Use of Cold Fluids in Postcardiac Arrest Therapeutic Hypothermia: A Safety Analysis

Introduction

Prevention of secondary brain injury is a major concern during postcardiac arrest care. Hypoxic–ischemic brain injury can lead to major disability and need of long-term healthcare resources. Therapeutic hypothermia (TH) after cardiac arrest has been studied since 1959 (Benson et al., 1959), but large randomized controlled trials proving its efficacy were first published in 2002 (Bernard et al., 2002; HACA Study Group, 2002). Cold fluids are widely used to induce hypothermia during TH, but concerns about the risks of excessive fluid resuscitation are growing in modern literature, both in terms of quantity and quality. Many studies report an association with higher mortality and morbidity among intensive care unit (ICU) patients. This specific question has been studied mainly in the septic shock population, in whom large volume of intravenous (IV) fluids is commonly used. A recent study showed, among a cohort of septic shock patients who required vasopressors, that a greater positive fluid balance after resuscitation was associated with higher mortality (Boyd et al., 2011). Higher positive fluid balance was also associated with higher mortality, increased length of mechanical ventilation, and increased ICU length of stay among patients with acute respiratory distress syndrome (NHLBI ARDS Clinical Trials Network, 2006). The content of resuscitation fluids, including high chloride level, has also been associated with unfavorable outcomes, such as kidney injury (Bellomo et al., 2012) and mortality (Raghunathan et al., 2014), challenging accepted dogmas of liberal fluid resuscitation.

There are two categories of cooling methods to induce and maintain TH. First, the external methods include ice bags, cooling fans, water circulating blankets, air circulating blankets, and water circulating gel-coated pads. Internal cooling methods can also be used, such as cold (4°C) IV fluids or intravascular devices, using circulating cold fluids. Different methods have been compared in the past regarding the efficacy of inducing and maintaining TH (Hoedemaekers et al., 2007). The specific question of the safety of large IV fluid volume in postcardiac arrest care has been addressed regarding the effect of fluids on respiratory outcomes (Jacobshagen et al., 2009), but their impact on patient survival is not clear in the literature. The goal of this study was to evaluate the impact of fluid volumes on clinical outcomes, including survival, of patients undergoing TH. Our hypothesis was that larger volumes of IV fluids would be associated with higher mortality and a deterioration of lung oxygenation.

Materials and Methods

We conducted a retrospective observational study, including all adults who underwent TH in their postcardiac arrest care, between November 2011 and November 2013 at a tertiary hospital center in Sherbrooke, Quebec, Canada. This ICU admits ∼1500 patients annually, including postcardiac arrest patients. During this time period, a TH protocol was used with 33°C as target temperature. Studied cardiac arrests could be either in or out-of-hospital. The local hypothermia protocol included an induction phase (22 mL/kg 4°C NaCl 0.9% bolus, external cooling with water-based cooling blankets [Medi-Therm III; Gaymar; NY], acetaminophen), a 24-hour maintenance phase (7 mL/kg 4°C NaCl 0.9% boluses as needed if target temperature not attained, as needed meperidine or neuromuscular blockers for chills), and a slow progressive (0.3°C per hour) rewarming phase. Sedation is maintained at a target Sedation–Agitation Scale (Riker et al., 1999) of 1–2 until rewarming is complete (defined as body temperature above 36°C), with choice of agent left to the treating physician's discretion.

A list of eligible patients was created by the local database (named CIRESSS). Adult patients who had sustained a cardiac arrest were included in the analysis if TH had been registered in their hospitalization summary between November 2011 and November 2013. Patients were excluded if TH was initiated in a referring hospital, the method to induce hypothermia was not described, or if the target body temperature was not achieved (32–34°C).

Data collected from electronic records included the following: age, sex, comorbid conditions, details about cardiac resuscitation (time of arrest, time of return of spontaneous circulation [ROSC], initial rhythm, in-hospital vs. out-of-hospital arrest, witnessed or unwitnessed arrest), details about TH (time from ROSC to hypothermia, method of cooling, amount of IV fluids given), and patient outcomes (SOFA [sequential organ failure assessment] score on days 1 and 3, and survival on day 28). We used the modified SOFA score, which is a validated score that may be more representative of current ICU practice regarding the cardiovascular component (Yadav et al., 2015). A higher score represents more severe dysfunction in six different organ systems with a maximum total score of 24.

The primary outcome measured was a 28-day survival rate according to the amount of cold IV fluids received. Secondary outcomes were the variation of the total SOFA score between days 1 and 3 and the variation of the respiratory component of the SOFA score between days 1 and 3.

We evaluated the association between IV fluid volume (continuous variable) and 28-day survival rate (dichotomic variable), using logistic regression. Multivariate exact logistic regression was used to determine the adjusted effect of IV fluid volume on 28-day survival rate (adjusted for first rhythm [shockable] and time to ROSC). We also evaluated the association between IV fluid volume and variation of total and respiratory component of SOFA score (continuous variable), using Spearman's rank correlation coefficient. Chi-square test and Fisher's exact test were used to examine the association between dichotomic patient's characteristics (known coronary artery disease, hypertension, diabetes, chronic kidney failure, history of cardiac failure, in-hospital vs. out-of-hospital arrest, and shockable vs. nonshockable first rhythm) and 28-day survival. All analyses were two tailed with the statistical significance level set at p < 0.05.

Results

The CIRESSS database identified 31 adults who underwent TH in their postcardiac arrest care between November 2011 and November 2013. Two patients were excluded from the cohort; one patient was transferred from a referring hospital with incomplete data about initiation of TH, and one patient died before reaching 34°C (early withdrawal of life support as per prior patient's wishes and family's demand). Mean (SD or standard deviation) patient age was 61.3 (14.6) years. Sixty-nine percent of the cohort were male. Most of the patients were known for cardiovascular disease (Table 1).

Cardiac arrest occurred out-of-hospital in 90% of patients and 93% were witnessed. Among the witnessed events, the mean (SD) duration of resuscitation was 26.2 (19) minutes. Ventricular fibrillation (VF) or ventricular tachycardia (VT) was the first recorded rhythm in 86% of patients.

Every patient who underwent TH was cooled with an external method, including ice bags or water circulating blankets. Every patient received IV fluids during the induction and maintenance phase of the protocol. The mean (SD) time from ROSC to hypothermia (defined as <34°C) was 385 (170) minutes. The mean (SD) duration of hypothermia was 32.2 (0.2) hours (including maintenance phase and rewarming), for a total of 38.6 hours between ROSC and normothermia during which the hypothermia protocol was applied. Five patients did not complete the TH protocol, four died during the protocol, and one patient regained consciousness during hypothermia. Deaths during protocol were attributable to early withdrawal of life support (as per prior patient's wishes and family's demand). The mean (SD) amount of fluids administered for the duration of TH protocol was 4.24 (2.87) L, with minimum of 0.5 and maximum of 11.2 L.

The 28-day survival rate was 51.7% (Table 2). Mean SOFA score on day 1 was 9.8, and the mean pulmonary component of the SOFA score was 2.4. Twenty-five patients survived until day 3, and at that time their mean SOFA score was 8.3 and the pulmonary component of the SOFA score was 1.9.

Patients with first recorded rhythm VF or VT, and patients with shorter duration of resuscitation (time between witnessed cardiac arrest and ROSC) had a higher rate of 28-day survival. Other patient characteristics did not influence the 28-day survival (Table 3). After adjusting for confounding variables, the 28-day survival rate was not influenced by the amount of fluids given to patients during TH protocol (odds ratio = 1.034; confidence interval 95% [0.741–1.464]; p = 0.85; Table 4). The amount of fluids did not influence either the variation of the total SOFA score between days 1 and 3 (ρ = −0.16, p = 0.46) or the variation of the pulmonary component of the SOFA score between days 1 and 3 (ρ = −0.2, p = 0.34).

Discussion

In our cohort, the amount of IV fluids did not influence 28-day mortality. The amount of IV fluids administered did not influence the variation of SOFA score between days 1 and 3, for both total SOFA score and the pulmonary subcomponent of the SOFA score. This observation is in line with other studies reporting no increase in pulmonary edema in patients receiving IV fluids during TH protocol (Bernard et al., 2003).

We found a nonstatistically significant increase of survival among patients who received more fluids. There are several potential explanations for that observation. First, a significant proportion (17%) of the cohort died before completion of TH protocol. Those patients did not receive as much fluids as the patients who survived, since they did not have time to complete the protocol. Also, some had a spontaneously low temperature after ROSC, likely explaining the wide range of IV cold fluids administered in our cohort (0.5–11.2 L). Patient admitted with a spontaneously low temperature after ROSC received less fluids, but are likely to have had a longer resuscitation, hence a possible worse prognosis. Finally, although counterintuitive, it is possible that giving more fluids to patients during postcardiac arrest care could be beneficial. It has been shown that during cardiac arrest, whole-body ischemia leads to endothelial activation and systemic inflammation in a “sepsis-like” syndrome (Adrie et al., 2002). This inflammatory state seems to aggravate patient's prognosis. During postcardiac arrest care, high levels of interleukin-6 and procalcitonin were associated with increased 30-day mortality (Bro-Jeppesen et al., 2015). As in sepsis, the systemic vasodilation induced by inflammatory cytokines could benefit from early IV repletion. In previous studies linking amount of fluids with unfavorable outcomes, causative association was difficult to establish, as sicker patients were more likely to be administered more fluids. In this study, since the incentive for clinicians to administer fluids may be more driven by a target temperature rather than clinical deterioration, it allows for a different view of the effect of IV fluids. This may also explain why no association was found in our study.

Given their low cost, wide availability, and effectiveness in inducing hypothermia, cold IV fluids appear to remain an acceptable option for TH protocols, although our sample size cannot exclude some deleterious effect. Newer intravascular cooling devices seem to offer advantage on cooling rapidity, but their higher price and invasiveness may limit their use.

Our study had several strengths. First, the primary outcome of patient survival is clearly a useful outcome, easy to interpret. The SOFA score chosen for secondary outcomes is a well-validated score, widely used in the ICU literature. Also, we reviewed every consecutive case of TH protocol between November 2011 and November 2013, avoiding risk of selection bias. Generalizability is good for North American patients in tertiary hospitals.

The major limitation of this investigation was its observational and retrospective approach with the associated risk of unaccounted confounding factors. Also, there were several missing data in the electronic records, especially for the calculation of the hepatic component of the SOFA score. The small sample size limits the power of the study, and findings should be reproduced in larger studies.

Conclusion

No association was demonstrated between quantity of cold IV fluids administered to postcardiac arrest patients in the setting of TH and excessive mortality or pulmonary morbidity. Given their low price and wide availability, cold IV fluids remain an acceptable option for TH induction and maintenance.