Adaptive support ventilation (ASV) is an advanced closed-loop mode of mechanical ventilation (MV) that maintains an operator preset minimum minute ventilation independent of patients' activity. ASV is based on a mode similar to pressure-controlled (Pinsp) synchronized intermittent mandatory ventilation. The settings for inspiratory pressure and mandatory rate are automatically and continuously adjusted according to continuous measurements of patients' lung mechanics and respiratory activity. Depending on the patient's status, ASV provides fully controlled ventilation (automatically managed pressure-controlled ventilation) or assisted ventilation (automatically managed pressure support ventilation; PS). One recent observational study shows that ASV adequately selects different tidal volumes (VT) and respiratory rate combinations based on respiratory mechanics in passive, mechanically ventilated patients.1
A respiratory weaning protocol based on ASV seems to simplify and accelerate ventilatory management.2 In line with these findings, ASV resulted in fewer changes and fewer manipulations of ventilator settings and ventilator alarms in cardiothoracic surgery patients.3 However, the patients in the relatively small studies mentioned above were so-called “fast-track” cardiothoracic surgery patients.4 Our institutional protocol is a non–fast-track protocol, thus allowing us to gather presently unavailable information on the effect of ASV on duration of intubation and ventilation in non–fast-track cardiothoracic surgery patients.
In the present study, we investigated the effect of ASV on time until tracheal extubation in non–fast-track patients after planned and uncomplicated coronary artery bypass grafting (CABG); this was the primary end point. Secondary end points were duration of assisted ventilation and switches from controlled ventilation to assisted ventilation. In addition, we studied the behavior of ASV with respect to applied VT, airway pressures, and respiratory rate and arterial blood gas (ABG) analysis results.
Patients and Setting
From October 2005 until July 2006, 128 consecutive post-CABG patients admitted to the 28-bed intensive care unit (ICU) of the Academic Medical Center, Amsterdam, The Netherlands, were included. The staffing consists of 140 ICU nurses, 8 full-time intensivists, 8 ICU fellows, and 10 residents of other specialties, such as internal medicine, anesthesiology, and surgery. The study protocol was approved by the local institutional review board, and preoperative written informed consent was obtained from eligible patients scheduled for elective CABG.
According to an open-label, randomized, controlled design, patients were assigned to receive MV involving either ASV (ASV group) or standard PC/PS ventilation (control group).
We created a homogenous group of adult (≥18 yr) patients after uncomplicated CABG, i.e., without a history of pulmonary disease or hemodynamic instability. We excluded patients with a history of pulmonary disease or pulmonary surgery. Patients who had an intraaortic balloon pump on admission to the ICU were excluded. Patients receiving inotropics and/or vasopressors at a rate higher than usual were also excluded. The upper limits in milligram per hour were dopamine (20), norepinephrine (0.5), dobutamine (25), and epinephrine at any rate. Patients were included and randomized on ICU arrival after CABG.
Cardiothoracic Surgery/Anesthesia Procedures
Patients were anesthetized according to our standard institutional protocol, beginning with 1 or 2 mg lorazepam as premedication and followed by etomidate, sufentanil, and rocuronium for induction of anesthesia and facilitation of intubation. During the surgical procedure, small doses of sufentanil are used as analgesic, and sevoflurane plus propofol is used to maintain anesthesia. Muscle relaxants are not given during the surgical procedure. Small doses of morphine and midazolam can be given at the end of the procedure.
Cardiopulmonary bypass was performed under moderate hypothermia (28°C–32°C), using a membrane oxygenator and nonpulsatile blood flow. At the end of anesthesia, all patients were transferred to the ICU with tracheal intubation. Anesthesiologists and surgeons in the operating room were blinded to inclusion and randomization of patients.
The postoperative ICU protocol was similar for both study groups and involved fluid resuscitation with normal saline and starch solutions, blood transfusion to maintain hemoglobin concentration (≥5.0 mmol/L), dopamine and norepinephrine in continuous infusion to achieve mean arterial blood pressure ≥70 mm Hg, and dobutamine and/or enoximone to achieve a cardiac index ≥2.5 L/min/m2 or a mixed venous oxygenation >60%. Sedation and analgesics were given according to the department's protocol for postoperative cardiothoracic surgery patients. Propofol was given for sedation via continuous infusion until core temperature was >36.0°C. When the target temperature was reached, propofol infusion was stopped. Acetaminophen (4 g/day) was started in all patients. Requirement for analgesia was assessed by attending ICU nurses during the entire ICU stay. Morphine was given in boluses of 1 to 2 mg IV. The boluses were repeated as needed. Postoperative shivering, if present, was treated with pethidine (25 mg IV). Neuromuscular blocking drugs were never used in the ICU.
During the study, all patients' lungs were ventilated by a Hamilton Galileo ventilator (software version GMP03.41f, GCP03.40a, GTP01.00; Hamilton Medical AG, Rhäzüns, Switzerland). Passive humidification of the ventilatory circuit was applied by means of an HME filter (Medisize Hygrovent S, Medisize, Hillegom, The Netherlands).
Initial Ventilator Settings.
Minute ventilation was set at 100% of the theoretical value based on predicted body weight (as a function of patient height)5 (“100% minute ventilation, equal to 100 mL · kg−1 · min−1”), oxygen inspiratory fraction (Fio2) of 50%, positive end-expiratory pressure (PEEP) of 10 cm H2O (maintained constant for 4 hr, after which the PEEP level was set at 5 cm H2O until extubation); maximum airway pressure was set 35 cm H2O (high-pressure pop-off). Thus, the absolute Pinsp limit for ASV was 25 H2O.
Flow trigger sensitivity was set at 2 L/s. An ABG analysis was performed 30 min after connection to the ventilator. If Paco2 was <3.5 kPa or >5.5 kPa, minute ventilation was decreased or, respectively, increased by (absolute) 10%. Any modification of the ventilator settings was to be controlled after 30 min by a new ABG analysis. Fio2 was adjusted to maintain arterial oxygen saturation (Sao2) of ≥95%. The trachea was extubated after achieving general tracheal extubation criteria (i.e., responsive and cooperative, urine output >0.5 mL · kg−1 · h−1, chest tube drainage <100 mL last hour, no uncontrolled arrhythmia, having a rectal temperature >36.0°C, a respiratory frequency of 10–20 breaths/min without machine-controlled breaths, Fio2 of 40%, and Pinsp 5–10 cm H2O for 30 min).
Initial Ventilator Settings.
PC ventilation, VT of 6–8 mL/kg predicted body weight, respiratory rate of 12–15 breaths/min, Fio2, PEEP, and flow trigger settings were identical to ASV group. An ABG analysis was performed after 30 min, and respiratory rate was decreased or increased by 2 breaths/min, to satisfy Paco2 criteria identical to the ASV group. Each modification of the ventilator settings was controlled 30 min later by another ABG analysis, and the adjustments were repeated in a manner similar to that of the ASV group. When patients tried to breathe spontaneously, the ventilatory mode was changed to PS ventilation set at 10 cm H2O and apnea security at 20 s. During PS ventilation, the flow trigger was set at 2.0 L/s, expiratory trigger sensitivity at 25%, and pressure ramp at 50 ms. In case of too many or too long apnoeic episodes, the patient returned to PC. Synchronized intermittent mandatory ventilation was never to be used. If satisfactory, the support level was decreased to a value between 5 and 10 cm H2O depending on VT. The trachea was extubated according to the criteria described for the ASV group.
The following data were collected. Patient characteristics: gender, age, weight and height; operation characteristics: number of bypasses, cardiopulmonary bypass time (pump time), and aortic clamp time; sedative and analgesic prescriptions: intraoperative and in the ICU.
The following outcome variables (see Definitions section below) were calculated for each patient: total duration of tracheal intubation; summed single assisted ventilation episodes; time to first single assisted ventilation episode; time to the assisted ventilation episode that was followed by tracheal extubation; and length of stay in the ICU.
In addition, ventilatory characteristics, including VT, PEEP, maximal airway pressure, Pinsp (in ASV group), and respiratory rate, were collected. These data were collected by a data logger connected to the ventilator (Hamilton data logger, version 3.27.1, Hamilton Medical AG).
Total duration of tracheal intubation was defined as the period from ICU admission until tracheal extubation. A single assisted ventilation episode in the ASV group was defined as an episode of ≥20 min during which the patient was breathing at least 5 breaths/min. A single assisted ventilation episode in the control group was defined as an episode of ≥20 min during which the ventilator was set in the PS mode. Pinsp was defined as the maximum airway pressure minus the level of PEEP.
Dose of Opiates and Sedatives
Opiate doses were all recalculated as morphine equipotent doses with the following formula: 10 mg of morphine = 0.1 mg fentanyl = 0.01 mg sufentanil.5 Doses of benzodiazepines were similarly converted to equipotent doses of diazepam using the following formula: 5 mg midazolam = 10 mg of diazepam = 50 mg oxazepam.6
The primary end point of this study was time until tracheal extubation. Secondary end points were the duration of assisted ventilation as the proportion of the total duration of MV and switches from controlled ventilation to assisted ventilation. In addition, we studied the behavior of ASV with respect to applied VT, airway pressures and respiratory rate, and ABG analysis results.
Sample Size Calculation
Based on a study of Petter et al.,3 in which a reduction of time until extubation from 3.2 (2.7–4.0) to 2.7 (2.1–4.2) hr with ASV was shown, we felt that a reduction of 10% of time until extubation would be a feasible and relevant difference between the groups. A sample size of 57 patients in each group was calculated to have 80% power to detect a 10% difference in means of time to extubation, assuming that the common standard deviation is 3 hr, and the mean time to extubation in the control group is 14 hr using a two-sided t test with a 0.05 two-sided significance level. Because of suspected technical difficulties with the data logger, we decided to include 64 patients in each group. Randomization was performed with envelopes. No stratification was performed.
Descriptive statistics were used to summarize patient characteristics. Categorical variables were compared between groups by χ2 tests. If normally distributed, continuous values were expressed as means ± sd; otherwise medians and interquartile ranges were used. Differences in % of time patients were breathing spontaneously were tested with a t test; all other differences were tested with nonparametric tests. Because the distribution of the data did not permit the use of parametric tests as originally planned, differences between time parameters from MV (duration) were compared by log rank tests and were expressed as medians (interquartile range). All analyses were performed using SPSS for windows (version 14.0, SPSS Inc., Chicago, IL).
One hundred twenty-eight patients were randomized, all of whom were analyzable with respect to the primary end points. For seven patients, data collection was incomplete because of data logger failures (four patients in the ASV group and three in the control group). Accordingly, only 121 patients were analyzable with respect to the secondary end points.
Patient characteristics in the ASV group and the control group are presented in Table 1. None of the patients from either group had to be reintubated or failed to wean. Dependency on inotropics and/or vasopressors was not different between the study groups (data not shown).
Duration of MV and Assisted Ventilation Episodes
No differences were found for total duration of tracheal intubation (Fig. 1). In the ASV group, time until tracheal extubation was 16.4 (12.5–20.8) hr. In the control group, time until tracheal extubation was 16.3 (13.7–19.3) hr (P = 0.97). The sum of time patients were on assisted ventilation, expressed as the median percentage of total duration of intubation and MV, was significantly shorter in the ASV group (43% [28–67]) compared with the control group (52% [33 – 75]) (P < 0.05).
Time until first episode of assisted ventilation (as defined in the Methods section) was not different between the ASV group (6.4 [2.6–9.9] hr) and the control group (7.0 [4.0–8.6] hr) (P = 0.95). Duration of the first assisted ventilation episode was shorter in the ASV group (0.9 [0.5–2.5] hr) than in the control group (2.8 [0.7–4.7] hr) (P = 0.002).
The number of assisted ventilation episodes was not different between groups: 2.5 (1.0–3.0) episodes in the ASV group and 2.0 (1.0–4.0) in the control group (P = 0.92). However, if we consider the frequency of all assisted ventilation episodes (without duration restriction, as defined in the Methods section), there were significantly more episodes in the ASV group (43.0 [14.0–74.0]) compared with the control group (4.0 [2.0–9.0]) (P < 0.001).
Time until start of the assisted ventilation episode that ended with tracheal extubation was 14.0 (9.3–16.7) hr in the ASV group and 10.9 (8.7–15.7) hr in the control group (P = 0.13).
VT were larger during assisted ventilation compared with controlled ventilation in both groups. With ASV, VT were significantly larger during controlled breathing episodes than with PC ventilation (Table 2). No differences for VT were found between the ASV group and control group during assisted ventilation. Pinsp was higher with both modes during controlled ventilation. Pinsp decreased with spontaneous breathing. In the control group, lower Pinsp was recorded as compared with the ASV group (Table 2). The respiratory rate was lower during controlled ventilation with ASV as compared with control. There were no differences in respiratory rate during assisted ventilation (Table 2). Compared with control patients, with ASV slightly lower Paco2 levels were recorded during both controlled and assisted ventilation.
No differences were found between the two groups regarding analgesics and sedatives administered during the surgical procedure. Indeed, patients in the ASV group received median 2.72 (1.83–3.54) mg/kg morphine equipotent doses and control patients received 2.74 (1.57–3.38) mg/kg morphine equipotent doses (P = 0.54). Total administration of diazepam equipotent doses was 0.35 (0.24–0.52) mg/kg in the ASV group and 0.34 (0.21–0.44) mg/kg in the control group (P = 0.54). Also, duration of propofol infusion in the ICU was not different in the ASV group (median 5.1 [2.8–7.6] hr) and the control group (5.3 [3.3–13] hr) (P = 0.93).
In this study of postoperative weaning of patients after planned and uncomplicated non–fast-track CABG, automatic weaning with ASV was feasible and safe. We found time until tracheal extubation with ASV to be similar to time until tracheal extubation with standard weaning. The duration of time patients were breathing with assisted ventilation, expressed as the percentage of total duration of intubation and MV, seemed to be significantly shorter in the ASV group. With ASV, however, significantly more episodes of assisted MV were started. Although VT were larger with ASV during controlled MV, no differences in VT were found during assisted ventilation.
Use of automated systems such as ASV is a way of addressing the current workforce shortage. Some other possible important reasons to use automated ventilation systems are cost containment, the need to provide a better medical product, and the prevention of human error.7 Built-in protocols and safety limits could/should lead to less noxious ventilation and more rapid weaning. More importantly, it is recommended to allow patients to breathe spontaneously for as long as possible for several reasons: first, with spontaneous ventilation more lung tissue will be kept recruited,8 and second, muscle atrophy of the diaphragm can occur rapidly.9 The present study shows, not only that the use of ASV is feasible and safe in the studied patient group, but also that ASV allows for frequent switches between controlled ventilation and assisted ventilation, thereby possibly preventing derecruitment and muscle atrophy.
Our hypothesis was that ASV would result in a reduction in tracheal intubation time and a larger percentage of assisted ventilation. However, this was not what we found. The reason we did not find a larger percentage of assisted ventilation during ASV might be that ASV patients were able to switch between PC and PS any time, in contrast to control patients, who were dependent on the attending ICU nurses and/or physicians for changes in ventilator settings. Consequently, and because of our definition of an assisted ventilation episode, this may have resulted in an underestimation of the percentage of assisted ventilations during ASV. There were more assisted breathing periods (without duration restriction), however, in ASV patients. Switching between controlled and assisted ventilation (and vice versa) with nonautomated weaning requires decision-making by individuals caring for the patient. This not only requires action by a caregiver but also an assessment of the clinical situation.
Another reason for finding a lower percentage of assisted ventilations with ASV might be that with the settings used, ASV slightly overventilates patients' lungs. Indeed, with ASV, Pco2 levels were lower than in control patients. This may have resulted in less drive of ASV patients to start (or continue) assisted ventilation during the weaning process.
In line with the previous comment, one important difference between our study and earlier ASV studies was that we did not allow minute ventilation to be changed other than for reasons of Paco2. By contrast, in a study on weaning of patients after cardiothoracic surgery, Sulzer et al.2 defined several phases: in phase 1 the ASV mode was set at 100% of the theoretical value of minute ventilation; when spontaneous breathing occurred, the ASV setting was reduced by 50% (phase 2) and again by 50% (phase 3), after which the trachea was extubated. In an observational study by Cassina et al.10 in 155 fast-track cardiothoracic surgery patients, a slightly different strategy was used. After the patients had resumed spontaneous breathing in the ASV mode, the ventilator was switched to PS with a low level of support. The fact that we did not use one of these strategies may not only explain the absence of a shorter time until extubation with ASV in our study, but may also explain the shorter duration of assisted ventilation with ASV: a “too-high” minute ventilation decreases the patient's need to trigger the ventilator, masking the ability to breathe spontaneously. The slightly but significantly lower Paco2 levels with ASV are in accordance with this suggestion. Additional studies are needed to determine the effect of a more proactive weaning protocol, such as the one used by Sulzer et al.2
Our findings are in contrast to those of a study on postoperative weaning of cardiothoracic surgery patients. Indeed, a significant reduction in tracheal intubation time was found for patients after fast-track surgery. In that particular study, tracheal intubation time with ASV was 3.2 (2.5–4.6) hr versus 4.1 (3.1–8.6) hr in the control group.2 The patients in our study, however, can be considered to be non–fast-track patients. This may explain, at least in part, the lack of a difference in tracheal intubation time between the two groups in our study. The advantageous effect of automatic weaning on total duration of tracheal intubation may have been lost because of the relatively long period of controlled ventilation in both groups.
The longer overall duration of ventilation in our study, as compared with other studies, may have been because of several other factors. One reason may be the relatively high dosages of intraoperative opioids in both groups. The use of continuous infusion of propofol, instead of bolus infusion of propofol linked to a target Ramsey score, may also have led to the longer duration of postoperative ventilation. One other explanation may be the Paco2 target of 3.5–5.5 kPa, which can be considered to be a low target for postoperative CABG patients; a higher target may have led to a faster reduction of the PS level in both groups and possibly to shorter ventilation times. The level of PS used in the assisted ventilation phase as a prerequisite for considering extubation is an additional possible factor, and related to the above-mentioned Paco2, again choosing a higher level could have led to a shorter time until extubation. Finally, in our department CABG patients are discharged to the stepdown facility no sooner than the following morning (i.e., the day after surgery). Thus, from an organizational point of view, there is no need to extubate the trachea “early.” To be even more explicit, the fact that 50% of our patients were not tracheally extubated after 16 hr might have been because of the fact that patients are left intubated overnight for this latter reason. These several factors hamper interpretation of our study results; thus, additional studies in centers with other postoperative strategies are mandatory to see whether ASV influences time until tracheal extubation.
In the study by Sulzer et al.,2 it was shown that the reduction in tracheal intubation time with ASV seemed to be mainly caused by a shortening of time between ICU admittance and start of assisted ventilation and not the period during which the work of breathing is progressively transferred from the ventilator to the patient. In contrast, in our study, time until the first episode of assisted ventilation was not different for the study groups. This may be due to the result of our definition of assisted ventilation episodes. The reason for choosing a 20-min period with at least 5 breaths/min was based, however, on another study with ASV.3 We consider this a clinically relevant period.
With ASV, VT were significantly larger than in the control group during controlled ventilation. Although most of the literature on lung-protective ventilation focuses on patients with acute lung injury/acute respiratory distress syndrome, the lungs of patients without lung injury might also benefit from lung-protective ventilation.11 Whether the differences in VT in our study are clinically relevant remains to be investigated in additional studies. The same is true for patients with chronic obstructive pulmonary disease, a group which we purposely excluded from this study. Theoretically, in the case of chronic obstructive pulmonary disease, ASV could have an advantage since the expiratory time constant is the most important determinant with this automatic ventilation mode. This needs to be addressed in future studies as well.
Apart from the fact that we did not adjust minute ventilation over time, as mentioned above, our study suffers from some important limitations. For practical reasons, as with most studies on MV, in our study MV was not performed in a blinded fashion. Another major shortcoming is that our weaning protocol mandates that most of the steps during weaning of uncomplicated cardiothoracic surgery patients are performed by ICU nurses. Indeed, in the majority of cases, the decision to proceed with tracheal extubation is initiated by ICU nurses. It could be that differences in organization of respiratory care lead to other results. Indeed, when decisions on MV are to be made by caregivers (such as respiratory therapists) who are not constantly at the bedside, ASV may be more beneficial. However, we explicitly wanted to determine the effect of ASV on duration of intubation in our clinical setting. As a consequence, this limits the generalizability of our results.
Although not statistically significant, there was a large difference in the median length of stay. The length of stay in our ICU, however, is highly dependent on external factors such as the availability of beds in the step-down facility.
In conclusion, weaning automation with ASV is feasible and safe in non–fast-track CABG patients. Time until tracheal extubation with ASV equals time until tracheal extubation with nonautomated weaning. ASV allows for frequent (automatic) switches from controlled to assisted ventilation. Future studies should evaluate whether a more proactive weaning strategy with ASV, aiming at earlier reduction of minute ventilation, increases duration of assisted ventilation and shortens time until tracheal extubation.