The sedation of critically ill patients remains an integral treatment in the intensive care unit (ICU). Historically, intravenous sedation (using propofol, benzodiazepines, or other compounds) has been the standard of care in the ICU; however, the development of an anesthetic-conserving device (AnaConDa, Sedana Medical, Co. Kildare, Ireland) enabled the use of volatile sedation. Thus, a historical technique was resurrected using a new technology. The high efficacy and reduced side effects of volatile sedation make it an attractive option for patients in the ICU and for those receiving support with extracorporeal membrane oxygenation (ECMO).
The Aim of This Study
We aimed to determine whether sedation influences the outcomes of patients supported using ECMO, and to compare intravenous sedation with volatile sedation in an ECMO setting. In addition, the ECMO run time, the sedation duration, the ventilator settings, and the incidence of delirium after sedation were analyzed. The results of this study might improve the management of patients receiving ECMO, thereby improving their outcomes.
Data were collected from 99 patients treated with ECMO between January 1, 2008 and December 31, 2014, in the 10 bed surgical ICU of a university hospital. The inclusion criterion was that ECMO was conducted for at least 24 hours. Eight patients did not meet this requirement and were not included in the evaluation (Table 1).
The average age of the 99 patients was 54 years (range = 16–86 years). The indications for ECMO were as follows: trauma (n = 48); extrapulmonary sepsis (n = 24); respiratory failure (n = 18); resuscitation (n = 6); and others (n =5) (Table 1).
The subgroup assignment was not randomized. The patients were assigned to different sedation regimes based on the timeline between January 1, 2008 and December 31, 2014. Isoflurane has been used as an alternative to propofol at our ICU since January 2011. The first instance of isoflurane sedation for ECMO was performed in September 2011. Subsequently, volatile sedation was always considered for patients receiving ECMO.
Data were collected retrospectively using the perfusionists' ECMO protocols and via a review of patient records. IBM SPSS Statistics, version 22 (International Business Machines Corporation, Armonk, NY) was used to analyze the descriptive data.
The correlations were analyzed using the χ2 test to determine whether a relationship existed between sedation type and “survival at hospital discharge.” Sedation type was the independent variable, and survival was the dependent variable. Nonparametric data were analyzed using the Mann–Whitney U test and the Kruskal–Wallis test; p < 0.05 was considered to indicate significance.
The objectives of the statistical analyses were to compare
- the patient outcomes before sedation with outcomes after propofol versus isoflurane use;
- the influence of sedation using propofol versus isoflurane on ventilation, length of stay, duration of therapy, and incidence of delirium.
In an additional analysis, the patients were divided into subgroups based on whether they 1) required bilevel positive airway pressure (BiPAP); 2) required continuous positive airway pressure (CPAP); or 3) were able to successfully transition to CPAP without requiring intermittent BiPAP. Because several patients died while intubated, the number of patients in these subgroups changed over time.
The incidence of delirium after sedation was analyzed based on the Ramsay score. For this analysis, only the patients who successfully transitioned to CPAP were included.
Each patient was pseudonymized with a specific study number. Only authorized individuals had access to the database.
Anesthetic-Conserving Device (AnaConDa)
The AnaConDa vaporizes liquid anesthetic agents to integrate them into mechanical ventilation (MV; Figure 1). This device facilitated the use of volatile anesthetics such as isoflurane and sevoflurane in an ICU setting. The volatile anesthetic is supplied in liquid form via a syringe pump into the evaporator (a porous rod). The anesthetic diffuses over the large surface of the rod is adsorbed, and vaporizes via the inspiratory gas flow for delivery into the lungs. Volatile anesthetics are not highly metabolized; therefore, a large amount of the anesthetic is present in the exhaled air (i.e., the end-tidal concentration is high). During expiration, the gas mixture is passed through activated carbon fibers, where the anesthetic is deposited. During inspiration, the anesthetic triggers the filter, and the anesthetic is resupplied to the patient. Therefore, the patient receives a mixture of newly released anesthetic from the evaporator and recycled anesthetic from the carbon fiber filter. The end-tidal concentration of anesthetic was monitored using a gas analyzer (Vamos monitor, Draeger, Lübeck, Germany).
The end-tidal concentration of isoflurane during volatile sedation has been determined to range between 0.2% and 0.6%. A bolus of 1.2 ml of anesthetic was administered to prime the line and support the initial anesthetic rate of 3 ml/h. Depending on the minute volume and the desired concentration, an anesthetic rate of 2–5 ml/h was selected to achieve an end-tidal isoflurane concentration of 0.3–0.5%. If the sedation was inadequate, then a bolus of 0.5 ml of liquid isoflurane was injected. The end-tidal concentration of isoflurane was decreased by stopping the pump syringe for a few minutes, and both the patient's reactions and the measured levels of isoflurane according to the gas analyzer were observed.
The Ramsay score was used to evaluate patients for delirium after sedation by applying one of the following 6 grades.
- 1 = The patient is anxious, agitated, or restless.
- 2 = The patient is cooperative, oriented, and quiet.
- 3 = The patient responds only to speech.
- 4 = The patient is asleep, but responds promptly to either a light tapping or a loud stimulus.
- 5 = The patient is asleep, but responds sluggishly to either a light tapping or a loud stimulus.
- 6 = The patient is asleep and unresponsive.
Extracorporeal Membrane Oxygenation and Ventilation Settings
In accordance with the ARDS Network Study Group guidelines, the ventilation parameters were selected to ensure lung-protective ventilation. The following indications for extracorporeal lung support (iLA/VV ECMO) were selected:
- Hypoxemia (pO2/FiO2 < 200 mm Hg)
- Tidal volume > 4–6 ml/kg ideal body weight (BW)
- Pinsp < 30 cm H2O
- Severe hypercapnia with increased pCO2, if pH < 7.2.
In addition, increased need for vasopressors (epinephrine > 1 mg/h) was the dynamic parameter used to estimate the severity of lung failure.
Interventional lung assist (iLA; NovaLung, Hechingen, Germany) was provided using a pumpless system, and CardioHelp HLS (Marquet, Raststatt, Germany) was used for venovenous (VV) or venoarterial (VA) ECMO.
Patients were always cannulated at the bedside in the ICU; for iLA, the patients were cannulated percutaneously via the femoral artery (15 French cannula) and femoral vein (17 French cannula). For ECMO, the patients were cannulated percutaneously into the right internal jugular vein and one of the femoral veins for VV or both femoral vessels for VA. A 31 French double-lumen cannula (Avalon, MAQUET, Rastatt, Germany) was inserted into the right jugular vein under fluoroscopy.
Heparin was administered continuously, and the need for heparin was monitored based on partial thromboplastin time (target > 50 s in the absence of bleeding risk due to trauma or postsurgery).
The blood flow after initiating ECMO was 100–120 ml/kg BW per minute. The flow was calibrated based on the arterial blood gases according to the arterial partial pressure. The target oxygen partial pressure (pO2) was 60 mm Hg, and the target arterial oxygen saturation (SaO2) was greater than 90%. Oxygen was the sweep gas, and the flow rates ranged from 1 to 15 L/minute. The flow was adjusted according to PaCO2 and pH. A pH greater than 7.2 was desired.
The tidal volume, minute ventilation, positive end-inspiratory pressure, and FiO2 were decreased to facilitate lung tissue healing after cannulation. To limit the potential risk for atelectasis, the positive end-expiratory pressure (PEEP) was not reduced below 10 cm H2O. Depending on the compliance of the respiratory system and the amount of PEEP, the peak pressure was chosen such that a tidal volume of 4 ml/kg ideal BW was strictly achieved. The maximum pressure limit was set to 30 cm H2O. The primary goal of maintaining ventilation and initiating weaning from the extracorporeal device and ventilator was to reduce the sedation target of the Ramsay score. Using a target Ramsay score of 2–3, spontaneous breathing efforts while on ventilatory support (CPAP-mode) were allowed during ongoing ECMO therapy.
Venoarterial ECMO-treated patients were included in the analysis although the VA and VV modes used different approaches. The aim of sedation was identical between VA and VV ECMO: a brief period of deep sedation (only as long as necessary) to facilitate early, continuous, spontaneous breathing on the ventilator as well as weaning from the ventilator.
To ensure that the groups were comparable, patient age and body mass index (BMI) were analyzed for significant differences. Neither age nor BMI significantly differed between the groups. The Simplified Acute Physiology Score (SAPS) II and III values were compared to assess the severity of critical illnesses. However, an analysis of the medians yielded no significant differences between the groups (Table 2).
The patients were subsequently divided into three nonrandomized subgroups. The first subgroup received isoflurane (n = 42) administered using the AnaConDa; the second subgroup received propofol (n = 43); and the third subgroup received an alternative form of sedation (n = 6), usually midazolam. In most cases, the additive sedative drug was either midazolam (n = 35) or ketamine (n = 32). Adequate analgesia was assured using sufentanil (n = 91).
Primary End Point
Of the 91 patients receiving ECMO, 45 (49.5%) patients survived. Of the patients who received propofol (n = 43), 24 (55.8%) survived to discharge. In addition, of the patients who received isoflurane (n = 42), 20 (47.6%) survived. Five of the six patients who received alternative sedation died during hospitalization. The significance of this difference in mortality was negligible because of the small sample size (Table 3).
The null hypothesis assumed that no correlation existed between the sedation drug and survival at hospital discharge (Table 4). Assuming a significance level of p < 0.05, the null hypothesis was retained in this analysis (p = 0.189). Therefore, sedation with propofol versus isoflurane was not correlated with survival.
Secondary End Points
First, the duration of various therapeutic measures and the lengths of stay in different settings were examined. The examined variables included the duration of ECMO therapy, the duration of sedation, and the total lengths of both hospitalization and ICU stay. Regarding the duration of sedation, the “alternative sedation” subgroup was excluded because of its small size. All the lengths of stay were expressed in days (Table 5).
No difference in the duration of ECMO therapy, the length of ICU stay, or the length of hospitalization was observed between sedation with propofol and sedation with isoflurane.
The Mann–Whitney U test was used to assess the relationship between sedation type and ventilation parameters. The transition to either BiPAP or CPAP (i.e., time to BiPAP or CPAP) was the first parameter analyzed, and the toleration of continuous CPAP (i.e., time to continuous CPAP) was the second parameter analyzed. The BiPAP/CPAP conversion refers to the time required for transition from ECMO to CPAP. “Continuous CPAP” reflects the interval in days between the first CPAP trial and the onset of CPAP not interrupted by BiPAP.
Neither the duration of ventilation nor the early adjustment of continuous CPAP significantly differed between the sedation subgroups. However, the interval to the first change-over of the ventilation mode was significantly shorter among the patients who received isoflurane (p = 0.030; Table 6).
In addition, the incidence of delirium was analyzed based on the Ramsay scores. The analyzed cohort included 57 patients because the remaining patients from the alternative sedation subgroup were included. The Mann–Whitney U test was used to determine the significance of the differences in the incidence of delirium between the subgroups (Table 7). No significant difference in the incidence of delirium was detected between the propofol and isoflurane subgroups.
Propofol and isoflurane sedation did not differentially influence patient outcomes in the setting of ECMO. In addition, the type of sedation did not appear to affect the secondary end-points examined in this study. However, differences in MV parameters were observed between sedation types, as patients who received isoflurane sedation transitioned from BiPAP to CPAP earlier than those receiving other types of sedation.
Neither ventilation duration nor time to implementation of continuous CPAP differed between the sedation subgroups. Complications such as pneumonia can occur in the context of MV.1 Therefore, clinicians must always strive to wean patients from MV as quickly as possible.2 Weaning from MV requires concomitant weaning from sedation. With respect to weaning, the safety profile of isoflurane is superior to that of propofol. However, the depth of sedation always plays a central role in weaning because even a prolonged wean can affect the outcomes of patients receiving ECMO, regardless of the type of sedation used.
Although no significant differences in side effects or complications were noted between isoflurane sedation and propofol sedation, total intravenous anesthesia has been associated with unfavorable side effects and complications. Intravenous sedatives interfere with hemodynamic regulation, potentially causing myocardial depression or bradycardia. Volatile anesthetics also affect hemodynamics by causing vasodilatation, and for this reason, volatile anesthetics are often used to adjust systemic vascular resistance during cardiopulmonary bypass and surgery. Propofol infusion syndrome (PRIS) is an additional complication associated with this type of sedation.3
Rhabdomyolysis, lactic acidosis, and kidney disease have also been linked to propofol sedation. The PRIS is a rare complication, but its mortality rate is staggering. In addition, one patient in this study developed allergic alveolitis attributed to a soy protein in propofol.
The high fat content of propofol also warrants the regular monitoring of the levels of triglycerides, lipase, and amylase because these factors might be elevated under conditions of long-term sedation.4 The membrane exchange rate was not affected in either the propofol or isoflurane sedation subgroup in our population. Propofol is markedly sequestered within the extracorporeal circuit in vitro. This sequestration may also affect the plasma propofol concentration in vivo.5 Lemaitre et al.6 observed important reductions in the propofol concentration in an ex vivo ECMO circuit primed with whole blood. In their model, the propofol concentration decreased rapidly; 70% of the baseline propofol concentration was lost after only 30 minutes. Further experiments demonstrated that oxygen exposure and contact with polyvinylchloride tubing were responsible for 70% and 85% of propofol loss after 45 minutes, respectively.
Volatile anesthetics represent an excellent alternative to propofol because of their high efficacy and reduced side effect profiles.7 Volatile anesthetics are potent bronchodilators. The advent of the AnaConDa has made the use of drugs such as isoflurane possible in the ICU. This advancement has been advantageous to clinicians who practice in these settings. Although rapid variations in the level of sedation are possible, isoflurane does not impair either liver or kidney function because only a small portion of the drug is metabolized and because the rapid exhalation of isoflurane prevents its accumulation in the body.
The inhalation route of isoflurane administration has one limitation. Polymethylpentene (PMP) in the oxygenator's membrane is not permeable to isoflurane; thus, it must be administered via the ventilator. The amount of sedative transferred into the patient's bloodstream is therefore affected by the tidal volume achieved at the ventilator and depends on the degree of respiratory failure experienced by the patient. In the current cohort, we followed the principle of adequate analgesia (sufentanil), anxiolysis, and delirium prophylaxis (low dose of haloperidol, clonidine, or lorazepam), with the lowest required level of sedation. This drug selection did not suppress spontaneous breathing activity and enabled early and continuous spontaneous breathing, which is a physiologic behavior.
When a mechanically ventilated or spontaneously breathing patient inhales isoflurane (volatile sedation), the gas washes into the functional residual capacity and the alveoli. The alveolar gas enters the pulmonary circulation, circulates through the body, and returns to the lungs partially depleted. The cycle repeats, gradually increasing the amount of anesthetic in the pulmonary circulation until matching the anesthetic concentration in the blood returning to the lungs, which is also the concentration of anesthetic to which the brain is exposed.8 It would be intriguing to examine whether isoflurane exchange continues in the lungs after oxygen and carbon dioxide transport have ceased. We do not have much data related to this line of investigation. Only one person survived after cessation of the MV. His lung function was only supported on VV ECMO. Drugs were titrated to allow for spontaneous breathing. The administration of isoflurane was stopped at the moment of MV cessation.
During the pre-PMP era, isoflurane was used as a sedative via a silicone oxygenator for patients of all ages on ECMO.9,10 Isoflurane withdrawal was recognized as an adverse event. Hallucinations and seizures after prolonged administration of isoflurane reported.11 No similar symptoms were noted in the patients of our cohort.
However, despite these advantages of isoflurane sedation, patient outcomes did not appear to be associated with the specific type of sedation in this study. Nevertheless, to improve the outcomes of patients receiving ECMO, new sedation methods must be developed when sedation is needed. The possibility exists that patients receiving ECMO might not require sedation because previous studies have demonstrated that “awake ECMO” can be used as a bridge to lung transplantation,12 thereby enabling both clinicians and patients to avoid the complications of sedation and MV. Future studies involving the use of “awake ECMO” are needed to demonstrate its effectiveness.
This article describes the results of a retrospective, observational, single-center study that investigated the effects of propofol sedation compared with isoflurane sedation on patient outcomes after ECMO. No differences in patient outcomes between isoflurane and propofol sedation were identified under the condition of ECMO; however, isoflurane appears to offer a slight advantage with respect to side effects and controllability. This report might improve our understanding of sedation in terms of patient outcomes after ECMO and provide insight into the effects of spontaneous breathing during ECMO.