisons fraught. Our study was uniquely prospective and therefore may suffer from

312

reporting bias. However, we believe this observation deserves further comment. In one large

313

North American cohort, WLST for neurological reasons, including brain death, within 72

314

hours of ROSC was associated with 44% of in-hospital deaths.7 By contrast, in our study,

315

WLST for any reason within the first three ICU days was associated with only 21% of in-

316

hospital deaths. Neurological reasons, including brain death, were recorded in only 7%.

317

Furthermore, among neurologically prognosticated patients, the median day of WLST was

318

ICU day 6, two to three days later than in other reports.11,17 Thus our reported WLST rate may

319

be an effect of the TTM trial protocol’s delayed approach to neurological prognostication.

320

This approach may have also contributed to the comparatively large proportion (9%) of

321

survivors with poor 6-month neurological outcomes in our study. In support of this

322

relationship, the PROPAC II study31, which allowed neurological examination 48 and 72

323

hours after cardiac arrest reported half this percentage. Conversely, a study from South Korea,

324

where WLST is prohibited, reported almost half the survivors having a poor neurological

325

outcome at 6 months.30

326

Our study has limitations. The TTM trial was not designed primarily to investigate the process

327

of neurological prognostication, related decision-making, nor the effects on outcome. We also

328

acknowledge that our results might not be applicable to all OHCA patients in the ICU since

329

the TTM trial did not enrol non-cardiac cause CA and in-hospital CA patients, sub-groups

330

with a worse prognosis.

Ac ce pt e

d

M

an

us

cr

ip t

307

13 Page 13 of 29

The responsible physician was not blinded to their patient’s TTM allocation, opening the

332

possibility of bias in continuing or withholding care on the basis of treatment allocation.

333

However, for ethical reasons, it is a good standard of care that two experienced senior medical

334

opinions are sought before a recommendation of withdrawal or limitation of life-sustaining

335

therapy.32 In our study, for patients undergoing neurological assessment, a second

336

experienced opinion was provided by a qualified neurologist or an intensivist in almost all

337

cases.

338

In conclusion, three-quarters of all in-hospital deaths in the TTM trial occurred after WLST.

339

While non-neurological reasons for WLST were much more common among those who died

340

prior to prognostic assessment, in patients who were assessed, WLST for neurological reasons

341

predominated. Our study suggests that although the clinical process of prognostication may be

342

linked to subsequent decisions about level-of-care, other factors can be influential. The three

343

different treatment recommendations in the TTM trial were associated with different lengths

344

of hospital stay and times to WLST. However, recommendations to not escalate care and to

345

withdraw active intensive care were associated with similar very high mortality at 6 months.

347 348 349

cr

us

an

M

d

Ac ce pt e

346

ip t

331

Author’s contribution

350

I.D. and T.C. designed the study and wrote the first draft of the manuscript. M.P.W., H.F., and

351

N.N. contributed to the present study concept and design. H.F. was senior investigator in the

352

TTM trial. M.P.W. and T.P were national investigators. J.C., N.A.S., G.G., A.W. were

353

principal investigators. N.N. was the chief-investigator of the TTM trial. R.R. was study

354

statistician. All authors of the manuscript were involved in the analysis and interpretation of

355

the data and reviewed and edited the manuscript. 14 Page 14 of 29

Funding

357

The TTM-trial and the present study was funded by independent research grants from the

358

non-profit or governmental agencies: Swedish Heart Lung Foundation (grant no. 20090275);

359

Arbetsmarknadens försäkringsaktiebolag AFA-Insurance Foundation (grant no.100001); The

360

Swedish Research Council (grant no. 134281, 296161, 286321); Regional research support,

361

Region Skåne; Governmental funding of clinical research within the Swedish NHS (National

362

Health Services) (grant no. M2010/1837, M2010/1641, 353301); Thelma Zoega Foundation;

363

Krapperup Foundation; Thure Carlsson Foundation; Hans-Gabriel and Alice Trolle-

364

Wachtmeister Foundation for Medical Research; Skåne University Hospital; Sweden, Tryg

365

Foundation; Denmark, and the European Clinical Research Infrastructures Network.

an

us

cr

ip t

356

366

369

Conflict of interest statement

d

368

M

367

I. Dragancea receives academic support from the County Council of Skåne.

371

M.P. Wise was funded 40% WTE during the study by a National Institute for Social Care and

372

Health Research (NISCHR) Academic Health Science Collaboration (AHSC) Clinical

373

Reserach Fellowship; received travel costs from the British Thoracic Society, Intensive Care

374

Society, Scottish Intensive Care Society, Orion Ltd.; Royalties from Willey Publishing;

375

Honorarium for lecturing at educational meeting Fisher & Paykel, and Merck; serves on the

376

advisory board for Kalobius Pharmaceuticals, Bard; Editor BMJ Open Respiratory Research.

377

N. Al-Subaie and J. Cranshaw have no conflicts of interests to report. H. Friberg receives

378

support from the Swedish National Health System (ALF), serves on the scientific advisory

379

board of QuickCool, and has received honoraria for lectures from from Natus Inc, and Bard

380

Medical. N. Nielsen receives support from the Swedish National Health System (ALF) and

Ac ce pt e

370

15 Page 15 of 29

has received honoraria from Bard Medical. G. Glover has received fees for consulting and for

382

travel from Bard Medical. R. Rylance have no conflicts of interest to report. T. Pellis has

383

received honorarium for lectures from Bard Medical. A Walden had received honorarium for

384

advice given to Bard. T. Cronberg receives academic support from the County Council of

385

Skåne and from the Swedish National Health System (ALF).

ip t

381

386

cr

387

us

388 389

an

390 391

395 396 397 398 399 400 401

d

394

Ac ce pt e

393

M

392

402 403 404 405

16 Page 16 of 29

406

Legends

407

Table 1 Demographic and clinical data.

408

Data are numbers of patients and percentages or medians and interquartile range (IQR). Awake before prognostication

Deceased before prognostication

Performed prognostication

Prognostication not performed

Number of patients1

939

452 (48%)

139 (15%)

313 (33%)

33 (3%)

Age

65 (56-73)

62 (54-71)

70 (62-79)

66 (58-73)

67 (58-73)

Male

761 (81%)

374 (83%)

103 (74%)

258 (82%)

24 (73%)

Location of CA Place of residence Public place Other

500 (53%) 385 (41%) 53 (6%)

204 (45%) 219 (49%) 29 (6%)

89 (64%) 43 (31%) 7 (5%)

187 (60%) 113 (36%) 13 (4%)

19 (58%) 10 (30%) 4 (12%)

Initial rhythm2 Shockable rhythm Asystole PEA Unknown

752 (80%) 113 (12%) 65 (7%) 8 (1%)

420 (93%) 16 (3%) 14 (3%) 2 (1%)

80 (58%) 32 (23%) 25 (18%) 2 (1%)

229 (73%) 56 (18%) 24 (8%) 4 (1%)

22 (67%) 9 (27%) 2 (6%) 0 (0%)

Time to ROSC (min)

25 (17-39)

20 (14-30)

35 (24-59)

30 (22-44)

29 (19-40)

Decision of WLST was taken3

312 (33%)

0 (0%)

105 (76%)

196 (63%)

10 (30%)

Time to WLST (days in ICU)

5 (3-6)

Not applicable

3 (2-3)

6 (5-8)

7 (4-7)

Time to ICU discharge, (hrs after CA)

120 (73-195)

116 (77-187)

44 (16-66)

162 (120-261)

103 (74-170)

140 (15%) 231 (25%) 143 (15%) 115 (12%) 310 (33%)

127 (28%) 213 (47%) 106 (23%) 2 (1%) 4 (1%)

0 (0%) 0 (0%) 0 (0%) 4 (3%) 135 (97%)

12 (4%) 17 (5%) 32 (10%) 90 (29%) 162 (52%)

1 (3%) 1 (3%) 5 (15%) 17 (52%) 9 (27%)

Time to hospital discharge (hrs after CA)

333 (185-564)

338 (195-512)

57 (34-84)

399 (214-930)

129 (64-229)

Time to death (days after CA)

5 (3-10)

20 (13-34)

2 (1-3)

7 (5-10)

7 (6-11)

378 (40%) 62 (7%) 37 (4%) 7 (1%) 449 (48%)

340 (75%) 47 (11%) 22 (5%) 0 (0%) 38 (9%)

0 (0%) 0 (0%) 0 (0%) 0 (0%) 139 (100%)

34 (11%) 13 (4%) 11 (4%) 7 (2%) 248 (79%)

4 (13%) 2 (6%) 4 (13%) 0 (0%) 22 (68%)

CPC at 6 months4 CPC1 CPC2 CPC3 CPC4 CPC5

cr

us

an

M

d

Ac ce pt e

CPC at ICU discharge CPC1 CPC2 CPC3 CPC4 CPC5

ip t

All patients

17 Page 17 of 29

409

CA = cardiac arrest. ROSC = return of spontaneous circulation. PEA = pulseless electric

410

activity. WLST = withdrawal of life-sustaining treatment. CPC = Cerebral Performance

411

Category.

412

1

413

patients in 33°C group.

414

2

415

36°C group.

416

3

One patient in whom data regarding neurological prognosis was missing had WLST.

417

4

The outcome at 6 month follow-up was missing in 6 patients (4 in the 33°C group and 2 in

418

36°C) but none belonged to the group of patients undergoing neurological prognostication.

ip t

Data on whether neurological prognostication was performed or not was missing in 2

an

us

cr

The initial rhythm was missing in one patient who was not neurologically evaluated in the

419

Table 2 Details of neurological prognostication and related treatment recommendations.

421

Table 2 Details of neurological prognostication and related treatment recommendations.

Ac ce pt e

d

M

420

18 Page 18 of 29

422

DNE n=55

WAIC n=141

61 (52%) 48 (41%) 8 (7%)

20 (36%) 27 (49%) 8 (15%)

61 (43%) 71 (51%) 8 (6%)

119 (96-143)

121 (107-139)

113 (87-131)

Recommendation was based on Clinical findings SSEP EEG CT MRI

98 (84%) 56 (48%) 67 (57%) 17 (15%) 5 (4%)

53 (96%) 24 (44%) 40 (73%) 20 (36%) 0 (0%)

WLST was performed

32 (27%)

34 (62%)

Days between recommendation and WLST 0 1 2 3 ≥4

1 (3%) 8 (25%) 5 (16%) 7 (22%) 11 (34%)

Reasons for WLST2 Neurological futility3 Failing circulation MOF Medical-comorbidity Ethical Other

Length of ICU stay (hours after CA) non-survivors survivors at ICU discharge Time to death (days after CA)

CPC at 6 months CPC1 CPC2 CPC3 CPC4 CPC5

117 (93-137)

us

280 (89%) 151 (48%) 213 (68%) 92 (29%) 13 (4%)

130 (92%)

196 (63%)

11 (33%) 10 (29%) 10 (29%) 1 (3%) 2 (6%)

106 (81%) 16 (12%) 6 (5%) 1 (1%) 1 (1%)

118 (60%) 34 (17%) 21 (11%) 9 (5%) 14 (7%)

24 (75%) 3 (9%) 5 (16%) 1 (3%) 3 (9%) 0 (0%)

29 (85%) 2 (6%) 3 (9%) 0 (0%) 4 (12%) 0 (0%)

113 (87%) 8 (6%) 10 (8%) 5 (4%) 23 (18%) 1 (1%)

166 (85%) 13 (7%) 18 (9%) 6 (3%) 30 (15%) 1 (0.5%)

191 (145-282) 291 (195-462)

172 (132-202) 221 (146-318)

120 (90-149) 144 (115-192)

137 (106-188) 218 (145-342)

14 (8-36)

8 (6-10)

5 (4-8)

7 (5-10)

34 (29%) 10 (8%) 9 (8%) 6 (5%) 58 (50%)

0 (0%) 2 (4%) 1 (2%) 1 (2%) 51 (92%)

0 (0%) 1 (1%) 1 (1%) 0 (0%) 139 (98%)

34 (11%) 13 (4%) 11 (4%) 7 (2%) 248 (79%)

d

M

an

129 (92%) 71 (50%) 106 (75%) 55 (39%) 8 (6%)

Ac ce pt e

Time to prognostication (hours after CA)

142 (45%) 146 (47%) 24 (8%)

cr

CC n=117 Who performed the recommendation1 Neurologist Intensivist Other

423

All patients n=313

ip t

Recommendation

Data are numbers of patients and percentages or medians and interquartile range (IQR).

19 Page 19 of 29

CC = continue active intensive care. DNE = do-not-escalate active intensive care.

425

WAIC = withdraw active intensive care. WLST = Withdrawal of life sustaining therapy.

426

CA = cardiac arrest. SSEP: somatosensory evoked potential; EEG: electroencephalography;

427

CT = computed tomography. MRI = magnetic resonance imaging. CPC = Cerebral

428

Performance Category.

429

1

430

patient in the group “WAIC”.

431

2

More than one reason could be recorded per patient.

432

3

Neurological futility comprises brain death, coma and absent N20 on SSEP, coma and

433

treatment refractory status epilepticus, myoclonus status and absent N20 on SSEP.

ip t

424

us

an

434

d

M

Table 3 Patients who were neurologically prognosticated before 108 hours after beginning

Ac ce pt e

435

cr

The data regarding who performed the neurological prognostication was missing in one

20 Page 20 of 29

436 437

Table 3 Patients who were neurologically prognosticated before 108 hours after beginning TTM and related treatment recommendations. Recommendation

All patients n=117

CC n=41

DNE n=14

WAIC n=62

25 (61%) 15 (37%) 1 (2%)

9 (64%) 2 (14%) 3 (22%)

35 (56%) 24 (39%) 2 (3%)

85 (62-97)

85 (70-97)

84 (67-95)

Recommendation was based on2 Clinical findings SSEP EEG CT MRI

33 (80%) 20 (49%) 24 (59%) 3 (7%) 0 (0%)

13 (93%) 3 (21%) 8 (6%) 2 (14%) 0 (0%)

58 (94%) 31 (50%) 41 (66%) 13 (21%) 1 (2%)

104 (89%) 54 (46%) 73 (62%) 18 (15%) 1 (1%)

WLST was performed

16 (39%)

11 (79%)

57 (92%)

84 (72%)

Reasons for WLST3 Neurological futility4 Failing circulation MOF Medical-comorbidity Ethical Other

14 (88%) 2 (13%) 2 (13%) 0 (0%) 2 (13%) 0 (0%)

8 (73%) 1 (9%) 1 (9%) 0 (0%) 3 (27%) 0 (0%)

48 (84%) 6 (11%) 7 (12%) 3 (5%) 14 (25%) 0 (0%)

70 (83%) 9 (11%) 10 (12%) 3 (4%) 19 (23%) 0 (0%)

8 (6-16)

5 (4-7)

4 (3-5)

5 (4-7)

14 (34%) 3 (7%) 1 (2%) 1 (2%) 22 (54%)

0 (0%) 1 (7%) 0 (0%) 0 (0%) 13 (93%)

0 (0%) 0 (0%) 1 (2%) 0 (0%) 61 (98%)

14 (12%) 4 (3%) 2 (2%) 1 (1%) 96 (82%)

Time to death (days after CA)

CPC at 6 months CPC1 CPC2 CPC3 CPC4 CPC5

ip t

69 (59%) 41 (35%) 6 (5%)

84 (66-97)

cr

us

an

M

Ac ce pt e

Time to prognostication ( hours after CA)

d

Who performed the recommendation1 Neurologist Intensivist Other

438

TTM and related treatment recommendations.

439

Data are numbers of patients and percentages or medians and interquartile range (IQR).

440

CC = “continue active intensive care”.

441

DNE = do-not-escalate active intensive care. WAIC = withdraw active intensive care.

442

1

443

patient in the group “WAIC”.

444

2

The data regarding who performed the neurological prognostication was missing in one

Multiple findings possible. 21 Page 21 of 29

445

3

More than one reason could be recorded per patient.

446

4

Neurological futility comprises brain death, coma and absent N20 on SSEP, coma and

447

treatment refractory status epilepticus, myoclonus status and absent N20 on SSEP.

448

Supplementary data 1 Demographic and clinical data stratified by TTM allocation.

450

Data are numbers of patients and percentages or medians and interquartile range (IQR).

451

CA = cardiac arrest. ROSC = return of spontaneous circulation. PEA = pulseless electric

452

activity. WLST = withdrawal of life- sustaining treatment. CPC = cerebral performance

453

category.

454

1

455

patient in the 36°C group.

456

2

457

36°C) but none belong to the group of patients who underwent neurological prognostication.

us

cr

ip t

449

an

The data regarding who performed the neurological prognostication was missing in one

M

The outcome at 6 month follow-up was missing in 6 patients (4 in the 33°Cgroup and 2 in

d

458

Supplementary data 2 Characteristics of patients in which recommendation on level of care

460

was based on a single prognostic modality.

461

Data are numbers of patients and percentages or medians and interquartile range (IQR).

462

* The ratio represents the number of patients with abnormal findings / number of patients in

463

whom the diagnostic tests were performed.

464

CA = cardiac arrest. WLST = withdrawal of life- sustaining treatment. PLR = pupillary light

465

reflexes.CR = corneal reflexes; GCS M = Glasgow Coma Scale Motor score. SSEP =

466

somatosensory evoked potential. EEG = electroencephalography.

467

GPED = generalized periodic epileptiform discharge. ICU = Intensive care unit.

468

MOF = Multi-organ failure.

Ac ce pt e

459

469

22 Page 22 of 29

Fig. 1

471

Study flow chart.

Ac ce pt e

472

d

M

an

us

cr

ip t

470

473

474

Fig. 2a

475

Kaplan-Meier curves showing the probability of withdrawal of life-sustaining therapy over

476

time by recommendation on level-of-care.The graf comprises neurologically prognosticated 23 Page 23 of 29

477

patients who had WLST. WLST = withdrawal of life-sustaining treatment. WAIC = withdraw

478

active intensive care.

an

us

cr

ip t

479

Fig. 2b

482

Kaplan-Meier curves showing the probability of death over time by recommendation on level-

483

of-care. The graf comprises neurologically prognosticated patients who died before 6- month

484

follow-up. WAIC = withdraw active intensive care

d

Ac ce pt e

485

M

480 481

24 Page 24 of 29

486

3a

489 490

Ac ce pt e

488

d

M

an

us

cr

ip t

487

Fig.

Cause of death for neurologically prognosticated patients.

25 Page 25 of 29

491

3b

494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513

Ac ce pt e

493

d

M

an

us

cr

ip t

492

Fig.

Cause of death for all patients.

26 Page 26 of 29

cr an

us 2.

3.

4.

5.

6.

7.

8.

9.

M

1.

Hasselqvist-Ax I, Riva G, Herlitz J, et al. Early cardiopulmonary resuscitation in out-ofhospital cardiac arrest. The New England journal of medicine 2015;372:2307-2315. Berdowski J, Berg RA, Tijssen JG, Koster RW. Global incidences of out-of-hospital cardiac arrest and survival rates: Systematic review of 67 prospective studies. Resuscitation 2010;81:1479-1487. Callaway CW, Schmicker RH, Brown SP, et al. Early coronary angiography and induced hypothermia are associated with survival and functional recovery after out-of-hospital cardiac arrest. Resuscitation 2014;85:657-663. Nielsen N, Wetterslev J, Cronberg T, et al. Targeted temperature management at 33 degrees C versus 36 degrees C after cardiac arrest. The New England journal of medicine 2013;369:2197-2206. Lilja G, Nielsen N, Friberg H, et al. Cognitive function in survivors of out-of-hospital cardiac arrest after target temperature management at 33 degrees C versus 36 degrees C. Circulation 2015;131:1340-1349. Lemiale V, Dumas F, Mongardon N, et al. Intensive care unit mortality after cardiac arrest: the relative contribution of shock and brain injury in a large cohort. Intensive care medicine 2013;39:1972-1980. Elmer J, Torres C, Aufderheide TP, et al. Association of early withdrawal of lifesustaining therapy for perceived neurological prognosis with mortality after cardiac arrest. Resuscitation 2016. Dragancea I, Rundgren M, Englund E, Friberg H, Cronberg T. The influence of induced hypothermia and delayed prognostication on the mode of death after cardiac arrest. Resuscitation 2013;84:337-342. Mulder M, Gibbs HG, Smith SW, et al. Awakening and withdrawal of life-sustaining treatment in cardiac arrest survivors treated with therapeutic hypothermia*. Critical care medicine 2014;42:2493-2499.

d

535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560

References

Ac ce pt e

534

ip t

514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533

27 Page 27 of 29

d

M

an

us

cr

ip t

10. Tomte O, Andersen GO, Jacobsen D, Draegni T, Auestad B, Sunde K. Strong and weak aspects of an established post-resuscitation treatment protocol-A five-year observational study. Resuscitation 2011;82:1186-1193. 11. Fugate JE, Wijdicks EF, White RD, Rabinstein AA. Does therapeutic hypothermia affect time to awakening in cardiac arrest survivors? Neurology 2011;77:1346-1350. 12. Cronberg T, Rundgren M, Westhall E, et al. Neuron-specific enolase correlates with other prognostic markers after cardiac arrest. Neurology 2011;77:623-630. 13. Geocadin RG, Buitrago MM, Torbey MT, Chandra-Strobos N, Williams MA, Kaplan PW. Neurologic prognosis and withdrawal of life support after resuscitation from cardiac arrest. Neurology 2006;67:105-108. 14. Al Thenayan E, Savard M, Sharpe M, Norton L, Young B. Predictors of poor neurologic outcome after induced mild hypothermia following cardiac arrest. Neurology 2008;71:1535-1537. 15. Sandroni C, Cariou A, Cavallaro F, 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. Resuscitation 2014;85:1779-1789. 16. Nolan JP, Soar J, Cariou A, et al. European Resuscitation Council and European Society of Intensive Care Medicine 2015 guidelines for post-resuscitation care. Intensive care medicine 2015;41:2039-2056. 17. Albaeni A, Chandra-Strobos N, Vaidya D, Eid SM. Predictors of early care withdrawal following out-of-hospital cardiac arrest. Resuscitation 2014;85:1455-1461. 18. Perman SM, Kirkpatrick JN, Reitsma AM, et al. Timing of neuroprognostication in postcardiac arrest therapeutic hypothermia*. Critical care medicine 2012;40:719-724. 19. Nielsen N, Winkel P, Cronberg T, et al. Detailed statistical analysis plan for the target temperature management after out-of-hospital cardiac arrest trial. Trials 2013;14:300. 20. Cronberg T, Lilja G, Horn J, et al. Neurologic Function and Health-Related Quality of Life in Patients Following Targeted Temperature Management at 33 degrees C vs 36 degrees C After Out-of-Hospital Cardiac Arrest: A Randomized Clinical Trial. JAMA neurology 2015;72:634-641. 21. Nielsen N, Wetterslev J, al-Subaie N, et al. Target Temperature Management after out-ofhospital cardiac arrest--a randomized, parallel-group, assessor-blinded clinical trial-rationale and design. American heart journal 2012;163:541-548. 22. Cronberg T, Horn J, Kuiper MA, Friberg H, Nielsen N. A structured approach to neurologic prognostication in clinical cardiac arrest trials. Scandinavian journal of trauma, resuscitation and emergency medicine 2013;21:45. 23. Westhall E, Rossetti AO, van Rootselaar AF, et al. Standardized EEG interpretation accurately predicts prognosis after cardiac arrest. Neurology 2016;86:1482-90. 24. Wijdicks EF, Hijdra A, Young GB, Bassetti CL, Wiebe S, Quality Standards Subcommittee of the American Academy of N. Practice parameter: prediction of outcome in comatose survivors after cardiopulmonary resuscitation (an evidence-based review): report of the Quality Standards Subcommittee of the American Academy of Neurology. Neurology 2006;67:203-210. 25. Metter RB, Rittenberger JC, Guyette FX, Callaway CW. Association between a quantitative CT scan measure of brain edema and outcome after cardiac arrest. Resuscitation 2011;82:1180-1185. 26. Scheel M, Storm C, Gentsch A, et al. The prognostic value of gray-white-matter ratio in cardiac arrest patients treated with hypothermia. Scandinavian journal of trauma, resuscitation and emergency medicine 2013;21:23.

Ac ce pt e

561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608

28 Page 28 of 29

d

M

an

us

cr

ip t

27. Golan E, Barrett K, Alali AS, et al. Predicting neurologic outcome after targeted temperature management for cardiac arrest: systematic review and meta-analysis. Critical care medicine 2014;42:1919-1930. 28. Dragancea I, Horn J, Kuiper M, et al. Neurological prognostication after cardiac arrest and targeted temperature management 33 degrees C versus 36 degrees C: Results from a randomised controlled clinical trial. Resuscitation 2015;93:164-170. 29. Geurts M, Macleod MR, van Thiel GJ, van Gijn J, Kappelle LJ, van der Worp HB. Endof-life decisions in patients with severe acute brain injury. The Lancet Neurology 2014;13:515-524. 30. Kim YJ, Ahn S, Sohn CH, et al. Long-term neurological outcomes in patients after out-ofhospital cardiac arrest. Resuscitation 2016;101:1-5. 31. Bouwes A, Binnekade JM, Kuiper MA, et al. Prognosis of coma after therapeutic hypothermia: a prospective cohort study. Annals of neurology 2012;71:206-212. 32. Ridley S, Fisher M. Uncertainty in end-of-life care. Current opinion in critical care 2013;19:642-647.

Ac ce pt e

609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624

29 Page 29 of 29