Introduction

During general anesthesia, core temperature (T_core) declines even in actively warmed patients [1]. Anesthesia-induced hypothermia is related to altered thermoregulation [2], exposure to a cold environment, and redistribution of body heat [3]. Inadvertent intraoperative hypothermia (defined as T_core < 36°C) increases perioperative adverse cardiac events,[4] perioperative transfusion [1, 5], coagulopathy [6], wound infection [7, 8], and prolonged hospitalization [1, 7].

Tissue oxygen saturation (StO₂) and StO₂-derived dynamic parameters during a vascular occlusion test (VOT) reflect tissue microcirculation and microvascular reactivity with local reperfusion reserve [9, 10]. StO₂ and its parameters have been shown to decrease in septic or cardiac surgical patients [11–13], and these alterations were related to poor clinical outcomes [13–17].

In previous investigations, peripheral microcirculation and microvascular reactivity were affected markedly during therapeutic hypothermia in newborns [18] and cardiac arrest patients [19]. However, the influence of intraoperative hypothermia on tissue microperfusion has not been investigated in cardiac surgical patients. We hypothesized that prewarming during the period of induction of anesthesia would attenuate intraoperative hypothermia and may enhance microcirculation in patients undergoing off-pump coronary artery bypass (OPCAB) surgery.

Methods

This study was approved by the Institutional Review Board of Seoul National University Hospital (ref # 1306-026-496; S1 File), and was registered at ClinicalTrials.gov (identifier, NCT02186210; S2 File). All patients gave written informed consent. The study was conducted in accordance with Good Clinical Practice guidelines and the principles of the Declaration of Helsinki.

Patient selection

Patients scheduled for OPCAB surgery at Seoul National University Hospital were screened for eligibility. Exclusion criteria were age < 20 or > 85 years, presence of anatomical anomaly or arteriovenous fistula in the upper extremities, peripheral vascular disease or recent cerebrovascular event (within 6 months), diabetes requiring insulin therapy, preoperative left ventricular (LV) ejection fraction (EF) < 35%, requirement of preoperative inotropes or ventricular assist devices, redo or combined surgery, other than a conventional operative procedure (such as a minimally invasive approach), and pregnancy (Fig 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. CONSORT flow diagram.

https://doi.org/10.1371/journal.pone.0159772.g001

After providing informed consent, patients were randomized 1:1 to the prewarming or control group. Block randomization (blocks of 4) was executed using a computer-generated program by an independent clinician and allocation concealment was kept until data analyses. Baseline patient data collection and laboratory tests including arterial blood gas analysis were performed before surgery. The operative risk was assessed and calculated using the European System for Cardiac Operative Risk Evaluation (EuroSCORE II) [20].

Study protocol

Without premedication, patients were monitored with five-lead electrocardiography, pulse oximetry, non-invasive blood pressure (BP), bispectral index, and cerebral oximetry using a non-infrared spectroscopic sensor in the operating room. After local analgesia with 1% lidocaine injection, the right or left radial artery was cannulated with a 20-G Angiocath™ (Becton Dickinson Medical Ltd., Tuas, Singapore) and connected to a FloTrac™ transducer and EV1000™ monitor (Edwards Lifesciences LLC, Irvine, CA, USA). From them, arterial BP, cardiac index (CI), and stroke volume variation were monitored continuously.

Anesthesia was induced with intravenous injection of midazolam 1.5 mg/kg, vecuronium 1.5 mg/kg, and sufentanil 1 μg/kg. After tracheal intubation, the lungs were ventilated with volume-controlled mechanical ventilation with a tidal volume of 7–8 mL/kg, fraction of inspired oxygen of 0.5, and a respiratory rate of 9–12/min, to maintain the arterial partial pressure of carbon dioxide at 35–45 mmHg without positive end-expiratory pressure. After placement of a central venous catheter (AVA HF; Edwards Lifesciences LLC) in the right or left internal jugular vein, a pulmonary artery (PA) catheter (Swan-Ganz CCOmbo V, model 774HF75; Edwards Lifesciences LLC) was inserted through the internal jugular lumen and was connected to a Vigilance™ II monitor (Edwards Lifesciences LLC) with in vivo calibration. CI and mixed venous saturation (SvO₂) were monitored through the Vigilance™ II monitor after placing the PA catheter.

Anesthesia was maintained with continuous infusion of propofol (Fresofol®2 MCT 2%; Fresenius Kabi, Graz, Austria) and remifentanil (Ultiva™; GlaxoSmithKline, San Polo di-Torrile, Italy) in an effect site target-controlled infusion (TCI) mode. For TCI, the effect site concentration (Ce) was set to 1–4 μg/mL for propofol and 5–12 ng/mL for remifentanil using a commercial infusion pump (Orchestra®; Fresenius Vial, Brezins, France). Ce was monitored with the Marsh pharmacokinetic model for propofol and the Minto model for remifentanil. Anesthetic depth was monitored using an electroencephalogram-based monitor (BIS VISTA™ monitor; Aspect Medical Systems, Norwood, MA, USA) to maintain a bispectral index between 30 and 50. Crystalloid was used for fluid maintenance and blood products were transfused according to the hemodynamic and laboratory variables. According to the center protocol, red cell transfusion was initiated considering hematocrit (to maintain > 27–30%), ongoing surgical bleeding, hemodynamic parameters, and underlying comorbidities. Administration of fluids and vasoactive agents was managed according to the attending anesthesiologist’s decision to maintain mean BP > 60 mmHg, CI > 2.0 L/min/m², and SvO₂ > 65%, according to the institutional practice. The primary choice of inotropic agent was noradrenalin in most cases. Intravenous infusates were administered using a fluid warming device (Level 1; Smiths Industries Medical System, Rockland, MA, USA) in both groups. Total intravenous anesthesia was maintained until the end of surgery and discontinued thereafter. All patients were transferred to a cardiopulmonary intensive care unit (ICU) after the surgery.

Before induction of anesthesia, T_core was measured at the tympanic membrane (T_tymp) using a tympanic thermometer (Thermoscan® IRT4520; Braun GmbH, Kronberg, Germany). The right and left T_tymp were measured, both twice, and the average of the measured temperatures was recorded. After insertion of a PA catheter, T_core was monitored through the PA catheter. Two skin surface temperatures were monitored with skin temperature probes placed at the tip of the index finger and on the forearm. From them, temperature of finger (T_finger) and forearm (T_forearm) were monitored, and the differences between T_core or T_forearm and T_finger (ΔT_core-finger and ΔT_{forearm-finger}) were recorded.

Patients in the prewarming group were heated using a forced-air warmer (Bair Hugger™ Model 505; 3M, St. Paul, MN, USA) connected to a full body blanket (Bair Hugger™ Model 30000; Arizant Healthcare Inc., Eden Prairie, MN, USA), also covered with a cotton blanket, with the air heating temperature set at 43°C. Patients were covered from the upper trunk to the lower leg except an arm for monitoring of skin temperatures. Prewarming was performed, after the placement of monitoring devices, until the end of induction of anesthesia. Patients in the control group were covered with a single cotton blanket without additional air heating. All patients used an automated coil-heated mattress (Blanketrol® II; Cincinnati Sub-Zero Products Inc., Cincinnati, OH, USA) placed on the operative table, set at 37°C.

T_core, T_finger, and T_forearm were recorded before the induction of anesthesia, and at 1, 2, and 3 h after induction of anesthesia, and at the end of surgery. The temperature of the operating room (T_OR) was kept between 20 and 23°C, automatically controlled by an operating room control system, and was recorded at the same time points. After the first 3 h post-induction, if the patients’ T_core reached 36.9°C, or above 36.7°C with a continuously increasing pattern with ≥ 0.1°C increase for three consecutive 10-min intervals, body surface cooling was applied using the automated mattress (temperature set at 23°C) until the T_core reached 36.5°C. Then, the mattress temperature was returned to the initial setting.

StO₂ and StO₂-derived parameters during VOT were monitored with an Inspectra™ StO₂ Tissue Oxygenation Monitor (model 650; Hutchinson Technology Inc., Hutchinson, MN, USA) with an Inspectra™ StO₂ Sensor (model 1615; Hutchinson Technology Inc.) attached to the thenar eminence of the hand of the opposite side for monitoring of T_finger and T_forearm. VOT was performed by inflating the pneumatic cuff placed at the upper arm up to 50 mmHg above the systolic BP. The cuff remained inflated until the StO₂ reached 40% (occlusion of vessel), and was then deflated rapidly. The occlusion slope was defined as the slope during 1 min after initiating cuff inflation. The recovery slope was calculated from the deflation of the cuff until the restoration of 85% of the baseline (Fig 2). VOT was performed before the induction of anesthesia, and at 1, 2, and 3 h after induction of anesthesia, and at the end of surgery. An experienced researcher (YJC), blinded to the randomization, analyzed the VOT graphs using the Inspectra Analysis Program (ver. 4.03; Hutchinson Technology Inc.) and recorded the VOT parameters.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Changes in tissue oxygen saturation during the vascular occlusion test.

StO₂, tissue oxygen saturation.

https://doi.org/10.1371/journal.pone.0159772.g002

Arterial blood gas analysis was performed at the same time points as the temperature recordings. Cardiac troponin I levels were measured before surgery, on ICU admission, 6 h after surgery, and on postoperative days 1, 2, 3, 5, and 7.

Results

Patient enrolment and data collection were performed between August 2014 and February 2015. Among the 64 patients screened for eligibility, 24 were excluded: 8 of these patients had arteriovenous fistulas for hemodialysis in their arms, 2 had poorly controlled diabetes requiring insulin injection, 1 had peripheral vascular obstructive disease, 3 had histories of cerebrovascular events within 6 months, preoperative LV EF was < 35% in 5 patients, co-operations for other organs were planned in 2 patients, minimally invasive procedures were planned in 2 patients, and 1 was a redo surgery. In total, 40 patients were randomized to the prewarming (n = 20) or control group (n = 20). All included patients completed the protocol and were analyzed (Fig 1 and S1 Fig).

Patient baseline and perioperative characteristics, including target vessels for coronary revascularization and main procedural (anastomosis) time, were similar between the groups (Table 1). The mean warming time in the prewarming group was 35 ± 6 min. There was no difference in the temperatures (T_OR, T_finger, T_forearm, ΔT_core-finger, and ΔT_{forearm-finger}) during surgery between the two groups except for T_core (Table 2). T_core was higher in the prewarming group than the control group during the first 3 h after induction (all adjusted p < 0.001, Table 2). At the end of surgery, T_core was comparable between the groups (adjusted p = 0.464, Table 2). The interaction between the time and group was significant for T_core (p < 0.001). The number of patients who developed intraoperative hypothermia (T_core < 36.0°C at any time during the surgery) was lower in the prewarming group (4/20 [20%] vs. 13/20 [65%], p = 0.004). The nadir T_core during surgery was 36.2 ± 0.3°C vs. 35.7 ± 0.2°C in the prewarming vs. control groups, respectively (p < 0.001). Body surface cooling was applied intraoperatively after 3 h after induction of anesthesia in 14/20 (70%) and 8/20 (40%) patients in the prewarming and control groups, respectively (p = 0.057). Additional forced-air warming was required intraoperatively in one patient in the control group, but in no patient in the prewarming group (p > 0.999). On ICU admission immediately after surgery, axillary temperature was 36.6 ± 0.4°C vs. 36.5 ± 0.4°C in the prewarming vs. control groups, respectively (p = 0.261).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Baseline and perioperative patients’ characteristics.

https://doi.org/10.1371/journal.pone.0159772.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Changes in temperatures, microcirculatory parameters and use of vasoactive agents of the prewarming and control group during off-pump coronary artery bypass surgery.

https://doi.org/10.1371/journal.pone.0159772.t002

Skin temperature measured at the finger was not different between the groups (p = 0.751), whereas temperatures at the forearm were higher in the prewarming group than the control group during the surgery (mean difference, 0.5°C [95% CI 0.0–0.9°C]; p = 0.037; Table 2). The interactions between time and group were not significant (p = 0.364 and p = 0.057 for T_finger and T_forearm, respectively). However, ΔT_core-finger and ΔT_{forearm-finger} were similar between the groups (p = 0.800 and p = 0.478, respectively; Table 2). The interactions between time and group were not significant (p = 0.550 and p = 0.739 for ΔT_core-finger and ΔT_{forearm-finger}, respectively).

The primary endpoint, decrease in recovery slope at 3 h after induction of anesthesia, did not differ between the groups (–0.6 ± 1.6 vs. –0.5 ± 2.0%/s in the prewarming vs. control groups, respectively, p = 0.953). Throughout the surgery, there was no difference in occlusion or in recovery slope between the groups (p = 0.367 and p = 0.229, respectively; Table 2). StO₂ values were similar between the groups during surgery (p = 0.516, Table 2). The interactions between time and group were not significant (p = 0.615, p = 0.415 and p = 0.834, for StO₂, occlusion and recovery slope, respectively).

Other hemodynamic and laboratory variables were similar between the groups except for SvO₂ (Table 3). For repeated measured values, SvO₂ was higher in the prewarming group than the control group (p = 0.040; mean difference, 3% [95% CI 0–6%]), although baseline values could not be obtained (Table 3). Intraoperative use of vasoactive agents, such as phenylephrine, vasopressin, and noradrenalin was comparable between the two groups (Table 2). None of the patients required adrenalin intraoperatively. The levels of postoperative troponin I until postoperative day 7 were also comparable between the groups. There was one mortality case in the control group. The patient developed cerebral infarction in the territory of the left middle cerebral artery on the first postoperative day. The patient underwent an immediate craniectomy, but developed multiple organ ischemic damage and eventually had cardiac arrest, which was not resuscitated, on the 39th postoperative day. There was no difference in the amount of perioperative transfusion, chest tube drain, time to extubate, postoperative complications including wound infection, or ICU or hospital stay between the groups (Table 4). At the 1-year follow-up after surgery, more than 94% of grafts were patent by computed tomographic or percutaneous coronary angiography in both groups; there was no difference between the groups (Table 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Hemodynamic and laboratory variables of the prewarming and control group during off-pump coronary artery bypass surgery.

https://doi.org/10.1371/journal.pone.0159772.t003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Postoperative course of patients in the prewarming and control group after off-pump coronary artery bypass surgery.

https://doi.org/10.1371/journal.pone.0159772.t004

Discussion

In this study, we demonstrated that prewarming during anesthesia induction effectively preserved T_core during the first 3 h after induction. It decreased the incidence of intraoperative hypothermia to less than one-third compared with the control group. However, regarding microcirculatory parameters, neither StO₂ nor recovery slope was affected by the prewarming. This is the first reported clinical trial to investigate the effect of prewarming on microvascular reactivity measured by StO₂ with VOT in patients undergoing cardiac surgery.

Inadvertent intraoperative hypothermia develops in about 40% of patients undergoing general anesthesia [21]. If not prevented, up to 70% of patients can become hypothermic after surgery [22]. Intraoperative hypothermia can increase perioperative morbid cardiac events, such as unstable angina, myocardial infarction, or cardiac arrest, in patients undergoing non-cardiac surgery [4]. Hypothermia also impairs platelet function and the coagulation cascade [6, 23], and increases blood loss and transfusion requirements during hip surgery [5]. It has also been related to surgical would infection, and delayed postoperative recovery and hospital stay after non-cardiac surgery [1, 7, 8].

The concept of prewarming in surgical patients started in the early 1990s [24, 25]. Patients undergoing kidney transplantation warmed with forced-air warmer or a circulating-water blanket, with warming starting on induction of anesthesia, were less hypothermic after 3 h of the study period than the control group [24]. Prewarmed patients undergoing hip arthroplasty, using an electric warming blanket before induction of anesthesia, developed less hypothermia than unwarmed patients intraoperatively [25].

The development of intraoperative hypothermia is largely attributable to the redistribution of body heat occurring during the induction of general anesthesia [3, 26]. Thus, to prevent heat loss, body surface warming during this redistribution period may be important. However, in many previous studies on prewarming, active warming was applied before the induction of anesthesia, or commenced on the ward, or in preanesthetic or operating room during the preinduction period [25, 27–31]. In our study, based on our previous study results [32] and to prevent any delay in the surgical procedure, we used a prewarming protocol that included the first 20–30 min after the injection of anesthetics. Thus, we believe that this reduced the incidence of intraoperative hypothermia compared with a previous study [30], in which the prewarming was provided before anesthesia induction for a longer period compared to our study: the incidence of hypothermia was 20% [4/20] vs. 32% [10/31], respectively, and the mean duration of warming time was 35 vs. 72 min. Furthermore, while prewarming provided for up to 50 min during the attachment of monitoring equipment resulted in a mean T_core < 36°C after 1 h of induction in another previous study [29], prewarming for 35 min focused on the induction period preserved the mean T_core > 36°C during the first 3 h of anesthesia in our study. Moreover, forced air warming started after surgical draping in a previous study did not prevent intraoperative hypothermia during the first hour of anesthesia in patients undergoing non-cardiac surgery [1].

Microcirculation measured using StO₂ has been advocated to reflect peripheral tissue perfusion [11]. The occlusion and recovery slope, StO₂-derived parameters following VOT, have been shown to indicate tissue oxygen consumption and microvascular reperfusion adequacy, respectively [10, 33]. Low StO₂ was associated with poor clinical outcomes in critically ill patients [34], and the reduced recovery slope was related to multiple organ dysfunction and increased mortality in septic patients [13]. In surgical patients, perioperative low StO₂ was associated with major complications and mortality after non-cardiac surgery [35]. Both the occlusion and recovery slope were higher in survivors than in non-survivors after open cardiac surgery using cardiopulmonary bypass [33]. Similarly, the recovery slope on postoperative day 1 was lower in patients with a composite of major complications than those without after cardiac surgery [17]. In newborn patients treated with therapeutic hypothermia, sluggish microcirculation was observed with a T_core of 34°C [18]. Furthermore, during mild hypothermia after cardiopulmonary resuscitation, the recovery slope was initially impaired during the first 24 h, and then recovered to the values of healthy subjects in the next 24 h [19]. Thus, we hypothesized that active warming focused on the period of anesthesia induction would effectively reduce inadvertent intraoperative hypothermia and furthermore enhance the peripheral microcirculation. However, microcirculatory parameters were not affected by the different T_core of the two treatment groups in the present study.

Vasoconstriction induced by the use of vasopressors can affect the vascular microcirculation. In previous studies, there were inconsistent data on vasomotor responses caused by the vasoactive agents administered. Administration of noradrenaline did not affect sublingual microperfusion during inhalational general anesthesia [36] or intestinal microcirculation in mice hemorrhagic shock models [37]. StO₂ measured at the thenar eminence was reduced after noradrenaline infusion but not after phenylephrine administration in normovolemic patients under balanced propofol-remifentanil anesthesia [38]. In the current study, the use of various preoperative and intraoperative cardiovascular drugs was comparable between the groups. Moreover, there was no difference between the groups in the use of vasoactive agents, such as phenylephrine and vasopressin, and the number of patients administered, mean duration, and the maximum dose at each time point for noradrenalin, during the first 3 h after induction of anesthesia as well as throughout the surgery. Thus, the effects of vasopressors on microcirculatory variables, especially on the primary end point–the recovery slope–in our results may be minimal.

Although this was a prospective randomized trial, more diabetic patients were included in the control group than in the prewarming group (12 vs. 6, p = 0.057; Table 1). Considering that diabetes mellitus can affect the microvasculature, the skewed distribution of included patients could be a confounding factor, which should be considered when interpreting our data. Moreover, we had more patients who received body surface cooling 3 h after induction in the prewarming group without statistical significance. This trend may have occurred because we strictly controlled body temperature to avoid both hypothermia and hyperthermia in this study, and there was no difference in temperature between the two groups at the end of study. However, it is difficult to draw any conclusion as to whether prewarming may induce hyperthermia because the study was not appropriately powered to address this. It is unclear and beyond of the scope of this study whether intraoperative hyperthermia may cause any clinical effect or whether surface cooling can prevent it during OPCAB surgery. Further studies are needed to investigate these questions.

SvO₂ has been used as a parameter to describe the adequacy of tissue oxygenation and a low SvO₂ was associated with poor prognosis in postoperative cardiovascular patients [39–41]. In a previous study, microvascular dysfunction was related to hyperlactatemia and reduced partial pressure of venous oxygen in septic patients [42]. In the current study, there was one mortality case in the control group, but none in the prewarming group. The main cause of death in the deceased case was a cerebral infarction that developed immediately postoperatively. In our results, microcirculatory parameters were similar between the groups; however, SvO₂ was higher in the prewarming group than the control group, although both groups maintained a mean SvO₂ ≥ 70% and comparable levels of lactate throughout the surgery. However, the baseline value of SvO₂ was not available practically in the current study protocol, and the deceased patient in the control group did not show hypothermia at any temperature measurement time point during the study. Thus, it is hard to draw any conclusion from our observation of higher SvO₂ in the prewarming group, and with respect to whether there is any relationship between prewarming and tissue oxygenation with clinical outcome in the current study.

This study had some limitations. First, the sample size was relatively small. A further study should be conducted to investigate the effects of prewarming on postoperative clinical outcomes and mortality after major cardiovascular surgery. Second, we did not fix the duration of warming time in the treatment group. Thus, our data do not provide results for any particular duration of prewarming. Instead, we used a practical protocol, which can be readily applied to the general practice of anesthesia and preparation for surgery. Prewarming focused on the period of anesthesia induction effectively prevented intraoperative inadvertent hypothermia compared with the control group in our study. Third, we excluded patients with decreased LV contraction. We tried to ensure the uniformity the enrolled population and expected that impaired cardiac function could affect the redistribution of body heat and microcirculation. Thus, the results of our study may not be applied to patients with impaired LV contraction.

In conclusion, prewarming using forced-air warmer during the induction of general anesthesia reduced the incidence of intraoperative hypothermia, but did not improve microvascular reactivity in patients undergoing OPCAB surgery.

Supporting Information

Acknowledgments

We thank the Medical Research Collaborating Center for their advice concerning the statistical analyses.