Article, Neurology

Serum potassium concentration predicts brain hypoxia on CT after avalanche-induced cardiac arrest

a b s t r a c t

Background: Brain anoxia after complete avalanche burial and cardiac arrest may occur despite adequate on- site triage.

Purpose: To investigate clinical and biological parameters associated with brain hypoxia in a cohort of avalanche victims with whole body computed tomographic (CT) scan.

Methods: Retrospective study of patients with CA and whole body CT scan following complete avalanche burial admitted in a level-I trauma center.

Main findings: Out of 19 buried patients with whole body CT scan, eight patients had refractory CA and 11 patients had pre-hospital return of spontaneous circulation. Six patients survived at hospital discharge and only two had Good neurologic outcome. Twelve patients had signs of brain hypoxia on initial CT scan, defined as Brain edema, loss of gray/white matter differentiation and/or hypodensity of basal ganglia. No clinical pre-hospital parameter was associated with brain anoxia. Serum potassium concentration at admission was higher in patients with brain anoxia as compared to patients with Normal CT scan: 5.5 (4.1-7.2) mmol/L versus 3.3 (3.0-4.2) mmol/L, respectively (P b .01). A threshold of 4.35 mmol/L serum potassium had 100% specificity to predict brain anoxia on Brain CT scan.

Conclusions: Serum potassium concentration had good predictive value for brain anoxia after complete avalanche burial. This finding further supports the use of serum potassium concentration for extracorporeal life support insertion at hospital admission in this context.

(C) 2016

Introduction

Mortality rate after complete avalanche burial is around 52% [1] due to asphyxia, severe trauma and/or deep hypothermia [2,3]. Asphyxia and severe trauma are the leading cause of Cardiac arrest in this context and are associated with Poor neurologic outcome [2-5]. Conversely, CA due to accidental hypothermia is rare but may confer ideal condition for successful neurological recovery despite prolonged avalanche burial [6,7]. On-scene triage of avalanche victims with CA aims at identifying patients with isolated accidental hypothermia to be resuscitated until rewarming. Triage algorithms are based upon

* Corresponding author at: Pole d’Anesthesie-Reanimation, Hopital Albert Michallon, BP 217, F-38043, Grenoble, France. Tel.: +33 4 76 76 92 88; fax: +33 4 76 76 51 83.

E-mail address: [email protected] (P. Bouzat).

duration of burial, airway conditions, body core temperature, initial car- diac activity and reported signs of life at extrication [8]. After on-scene triage, updated recommendations for Extracorporeal life support relies on body core temperature lower than 30 ?C, duration of burial longer than 60 minutes, no severe trauma and serum potassium concentration lower than 8 mmol/L at hospital admission [9]. Despite adequate adherence to algorithms, patients with brain anoxia are still admitted to the Emergency Department (ED), which challenges the usefulness of the applied criteria.

Only limited data are available regarding CT scan findings after com- plete avalanche burial and mainly focus on traumatic injuries [10,11]. However, signs of brain anoxia on cerebral CT scan would be relevant to further explore the association between clinical/biological parame- ters and Neurological prognosis. Moreover, snow aspiration signs on thoracic CT scan maybe also helpful to test the relevancy of airway pat- tern assessment in the field. As whole body imaging was performed in

http://dx.doi.org/10.1016/j.ajem.2016.01.037

0735-6757/(C) 2016

our center to assess Associated injuries and potential brain anoxia in these patients, we decided to test whether CT signs of brain hypoxia were associated with clinical and biological parameters used for triage of avalanche victims with CA [12,13].

Methods

We retrospectively studied consecutive avalanche patients admitted in one level-I trauma center (Grenoble University Hospital, Northern French Alps, France) from 2002 to 2014. Patients with the following criteria were finally included: (1) complete avalanche burial, defined as a burial concerning at least head and chest (2) on-scene cardiac arrest, and (3) whole-body CT scan including cerebral, cervical, thoracic, abdominal and pelvic regions. Exclusion criteria were: (1) partial burial

(2) no CT scan, and (3) post-mortem CT scan. All these patients were part of a larger cohort used for survival assessment after avalanche induced cardiac arrest [14]. The Regional Institutional Ethics Committee approved the study design (Comite d’Ethique des Centers d’Investigation Clinique de l’inter-region Rhone-Alpes-Auvergne, IRB number 5891, approval on September 1, 2014).

Pre-hospital triage was done on-site by emergency physicians using recommendations for avalanche victims’ management edited before 2014 [1,3]. Briefly, following CA, patients were transported to the ED if

(1) return of spontaneous circulation (ROSC) was obtained in the field

or (2) refractory cardiac arrest by isolated accidental hypothermia was suspected: duration of burial longer than 35 minutes and body core temperature lower than 32 ?C with any signs of life reported by wit- nesses prior to CA, Cardiac activity at extrication including ventricular fi- brillation, or asystole with patent airway. At hospital admission, ECLS was inserted in patients with serum potassium concentration lower than 12 mmol/L associated with body core temperature (measured by esophageal probe) lower than 32 ?C and no obvious signs of trauma.

Clinical and biological data were extracted from the registry of the Trauma system of the Northern French Alps and completed using pa- tients’ files if necessary. Pre-hospital data included duration of burial, duration of no flow and low flow, body core temperature measured by esophageal probe, initial cardiac rhythm, signs of life preceding CA, air- way patency (air pocket, airway obstruction) and signs of severe trau- ma. Biological data were collected upon hospital admission on a central line: arterial blood gases, serum potassium concentration, serum lactate concentration, coagulation parameters (activated partial thromboplastin ratio, prothrombin ratio, fibrinogen, platelets count) and hemoglobin level. Injury Severity Score was also recorded. Survival was reported at hospital discharge and neurologic outcome of survivors was assessed using Cerebral Performance Category at 3 months.

Whole-body CT scans were done under the supervision of intensivists. The aim of whole-body imaging was to assess associated in- juries and potential brain anoxia. CT scans were conducted using Sie- mens Sensation 16 (Siemens, Erlangen, Germany) or Philips Brilliance 40 and 64 (Philips Medical Systems, Eindhoven, The Netherlands). Whole body CT protocol consisted of the following acquisitions: (1) a non-enhanced encephalic CT scan, (2) a non-enhanced CT scan of the neck, from the base of the skull to the level of the second thoracic verte- bra, (3) a contrast-enhanced CT scan of the thorax, abdomen and pelvic regions from the level of the sixth cervical vertebra to the lesser tro- chanter, (4) a contrast-enhanced scan of the abdomen from the dia- phragmatic dome to the lesser trochanter. Arms were placed above the head after the CT scan of the head and neck. Optionally, mainly when a severe cervical traumatism was suspected, a CT angiography of arterial supra-aortic vessels was added. Regarding contrast-medium injection, a 120-mL bolus of iso-osmolar, non-ionic iodinated Contrast material [350 mg iodine/ml, Iohexol (Omnipaque 350; GE Healthcare)] followed by a saline flush of 40 mL was injected into an antecubital vein at a flow rate of 4 mL/s. The data acquisition was initiated 6 seconds after 100 Hounsfield units (HU) attenuation in the descending thoracic aorta. After a further delay of 45 s, the abdominal portal-venous

enhanced phase was acquired. For supra-aortic vessel acquisition, an additional injection of 120 mL of the same contrast medium was done, and acquisition was triggered at 75 HU attenuation in Ascending aorta. Three trained radiologists reviewed all whole body CT scans. In- volved physicians included one neuroradiologist, one chest radiologist, and one general radiologist. For each patient, radiologists interpreted traumatic and non-traumatic lesions according to the methodology de- scribed in Supplemental file no. 1. Brain hypoxia was defined as brain edema, loss of white/gray matter differentiation, and/or hypodensity of basal ganglia. In order to further study radiologic pattern of snow as- piration, and since there is no gold standard for this condition, we pro- posed defining snow aspiration using the following methodology. First we identified all patients with lung parenchyma abnormalities. We ex- cluded those with isolated gravity-related lung opacities. We checked CT scans of remaining patients for signs of Thoracic traumatic injuries, which were defined as fractures of ribs, sternum, clavicles, thoracic ver- tebrae and the presence of a pneumo- or hemothorax. Traumatic inju- ries associated with resuscitation were defined as isolated fractures of anterolateral ribs and/or sternum in patients with known cardiac resus- citation [15]. Lung opacities with no associated signs of thoracic trauma (not including traumatic injuries due to resuscitation) were considered as snow aspiration. Traumatic injuries in the whole body were also an-

alyzed using the CT features described in Supplemental file no. 1.

Data were expressed as median and 25th to 75th percentiles. Cate- gorical variables were compared using the Fisher exact test for two- by-two tables and the Freeman-Halton extension for 2 x 3 tables. Con- tinuous variables were compared using Wilcoxon rank sum test. Multi- variate analysis was not conducted given the low number of patients. The properties of serum potassium concentration on admission for brain hypoxia prediction were also tested using receiver operating curve (ROC) analysis. Maximization of the Youden index [16] was used to determine the best threshold. Density function was employed to generate a smooth kernel density ROC curve [17]. 95% confidence in- tervals of the areas under the curve were yielded using Delong method for the empirical curve [18] and stratified bootstrap from 1000 repli- cates for the smooth curve [19]. Statistical analysis was done with Stata 12 software. P b .05 was declared statically significant.

Results

Thirty-nine patients with whole-body imaging were admitted to the ED following avalanche burial within the study period. Only 19 patients met the inclusion criteria for the final analysis (see flow chart in Fig. 1). Pre-hospital characteristics of the study population are shown in Table 1. The median duration time of cardiopulmonary resuscitation was 70 minutes (15-420 minutes). Eight patients were transferred to the ED with refractory CA and 11 patients had pre-hospital ROSC. All pa- tients with refractory CA received ECLS at hospital admission. Eight pa- tients had traumatic injuries but only 2 patients presented Injury Severity Score higher than 15. Detail of injuries is described in Supple- mental file no. 2. Biological parameters are reported in Table 2. Six pa- tients (32%) survived at hospital discharge. CPC score was 1 (no disability) for 2 patients, 3 (severe disability) for 1 patient and 4 (vege- tative state) for 3 patients. The 2 patients with good neurologic outcome had initial pulseless electrical activity at extrication before refractory CA and ECLS implantation. All patients with initial asystole and refractory CA did not survive (n = 6 patients). In the Non-survivor patients (n = 13 patients), causes of death were brain death for 11 patients (85%) and multiple organ failure for 2 patients (15%).

Out of 19 patients, 12 (63%) patients had signs of brain hypoxia on cerebral imaging. One patient had isolated brain edema whereas 11 pa- tients had diffuse edema associated with loss of gray/white matter dif- ferentiation. Univariate analysis between patients with brain hypoxia (n = 12 patients) and patients with normal cerebral CT scan (n = 7 pa- tients) is presented in Table 3. No clinical prehospital parameter was

Fig. 1. Flow chart of the study population.

associated with brain hypoxia on CT scan. For instance, only 1 patient out of 12 patients with brain hypoxia presented with an airway obstruc- tion in the field. Snow aspiration assessed by lung imaging was also not different between the 2 groups. Regarding biological data, serum potassium concentration was higher in patients with brain hypoxia as compared to normal CT scan patients: 5.5 (4.1-7.2) mmol/L vs 3.3 (3.0-4.2) mmol/L, respectively (P b .01). Patients with brain hypoxia had also more coagulation disorders as compared to the normal CT scan patients.

ROC curve analysis to predict brain hypoxia with serum potassium concentration at hospital admission is represented in Fig. 2. Area under the empirical ROC curve was 0.91 (95% CI 0.78-1.00). The thresh- old of 4.35 mmol/L had a sensitivity of 67% and specificity of 100% for

Table 1

Pre-hospital characteristics of the 19 studied patients?

Variables Values

Age (years) 36 (23-41)

Male gender (number)? 16

Prehospital body core temperature (?C)? 27.9 (26.0-30.2)

Burial time (min)? 30 (20-45)

Maximum duration of no flow (min)? 25 (20-40)

Duration of low flow (min)? 20 (10-110) Prehospital cardiac rhythm (number):

Asystole 15

Ventricular fibrillation 1

Pulseless electrical activity 3

Return of spontaneous circulation (number) 11

Transport under CPR (number) 8

Vital signs preceding CA (number) 3

Prehospital trauma signs (number) 4

Prehospital airway obstruction (number) 1

Air pocket (number) 3

* Values are expressed as median and 25-75th percentiles. CPR, cardiopulmonary re- suscitation; CA, cardiac arrest.

the diagnosis of brain hypoxia. Although jagged, empirical ROC curve was consistent with the smooth kernel density ROC curve and its bootstrapped 95% confidence interval, ie, 0.88 (95% CI 0.78-0.93).

Discussion

In this series of avalanche victims with CA and whole body CT scan, clinical parameters in the field were not associated with brain hypoxia on cerebral CT scan. At hospital admission, serum potassium concentra- tion was found to be a good predictor of brain hypoxia. The threshold of

4.35 mmol/L reached 100% specificity. These findings highlighted the role of serum potassium concentration on admission to help clinical de- cision for ECLS insertion at hospital.

Overall mortality in our study was 68% and was mainly related to global cerebral ischemic injuries occurring in 12 out of 19 patients. This rate was higher than previously reported mortality. For instance, in a cohort of 1886 Swiss avalanche victims, overall mortality was 23%. However, this proportion rose to 52.4% in patients with complete burial [1]. In our cohort, elevated mortality maybe related to specific se- lection of patients with CA and complete avalanche burial. This finding was in line with recent publication focusing on CA after avalanche burial [20]. Brain hypoxia was the leading cause of mortality in our cohort. Loss of gray/white matter differentiation was the prominent cerebral imag- ing pattern, also known as the “loss of boundary” sign. This specific ce- rebral finding has been described after CA and was found to be strongly associated with time to ROSC. Indeed, the incidence of this pattern was 23% when ROSC occurred within 20 minutes and increased up to 83% after 20 minutes [21]. Interestingly, signs of cerebral edema on CT scan were frequently found after CA related to asphyxia [22]. In a cohort of 51 patients with out-of-hospital CA, loss of gray/white differentiation was also associated with poor neurologic outcome [23]. Taken together, these findings confirmed the high proportion of asphyxia in our series resulting in poor outcome.

Clinical pre-hospital parameters used for on-scene triage were not associated with brain hypoxia. This finding underlined the poor predic- tive value of each clinical sign to rule out asphyxia after complete ava- lanche burial and CA. This concept was further supported by international guidelines, which recommended transferring patients to hospital when deep hypothermia was strongly suspected in the field re- gardless of pre-hospital clinical parameters [3]. No association was also found between snow aspiration on lung CT scan and brain hypoxia. In- deed, asphyxia may be more related to rebreathing of expired CO2 due to limited or no air pocket rather than acute airway obstruction by snow and/or ice mask. We also believe that snow aspiration maybe underestimated on admission CT since snow water is hypotonic and gets quickly absorbed through pulmonary and systemic circulation, similarly to fresh water drowning [24].

Table 2

Biological data of the 19 studied patients on hospital admission

Variables Values

Serum potassium (mmol/L)

4.2 (3.9-5.7)

serum sodium (mmol/L)

138 (137-143)

arterial pH

7 (6.6-7.2)

PaCO2 (mmHg)

40 (32-60)

PaO2 (mmHg)

361 (150-558)

serum bicarbonate (mmol/L)

10 (5-14)

Serum lactate (mmol/L)

11.8 (8.1-14.2)

Hemoglobin (g/L)

146 (128-160)

Platelets (G/L)

154 (121-204)

Activated Partial Thromboplastin ratio

1.82 (1.31-4.1)

Prothrombin ratio (%)

52 (30-61)

Serum fibrinogen (g/L)

1.6 (0.8-1.9)

Serum creatinine (umol/L)

140 (118-184)

Serum glucose (mmol/L)

15.1 (11.3-18.6)

Values are expressed as median and 25-75th percentiles.

Table 3

Univariate analysis between patients with brain hypoxia on cerebral CT scan (n = 12 patients) and patients with normal cerebral CT scan (n = 7 patients).

Variable

No brain hypoxia (n = 7 patients)

Brain hypoxia (n = 12 patients)

p-value

Burial time (min)

35 (22-360)

27.5 (20-42.5)

0.42

Core temperature (?C)

26.5 (23.3-30.1)

28.0 (26.8-32.0)

0.24

Signs of life, n

2

1

0.52

Airway obstruction, n

0

1

1

Air pocket, n

2

1

0.52

Prehospital cardiac rhythm

0.70

Asystole, n

5

10

Ventricular fibrillation, n

0

1

Pulseless electrical activity, n

2

1

prehospital ROSC, n

5

6

0.63

ECLS, n

2

6

0.63

Serum potassium (mmol/L)

3.3 (3.0-4.2)

5.5 (4.1-7.2)

b0.01

pH

7.2 (6.8-7.2)

6.9 (6.5-7.1)

0.19

PaO2 (mmHg)

432 (334-664)

275 (64-486)

0.18

PaCO2 (mmHg)

36 (35-50)

53 (31-133)

0.31

Bicarbonate (mmol/L)

15 (5-18)

10 (7-13)

0.15

Lactate (mmol/L)

8.1 (6.4-21.3)

12.6 (9.8-14.2)

0.18

Hemoglobin (g/dl)

146 (124-158)

148 (133-160)

0.61

Platelet (G/L)

166 (132-214)

143.5 (120.0-184.5)

0.29

APT ratio

1.3 (0.8-1.6)

2.8 (1.7-5.2)

0.03

PR (%)

58 (55-87)

41 (28-56)

0.01

Fibrinogen (g/L)

1.7 (1.6-2.7)

1.3 (0.7-1.8)

0.05

CPC 1-2, n

2

0

0.12

Death, n

1

12

b0.001

Snow aspiration imaging, n

5

9

1

Values are expressed as median and 25-75th percentiles. ROSC, return of spontaneous circulation; ECLS, extracorporeal life support; APT, Activated Partial Thromboplastin ratio; PR, pro- thrombin ratio; CPC, cerebral performance category.

Regarding biological parameters, serum potassium concentration on hospital admission was associated with brain hypoxic damage in our co- hort. This biological parameter also showed good AUC to predict brain hypoxia. Serum potassium concentration has been extensively studied in avalanche victims for in-hospital triage in retrospective case- control studies [25-27] and case reports [6,28-30]. In a retrospective study of 32 hypothermic avalanche victims, serum potassium at

Fig. 2. Empirical receiving operator curve (ROC, black line) and smooth kernel density ROC curve (dash line) of serum potassium concentration to predict brain hypoxia. Serum po- tassium concentration was measured at hospital admission on a central line. Brain hypoxia was defined by experienced neuroradiologist as isolated brain edema, loss of white/gray matter differentiation and/or hypodensity of basal ganglia.

hospital admission was lower in survivors compared to non survivors (4.25 +- 4.9 mmol/L vs 9.95 +- 4.9 mmol/L, respectively). The highest ad- mission serum potassium associated with survival after avalanche- induced cardiac arrest was 6.4 mmol/L but no information was given re- garding neurologic outcome [25]. To our knowledge, our study was the first to demonstrate an association between serum potassium concen- tration and cerebral CT signs of hypoxia. Guidelines for avalanche vic- tims’ management now recommend stopping resuscitation in patients with refractory CA when serum potassium concentration is higher than 8 mmol/L [9]. ROC analysis in our study revealed that the threshold for irreversible brain damage was lower. Indeed, serum potassium con- centration of 4.35 mmol/L had 100% specificity to diagnose brain ische- mia on CT scan. This result supported the lowering of serum potassium cut-off for in-hospital triage. Nevertheless, the small sample size of our study did not allow us to draw firm conclusion regarding this threshold. We also found Coagulation disorders to be associated with brain hypox- ia. Interestingly, initial coagulopathy was associated with poor neuro- logic outcome in a cohort of 252 patients with out-of-hospital CA [31]. However, coagulation assessment could be challenging at the bedside and the role of coagulopathy for in-hospital triage deserves further exploration.

We acknowledge several limits to our study. This work was a small sample size retrospective study and the present case series illustrated Rare conditions since only hemodynamically stable avalanche victims were transported to CT scan. The limited number of patients in each cat- egory should temper our conclusions down. Accordingly, the serum po- tassium threshold should not be generalized for every avalanche patient. However, performing a large study is challenging due to low in- cidence of complete avalanche burial and cardiac arrest. Data regarding whole body imaging after avalanche burial are scarce and this study adds further insights about factors associated with brain hypoxia in this context. The implementation of an international registry will over- come this limitation in future studies.

In conclusion, serum potassium concentration had good predictive value for brain hypoxia prediction after complete avalanche burial and cardiac arrest. This biological parameter is a key element for in- hospital triage to insert ECLS after refractory CA. Pre-hospital clinical pa- rameters were not associated with brain hypoxia in our cohort, which

challenged the relevancy of such parameters to predict cerebral ischemic damages.

Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.ajem.2016.01.037.

Conflict of interest

None.

Acknowledgments

None.

References

  1. Brugger H, Durrer B, Adler-Kastner L, Falk M, Tschirky F. Field management of ava- lanche victims. Resuscitation 2001;51(1):7-15.
  2. Boyd J, Haegeli P, Abu-Laban RB, Shuster M, Butt JC. Patterns of death among ava- lanche fatalities: a 21-year review. CMAJ 2009;180(5):507-12.
  3. Brugger H, Durrer B, Elsensohn F, Paal P, Strapazzon G, Winterberger E, et al. Resus- citation of avalanche victims: Evidence-based guidelines of the international com- mission for mountain emergency medicine (ICAR MEDCOM): intended for physicians and other advanced life support personnel. Resuscitation 2013;84(5): 539-46.
  4. Hohlrieder M, Brugger H, Schubert HM, Pavlic M, Ellerton J, Mair P. Pattern and se- verity of injury in avalanche victims. High Alt Med Biol 2007;8(1):56-61.
  5. McIntosh SE, Grissom CK, Olivares CR, Kim HS, Tremper B. Cause of death in ava- lanche fatalities. Wilderness Environ Med 2007;18(4):293-7.
  6. Boue Y, Lavolaine J, Bouzat P, Matraxia S, Chavanon O, Payen JF. Neurologic recovery from profound accidental hypothermia after 5 hours of cardiopulmonary resuscita- tion. Crit Care Med 2014;42(2):e167-70.
  7. Boue Y, Payen JF, Torres JP, Blancher M, Bouzat P. Full neurologic recovery after prolonged avalanche burial and cardiac arrest. High Alt Med Biol 2014;15(4):522-3.
  8. Kottmann A, Blancher M, Spichiger T, Elsensohn F, Letang D, Boyd J, et al. The ava- lanche victim resuscitation checklist, a new concept for the management of ava- lanche victims. Resuscitation 2015;91:e7-8.
  9. Truhlar A, Deakin CD, Soar J, Khalifa GE, Alfonzo A, Bierens JJ, et al. European Resus- citation Council guidelines for resuscitation 2015: section 4. cardiac arrest in special circumstances. Resuscitation 2015;95:148-201.
  10. Grosse AB, Grosse CA, Steinbach LS, Zimmermann H, Anderson S. Imaging findings of avalanche victims. Skeletal Radiol 2007;36(6):515-21.
  11. Wick MC, Weiss RJ, Hohlrieder M, Tecklenburg K, Jaschke W, Rieger M. Radiological aspects of injuries of avalanche victims. Injury 2009;40(1):93-8.
  12. Boyd J, Brugger H, Shuster M. Prognostic factors in avalanche resuscitation: a sys- tematic review. Resuscitation 2010;81(6):645-52.
  13. Mair P, Brugger H, Mair B, Moroder L, Ruttmann E. Is extracorporeal rewarming in- dicated in avalanche victims with unwitnessed hypothermic cardiorespiratory ar- rest? High Alt Med Biol 2014;15(4):500-3.
  14. Boue Y, Payen JF, Brun J, Thomas S, Levrat A, Blancher M, et al. Survival after avalanche-induced cardiac arrest. Resuscitation 2014;85(9):1192-6.
  15. Kashiwagi Y, Sasakawa T, Tampo A, Kawata D, Nishiura T, Kokita N, et al. Computed tomography findings of complications resulting from cardiopulmonary resuscita- tion. Resuscitation 2015;88:86-91.
  16. Youden WJ. Index for rating Diagnostic tests. Cancer 1950;3(1):32-5.
  17. Zou KH, Hall WJ, Shapiro DE. Smooth non-parametric receiver operating characteristic (ROC) curves for continuous diagnostic tests. Stat Med 1997;16(19): 2143-56.
  18. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Bio- metrics 1988;44(3):837-45.
  19. Carpenter J, Bithell J. Bootstrap confidence intervals: when, which, what? A practical guide for medical statisticians. Stat Med 2000;19(9):1141-64.
  20. Moroder L, Mair B, Brugger H, Voelckel W, Mair P. Outcome of avalanche victims with out-of-hospital cardiac arrest. Resuscitation 2015;89:114-8.
  21. Inamasu J, Miyatake S, Suzuki M, Nakatsukasa M, Tomioka H, Honda M, et al. Early CT signs in out-of-hospital cardiac arrest survivors: temporal profile and prognostic significance. Resuscitation 2010;81(5):534-8.
  22. Morimoto Y, Kemmotsu O, Kitami K, Matsubara I, Tedo I. Acute Brain swelling after out-of-hospital cardiac arrest: pathogenesis and outcome. Crit Care Med 1993; 21(1):104-10.
  23. Kim SH, Choi SP, Park KN, Youn CS, Oh SH, Choi SM. Early brain computed tomogra- phy findings are associated with outcome in patients treated with therapeutic hypo- thermia after out-of-hospital cardiac arrest. Scand J Trauma Resusc Emerg Med 2013;21:57.
  24. Christe A, Aghayev E, Jackowski C, Thali MJ, Vock P. Drowning-post-mortem imag- ing findings by computed tomography. Eur Radiol 2008;18(2):283-90.
  25. Locher T, Walpoth BH. Differential diagnosis of circulatory failure in hypothermic av-

    alanche victims: retrospective analysis of 32 avalanche accidents. Praxis (Bern 1994) 1996;85(41):1275-82.

    Mair P, Kornberger E, Furtwaengler W, Balogh D, Antretter H. Prognostic markers in patients with severe accidental hypothermia and cardiocirculatory arrest. Resuscita- tion 1994;27(1):47-54.

  26. Schaller MD, Fischer AP, Perret CH. Hyperkalemia. A prognostic factor during acute severe hypothermia. JAMA 1990;264(14):1842-5.
  27. Dobson JA, Burgess JJ. Resuscitation of severe hypothermia by extracorporeal rewarming in a child. J Trauma 1996;40(3):483-5.
  28. Oberhammer R, Beikircher W, Hormann C, Lorenz I, Pycha R, Adler-Kastner L, et al. Full recovery of an avalanche victim with profound hypothermia and prolonged cardiac arrest treated by extracorporeal re-warming. Resuscitation 2008;76(3): 474-80.
  29. von Segesser LK, Garcia E, Turina M. Perfusion without systemic heparinization for rewarming in accidental hypothermia. Ann Thorac Surg 1991;52(3):560-1.
  30. Kim J, Kim K, Lee JH, Jo YH, Kim T, Rhee JE, et al. Prognostic implication of initial co- agulopathy in out-of-hospital cardiac arrest. Resuscitation 2013;84(1):48-53.