Wake-up times following sedation with sevoflurane versus propofol after cardiac surgery

Abstract

Objectives. Intravenous sedation in the intensive care unit (ICU) may contribute to altered consciousness and prolonged mechanical ventilation. We tested the hypothesis that replacing intravenous propofol with inhaled sevoflurane for sedation after cardiac surgery would lead to shorter wake-up times, quicker patient cooperation, and less delusional memories. Design. Following coronary artery bypass surgery with cardiopulmonary bypass, 100 patients were randomized to sedation with sevoflurane via the anesthetic conserving device or propofol. Study drugs were administered for a minimum of 2 hours until criteria for extubation were met. Primary endpoints were time from drug stop to extubation and to adequate verbal response. Secondary endpoints were adverse recovery events, memories reported in the ICU Memory Tool test, and ICU/hospital stay. Results. Median time from drug stop to extubation (interquartile range/total range) was shorter after sevoflurane compared to propofol sedation; 10 (10/100) minutes versus 25 (21/240) minutes (p <0.001). Time from extubation to adequate verbal response was shorter (p =0.036). No differences were found in secondary endpoints. Conclusions. Sevoflurane sedation after cardiac surgery leads to shorter wake-up times and quicker cooperation compared to propofol. No differences were seen in ICU-stay, adverse memories or recovery events in our short-term sedation.

Key words::

• randomized controlled trial
• sevoflurane
• propofol
• sedation
• postoperative period
• extubation
• memories

Introduction

Short-term postoperative mechanical ventilation is a frequent practice in cardiac surgery where intravenous sedation is typically used (1). The use of sedative infusions in intensive care unit (ICU) patients – in order to improve tolerance to mechanical ventilation and the endotracheal tube – may contribute to prolonged mechanical ventilation and unpredictable wake-up times (2–5). Benzodiazepines, but also propofol, may also contribute to the development of delusional memories and delirium (6–9). Delusional memories from the ICU stay have been associated with later posttraumatic stress symptoms (10). Patients’ memories from the ICU stay appear to impact the later development of adverse psychological symptoms, with the lack of factual memories combined with delusional memories appearing to have the worst outcome (11–13). The use of inhaled isoflurane for ICU sedation appears not to contribute to delusional memories to the same extent as midazolam (14).

Sevoflurane – administered with the anesthetic conserving device (AnaConDa®) in the ICU – is a feasible alternative to propofol short-term sedation following coronary artery bypass surgery (CABG) (15,16). To our knowledge, there is no study of patients’ recollections after short-term sedation with sevoflurane versus propofol.

In a recently published paper (16), we compared the cardiac effects of sevoflurane sedation with propofol sedation following CABG. The present, a priori planned substudy, with focus on the effects of sedative choice on awakening and ICU memories was performed in parallel. Here we tested the hypothesis that sevoflurane sedation would lead to shorter wake-up times, quicker patient cooperation, and less adverse memories after awakening with comparable adverse events during emergence.

Material and methods

This randomized controlled study was registered as a clinical trial (Clinicaltrials.gov identifier NCT00484575) and took place in the Cardiothoracic ICU at the Karolinska University Hospital, Solna. The study was approved by the Institutional Review Board and The European Medicines Agency. A schematic flowchart of the study is presented in Figure 1.

Figure 1. Flowchart of the study. CABG, coronary artery bypass grafting; CPB, cardio pulmonary bypass; ICU-MT, Intensive Care Unit Memory Tool; n, number.

Inclusion and exclusion criteria

Written informed consent was obtained in 107 patients scheduled to undergo elective or subacute CABG. Seven patients were excluded before randomization. Exclusion criteria were off-pump CABG (three patients), additional surgery performed beyond the scheduled CABG (three patients) or need of mechanical circulatory support for per-operative cardiac failure (one patient).

Perioperative management and randomization

An oral benzodiazepine and/or an intramuscular opioid were given for premedication, at the discretion of the attending anesthesiologist. After induction of anesthesia with midazolam (0–4 mg), fentanyl (0.2–0.4 mg), propofol (70–200 mg), and atracurium (35–50 mg) patients were intubated. Sevoflurane (end-tidal concentration 0.5–2.5%) and fentanyl (0.3–0.5 mg) were used for maintenance of anesthesia and analgesia. Propofol (4–5 mg/kg/h) was used during the cardiopulmonary bypass (CPB) period for maintenance of anesthesia. CPB was conducted in mild hypothermia or normothermia. Patients were allocated to treatment via randomization in the Cardiothoracic ICU at the time of chest closure, by an ICU nurse who drew a randomized sealed envelope containing a code for either inhaled sevoflurane via the AnaConDa^® or intravenous propofol.

Postoperative management

Patients randomized to sevoflurane (Sevorane^®, Abbott Scandinavia AB, Solna, Sweden) received this treatment upon arrival to the ICU, where the propofol infusion from the short transportation was discontinued. The AnaConDa^® (Sedana Medical AB, Uppsala, Sweden) is a modified heat–moisture exchanger that facilitates delivery of sevoflurane or isoflurane with any ICU ventilator (17). The AnaConDa^® is placed between the Y-piece of the respiratory circuit and the endotracheal tube. The anesthetic agent is infused to an evaporator rod in the AnaConDa^® and vaporized during inspiration of oxygen/air. At expiration, approximately 90% of exhaled anesthetic agent is adsorbed to an active carbon filter in the AnaConDa^® and recycled to the patient with the next breath. Scavenging is performed actively or passively from the ventilator and the gas analyzer. Ambient anesthetic concentrations with this method have been below recommended staff exposure limits (18,19).

Sevoflurane was infused to produce a desired end-tidal concentration of 0.5–1.0%, measured with a gas analyzer (Ultima, Datex-Ohmeda AB, Bromma, Sweden). The initial infusion rate (2–6 ml/h) was prescribed from a nomogram of the required infusion rate calculated from the minute ventilation. A bolus of 0.5 ml sevoflurane to the AnaConDa^® was used for increasing sedative depth. Adjustments of 10–20% of the infusion rate were made when needed.

Patients randomized to propofol (Propofol 20 mg/ml, Braun Medical AB, Danderyd, Sweden) continued receiving the infusion used for transport to the ICU. The initial infusion rate was set to 2 mg/kg/h. We selected this starting dose as it was in the range of reported average steady-state infusion rates for sedation following cardiac surgery (20) and the routine starting dose in our unit. A bolus of 20–50 mg of propofol was used for increasing sedative depth. Adjustments of 10–20% of the infusion rate were made when needed.

The study drugs were administered to achieve a Motor Activity Assessment Scale score (21) of 2–3 for 2 hours and thereafter until defined criteria for waking the patient were met. The AnaConDa^® was removed from the breathing circuit or the propofol infusion was stopped when the following criteria were fulfilled:

1. Arterial oxygen tension >10 kPa and arterial carbon dioxide tension <7 kPa with an inspired oxygen fraction of ≤0.4.

2. Positive end-expiratory pressure of ≤5 cm H₂O.

3. Body temperature above 36.5°C.

4. Bleeding in drains less than 100 ml/h for two consecutive hours.

Extubation criteria were spontaneous respiratory rate ≥10 per minute, tidal volume ≥5 ml/kg, and patient's opening eyes and hand grip on verbal demand.

The duration of study drug administration and the time from ending sedative delivery to extubation were noted. Every 15 minutes up to 60 minutes after extubation, nurses asked patients to state their birth date and noted the time when they were able to answer correctly.

Postoperative pain was managed with a continuous morphine infusion (0.5–2 mg/h during sedation and 0.5–4 mg/h following extubation). This infusion was started at 1 mg/h at the same time as the study drugs. Following extubation, opioid dosage was adjusted to target a numeric rating scale (0–10) value ≤3. Patients received paracetamol (1 g) within the first hour of ICU stay and then every 6 hours. Nausea and/or vomiting were treated with ondansetron. Normally, drains were removed in the morning after surgery and patients were discharged to the surgical ward before midday.

Adverse events following sedation

During the first 12 hours following termination of sedation and mechanical ventilation, ICU staff noted the following adverse events potentially associated to sedative choice:

1. Vomiting and/or nausea treated with ondansetron.

2. Severe pain (numeric rating scale (0–10) ≥8).

3. Shivering.

4. Agitation requiring pharmacological treatment.

Need of non-invasive ventilation (CPAP or BIPAP) the first 12 hours to maintain respiratory parameters stated in drug stop criteria.

Memories from awakening and ICU stay

In order to measure patient's memories from awakening and the stay in the ICU, the ICU Memory Tool (11,22,23) was completed by the patient on the day of hospital discharge. This validated questionnaire includes specific questions regarding ICU memories, which is divided into three categories:

1. Factual memories (alarms, voices, lights, faces, endotracheal tube, suctioning, darkness, clock, family, tube in the nose, and ward round).

2. Memories of unpleasant feelings (pain, panic, being uncomfortable, feeling confused, feeling down, and feeling anxious/frightened).

3. Delusional memories (hallucinations, nightmares, dreams, and/or feeling that people were trying to hurt them).

ICU stay, re-admission, and hospital stay

Length of ICU stay, re-admission, and total length of hospital stay before discharge were noted. Criteria for hospital discharge were: (1) stable hemodynamics, (2) stable heart rhythm, (3) patient afebrile and surgical wounds uninfected, (4) no intravenous drugs, and (5) patient walking, able to eat and pass stools.

Statistics

For the study evaluating cardiac effects of sedative choice, a power calculation was performed, based on anticipated differences in troponin between sedative regimens (16).

For the present substudy, a power calculation was performed with regard to differences in wake-up times. Based on data from a trial with desflurane versus propofol for postoperative sedation (24), we anticipated at least a 5-minute difference in extubation time (standard deviation 5 minutes) between sevoflurane and propofol sedation. With 50 patients in each group, this rendered a power of >90% at a 5% significance level. Data were analyzed per protocol in order to best describe the efficacy of the two drugs on sedation and recovery. Duration of mechanical ventilation, duration of drug administration, and median time from drug stop to extubation were analyzed with Mann–Whitney U-test as these parameters were not normally distributed. Time to adequate verbal response following extubation and the median ICU/hospital stay were also analyzed with Mann–Whitney U-test. Continuous parameters were analyzed with Student's T-test and binominal data with Fisher's exact test. IBM SPSS 18.0 (SPSS Inc., Chicago, IL) was used to perform the statistical analysis.

Results

A schematic flowchart of the study is presented in Figure 1. Pre- and perioperative demographics are presented in Table I. Study drug dose, treatment time, and time to extubation are presented in Table II and Figure 2. The time from ending sedative delivery to extubation was shorter in the sevoflurane group; median 10 (interquartile range (IQR)10/range 100) minutes versus 25 (IQR 21/range 240) minutes after propofol (p <0.001). Patients in the sevoflurane group responded earlier with their birth date after terminated sedation than patients in the propofol group (p =0.036) (Figure 3). Opioid requirements were similar between groups during surgery and following sedation (Tables I and II). Adverse events during recovery from sedation, ICU and hospital length of stay, and ICU re-admission rates were comparable between groups (Tables III and IV). One patient in the sevoflurane group died within the first 30 days after surgery, unrelated to the ICU stay and after hospital discharge. No significant differences were seen between groups in the types of reported memories as assessed with the ICU Memory Tool, with a similar incidence of delusional memories in both groups (Table V).

Table I. Pre- and perioperative patient group characteristics.

Data	Sevoflurane (n =49)	Propofol (n =50)	P-value
Age, years (SD)	65 (8)	66 (11)	0.495
BMI (SD), index kg/m^2 (SD)	28.8 (2)	27.3 (4.6)	0.729
Female, n	12	8	0.298
Euroscore, score number (SD)	3.8 (2.1)	4.3 (2.8)	0.276
Euroscore logistic mortality, % (SD)	3.31 (2.6)	4.53 (4.0)	0.078
CPB time, minutes (SD)	62 (25)	70 (19)	0.077
Duration of anesthesia, minutes (SD)	206 (39)	212 (48)	0.509
Total dose of fentanyl, mg (SD)	0.50 (0.14)	0.52 (0.11)	0.982
Total dose of midazolam, mg (SD)	1.5 (1.1)	1.6 (1.1)	0.615

BMI, body mass index; CPB, cardio-pulmonary bypass; n, number; SD, standard deviation.

Table II. Drug sedation and extubation.

Drugs and mean dose	Sevoflurane (n =49)	Propofol (n =50)	P-value
Mean propofol dose, mg/kg/h (SD)	–	2.0 (0.7)	–
Mean sevoflurane dose, ml/h (SD)	3.8 (0.9)	–	–
Mean endtidal sevoflurane, % (SD)	0.8 (0.18)	–	–
Median time study drug administration, minutes (IQR/total range)	165 (69/238)	185 (116/830)	0.268
Mean morphine infusion, mg/h first 6 hours (SD)	1.44 (0.47)	1.36 (0.45)	0.355
Median time drug stop to extubation, minutes (IQR/total range)	10 (10/100)	25 (21/240)	<0.001
Median ICU mechanical ventilation, minutes (IQR/total range)	185 (74/230)	215 (108/1056)	0.056

IQR, interquartile range; n, number; SD, standard deviation.

Figure 2. Time to extubation. •, outlying point more than 1.5 box widths.

Figure 3. Time to adequate verbal response.

Table III. Adverse recovery events first 12 hours following extubation.

Effect	Sevoflurane (n =49)	Propofol (n =50)	P-value
Postoperative nausea and vomiting, n	12	9	0.470
Severe pain after extubation of NRS ≥8, n	3	1	0.362
Shivering, n	2	1	0.617
Agitation requiring pharmacological treatment, n	1	0	0.495
Need of non-invasive ventilation, n	3	5	0.715

NRS, numeric rating scale 0–10; n, number.

Table IV. ICU and hospital stay.

Effect	Sevoflurane (n =49)	Propofol (n =50)	P-value
Median ICU stay, hours (IQR/total range)	22 (5/74)	22 (4/246)	0.364
Median hospital stay, days (IQR/total range)	6 (2/8)	6 (2/44)	0.866
ICU stay ≥24 hours, n	10	10	1.0
ICU stay ≥48 hours, n	5	5	1.0
Re-admission to ICU, n	1	5	0.204

n, number; IQR, interquartile range; SD, standard deviation.

Table V. ICU Memory Tool data.

Data	Sevoflurane (n =49)	Propofol (n =50)	P-value
Questionnaire responders in study groups, % (n)	86.0 (44)	92.0 (46)	0.74
Clear memories of the hospital admission, %* (n)	100 (44)	93.5 (43)	0.24
Clear memories from the ICU stay, %* (n)	38.6 (17)	50.0 (23)	0.30
Factual memories from the ICU stay, %* (n)	84.1 (37)	84.8 (39)	1.00
Endotracheal tube, %* (n)	4.5 (2)	10.9 (5)	0.44
Delusional memories from the ICU stay, %* (n)	20.5 (9)	28.3 (13)	0.47
Dreams, %* (n)	15.9 (7)	13.0 (6)	0.77
Nightmares, %* (n)	9.1 (4)	17.4 (8)	0.36
Hallucinations, %* (n)	6.8 (3)	13.0 (6)	0.49
Paranoia, %* (n)	4.5 (2)	4.3 (2)	1.00
Memories of feelings from the ICU stay %* (n)	54.5 (24)	60.9 (28)	0.67
Confusion, %* (n)	29.5 (13)	15.2 (7)	0.13
Anxiety, %* (n)	18.2 (8)	19.6 (9)	1.0
Panic, %* (n)	4.5 (2)	0 (0)	0.24
Feeling uncomfortable, %* (n)	18.2 (8)	21.7 (10)	0.79
Feeling down, %* (n)	6.8 (3)	17.4 (8)	0.20
Pain, %* (n)	20.9 (9)	28.3 (13)	0.47

%*, percent within group of completed ICU Memory Tool (excluding loss to follow-up); n, number; SD, standard deviation.

One patient in the sevoflurane group did not have the AnaConDa^® removed, but merely the sevoflurane infusion discontinued, leading to slow awakening because of rebreathing of sevoflurane. This patient represents the extreme outlier (100 minutes) in the sevoflurane group seen in Figure 2. The extreme outlier in the propofol group (240 minutes) had an average propofol infusion rate of 1.4 mg/kg/h and a body mass index of 38, possibly leading to drug accumulation in fatty tissues. One patient randomized to sevoflurane failed to receive allocated treatment due to the need for immediate return to the operating room due to bleeding. The patient returned from re-surgery during on-call hours and was given propofol for sedation. This patient was excluded from analysis. Median study drug administration was 165 minutes in the sevoflurane group and 185 minutes in the propofol group (Table II).

Discussion

The main finding of our study is that sevoflurane for sedation in the ICU following cardiac surgery is associated with shorter time to extubation and to adequate verbal response than propofol, with comparable memories after sedation and awakening, as well as comparable incidence of adverse events during emergence from sedation.

The shorter range in time from sedation stop to extubation after sevoflurane sedation (100 minutes vs. 240 minutes in the propofol group) implies that this sedative strategy makes it easier to predict and plan for extubation and continued respiratory care once sedation is stopped. While important clinical benefits of predictable wake-up times are still to be proven, physiotherapists in our ICU's are normally available only in daytime and are often involved and of significant help together with our ICU nurses in the continued respiratory and physiotherapy evaluations and interventions after extubation.

A likely explanation to the short wake-up times found with inhaled anesthetics for sedation is the pharmacokinetics of these drugs. Elimination of sevoflurane is independent of the patient's capacity to metabolize drugs, a feature not shared by intravenous sedatives. Patients that theoretically would benefit from a sedative with these features could be those with impaired renal or hepatic function and affected drug metabolism. Reliable neurological assessments and planned extubation after prolonged intravenous sedation may be difficult in such patients due to residual sedation (4,25). Indeed, a recently published study comparing sevoflurane with propofol and midazolam for sedation in critically ill patients demonstrated shorter wake-up times for ICU patients sedated with sevoflurane than for those receiving propofol or midazolam (26). A study of 70 patients by Rohm et al. (15) compared sevoflurane with propofol sedation following CABG. Rohm reported shorter and more predictable extubation times after sevoflurane sedation compared with propofol sedation, but also shorter hospital stay for sevoflurane sedated patients. We did not find any differences in ICU or hospital length of stay between groups. In our unit, patients remain in the ICU until the next morning per clinical routine. Moreover, patient discharge to rehabilitation units is often influenced by their bed occupancy, making it difficult to identify possible differences between groups with regard to hospital length of stay.

Titrating drug dosage, monitoring sedation, and sedation-minimizing-strategies have an impact on wake-up times. When reviewing trials of sedation following cardiac surgery, different propofol (average) infusion rates in the range of 0.64–2.71 mg/kg/h have been applied (20). Regardless the dosing tradition, it could be an advantage to use a low starting dose regime followed by early and repeated dose adjustments.

We did not find any differences between groups in adverse memories after sedation. A study comparing isoflurane with midazolam for long-term sedation indicated that delusional memories after isoflurane sedation are uncommon (14). In a recent sedation study of critically ill patients by Mesnil (26), hallucinations were noted in one-third of propofol or midazolam sedated patients and no hallucinations in those sedated with sevoflurane. Ringdal et al. (27) found an association between high daily doses of propofol in critically ill patients and the incidence of delusional memories, indicating that high doses of propofol for sedation may influence the incidence of delusions. In our study, the shorter wake-up times after sevoflurane sedation – with a potentially shorter transition between the sedated state and the fully awake state – were not associated with a lower incidence of delusional memories compared with propofol. The lack of difference between groups regarding delusional memories may depend on a true lack of difference in potential residual sedative effects. Furthermore, the administration of the study drugs in our trial was much shorter than in previous studies evaluating memories after inhaled sedation (14,27). Additionally, all patients in our study underwent CPB, a procedure that potentially may influence postoperative memory and cognitive function (28).

Our study has some limitations. The study was not double-blinded, due to practical reasons with the new sevoflurane delivery device. There are different traditions of dosing propofol following cardiac surgery, and our propofol starting dose might have been higher than in some other units, potentially affecting wake-up times. The loss to follow-up in the ICU Memory Tool test could possibly have been reduced with an additional follow-up after hospital discharge. Finally, while we classified patients as agitated only if they required treatment for agitation, we did not use a validated delirium instrument.

In conclusion, short-term sevoflurane sedation following CABG with CPB leads to a shorter and more predictable time to extubation and cooperation, with comparable adverse events and memories during emergence and ICU stay as propofol. Postoperative sevoflurane sedation was not associated with shorter ICU or hospital length of stay in our setting.

Acknowledgements

We thank Registered Nurse Helena Pietrzyk at the surgical ward and all nurses at the ICU for invaluable help in conducting this study.

Declaration of interest: The regional agreement on medical training and clinical research (ALF) between Stockholm County Council and Karolinska Institutet provided financial funding. Part of the funding was also provided by Lena and Per Sjöberg scholarship. Abbott Scandinavia AB sponsored purchase of sevoflurane (Sevorane^®) and Sedana Medical AB supplied the anesthetic conserving device (AnaConDa^®). Peter Sackey has received honoraria as a lecturer for Abbott Scandinavia AB and has been an Advisory Board participant for Baxter International Inc. The other authors state no conflicts of interest. The authors alone are responsible for the content and writing of the paper.