Article, Neurology

Temperature variability during targeted temperature management is not associated with neurological outcomes following cardiac arrest

a b s t r a c t

Introduction: Recent studies on comatose survivors of cardiac arrest undergoing targeted temperature manage- ment (TTM) have shown similar outcomes at multiple target temperatures. However, details regarding core tem- perature variability during TTM and its prognostic implications remain largely unknown. We sought to assess the association between core temperature variability and neurological outcomes in patients undergoing TTM follow- ing cardiac arrest.

Methods: We analyzed a prospectively collected cohort of 242 patients treated with TTM following cardiac arrest at a tertiary care hospital between 2007 and 2014. Core temperature variability was defined as the statistical var- iance (i.e. standard deviation squared) amongst all core temperature recordings during the maintenance phase of TTM. Poor neurological outcome at hospital discharge, defined as a Cerebral Performance Category score N 2, was the primary outcome. Death prior to hospital discharge was assessed as the secondary outcome. Multivariable logistic regression was used to examine the association between temperature variability and neu- rological outcome or death at hospital discharge.

Results: A poor neurological outcome was observed in 147 (61%) patients and 136 (56%) patients died prior to hospital discharge. In multivariable logistic regression, increased core temperature variability was not associated with increased odds of poor neurological outcomes (OR 0.38, 95% CI 0.11-1.38, p = 0.142) or death (OR 0.43, 95% CI 0.12-1.53, p = 0.193) at hospital discharge.

Conclusion: In this study, individual core temperature variability during TTM was not associated with poor neu- rological outcomes or death at hospital discharge.

Introduction

Cardiac arrest is a major cause of morbidity and mortality in the United States, affecting approximately 400 000 patients annually and accounting for 15% of all-cause mortality [1,2]. The rate of survival to hospital discharge remains exceedingly low and b 10% of cardiac arrest

? Authorship disclosure: All authors take responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation.

?? Conflicts of interest: No potential conflicts of interest are disclosed. No grant support

was used in the preparation of this work.

* Corresponding author.

E-mail addresses: [email protected] (A. Nayeri), [email protected] (N. Bhatia), [email protected] (B. Holmes), [email protected] (N. Borges), [email protected] (W. Armstrong), Meng.xu@ vanderbilt.edu (M. Xu), [email protected] (E. Farber-Eger), [email protected] (Q.S. Wells), [email protected] (J.A. McPherson).

victims are discharged with a favorable neurological outcome [3]. Neu- rological injury remains the leading cause of death in this patient popu- lation and accounts for approximately two-thirds of all-cause mortality [4].

In an effort to minimize ongoing neurological injury, the use of Targeted temperature management (TTM) in comatose survivors of car- diac arrest has become the standard of care [5,6]. Modifiable factors with regard to TTM include duration of treatment, mechanism of cooling, rate of rewarming, and target temperatures. There is a paucity of observational data that evaluate most of these variables as prognostic indicators. However, with regard to target temperatures, the argument for more aggressive TTM with lower temperatures was refuted by the findings of a large international randomized trial, which showed no dif- ferences in neurological outcome or survival based on a target temper- ature of 33 ?C or 36 ?C [7,8].

The effects of temperature variation outside the range of 33-36 ?C have only been evaluated in observational studies. There is some

http://dx.doi.org/10.1016/j.ajem.2017.01.058

890 A. Nayeri et al. / American Journal of Emergency Medicine 35 (2017) 889-892

evidence that hyperthermia in the first 48 h following cardiac arrest is associated with worse neurological outcomes and lower odds of surviv- al [9]. The effects of significant hyperthermia or hypothermia exclusive- ly in patients undergoing TTM following cardiac arrest are not as clear [10]. Moreover, to our knowledge there has been no previous study that evaluates the prognostic significance of continuous variations in core temperatures during TTM. We sought to better characterize core temperature variability during the maintenance phase of TTM and to as- sess for any relationship between increased temperature variability and neurological outcomes.

Methods

Patient population & study design

The study population included 242 consecutive patients treated at a tertiary care hospital between 2007 and 2014 who met the following criteria: cardiac arrest from non-traumatic etiology, comatose following successful return of spontaneous circulation (ROSC), and treatment with the institution’s TTM protocol for at least 24 h. All patients were cooled externally using an active surface-cooling device, the Arctic Sun System, to maintain a core body temperature of 33 ?C during the main- tenance phase of TTM [11]. This was followed by active rewarming with a goal rate of 0.25 ?C per hour.

Following approval from the institutional review board, demograph- ic and clinical data were collected for each patient in a prospective man- ner and stored in a REDCap database [12]. These variables included age, sex, location of arrest, initial rhythm at time of arrest, receipt of bystand- er cardiopulmonary resuscitation (CPR), time to ROSC, and Cerebral Per- formance Category (CPC) scores at hospital discharge. Initial rhythm was assessed as either a shockable [Ventricular tachycardia or Ven- tricular fibrillation ] or Non-shockable rhythm. Presence of shock at admission was defined as systolic blood pressures (SBP) b 90 or by the use of vasopressors. Mechanical circulatory support was defined as the use of either (1) an intra-aortic balloon pump (IABP), (2) extracorporeal membrane oxygenation , or (3) ventricular assist device . Time to the initiation of TTM and time to reach target temperature were calculated from the time to ROSC. Time at target temperature was calcu- lated via a review of the vital signs in the electronic health record to estimate the duration of time each patient’s core temperature was maintained at a goal of 33 ?C. Troponin I concentrations were recorded where available and are reported as ug/l (normal b 0.05 ug/l).

Each patient’s course of TTM was conceptualized in four parts: in-

duction, maintenance, rewarming, and normothermia. The mainte- nance phase was defined as starting with first core temperature at or below the target and ending with the first attempt at active rewarming. Core temperature variability was defined as the statistical variance (i.e.

minimum temperatures within 48 h of cardiac arrest. A similar analysis was done to assess the association between core temperature variability and survival to hospital discharge. Odds ratios (OR) with 95% confi- dence intervals (CI) are presented. All tests were two-tailed and a p- value of less than or equal to 0.05 was considered statistically significant.

Results

Baseline demographic and clinical characteristics of the cohort are presented in Table 1. The median core temperature during the mainte- nance was 32.9 ?C (IQR 32.4 ?C-33.3 ?C). Core temperature variability was calculated as the statistical variance amongst core temperature re- cordings during the maintenance phase of TTM. Amongst all patients, the median variability during the maintenance phase was 0.22 ?C² (IQR 0.08-0.42 ?C²). A comparison of core temperature variability with CPC scores at hospital discharge is provided in Fig. 1.

A poor neurological outcome was observed in 147 (61%) patients and 136 (56%) patients died prior to hospital discharge. Multivariable logistic regression was used to test for an association of core tempera- ture variability with poor neurological outcomes and death at hospital discharge. The covariates in the models were age, receipt of bystander CPR, initial rhythm, initial temperature, location of arrest, time spent at target temperature, time to ROSC, time to TTM, and the highest and lowest recorded temperatures within 48 h of arrest. Higher levels of core temperature variability were not associated with poor neurological outcomes at hospital discharge (p = 0.142). Advanced age (p = 0.046) and increasing time to ROSC (p b 0.001) were associated with higher odds of poor neurological outcomes. Shockable rhythms (p b 0.001) and in-hospital arrest (p = 0.036) predicted lower odds of poor neuro- logical outcomes, Fig. 2. Increased core temperature variability during TTM was not significantly associated with death prior to hospital dis- charge (p = 0.193). Lengthier times to ROSC were associated with higher odds of death (p b 0.001). Shockable rhythms (p b 0.001) and in-hospital arrest (p = 0.039) were associated with lower odds of death, Table 2.

Discussion

The primary findings of this observational study are as follows:

(1) individual variation in core temperatures amongst patients treated

Table 1

Baseline characteristics. Data are presented as median (IQR) for continuous variables and number (percentage) of patients for categorical variables. N represents the number of non-missing values. CPR, cardiopulmonary resuscitation; ROSC, return of spontaneous cir- culation; T-max, maximum temperature; T-min, minimum temperature; TTM, targeted temperature management.

standard deviation squared) amongst all core temperature recordings

Characteristic	Overall	N
Age (years)	61 (51-69)	242
Male (%)	145 (60%)	242
In-hospital arrest (%)	44 (18%)	242
Shockable rhythm (%)	131 (56%)	235
witnessed arrest (%)	190 (79%)	241
Received bystander CPR (%)	121 (50%)	242
Time to ROSC (min)	20 (15-34)	229
Time to initiation of TTM (min)	122 (65-240)	232
Time to reach target temperature (min)	165 (75-270)	242
Initial body temperature (?C)	36 (34.7-36.7)	242
Time at target temperature (h)	21 (18-24)	242
Core temperature variability (?C²)	0.22 (0.08-0.42)	242
T-max within 48 h of arrest (?C)	37.6 (37.2-38.2)	242
T-min within 48 h of arrest (?C)	31.9 (31.5-32.4)	242
ST-segment elevation myocardial infarction (%)	56 (23%)	239
Peak Troponin I (ug/l)	2.74 (0.42-16.94)	232
Coronary angiography (%)	142 (59%)	242
Percutaneous coronary intervention (%)	65 (27%)	242
Shock on admission (%)	95 (39%)	242
Mechanical circulatory support device (%)	37 (15%)	242

during the maintenance phase.

The primary outcome of this study was poor neurological outcome

at hospital discharge, defined as a CPC score N 2 [13]. A favorable neuro- logical outcome (Neurologically intact survival) was defined as a CPC score of 1 or 2. Death prior to hospital discharge was assessed as the sec- ondary outcome.

2.2. Statistical analysis

All information was de-identified prior to statistical analysis in Stata Statistical Software: Release 14 (College Station, TX, USA) [14]. Descrip-

tive statistics were calculated as the median with interquartile ranges (IQR) for continuous variables. Frequencies (percentages) are depicted for categorical variables. Multivariable logistic regression was used to assess for an association between core temperature variability and neu- rological outcomes while adjusting for age, location of arrest, receipt of

bystander CPR, initial rhythm, initial temperature, time to ROSC, time to TTM, time at target temperature, in addition to maximum and

A. Nayeri et al. / American Journal of Emergency Medicine 35 (2017) 889-892 891

Fig. 1. CPC scores at hospital discharge versus core temperature variability. An arrow indicates two additional patients with a CPC score of 1 at hospital discharge and core temperature variability of 1.58 ?C² and 2.83 ?C², respectively.

with TTM is common, and (2) this variation does not predict poor neu- rological outcomes.

The prognostic implications of temperature variability following car- diac arrest have been previously reported with somewhat conflicting results. In 2009 Suffoletto et al. examined the association of body tem- perature changes with Short-term outcomes [15]. This observational study included roughly 3400 patients, only 58 of which underwent in- duced TTM. The principal finding of the study was that hyperthermia following cardiac arrest was associated with death and unFavorable neurological outcomes. Temperature lability and hypothermia were both associated with lower odds of survival, although no statistically significant association with neurological outcomes was seen. There are a number of key differences in study design and variable definitions be- tween our study and the one mentioned above that are important to ad- dress. Suffoletto et al. studied a patient population where TTM was used

Fig. 2. Forest plot representation of predictors of poor neurological outcomes. For each variable, the odds ratio is represented by a black box and the 95% confidence interval is represented by a horizontal line. CPR, cardiopulmonary resuscitation; ROSC, return of spontaneous circulation; T-max, maximum.

Table 2 Odds ratios of poor neurological outcomes and death. CI, confidence interval; CPR, cardio- pulmonary resuscitation; OR, odds ratio; ROSC, return of spontaneous circulation; T-max, maximum temperature; T-min, minimum temperature; TTM, targeted temperature management.

Characteristic	OR	95% CI	p value
Poor neurological outcome at hospital discharge
Age (years)	1.02	1.01-1.05	0.046
Bystander CPR	0.84	0.40-1.78	0.654
Core temperature variability	0.38	0.11-1.38	0.142
Shockable rhythm	0.15	0.07-0.33	b0.001
Initial temperature	1.11	0.88-1.40	0.373
In-hospital arrest	0.37	0.14-0.94	0.036
Time at target temperature (h)	1.02	0.93-1.11	0.730
Time to ROSC (min)	1.06	1.03-1.09	b0.001
Time to TTM (min)	1.00	0.99-1.00	0.397
T-max within 48 h of arrest	0.79	0.54-1.16	0.227
T-min within 48 h of arrest	0.74	0.41-1.36	0.335
Death prior to hospital discharge
Age (years)	1.02	0.99-1.04	0.155
Bystander CPR	0.80	0.39-1.64	0.534
Core temperature variability	0.43	0.12-1.53	0.193
Shockable rhythm	0.18	0.09-0.37	b0.001
Initial temperature	1.04	0.83-1.30	0.712
In-hospital arrest	0.39	0.16-0.95	0.039
Time at target temperature (h)	1.01	0.93-1.11	0.745
Time to ROSC (min)	1.05	1.02-1.07	b0.001
Time to TTM (min)	1.00	0.99-1.00	0.341
T-max within 48 h of arrest	0.76	0.53-1.10	0.150
T-min within 48 h of arrest	0.80	0.44-1.44	0.454

in a small minority of patients and most of the observed cases of hypo- thermia were passive in nature. Moreover, hypothermia was defined as a core temperature below 36 ?C, temperature lability was defined as a categorical variable, and only patients with in-hospital cardiac arrest were evaluated.

In 2015, Nobile et al. reported the first study on temperature vari- ability exclusively in comatose survivors of cardiac arrest undergoing TTM [16]. Amongst a cohort of 229 patients in Belgium, increased tem- perature variability during TTM was not associated with poor neurolog- ical outcomes at 3 months post-arrest. Of note, similar to this study, there was a non-significant directional association between increased temperature variability and favorable neurological outcomes. Despite the similar results, there is a key difference in study design and variable definition between the two studies. Nobile et al. assessed temperature variability during TTM as the standard deviation of core temperature re- cordings and arbitrarily chose a deviation N 1 ?C as the definition of “high variability”. This categorical measure of temperature variability was used in the primary analyses. In contrast, we assessed core temper- ature variability as a continuous measure and were able to demonstrate a lack of prognostic significance across all levels of observed variation.

Regardless of being defined as a categorical or continuous variable, temperature variation during TTM is common [16]. The underlying causes of these variations and opportunities for intervention remain poorly understood. Temperature variation during TTM may in some cases represent a physiological response to an underlying stressor. In- creased risk of bacteremia from translocation in ischemic gut or pulmo- nary aspiration may in fact contribute to temperature fluctuations during the maintenance phase of TTM [17,18]. There may also likely exist a number of other host-related factors that lead to increased tem- perature fluctuations following ROSC and during TTM. Iatrogenic vari- ables, particularly variability in TTM protocols and delivery methodology amongst institutions, also likely contribute to tempera- ture variability in these patients. It remains unknown to what extent en- dogenous and exogenous factors contribute to temperature fluctuations during TTM and concordantly whether or not there is any difference in prognostic implication between them. Moreover, due to the observa- tional nature of this study, the utility or danger of therapeutic control over temperature variation remains to be examined.

892 A. Nayeri et al. / American Journal of Emergency Medicine 35 (2017) 889-892

Regardless of the underlying cause of variation, there are a number of viable concerns in patients with marked core temperature variability. electrolyte disturbances, particularly those of potassium, during the in- duction and rewarming phases of TTM are well-known [19,20]. In- creased core temperature variability during the maintenance phase could lead to increased electrolyte shifts in these patients. Particularly in patients with lower target temperatures (33 ?C) where episodes of hypokalemia are more common, increased temperature variability might precipitate such electrolyte disturbances [7]. For patients with a target temperature of 36 ?C, significant degrees of core temperature var- iability might lead to periods of normothermia or even hyperthermia. Although core temperature variability was not associated with poor outcomes in our cohort, the target core temperature for all patients in- cluded in the study was 33 ?C. There is a need for further assessment of temperature fluctuations during the TTM as a potential prognostic in- dicator in cohorts with higher target temperatures.

Overall, despite rigorous research efforts, little remains known re- garding the optimal delivery of TTM following cardiac arrest with the exceptions of active avoidance of hyperthermia for the first 48 h and the lack of outcome difference based on a target temperature of 33 ?C to 36 ?C. Even after the recent advances in CPR and post-resuscitation care, the overall prognosis following cardiac arrest remains dismal. In our select cohort of patients who achieved ROSC and subsequently underwent TTM, only 39% experienced a favorable neurological out- come. For a number of these patients, particularly those with lengthier times to ROSC and those without shockable rhythms, a poor outcome might be pre-determined prior to any in-hospital intervention [21,22]. For the remaining subset with better odds of a favorable neurological outcome, it remains unknown whether a number of technique modifi- cations can increase the clinical utility of TTM. That is, with regard to the length of treatment, mechanism of Temperature control, and thera- peutic control over temperature variation, randomized trials are neces- sary to better evaluate any association with outcomes.

Study limitations

This study has a number of limitations. The design as a retrospective cohort study only allows for inference of association instead of causa- tion. While we attempted to account for a number of potential con- founders in the multivariable analyses, it is plausible that other unidentified variables may have influenced the results. Given the rela- tively small cohort of patients, insufficient statistical power is a concern. As a single institution study, the findings presented here may not be ap- plicable to patients in other institutions with different TTM protocols. The patients in this study were externally cooled with the Arctic Sun System and our descriptions of core temperature variability may likely not apply to other mechanisms of achieving therapeutic hypothermia. Additionally, all patients in this cohort had a target temperature of 33 ?C and the effects of temperature variation in cohorts with a target temperature of 36 ?C remain unknown. We were also unable to assess temperature variation in each patient as a product of physiological ther- moregulation or iatrogenic processes. However, this will be very diffi- cult to do even in a prospective study or randomized trial. Finally, due to the observational nature of the study, we are unable to comment on the potential benefit or harm of therapeutic control over tempera- ture variation.

Conclusions

In this study of comatose survivors of cardiac arrest undergoing TTM, there was no association between individual core temperature variabil- ity during TTM and poor neurological outcomes or death at hospital dis- charge. There is a great need for prospective studies and randomized trials to better evaluate the association between particular details of TTM delivery and patient outcomes.

References

AJEM