a b s t r a c t

Background: Targeted temperature management post-cardiac arrest is currently implemented using various methods, broadly categorized as internal and external. This study aimed to evaluate survival-to-hospital dis- charge and neurological outcomes (Glasgow-Pittsburgh Score) of post-cardiac arrest patients undergoing inter- nal cooling verses external cooling.

Methodology: A randomized controlled trial of post-resuscitation cardiac arrest patients was conducted from Oc- tober 2008-September 2014. Patients were randomized to either internal or external cooling methods. historical controls were selected matched by age and gender. Analysis using SPSS version 21.0 presented descriptive statis- tics and frequencies while univariate logistic regression was done using R 3.1.3.

Results: 23 patients were randomized to internal cooling and 22 patients to external cooling and 42 matched con- trols were selected. No significant difference was seen between internal and external cooling in terms of survival, neurological outcomes and complications. However in the internal cooling arm, there was lower risk of develop- ing overcooling (p = 0.01) and rebound hyperthermia (p = 0.02). Compared to normothermia, internal cooling had higher survival (OR = 3.36, 95% CI = (1.130, 10.412), and lower risk of developing cardiac arrhythmias (OR

= 0.18, 95% CI = (0.04, 0.63)). Subgroup analysis showed those with cardiac cause of arrest (OR = 4.29, 95% CI = (1.26, 15.80)) and sustained ROSC (OR = 5.50, 95% CI = (1.64, 20.39)) had better survival with internal cooling compared to normothermia. Cooling curves showed tighter Temperature control for internal compared to exter- nal cooling.

Conclusion: Internal cooling showed tighter temperature control compared to external cooling. Internal cooling can potentially provide better survival-to-hospital discharge outcomes and reduce cardiac arrhythmia complica- tions in carefully selected patients as compared to normothermia.

(C) 2017

Abbreviations: TTM, targeted temperature management; OHCA, out-of-hospital cardiac arrest; ROSC, return of spontaneous circulation; ED, emergency department; MICU, medical intensive care unit; CCU, coronary care unit; R2TH, ROSC to initiation of TTM; R2TT, ROSC to target temperature; T2TT, time from initiation of TTM to target temperature; CPR, cardiopulmonary resuscitation; PEA, pulseless electrical activity; VF/ VT, Ventricular fibrillation/ventricular tachycardia; OR, odds ratio; CI, confidence interval; PCI, percutaneous coronary intervention; AICD, automatic Implantable cardioverter defibrillator; CPC, cerebral performance category; OPC, overall performance category; IQR, interquartile range.

* Corresponding author at: c/o Department of Emergency Medicine, Singapore General Hospital, Outram Road, 169608, Singapore.

E-mail address: [email protected] (M.E.H. Ong).

Introduction

In the USA, the incidence of out-of-hospital cardiac arrest (OHCA) is estimated at 1.89/1000 person-years [1], with an estimated 400- 460,000 people that die every year from OHCA [2]. In Singapore, over 1800 people suffer out-of-hospital cardiac arrest (OHCA) every year and the average chance of survival is a dismal 2.7% [3]. Survival rates in other Asia-Pacific countries are also low ranging from 0.5% in Malay- sia to 8.5% in Korea [4]. Targeted temperature management is a treat- ment recommended by the American Heart Association as part of post-Cardiac arrest management guidelines [5]. The guidelines include

http://dx.doi.org/10.1016/j.ajem.2017.07.017

0735-6757/(C) 2017

achieving target temperature of 32-36 ?C in as short as possible a time and maintained over 24 h. This recommendation stemmed from 2002 randomized controlled trials that showed decreased mortality and im- proved neurological outcomes in comatose patients after successful re- suscitation with TTM compared to Standard therapy [6,7]. TTM has been shown to decrease inflammation, reduce cerebral, myocardial and cellu- lar metabolism and promote heart epicardial flow [8]. Proposed mecha- nisms for the protective effect of hypothermia include reduced cerebral Oxygen metabolism and ischemia [9-13], suppression of reperfusion in- jury [11,14-16] and by suppression of Inflammatory mediators [17,18]. However a recent trial involving OHCA with TTM of 33 ?C compared to 36 ?C showed no difference in outcomes [19]. The main difference in this trial and the 2002 trials was in the active prevention of hyperther- mia in the control group. This offers a possible interpretation that in temperature management post-cardiac arrest, prevention of hyperther-

mia may be more important than an induced hypothermia.

Published results of randomized trials comparing between methods of TTM after cardiac arrest in terms of mortality and neurological mor- bidity outcomes are lacking. Invasive/internal cooling methods such as endovascular cooling catheter have been proposed to have better tem- perature control during the therapy compared to external cooling methods [20-21]. There have been other observational studies between internal and external cooling which did not find any difference in neu- rological outcomes and further suggested that it might be worth looking at rewarming phase as contributor to outcome [21,22]. In this study, we aimed to evaluate survival-to-hospital discharge and neurological out- comes (Glasgow-Pittsburgh Score) of post-cardiac arrest patients un- dergoing TTM with internal cooling verses external cooling. The secondary objective was to compare survival outcomes for TTM (either method) and normothermia (historical controls). Other secondary out- comes included looking at each TTM method’s performance and complications.

Methods

The study was designed as a single-centre phased, prospective, clin- ical study with a nested, randomized controlled comparison between external and internal cooling for post cardiac patients. The patients in the intervention arm were randomized in 1:1 ratio to receive either in- ternal cooling using an endovascular catheter or external cooling using gel pads with water-based circulating system.

Study participants were consecutive patients who suffered out-of- hospital or in-hospital cardiac arrest achieving return of spontaneous circulation (ROSC) for N 30 min, admitted to one of two intensive care units capable of providing TTM. Other inclusion criteria included pa- tients aged between 18 and 80 years old who remained comatose or un- responsive post-resuscitation but otherwise hemodynamically stable (systolic blood pressure above 90 mm Hg with or without inotropic support).

Exclusion criteria included patients who had Traumatic causes of ar- rest, cardiac arrest due to intracranial hemorrhage, remained hemody- namically unstable despite fluid and/or vasopressor support, female aged below 50 years with positive pregnancy test or patients with known poor premorbid status (bedbound and uncommunicative).

Patients were recruited from the Emergency Department (ED), Cor- onary Care Unit (CCU) and Medical Intensive Care Unit (MICU) from Oc- tober 2008 to September 2014. Informed consent was obtained from all patients enrolled in the prospective interventional trial.

25 retrospective (2006-2008) and 17 contemporary controls (2012-2014) were chosen from cardiac arrest patients who did not re- ceive TTM (they received conventional normothermia treatment), but who would otherwise have met all criteria for TTM treatment. Besides the retrospective controls, the contemporary controls had not been en- rolled into the prospective trial either because they had not been iden- tified for enrollment or they had been admitted into an overflow ICU

where nursing staff were unable to support TTM. These controls were matched by age and gender to intervention cases.

The protocol and consent procedures were approved by the ethics committee of a public tertiary hospital and also registered under ClinicalTrials.gov registry (NCT00827957). Delayed consent had to be sought from relatives as participants enrolled were in a life-threatening situation, unconscious and unable to provide consent for trial enrollment. Post-resuscitated cardiac arrest patients that met all the inclusion criteria in the ED were randomized to either TTM arm by sealed opaque envelopes. ED physicians would then initiate hypothermia treatment by cold saline infusion and cold ice packs around the patient’s groin and ax- illa. A Foley or esophageal temperature probe was also inserted to mon- itor and chart temperature. Cooling treatment by either surface or internal cooling device was then continued at the MICU or CCU. Consent was taken from relatives either at the ED if they were present or subse- quently at the MICU/CCU. Eligible post-resuscitation patients from MICU/CCU after in-hospital cardiac arrest were also randomized to ei-

ther arm of therapy by contacting the study coordinator.

Prior to commencement of the therapy, patients were first sedated and paralyzed to avoid shivering and then crash-cooled to target tem- perature of 34 ?C. After which, they were maintained at that tempera- ture for 24 h before rewarming passively at 1 ?C every 4 h (0.25 ?C/h) to 36.5 ?C.

Internal cooling was achieved with the Alsius Thermogard XP(TM) In- travascular Temperature Management System (ZOLL Medical Corp), which requires a triple lumen catheter to be inserted into the central ve- nous system of the patient via the femoral approach. The thermal regu- lation system controls the temperature of saline and circulates cool or warm saline through the catheter balloons without infusing saline into the patient’s body. External cooling and rewarming was achieved with the Arctic Sun(R) 2000 Temperature Management System (Medivance Pte Ltd.) which uses ArcticGel gel-coated pads to maintain contact with the patient’s skin throughout the treatment.

Only physicians with experience and skills in central venous cannu- lation were allowed to insert the internal cooling catheter. Only MICU/ CCU physicians and nurses who were trained in the usage of both tem- perature management systems by completing the manufacturer’s train- ing program and who were familiar with the protocol were allowed to use the devices.

Definition of outcomes and other variables collected followed the Utstein recommendations for reporting [23-25]. The primary outcome assessed in this study was survival-to-hospital discharge and neurolog- ical outcomes. Survival to hospital discharge was defined as patient sur- viving the primary event and discharged from the hospital alive. Return of spontaneous circulation (ROSC) was defined as the presence of any palpable pulse, which was detected by manual palpation of a major ar- tery. neurological status on discharge was assessed using the Glasgow- Pittsburgh outcome categories to evaluate quality of life after successful resuscitation. The Cerebral Performance Categories evaluate only the ce- rebral performance capabilities while the overall performance catego- ries reflect both cerebral and non-cerebral status.

The primary outcome of the study was survival-to-hospital dis- charge and neurological outcomes on discharge between the internal cooling arm and external cooling arm. The secondary objective was to compare Survival outcomes for TTM (either method) and normother- mia (historical controls). Other secondary outcomes included looking at each TTM method’s performance and complications. Time and tem- perature data were collected in the data-logger of each TTM device by a bladder-temperature probe (Foley catheter temperature sensor, Smiths Medical) or esophageal probe (Smiths Medical). Incidents of overcooling b 33 ?C and undercooling N 34.2 ?C after target temperature was achieved and post-rewarming rebound hyperthermia >=38 ?C were noted. Time data correlated with temperature curves were used to de- termine the time from ROSC to initiation of TTM (R2TH), time from ROSC to target temperature (R2TT) and time from initiation of TTM to target temperature (T2TT).

Tympanic temperature at ROSC, time of first and sustained ROSC to time of therapy initiation, other interventions started (e.g. percutaneous coronary intervention) and complications related to cooling such as co- agulopathies (venous thrombosis), electrolyte abnormalities (potassi- um derangements), cardiac arrhythmias, abnormal brain activity

Table 1

Characteristics of study participants.

Characteristics Control

N = 42

TTM N = 45

External

N = 22

Internal

N = 23

(seizures) and skin injuries were also collected. Data was entered in Microsoft Excel 2010 (version 14) and analyzed	Median age (IQR)		65.4 (56.7, 72.0)	62.8 (54.0, 67.2)	62.0 (55.5, 68.0)
using SPSS version 21.0 (SPSS, Chicago, IL) and R 3.1.3 (www.r-project.	Gender, male (%)		34 (81.0)	19 (86.4)	16 (69.6)
org). Frequency tables and descriptive statistics were calculated for all	Race (%)	Chinese	33 (78.6)	16 (72)	18 (78.3)

variables. Median together with interquartile range (IQR) was reported for continuous variables, while frequency together with percentage was

	Malay	5 (11.9)	4 (18.2)	4 (17.4)
	Indian	3 (7.1)	2 (9.1)	1 (4.3)
Others		1 (2.4)	0 (0.0)	0 (0.0)

reported for categorical variables. Kruskal-Wallis test was carried out to	Past medical	Diabetes mellitus	17 (40.5)	7 (31.8)	10 (43.5)
compare continuous variables among controls, patients with external	history (%)	Hypertension	25 (59.5)	10 (45.5)	14 (60.9)
cooling and those with internal cooling, while Fisher’s exact test was		Dyslipidemia	14 (33.3)	11 (50.0)	9 (39.1)

performed to compare the distribution of categorical data among these groups of patients. Univariable logistic regression was carried out to compare the development of different events (survival, neurolog- ical function, any complications, and any cooling issues) between each pair of groups. Statistical significance was generally set at a two-tailed P value of b 0.05.

		No medical history	8 (19.0)	2 (9.5)	3 (13.0)
	Cardiac arrest	Out-of-hospital	34 (81.0)	17 (77.3)	21 (91.3)
3. Results	location (%)	In-hospital	8 (19.0)	5 (22.7)	2 (8.7)
	Witnessed	Unwitnessed	7 (17.1)a	2 (9.1)	2 (8.7)

Cerebrovascular	4 (9.5)	1 (4.5)	4 (17.4)
accident Cancer	2 (4.8)	2 (9.1)	2 (8.7)
Cardiac conditions	24 (57.1)	10 (45.5)	9 (39.1)
Respiratory	8 (19.0)	5 (22.7)	4 (17.4)
conditions Renal conditions	10 (23.8)	5 (22.7)	4 (17.4)

From October 2008 to September 2014, there were 96 cardiac arrest patients who were admitted to MICU/CCU, of which 26 were excluded because they were found to be ineligible while another 25 cases were eligible but were not randomized/recruited for the interventional trial, eg they had been admitted to an overflow ICU, or the on-duty staff did not randomize/enrolled them. The remaining 45 patients were random-

collapse (%)

collapse

Bystander 23 (56.1)^a

Paramedic	5 (11.9)a	1 (4.5)	1 (4.3)
Hospital staff	6 (14.6)a	5 (22.7)	1 (4.3)
Bystander CPR (%)	11 (26.8)a	4 (18.2)	10 (43.5)
Prehospital defibrillation (%)	9 (22.0)a	7 (31.8)	11 (47.8)
Any defibrillation (%)	19 (46.3)a	11 (50.0)	14 (60.9)
Initial rhythm at Asystole	18 (42.9)	8 (36.4)	6 (26.1)

14 (63.6) 19 (82.6)

ized in 1:1 ratio to either the internal cooling arm or the external cooling arm, resulting in 23 (51.1%) in the former and 22 (48.9%) in the latter. These cases were matched by age and gender to 42 patients controls chosen from cardiac arrest patients from 2006 to 2014 who received conventional normothermia treatment but who would otherwise have met all criteria for TTM treatment.

The characteristics of the study participants are listed in Table 1.

ED (%)

Pulseless electrical activity Ventricular Fibrillation Ventricular tachycardia

Sinus rhythm (ROSC en-route)

14 (33.3) 7 (31.8) 8 (34.8)

3 (7.1) 2 (9.1) 3 (13.0)

1 (2.4) 0 (0.0) 0 (0.0)

5 (11.9) 5 (22.7) 6 (26.1)

There were differences in baseline characteristics between cases and

Others 1 (2.4) 0 (0.0) 0 (0.0)

controls in terms of age, gender, past medical history, witnessed, by- stander cardiopulmonary resuscitation (CPR) and either prehospital and/or ED defibrillation. In the control group, 18 (42.9%) had initial

Median time in minutes to first ROSC (IQR)

Median time in minutes to sustained ROSC (IQR)

24.0 (13.5,

42.0)^c

27.0 (14.5,

43.0)^c

25.0 (15.3,

39.3)

27.0 (19.0,

46.0)^d

26.0 (12.3,

40.3)^e

28.5 (12.3,

50.8)^e

rhythm at the ED of asystole, 14 (33.3%) Pulseless electrical activity and 4 (9.5%) ventricular fibrillation/tachycardia (VF/VT). In the TTM group, between the internal versus the external cooling arm, the initial rhythm at ED were 6 (26.1%) versus 8 (36.4%) for asystole, 8 (34.8%) versus 7 (31.8%) for PEA and 3 (13.0%) versus 2 (9.1%) for VF/VT. Table 2 shows the comparisons of survival-to-hospital discharge and neurological outcomes among controls, internal cooling arm and exter- nal cooling arm. Survival was higher in the internal cooling arm of the intervention group compared to control [11 (47.8%) internal cooling versus 9 (24.1%) control, OR 3.36, 95% CI 1.12-10.11]. However, this

was not seen when assessing neurological outcomes.

There were no significant differences seen in complications of potas- sium electrolyte disturbances or seizures when comparing between control and intervention groups or between both arms of the interven- tion groups. However, when looking at cardiac arrhythmias, there was greater occurrence in control versus internal intervention group [19 (45.2%) control vs 3 (13.0%) internal cooling, odds ratio (OR) 0.18, 95% confidence interval (CI) 0.05-0.71] and overall intervention group [19 (45.2%) control vs 8 (17.8%) intervention, OR 0.26, 95% CI 0.10-0.70].

Table 3 compares the two arms of the intervention group, namely the internal and external cooling methods. There were no significant differences in temperature at ROSC or prior to therapy initiation in both groups. The median time from first R2TH, sustained R2TH, T2TT, first R2TT or sustained R2TT also did not show significant differences. There were more cooling issues seen in external cooling methods com- pared to internal cooling: overcooling b 33 ?C [11 (50.0%) external vs 5

Recurrent ROSC (%) 4 (10.0)^b 4 (19.0)^d 6 (26.1)

PCI performed 11 (26.2) 5 (22.7) 5 (21.7)

AICD implanted 1 (2.4) 0 (0.0) 1 (4.3)

CPR = cardiopulmonary resuscitation, ROSC = Return of Spontaneous Circulation, ED = Emergency Department, PCI = Percutaneous Coronary Intervention, AICD = Automatic Implantable Cardioverter Defibrillators.

^a1 patient from control arm left out as no data was available, ^b2 patients from control arm

left out due to missing data, ^c3 patients from control arm left out due to missing data, ^d1 patient from TTM (external) arm left out as no data was available, ^e3 patients from TTM (internal) arm left out due to missing data.

(21.7%) internal, p = 0.01] and undercooling N 34 ?C during mainte- nance phase [19 (86.4%) external vs 12 (52.2%) internal, p = 0.83]. Re- bound hyperthermia >= 38 ?C was seen to be significant between two groups [6 (27.3%) external vs 3 (13.0%) internal, p = 0.02].

Fig. 1 shows the distribution of the study participants. Figs. 2 and 3 show the median and interquartile range of temperature during first 24 h of cooling and the initial 12 h of rewarming phase in internal and external cooling methods.

In univariable analysis of variance, factors such as age [OR 0.92, 95% CI 0.86-0.97], duration of collapse to sustained ROSC [OR 0.96, 95% CI 0.93-0.99], incidence of recurrent ROSC [OR 0.11, 95% CI 0.01-0.69],

overcooling [OR 0.11, 95% CI 0.02-0.48], rebound hyperthermia [OR 7.64, 95% CI 1.55-57.2] were shown to be significant towards survival to discharge.

Multivariable analysis showed significance in the following factors for survival: age [OR 0.87, 95% CI 0.76-0.95], duration of collapse to

Table 2

Comparison of survival to hospital discharge or 30-days post-arrest, neurological outcomes and complications.

Control

Intervention N = 45 Intervention

External cooling vs

Internal cooling vs

Internal cooling vs external

Intervention vs

	N = 42	Internal N = 23	External N = 22	N = 45	control OR (95% CI)	control OR (95% CI)	cooling OR (95% CI)	control OR (95% CI)
Survival (%)	9 (21.4)	11 (47.8)	7 (31.8)	18 (40.0)	1.71 (0.52-5.49)	3.36 (1.13-10.41)	1.96 (0.59-6.86)	2.44 (0.95-6.30)
CPC 1-2 (%)	8 (19.0)	7 (30.4)	5 (22.7)	12 (26.7)	1.25 (0.36-4.41)	1.86 (0.57-6.02)	1.49 (0.39-5.65)	1.55 (0.56-4.26)
OPC 1-2 (%)	8 (19.0)	7 (30.4)	5 (22.7)	12 (26.7)	1.25 (0.36-4.41)	1.86 (0.57-6.02)	1.49 (0.39-5.65)	1.55 (0.56-4.26)
Any cardiac arrhythmias	19 (45.2)	3 (13.0)	5 (22.7)	8 (17.8)	0.36 (0.10-1.09)	0.18 (0.04-0.63)	0.51 (0.09-2.39)	0.26 (0.10-0.70)
Any seizures	6 (14.3)	1 (4.3)	4 (18.2)	5 (11.1)	1.33 (0.31-5.28)	0.27 (0.01-1.74)	0.21 (0.01-1.54)	0.75 (0.21-2.67)
Any hyperkalemia	10 (23.8)	3 (13.0)	2 (9.1)	5 (11.1)	0.32 (0.05-1.38)	0.48 (0.10-1.79)	1.50 (0.23-12.33)	0.40 (0.12-1.29)

CPC = cerebral performance category, OPC = overall performance category, OR = odds ratio, CI = confidence interval.

sustained ROSC [OR 0.93, 95% CI 0.85-0.98] and rebound hyperthermia [OR 10.44, 95% CI 1.28-136.83].

Discussion

In this study, we found that the internal cooling method of providing TTM resulted in better survival outcomes compared to controls. Howev- er we were unable to show a significant difference in outcomes for ex- ternal cooling compared to internal cooling. Internal cooling also provided better temperature control compared to external cooling.

Our overall findings are consistent with previous clinical trials inves- tigating the efficacy of TTM versus normothermia in out-of-hospital car- diac arrest survivors [6,7]. We had studied an additional component by randomizing our intervention group to receive either internal or exter- nal cooling. Similarly with other studies [21,26,27], our results showed that internal cooling provides better control of temperature mainte- nance during cooling and a lower incidence of overcooling.

Each of the methods developed to carry out TTM has their respective advantages and disadvantages as described in several papers [21,28,29]. The advantages accorded to internal cooling included the ability to achieve rapid cooling, tighter maintenance of temperature with lesser shivering. However, internal cooling method is also invasive, increases risk of infection and bleeding [30]. Additionally, there are more logistics involved and catheter insertion requires trained personnel. While exter- nal cooling is able to be avoid invasive procedures, previously reported disadvantages include longer cooling time, difficulty in temperature ti- tration, risk of skin injuries as well as more shivering complications [21,28].

There had been concerns that internal cooling with its requirement for specialized skills needed for catheter insertion could result in a time delay towards initiation of therapy. We found no significant differ- ence between the two groups in the time taken to reach target temper- ature from initiation of TTM therapy. However we note that in our

study, time to initiation of TTM tended to be relatively long, and was usually initiated at the intensive care unit, rather than at ED.

Beyond the variables affecting initiation time of each TTM method, our study also looked into the maintenance phase and rewarming phase. Figs. 2 and 3 show the fluctuations of temperature in the first 24 h of maintenance and the initial 12 h of rewarming. Temperature fluctuations were present more in external cooling as compared to in- ternal cooling. Furthermore, this was substantiated when we analyzed for any overcooling, undercooling or rebound hyperthermia events. Ex- ternal cooling seemed to fare worse for all of the events compared to in- ternal cooling, although it did not reach statistical significance.

TTM in post-cardiac arrest patients is not without complications that can arise from the poor condition of patient or from the treatment. Cold temperature can increase coagulopathies, result in vasoconstriction and thus poor blood flow to organs and peripheries [31]. This could manifest as vessel occlusions (deep vein thrombosis), skin complications from di- rect cold exposure, stress ulcers due to long recumbence, hyperkalemia with breakdown of cells and organ dysfunction such as cardiac arrhyth- mias and seizures. Previous studies done had shown some side effects of hypothermia like immunosuppression, Electrolyte disorders, coagulop- athy and hypovolemia [28,31].

In our study, there were no incidences of deep venous thrombosis, stress ulcers or skin complications in any of the three groups that were seen in other studies. Seizures and hyperkalemia complications al- though more prevalent in the control group, did not reach significance statistically. Cardiac arrhythmia complications were seen less in the group with TTM compared to controls. This decreased occurrence is also seen when internal cooling is compared against controls. As such, our study suggests that TTM, in particular internal cooling, may actually decrease complications of cardiac arrhythmias as compared to controls. Fever is commonly seen after hypoxic-ischemic brain injury and temperature control has been shown to reduce the damage cause [25]. Our multivariate analysis suggests age, duration of collapse to sustained ROSC and rebound hyperthermia are significant factors towards

Table 3

Comparison between internal and external cooling methods of targeted temperature management.

	Intervention cooling methods		P value
	Internal N = 23	External N = 22
T0C at ROSC, median (IQR)	36.05 (35.05-36.28) (N = 16)	36.60 (35.00-36.23) (N = 16)	0.47
Pretreatment T0C, median (IQR)	35.30 (34.70-37.20) (N = 23)	34.70 (34.03-37.55) (N = 21)	0.65
First R2TH in minutes, median (IQR)	304.50 (244.00-424.00) (N = 22)	333.00 (257.00-450.00) (N = 21)	0.74
Sustained R2TH in minutes, median (IQR)	285.50 (246.25-424.00) (N = 22)	333.00 (257.00-469.00) (N = 21)	0.60
T2TT in minutes, median (IQR)	109.00 (68.00-205.25) (N = 22)	118.00 (61.25-227.25) (N = 20)	0.92
First R2TT in minutes, median (IQR)	469.00 (319.50-675.00) (N = 21)	505.00 (385.00-758.00) (N = 19)	0.65
Sustained R2TT in minutes, median (IQR)	469.00 (305.00-685.00) (N = 21)	505.00 (385.00-758.00) (N = 19)	0.57
Completion of at least 12 h of cooling (%)	22 (95.7)	22 (100.0)	1.00
Overcooling b 33.0 ?C	5 (21.7)	11 (50.0)	0.01
Undercooling N 34.0 ?C during maintenance phase (+-0.2 ?C)	12 (52.2)	19 (86.4)	0.83
Rebound hyperthermia >= 38 ?C	3 (13.0)	6 (27.3)	0.02

IQR = interquartile range, R2TH = time from ROSC to initiation of TTM, R2TT = time from ROSC to target temperature of 34 ?C, T2TT = time from initiation of TTM to target temperature of 34 ?C, OR = odds ratio, CI = confidence interval.

Note: External cooling technique is taken as reference point for calculation of odds ratio.

Fig. 1. Distribution of study participants.

Fig. 2. Average core body temperature over first 24 h of cooling phase using internal or external cooling method.

Fig. 3. Average core body temperature over 12 h from initiation of rewarming phase in internal or external cooling method.

survival. As with previous studies, there is suggestion that the post- cooling phase could be more important than the cooling phase itself. Better temperature control with internal cooling, might be an advantage to prevent hyperthermia and control body temperature post-cardiac rewarming.

Limitations of our study include our small sample size, which limits the power of our analysis when looking at survival outcomes and thus potentially important clinical differences between two therapies cannot be ruled out until a larger trial is undertaken. While there was signifi- cant difference between internal cooling and controls for survival out- comes, we are still unable to decisively conclude any significant difference between internal and external cooling groups. While EEG monitoring was performed as needed for patients with seizures, this was intermittent and not routine, and may have under detected sei- zures, especially in TTM patients that had been paralyzed.

Also, healthcare professionals and Research staff could not be blinded to the type of therapy assigned to the individual, although the assign- ment of the therapy was random via sealed envelopes. Thus, observer bias cannot be ruled out in the determinations of futility resulting in ear- lier withdrawal of life support.

There may have been some selection bias for the controls, which were not part of the Randomized Controlled Trial (RCT). The nested RCT design was chosen due to the expected difficulties with recruitment and small sample size associated from a single center study. In addition, only 2 ICUs in the hospital (CCU and MICU) were equipped and had staff trained to perform TTM. Thus the nested design with retrospective/con- current controls was chosen to optimize the number of cases. Although every effort was made to consistently recruit cases for TTM who pre- sented to CCU/MICU, overflow cases and cases admitted to the other ICUs would not have been enrolled and these would be eligible for se- lection as controls. Admission to overflow units also adds another

variable to the control group since they were treated in different units than the TTM groups.

Lastly, we should be aware of survival bias that may be present due to our eligibility criteria in recruiting participants. Participants had to meet inclusion criteria such as hemodynamic stability and ability to maintain ROSC for at least 30 min. This would exclude patients who were more unstable and had potentially worse prognosis with greater Disease burden. As such, the findings may not be generalizable to every cardiac arrest patient based on this study alone.

There are still unanswered questions regarding which method may be superior in achieving better Survival and neurological outcomes. The importance of normothermia post-cooling in recent studies and in our results may also raise questions regarding the ideal duration of TTM needed for best outcomes.

Conclusion

In post-resuscitated cardiac arrest patients who are hemodynami- cally stable with or without inotropes support and able to maintain ROSC for at least 30 min, internal cooling compared to normothermia was associated with a survival-to-hospital discharge benefit, fewer car- diac arrhythmia complications. With its advantage in tighter tempera- ture modulation, internal cooling may be beneficial in ensuring good temperature control post-cardiac arrest and the avoidance of rebound hyperthermia.

Funding source

This study was supported by grants from Department of Clinical Research, Singapore General Hospital (DCR/P08/2009) and Singapore General Hospital research grant (SRG#11/2010).

Conference presentations

4th Asian EMS Conference (EMS ASIA), August 2016, Seoul, Korea (Poster presentation)

4th SingHealth Duke-NUS Scientific Congress, September 2016, Singapore (Poster presentation)

Conflict of interest

A/Prof Ong has licensing agreement and patent filing (Application no: 13/047,348) with ZOLL Medical Corporation for a study titled ‘Meth- od of predicting acute cardiopulmonary events and survivability of a pa- tient’. No further conflict of interests for other authors.

Acknowledgements

We would like to thank the following physicians for their contribu- tion and also acknowledge the support given by other physicians and nurses from their department.

• Dr. Juliana Poh, Department of Emergency Medicine, Singapore Gen- eral Hospital
• Dr. Aaron Sung Lung Wong, Coronary Care Unit, Department of Cardi- ology, National Heart Centre Singapore
• Dr. Chee Tang Chin, Coronary Care Unit, Department of Cardiology, National Heart Centre Singapore
• Intensive Care Unit, Department of Respiratory and Critical Care Med- icine, Singapore General Hospital

References

1. Rea TD, Pearce RM, Raghunathan TE, Lemaitre RN, Sotoodehnia N, Jouven X, et al. In- cidence of out-of-hospital cardiac arrest. Am J Cardiol 2004;93:1455-60.
2. Zheng ZJ, Croft JB, Giles WH, Mensah GA. Sudden cardiac death in the United States, 1989 to 1998. Circulation 2001;104:2158-63.
3. Lai H, Choong CV, Fook-Chong S, Ng YY, Finkelstein EA, Haaland B, et al. Interven- tional strategies associated with improvements in survival for out-of-hospital cardi- ac arrests in singapore over 10 years. Resuscitation 2015;89:155-61.
4. Ong ME, Shin SD, De Souza NN, Tanaka H, Nishiuchi T, Song KJ, et al. Outcomes for out-of-hospital cardiac arrests across 7 countries in asia: the pan asian resuscitation outcomes study (paros). Resuscitation 2015;96:100-8.
5. Peberdy MA, Callaway CW, Neumar RW, Geocadin RG, Zimmerman JL, Donnino M, et al. Part 9: post-cardiac arrest care: 2010 American heart association guidelines for cardiopulmonary resuscitation and emergency cardiovascular care. Circulation 2010;122:S768-86.
6. Bernard SA, Gray TW, Buist MD, Jones BM, Silvester W, Gutteridge G, et al. Treatment of comatose survivors of out-of-hospital cardiac arrest with induced hypothermia. N Engl J Med 2002;346:557-63.
7. Hypothermia after Cardiac Arrest Study G. Mild therapeutic hypothermia to improve the neurologic outcome after cardiac arrest. N Engl J Med 2002;346:549-56.
8. Nolan JP, Morley PT, Vanden Hoek TL, Hickey RW, Kloeck WG, Billi J, et al. Interna- tional liaison committee on R. Therapeutic hypothermia after cardiac arrest: An ad- visory statement by the advanced life support task force of the international liaison committee on resuscitation. Circulation 2003;108:118-21.
9. Steen PA, Newberg L, Milde JH, Michenfelder JD. Hypothermia and barbiturates: in- dividual and combined effects on canine cerebral oxygen consumption. Anesthesiol- ogy 1983;58:527-32.
10. Mori K, Itoh Y, Saito J, Takeyama Y, Kurata Y, Kaneko M, et al. Post-resuscitative hy- pothermic bypass reduces ischemic brain injury in swine. Acad Emerg Med 2001;8: 937-45.
11. Ginsberg MD, Sternau LL, Globus MY, Dietrich WD, Busto R. Therapeutic modulation of brain temperature: relevance to ischemic brain injury. Cerebrovasc Brain Metab Rev 1992;4:189-225.
12. Ji Y, Liu J. Numerical studies on the effect of lowering temperature on the oxygen transport during brain hypothermia resuscitation. Comput Biol Med 2002;32: 495-514.
13. Laptook AR, Corbett RJ. The effects of temperature on hypoxic-ischemic brain injury. Clin Perinatol 2002;29:623-49 [vi].
14. Safar PJ, Kochanek PM. Therapeutic hypothermia after cardiac arrest. N Engl J Med 2002;346:612-3.
15. Colbourne F, Sutherland G, Corbett D. Postischemic hypothermia. A critical appraisal with implications for clinical treatment. Mol Neurobiol 1997;14:171-201.
16. D’Cruz BJ, Fertig KC, Filiano AJ, Hicks SD, DeFranco DB, Callaway CW. Hypothermic reperfusion after cardiac arrest augments brain-derived neurotrophic factor activa- tion. J Cereb Blood Flow Metab 2002;22:843-51.
17. Guo Q, Tan X, Cai H, Lei B, Xu Q. Influence of moderate hypothermia on the contents of 6-keto-pgf1 alpha and txb2 in brain tissues after cardiac arrest and resuscitation in dogs. Hunan Yi Ke Da Xue Xue Bao 1998;23:369-71.
18. Kentner R, Rollwagen FM, Prueckner S, Behringer W, Wu X, Stezoski J, et al. Effects of Mild hypothermia on survival and serum cytokines in uncontrolled hemorrhagic shock in rats. Shock 2002;17:521-6.
19. Nielsen N, Wetterslev J, Cronberg T, Erlinge D, Gasche Y, Hassager C, et al. Targeted temperature management at 33 ?C versus 36 ?C after cardiac arrest. N Engl J Med 2013;369:2197-206.
20. Hoedemaekers CW, Ezzahti M, Gerritsen A, van der Hoeven JG. Comparison of cooling methods to induce and maintain normo- and hypothermia in intensive care unit patients: a prospective intervention study. Crit Care 2007;11:R91.
21. Tomte O, Draegni T, Mangschau A, Jacobsen D, Auestad B, Sunde K. A comparison of intravascular and Surface cooling techniques in comatose cardiac arrest survivors. Crit Care Med 2011;39:443-9.
22. Deye N, Cariou A, Girardie P, Pichon N, Megarbane B, Midez P, et al. Clinical, econom- ical impact of endovascular cooling in the management of cardiac arrest study G. Endovascular versus external targeted temperature management for patients with out-of-hospital cardiac arrest: a randomized, controlled study. Circulation 2015; 132:182-93.
23. Cummins RO, Chamberlain DA, Abramson NS, Allen M, Baskett P, Becker L, et al. Rec- ommended guidelines for uniform reporting of data from out-of-hospital cardiac ar- rest: the Utstein style. Task force of the american heart association, the european resuscitation council, the heart and stroke foundation of canada, and the australian resuscitation council. Ann Emerg Med 1991;20:861-74.
24. Cummins RO, Chamberlain DA, Abramson NS, Allen M, Baskett PJ, Becker L, et al. Recommended guidelines for uniform reporting of data from out-of-hospital cardiac arrest: the utstein style. A statement for health professionals from a task force of the American heart association, the European resuscitation council, the heart and stroke foundation of Canada, and the Australian resuscitation council. Circulation 1991;84: 960-75.
25. Perkins GD, Jacobs IG, Nadkarni VM, Berg RA, Bhanji F, Biarent D, et al. Cardiac arrest and cardiopulmonary resuscitation outcome reports: Update of the utstein resusci- tation registry templates for out-of-hospital cardiac arrest: A statement for healthcare professionals from a task force of the international liaison committee on resuscitation (American Heart Association, European Resuscitation Council, Aus- tralian and New Zealand Council on Resuscitation, Heart and Stroke Foundation of Canada, Interamerican Heart Foundation, Resuscitation Council of Southern Africa, Resuscitation Council of Asia); and the American Heart Association Emergency Car- diovascular Care Committee and the Council On Cardiopulmonary, Critical Care, Perioperative and Resuscitation. Resuscitation 2015;96:328-40.
26. Flint AC, Hemphill JC, Bonovich DC. Therapeutic hypothermia after cardiac arrest: performance characteristics and safety of surface cooling with or without endovascular cooling. Neurocrit Care 2007;7:109-18.
27. Pichon N, Amiel JB, Francois B, Dugard A, Etchecopar C, Vignon P. Efficacy of and tol- erance to Mild induced hypothermia after out-of-hospital cardiac arrest using an endovascular cooling system. Crit Care 2007;11:R71.
28. Polderman KH, Herold I. Therapeutic hypothermia and controlled normothermia in the intensive care unit: practical considerations, side effects, and cooling methods. Crit Care Med 2009;37:1101-20.
29. Jordan JD, Carhuapoma JR. Hypothermia: comparing technology. J Neurol Sci 2007; 261:35-8.
30. Geri G, Cariou A. Invasive versus noninvasive cooling after in- and out-of-hospital cardiac arrest. Future Cardiol 2013;9:615-8.
31. Polderman KH. Mechanisms of action, physiological effects, and complications of hy- pothermia. Crit Care Med 2009;37:S186-202.