a b s t r a c t

Purpose: A recent study showed that increased blood glucose variability was an independent predictor of mortality in cardiac arrest survivors treated with Therapeutic hypothermia . We hypothesized that the association of blood glucose variability with outcomes would differ depending on the TH phase, as body temperature affects glucose homeostasis.

Methods: A retrospective cohort of 147 consecutive cardiac arrest patients treated with TH was analyzed. Mean absolute glucose change (MAGC) was calculated using blood glucose values during the entire TH period and during each TH phase (induction, from the TH initiation to the achievement of the target temperature; maintenance, 24 hours from the end of induction; and rewarming, from the end of the maintenance to the achievement of 36.5?C). The primary and secondary outcomes were mortality and neurological outcome at 30 days. Multivariate regression analyses were performed with variables with a significance level b 0.1 on univariate analyses.

Results: The hypoglycemia rate increased significantly during the rewarming phase compared with the maintenance phase (P = .003). The MAGC during the TH maintenance phase was an independent predictor of mortality (OR = 1.056, 95% CI 1.008-1.107, P = .023) and unFavorable neurologic outcome (OR = 1.202, 95% CI 1.043-1.384, P = .038), while the MAGC during the rewarming phase and the entire TH period were not. Conclusion: The increased MAGC during the TH maintenance phase was associated with mortality and unfavorable neurologic outcome. However, this study cannot prove a causal association due to the retrospective design. In addition, we showed that the hypoglycemia rate increased significantly during the rewarming phase.

(C) 2013

1. Introduction

Ischemia-reperfusion injury following cardiac arrest frequently results in metabolic derangements such as hyperglycemia [1]. It is well known that high blood glucose levels are associated with unfavorable neurologic outcome or mortality in Cardiac arrest survivors [2-4]. Several studies in intensive care unit (ICU) patients indicate that intensive insulin therapy may cause harm compared with conventional glucose control [5-7]. Based on these studies, the present post-resuscitation care guidelines recommend maintaining blood glucose between 144 and 180 mg/dL [8]. To date, however, neither the optimal blood glucose target nor the optimal method of Glycemic control is known.

? Funding Sources/Disclosures: The authors have no relevant financial information or potential conflicts of interest to disclose.

* Corresponding author. Tel.: +82 62 220 6809; fax: +82 62 228 7417.

E-mail address: [email protected] (K.W. Jeung).

Recently, several investigators reported that increased blood glucose variability, rather than high blood glucose itself, was an independent predictor of in-hospital mortality in various subgroups of critically ill patients [9-13]. In a retrospective cohort study of 5728 ICU patients, Hermanides et al reported that glucose variability was highly associated with ICU death in high and low ranges of mean glucose [9]. Recent investigations suggest that increased glucose variability by itself can adversely affect outcomes [14-17]. Therapeutic hypothermia is now recommended as a standard treatment in comatose cardiac arrest survivors. To our knowledge, few clinical studies have evaluated blood glucose dynamics and the association between glucose variability and outcomes in comatose cardiac arrest survivors treated with TH [18]. In a study of 220 comatose cardiac arrest survivors treated with TH, Cueni-Villoz et al reported that increased blood glucose variability was an independent predictor of in-hospital mortality irrespective of mean blood glucose levels or of the severity of injury [18]. In that study, blood glucose data during the TH maintenance phase were compared with those during the post- rewarming normothermic phase. We hypothesized that the

0735-6757/$ - see front matter (C) 2013 http://dx.doi.org/10.1016/j.ajem.2012.11.002

hypothermia protocol“>association of glucose variability with outcomes would differ depending on the TH phase as body temperature affects glucose homeostasis [19,20]. At present, it is not known in which time period the blood glucose variability is optimal in predicting outcomes: during a single individual phase or during the entire TH period. The objectives of this study were to analyze glucose dynamics during TH and to investigate the association of glucose variability in each TH phase and the entire TH period with outcomes in cardiac arrest survivors.

1. Methods
   1. Study design

This study was a retrospective observational cohort study. This study was approved and the requirement for informed consent was waived by the institutional review board of Chonnam National University hospital.

Study setting and population

The subjects included in this study were consecutive comatose patients resuscitated from cardiac arrest, admitted to the Chonnam National University Hospital, Gwangju, Republic of Korea, who were treated with TH from January 2008 to December 2011. Chonnam National University Hospital is a university-based, tertiary care hospital that serves a population of approximately 1467000 people in Gwangju Metropolitan City.

Patients were excluded if: (1) TH was interrupted due to hemodynamic instability, recurrent Lethal arrhythmia, and death; (2) target temperature was not maintained; (3) they received a different TH protocol regarding target temperature (b 32?C or N 34?C) or hypother- mia duration (48 or 72 hours); (4) the glucose data were incomplete.

Therapeutic hypothermia protocol

In all patients, TH was applied according to a written institutional TH protocol. In brief, TH was applied to all comatose cardiac arrest cases excluding the following: (1) refusal of TH by next-of-kin; (2) terminal illness; (3) recovery of consciousness before TH was started;

(4) hemodynamic instability refractory to intensive care; (5) hemorrhagic stroke on brain computed tomography; (6) significant gastrointestinal bleeding.

TH was induced with ice packs, intravenous cold saline, and cooling devices (Blanketrol II, Cincinnati Subzero Products, Cincinnati, OH, or COOLGARD 3000 Thermal Regulation System, Alsius Corpora- tion, Irvine, CA). The target temperature of 33 +- 1?C was maintained for 24 hours. Upon completion of the TH maintenance phase, patients were rewarmed to 36.5?C at a rate of 0.25?C/h. During TH, temperature was monitored using an esophageal temperature probe. Midazolam and remifentanil were used for sedation and analgesia. Atracurium was administered to control shivering, if required. All patients received standard intensive care according to our institutional ICU protocol.

Glucose management protocol

Blood glucose was managed by ICU nurses according to a written algorithm (Fig. 1). Blood glucose was measured every hour from blood samples obtained from an arterial catheter using Accu-check (Roche/ Hitachi, Basel, Switzerland), except if hypoglycemia (blood glucose b 60 mg/dL) was observed, in which case 50 ml of 20% dextrose in water was given intravenously and blood glucose was sampled earlier. The target glucose level was 80 to 200 mg/dL, which was achieved using Intravenous insulin. Glucose-containing fluids were avoided whenever possible during TH. Enteral feeding or parenteral nutrition was started at the end of TH.

Fig. 1. Glycemic control protocol. D20W; dextrose 20% in water.

Fig. 2. Study participants.

Measurements

Patients who met the Inclusion/exclusion criteria were identified from electronic medical records; the medical records of these patients were reviewed by one investigator. The following variables were obtained for each patient: age, gender, pre-existing disease, presence of a witness on collapse, first monitored rhythm, etiology of cardiac arrest, location of cardiac arrest (out-of-hospital versus in-hospital), dose of epinephrine administered during cardiopulmonary resuscita- tion (CPR), dose of insulin administered during TH, time from collapse to restoration of spontaneous circulation (ROSC), time from ROSC to initiation of TH, time from initiation of TH to the achievement of target temperature, time from the end of the maintenance phase to the achievement of 36.5?C, and the Sequential Organ Failure Assessment score at the time of ICU admission.

The baseline blood glucose value before the TH induction and the blood glucose values during the entire TH period, which were measured at an interval of 1 hour, were collected for each patient. Mean and median blood glucose levels were calculated using blood glucose values during the entire TH period and during each TH phase: induction (from the initiation of TH to the achievement of the target temperature), maintenance (24 hours from the end of the induction phase), and rewarming (from the end of the maintenance phase to the achievement of 36.5?C). To measure glucose variability, the SD and mean absolute glucose change per patient per hour (MAGC) were calculated [9]. The MAGC was calculated by taking the sum of all absolute glucose changes and dividing it by the total time spent in hours. Hypoglycemia was defined as a blood glucose value below 60 mg/dl. The hypoglycemia rate was calculated as the number of blood glucose measurements showing hypoglycemia divided by the total number of glucose measurements during TH in each patient.

The primary outcome was mortality at 30 days. The secondary outcome was neurologic outcomes at 30 days. Neurologic outcome was assessed using the Glasgow-Pittsburgh Cerebral Performance Categories (CPC), according to recommendations for outcome assessment in comatose cardiac arrest patients, and recorded as CPC 1 (good performance), CPC 2 (moderate disability), CPC 3 (severe disability), CPC 4 (vegetative state), and CPC 5 (brain death or death) [21]. Neurologic outcome was dichotomized as either favorable (CPC 1 and CPC 2) or unfavorable (CPC 3 to CPC 5).

Data analysis

Continuous variables were reported as the median and inter- quartile range (IQR) because all continuous variables showed a non- normal distribution. The Mann-Whitney U test was conducted for unpaired comparisons, and the Wilcoxon signed-rank test was performed for paired comparisons. The Kruskal-Wallis test was performed to compare the distribution of blood glucose values at each TH phase, and the Mann-Whitney U test with the Bonferroni correction was conducted for post hoc analysis. The Friedman test was used to compare repeated measures of blood glucose at each TH phase, and then the Wilcoxon signed-rank test with the Bonferroni correction was performed for post hoc analysis. Categorical variables are shown as numbers of cases with percentages. Comparisons of categorical variables were performed using the ?² test or Fisher’s exact test, as indicated. Multivariate binary logistic regression analysis was used to assess independent predictors of mortality and neurologic outcome. All variables with a significance level b 0.1 in univariate analyses were included in a multivariable logistic regres- sion model. Colinearity between variables was excluded before modeling. Forward selection was used to reach the final model.

Table 1 Patient demographics and clinical characteristics
	Mortality at 30 days		P	Neurologic outcome at	30 days	P
	Survivors	Nonsurvivors		Favourable	Unfavourable
	(n = 98)	(n = 49)		(n = 51)	(n = 96)
Age, y, median (IQR)	52.0 (41.0-67.0)	65.0 (48.5-72.5)	.006	49.0 (35.0-57.0)	63.5 (48.0-72.8)	b .001
Male sex	70 (71.4)	32 (65.3)	NS	36 (70.6)	66 (68.8)	NS
Pre-existing illness
Coronary artery disease	9 (9.2)	7 (14.3)	NS	4 (7.8)	12 (12.5)	NS
Heart failure	3 (3.1)	3 (6.1)	NS	1 (2.0)	5 (5.2)	NS
Hypertension	28 (28.6)	23 (46.9)	.042	11 (21.6)	40 (41.7)	.018
Diabetes	22 (22.4)	17 (34.7)	NS	6 (11.8)	33 (34.4)	.003
Pulmonary disease	9 (9.2)	6 (12.2)	NS	2 (3.9)	13 (13.5)	NS
Renal impairment	6 (6.1)	5 (10.2)	NS	1 (2.0)	10 (10.4)	NS
Cerebrovascular accident	10 (10.2)	7 (14.3)	NS	2 (3.9)	12 (15.6)	NS
Hepatic disease	5 (5.1)	1 (2.0)	NS	1 (2.0)	5 (5.2)	NS
Witness of collapse	81 (82.7)	31 (63.3)	.013	45 (88.2)	67 (69.8)	.014
First monitored rhythm			.006			b .001
Shockable	29 (29.6)	4 (8.2)		26 (51.0)	7 (7.3)
Non-shockable	69 (70.4)	45 (91.8)		25 (49.0)	89 (92.7)
Location of arrest			NS			NS
Out-of-hospital	79 (80.6)	41 (83.7)		40 (78.4)	80 (83.3)
In-hospital	19 (19.4)	8 (16.3)		11 (21.6)	16 (16.7)
Aetiology			.003			b .001
Cardiac	61 (62.2)	17 (34.7)		42 (82.4)	36 (37.5)
Non-cardiac	37 (37.8)	32 (65.3)		9 (17.6)	60 (62.5)
Epinephrine administered during CPR, mg, median (IQR)	3.0 (1.0-5.0)	3.0 (2.0-7.0)	NS	2.0 (1.0-5.0)	2.0 (4.0-7.0)	.002
Time from collapse to ROSC, min, median (IQR)	25.0 (15.0-40.0)	30.0 (20.5-40.0)	NS	20.0 (12.8-35.5)	30.0 (20.0-40.0)	.009
Time from ROSC to initiation of TH, min, median (IQR)	240.0 (155.0-300.0)	278.0 (175.0-332.5)	NS	230.0 (160.0-300.0)	240.0 (170.0-320.0)	NS
Time from initiation of TH to achieving target temperature, h,	3.3 (2.0-5.5)	2.0 (1.0-4.3)	.017	4.0 (2.5-6.5)	2.5 (1.5-4.0)	.001
median (IQR) Time from the end of maintenance phase to achievement of	10 (7-12)	10 (7-13)	NS	9 (7-12)	11 (7-13)	NS
36.5?C, h, median (IQR) Cooling device used			NS			NS
Surface cooling	65 (66.3)	34 (69.4)		31 (60.8)	68 (70.8)
Endovascular cooling	33 (33.7)	15 (30.6)		20 (39.2)	28 (29.2)
SOFA score	9.0 (7.0-11.0)	10.0 (9.0-12.0)	.001	9 (6-10)	10 (9-12)	.001

NS, not significant.

Goodness of fit of the final model was evaluated with the Hosmer- Lemeshow test. Data were analyzed using PASW/SPSSTM software, version 18 (IBM Inc, Chicago, IL). Significance was set at P b .05.

1. Results

A total of 147 patients were included (Fig. 2). Clinical characteristics are shown in Table 1. Compared to survivors, nonsurvivors were older. The proportion of patients with hyperten- sion was higher in the nonsurvivors group. Nonsurvivors had a lower proportion of witnessed collapse, higher proportion of non- shockable rhythm, higher proportion of non-cardiac etiology, shorter time from the initiation of TH to achieving the target temperature, and higher SOFA score. Patients with a favorable neurologic outcome were significantly younger. Patients with hypertension and diabetes were more likely to have an unfavorable neurologic outcome. Favorable neurologic outcome was significantly associated with a higher proportion of witnessed collapse, higher proportion of shockable rhythm, cardiac etiology, lower dose of epinephrine administered during CPR, shorter time from collapse to ROSC, longer time from the initiation of TH to achievement of the target temperature, and lower SOFA score.

A total of 4829 glucose values, which consisted of 426 values in the induction phase, 3,528 values in the maintenance phase, and 875 values in the rewarming phase, were included in the analysis. The median blood glucose levels were 140.0 (108.8-196.0) mg/dL, 127.0

(103.0-163.0) mg/dL, and 107.0 (88.0-129.0) mg/dL during the TH induction, maintenance, and rewarming phases, respectively. The levels differed significantly among the three TH phases (P b .001, Kruskal-Wallis). Multiple pair-wise comparisons with the Bonferroni

correction showed significant differences in blood glucose values between the two groups (induction versus maintenance, P b .001; maintenance versus rewarming, P b .001; induction versus rewarm- ing, P b .001). Each patient’s mean and median glucose values decreased substantially from the induction phase to the maintenance phase as well as from the maintenance phase to the rewarming phase (P b .001) (Fig. 3). The SD and MAGC were significantly lower during the rewarming phase compared with the maintenance phase (P b

.001) (Fig. 3). In this study, the SD and MAGC during the induction phase were not included in the analysis because the duration of the induction phase was less than 2 hours in 39.5% of patients (58/147), and longer than 3 hours in only 46.3% of patients (68/147). The hypoglycemia rate was significantly higher during the rewarming phase compared with the maintenance phase (P = .003, Fig. 4).

The relationships between the glucose variables and outcomes are presented in Table 2. Initial blood glucose was not associated with outcomes. Compared to survivors, nonsurvivors showed higher mean and median glucose levels during the entire TH period, the induction phase, and the maintenance phase. The SD and MAGC during the maintenance phase were significantly lower in survivors. The mean and median glucose levels during the maintenance phase were significantly higher in patients with unfavorable neurologic outcomes. A multivariate logistic regression analysis revealed that non- cardiac etiology, the SOFA score and the MAGC during the mainte- nance phase were significant predictors of mortality at 30 days (Table 3). The multivariate logistic regression analysis demonstrated that age, presence of diabetes, presence of previous cerebrovascular accident, witness of collapse, Noncardiac etiology, epinephrine administered during CPR, and MAGC during the maintenance phase

were significant predictors of unfavorable neurologic outcome.

Fig. 3. Calculated blood glucose variables. Multiple comparisons with the Bonferroni correction also show that (a) the mean glucose level and (b) the median glucose level differed significantly between phases (induction versus maintenance and maintenance versus rewarming). In this study, the standard deviation and the mean absolute glucose change during the induction phase were not included in the analysis, since the induction phase lasted less than 2 hours in 39.5% of patients (58/147). *P b .05.

1. Discussion

In this study, the MAGC during the TH maintenance phase was an independent predictor of mortality and neurologic outcome at 30 days, while the MAGC during the rewarming phase or the entire TH period was not associated with outcomes.

Several previous studies have indicated an association between glucose variability and in-hospital mortality in critically ill patients [9-13,18]. In a study by Krinsley et al which included 3,252 ICU patients, blood glucose variability, which was measured using SD, was a predictor of mortality within different ranges of the mean glucose and a stronger predictor than the mean glucose itself [13]. Consistent with these studies, the blood glucose variability, rather than the mean blood glucose level, was a predictor of mortality in this study. At present, it is unclear why the MAGC during the TH maintenance phase only, rather than that during the entire TH period or the rewarming period, was an independent predictor of outcomes. We suspect that it is related to the decrease in the MAGC over time. Glucose variability is thought to decrease over time because the blood glucose value reaches the target range over time. Consistently, in this study, the blood glucose variability during the maintenance phase was higher compared to that during the later rewarming phase. Thus, the glucose variability is expected to predict outcomes more accurately if it is measured as early as possible after ROSC.

In the present study, the MAGC during the maintenance phase was also an independent predictor of neurologic outcome. To our knowledge, few studies have examined the association of blood glucose variability with neurologic outcome. In the study by Cueni- Villoz et al, multivariate analysis for identifying independent pre- dictors of neurologic outcome was not provided, although univariate analysis showed a significant association between increased blood glucose variability and worse Neurologic recovery [18]. A recent study

in patients with traumatic brain injury showed findings similar to those of the present study. In a study by Matsushima et al that included 109 patients with traumatic brain injury, blood glucose variability, defined by SD and percent of excursion, was significantly associated with poor long-term functional outcome [22].

Fig. 4. Hypoglycemia rate in each phase of therapeutic hypothermia.

Table 2

Relationships between blood glucose variables and outcomes

	Mortality at 30 d		P	Neurologic outcome at 30	d	P
	Survivors (n = 98)	Nonsurvivors (n = 49)		Favourable (n = 51)	Unfavourable (n = 96)
Initial glucose level	227.5 (160.8-308.4)	251.0 (183.5-318.0)	NS	234.0 (163.0-307.0)	239.0 (172.3-313.8)	NS
Glucose level during entire TH Mean glucose level, mg/dl	125.1 (104.5-148.3)	132.3 (117.1-165.6)	.046	121.8 (103.3-147.4)	132.2 (110.6-162.4)	NS
Median glucose level, mg/dl	120.5 (100.4-144.2)	128.0 (113.5-158.0)	.043	117.0 (99.5-141.0)	127.3 (108.2-154.0)	NS
Glucose variability during entire TH Standard deviation, mg/dl	27.3 (19.2-40.3)	33.2 (23.6-40.4)	NS	27.5 (19.5-37.3)	31.6 (22.1-41.9)	NS
MAGC, mg/dl/h	13.2 (8.6-18.0)	14.6 (10.4-20.5)	NS	13.4 (8.4-17.9)	13.5 (10.0-19.2)	NS
Glucose level during induction Mean glucose level, mg/dl	140.3 (113.8-191.8)	191.0 (139.1-255.4)	.001	141.7 (122.0-190.7)	164.0 (112.0-226.5)	NS
Median glucose level, mg/dl	142.0 (107.0-193.3)	192.5 (133.5-256.5)	b .001	142.0 (113.0-185.0)	162.1 (114.5-221.6)	NS
Glucose level during maintenance Mean glucose level, mg/dl	127.8 (104.0-156.7)	137.2 (121.1-172.2)	.036	120.9 (99.3-159.4)	134.7 (114.3-163.4)	.042
Median glucose level, mg/dl	125.4 (100.4-149.6)	133.5 (116.8-176.0)	.044	119.0 (96.0-149.0)	130.0 (110.3-166.4)	.036
Glucose variability during maintenance Standard deviation, mg/dl	23.1 (15.2-32.5)	25.2 (18.5-38.0)	.037	22.8 (15.3-29.4)	23.9 (17.6-39.3)	NS
MAGC, mg/dl/h	13.0 (9.1-17.7)	15.8 (10.1-22.5)	.021	12.4 (9.1-16.7)	14.1 (10.0-20.3)	NS
Glucose level during rewarming Mean glucose level, mg/dl	103.7 (88.2-122.4)	106.5 (87.6-124.6)	NS	101.1 (88.4-117.0)	108.0 (90.5-136.1)	NS
Median glucose level, mg/dl	106.5 (87.6-124.6)	108.0 (95.3-141.3)	NS	106.5 (89.0-121.0)	107.0 (91.3-137.8)	NS
Glucose variability during rewarming Standard deviation, mg/dl	13.1 (9.1-17.6)	13.5 (7.0-23.6)	NS	12.4 (8.7-17.7)	13.7 (8.1-21.2)	NS
MAGC, mg/dl/h	6.5 (3.4-10.2)	7.1 (4.6-12.1)	NS	6.8 (3.4-10.8)	6.5 (4.2-11.6)	NS
Dose of insulin administered during TH, U	4 (0-14)	8 (2-22)	NS	4 (0-14)	6 (0-18)	NS
Hypoglycemia rate, %	0.0 (0.0-4.5)	0.0 (0.0-2.7)	NS	0.0 (0.0-3.2)	0.0 (0.0-5.0)	NS

CVA, cerebrovascular accident.

In this study, the hypoglycemia rate increased significantly during the rewarming phase compared with the maintenance phase. Escolar et al reported that insulin secretion increased during rewarming in an ex-vivo study [19]. Lehot et al reported that the insulin level increased during rewarming in patients who had undergone hypothermic cardiopulmonary bypass [23]. The hypoglycemia rate was not associated with outcomes in the present study. However, in a recent study by the NICE-SUGAR study investigators, hypoglycemia was significantly associated with an increased risk of death in critically ill patients [7]. At present, the reasons for the difference in the association of hypoglycemia with outcomes between the present study and the aforementioned study are not readily apparent, but this may partly be due to differences in the incidence of hypoglycemia, study population, target blood glucose range, and glucose control technique between the studies. Hypoglycemia can be especially dangerous in this subset of patients because hypoglycemia symptoms may be recognized late due to a lack of symptoms during sedation. Therefore, special attention should be paid during the rewarming phase to avoid the occurrence of hypoglycemia.

Although whether high glucose variability is harmful or just a marker for high mortality is uncertain, several studies have indicated that high glucose variability by itself mediates harm by increasing oxidative stress, endothelial cell damage, mitochondrial damage, and coagulation activation [14,15]. In an in vitro study by Risso et al that explored the effect of fluctuating glucose on human umbilical vein Endothelial cells, apoptosis was enhanced in endothelial cells exposed to intermittent, rather than constant, high glucose concentration [14]. In a case-control study of 21 patients with diabetes compared with age- and sex-matched controls, acute glucose fluctuations were significantly associated with urinary excretion of 8-isoprostaglan- din-F2?, a marker of oxidative stress, while this association was not affected by HbA_1c or mean daily glucose concentrations, which are

markers of sustained hyperglycemia [16]. These studies indicate the possibility of reducing glucose variability as a future component of	Epinephrine administered during CPR	1.202 (1.043-1.384)	.011
glucose management in patients resuscitated from cardiac arrest.	MAGC during maintenance	1.202 (1.043-1.384)	.038
Further studies comparing glucose management protocols that aim at	CVA, cerebrovascular accident.

lowering glucose variability rather than glucose value with the current glucose control protocol are required.

Previous studies have proposed various methods for defining glucose variability [9,10,12,13,16,18,24]. The mean amplitude of glycemic excursions using continuous glucose monitoring has been regarded as the gold standard for assessing glucose fluctuation [25]. Continuous glucose sensors, however, are not widely available in many countries. Thus, methods using intermittent glucose measure- ments such as SD, range, and MAGC have been used instead of measures requiring continuous glucose monitoring. Egi et al and Krinsley et al used SD, and Cueni-Villoz et al used range [11,13,18]. However, SD and range may not be appropriate for defining the glucose variability of repeated measurements of glucose, since SD and range do not take change over time into account [9]. Furthermore, SD may not be the optimal measure of glucose variability in cases of non- normally distributed glucose values. We calculated the MAGC to take into account variability over time. In this study, the MAGC during the maintenance phase showed a significant association with mortality

Table 3

Results of the logistic regression analysis for independent factors associated with unfavourable neurologic outcome and mortality

Odds ratio (95% confidence interval) P

Mortality at 30 days
Non-cardiac aetiology	6.806 (2.631-17.610)	b .001
SOFA score	1.360 (1.113-1.662)	.003
MAGC during maintenance	1.056 (1.008-1.107)	.023

Unfavourable neurologic outcome at 30 days

Age 1.077 (1.037-1.119) b .001

Diabetes mellitus 4.505 (1.248-16.265) .022

Previous CVA 18.025 (1.839-176.696) .013

Witness of collapse 0.155 (0.034-0.700) .015

Non-cardiac aetiology 15.311 (4.764-49.215) b .001

and neurologic outcome, while SD did not. However, the MAGC is an inappropriate measure for defining glucose variability with relatively Short duration, such as the TH induction phase in the present study. This study has many limitations. First, it was retrospective, and thus potentially subject to selection bias. Second, this study was performed in one center only, and our findings may not be generalizable. Third, the blood glucose target and the blood glucose control technique themselves can affect blood glucose dynamics and glucose variability. However, we did not examine the differences among blood glucose control methods. Fourth, we sought to investigate the association of the glucose variability in each TH phase with outcomes. However, the MAGC during the induction phase was not analyzed because of the short duration of the induction phase. Continuous blood glucose monitoring may be required to obtain blood glucose variability during the induction phase. Fifth, we included a heterogeneous group of cardiac arrest survivors in terms of first monitored rhythm, etiology, and location of arrest. These factors might affect the significance of glucose variability, although we used

regression analyses to adjust for related factors.

Our study identified an association of blood glucose variability during the TH maintenance phase with mortality and unfavorable neurologic outcome in comatose cardiac arrest survivors treated with TH. However, given the limitations of our study, prospective, multicen- ter studies using well-defined glucose control protocol and continuous blood glucose monitoring are required to confirm our findings.

1. Conclusions

In this study, we found that the MAGC during the TH maintenance phase was an independent predictor of mortality and neurologic outcome at 30 days, while that during the rewarming phase or the entire TH period was not associated with mortality or neurologic outcome. In addition, we have shown that blood glucose values decreased over time during TH, and that the incidence of hypogly- cemia increased significantly during the rewarming phase compared with the maintenance phase. Further study is required to assess the effect of decreasing glucose variability on outcomes.

References

1. Beiser DG, Carr GE, Edelson DP, et al. Derangements in blood glucose following Initial resuscitation from in-hospital cardiac arrest: a report from the national registry of cardiopulmonary resuscitation. Resuscitation 2009;80:624-30.
2. Mullner M, Sterz F, Binder M, et al. Blood glucose concentration after cardiopulmonary resuscitation influences functional neurological recovery in human cardiac arrest survivors. J Cereb Blood Flow Metab 1997;17:430-6.
3. Langhelle A, Tyvold SS, Lexow K, et al. In-hospital factors associated with improved outcome after out-of-hospital cardiac arrest. A comparison between four regions in Norway. Resuscitation 2003;56:247-63.
4. Skrifvars MB, Pettila V, Rosenberg PH, et al. A multiple logistic regression analysis of in-hospital factors related to survival at six months in patients resuscitated from out-of-hospital ventricular fibrillation. Resuscitation 2003;59:319-28.
5. NICE-SUGAR Study Investigators. Intensive versus conventional glucose control in critically ill patients. N Engl J Med 2009;360:1283-97.
6. Wiener RS, Wiener DC, Larson RJ. Benefits and risks of tight glucose control in critically ill adults: a meta-analysis. JAMA 2008;300:933-44.
7. NICE-SUGAR study investigators. Hypoglycemia and risk of death in critically ill patients. N Engl J Med 2012;367:1108-18.
8. Peberdy MA, Callaway CW, Neumar RW, 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.
9. Hermanides J, Vriesendorp TM, Bosman RJ, et al. Glucose variability is associated with intensive care unit mortality. Crit Care Med 2010;38:838-42.
10. Ali NA, O’Brien Jr JM, Dungan K, et al. Glucose variability and mortality in patients with sepsis. Crit Care Med 2008;36:2316-21.
11. Egi M, Bellomo R, Reade MC. Is reducing variability of blood glucose the real but hidden target of intensive insulin therapy? Crit Care 2009;13:302.
12. Lundelin K, Vigil L, Bua S, et al. Differences in complexity of glycemic profile in survivors and nonsurvivors in an intensive care unit: a pilot study. Crit Care Med 2010;38:849-54.
13. Krinsley JS. Glycemic variability: a strong independent predictor of mortality in critically ill patients. Crit Care Med 2008;36:3008-13.
14. Risso A, Mercuri F, Quagliaro L, et al. Intermittent high glucose enhances apoptosis in human umbilical vein endothelial cells in culture. Am J Physiol Endocrinol Metab 2001;281:E924-30.
15. Quagliaro L, Piconi L, Assaloni R, et al. Intermittent high glucose enhances ICAM-1, VCAM-1 and E-selectin expression in human umbilical vein endothelial cells in culture: the distinct role of protein kinase C and mitochondrial superoxide production. Atherosclerosis 2005;183:259-67.
16. Monnier L, Mas E, Ginet C, et al. Activation of oxidative stress by acute glucose fluctuations compared with sustained chronic hyperglycemia in patients with type 2 diabetes. JAMA 2006;295:1681-7.
17. Ceriello A, Esposito K, Piconi L, et al. Oscillating glucose is more deleterious to endothelial function and oxidative stress than mean glucose in normal and Type 2 Diabetic patients. Diabetes 2008;57:1349-54.
18. Cueni-Villoz N, Devigili A, Delodder F, et al. Increased blood glucose variability during therapeutic hypothermia and outcome after cardiac arrest. Crit Care Med 2011;39:2225-31.
19. Escolar JC, Hoo-Paris R, Castex C, et al. Effect of Low Temperatures on glucose- induced insulin secretion and ionic fluxes in rat pancreatic islets. J Endocrinol 1987;115:225-31.
20. Helman A, Gilbert M, Pfister-Lemaire N, et al. Glucagon and insulin secretion and their biological activities in hypothermic rats. Endocrinology 1984;115:1722-8.
21. Booth CM, Boone RH, Tomlinson G, et al. Is this patient dead, vegetative, or severely neurologically impaired? Assessing outcome for comatose survivors of cardiac arrest. JAMA 2004;291:870-9.
22. Matsushima K, Peng M, Velasco C, et al. Glucose variability negatively impacts long-term functional outcome in patients with traumatic brain injury. J Crit Care 2012;27:125-31.
23. Lehot JJ, Piriz H, Villard J, et al. Glucose homeostasis. Comparison between hypothermic and normothermic cardiopulmonary bypass. Chest 1992;102: 106-11.
24. Egi M, Bellomo R, Stachowski E, et al. Blood glucose concentration and outcome of critical illness: the impact of diabetes. Crit Care Med 2008;36:2249-55.
25. Service FJ, O’Brien PC, Rizza RA. Measurements of glucose control. Diabetes Care 1987;10:225-37.