A Randomized Controlled Trial of Adaptive Support Ventilation Mode to Wean Patients after Fast-track Cardiac Valvular Surgery

ADAPTIVE support ventilation (ASV) is a closed-loop mode of ventilation in which the ventilator adjusts the inspiratory pressure to achieve a tidal volume that minimizes the work of breathing based on the Otis equation. In addition, the ventilator switches between control and supports breaths based on the presence or absence of spontaneous breathing effort. Previous studies have demonstrated that automated weaning using ASV mode results in faster^1,2  or equivalent weaning time³  after fast-track coronary artery bypass surgery compared with protocolized weaning using synchronized intermittent mandatory ventilation (SIMV) with pressure support^2,3  or pressure-regulated volume control with automode.¹  However, there are no studies examining automated weaning in patients undergoing valvular surgery. Morbidity and mortality are higher in this group, and they are forming an increasingly large proportion of cardiac surgical patients.⁴  Despite this, there is a dearth of data on fast-track cardiac valvular surgery. Almost all randomized controlled trials of patients undergoing fast-track cardiac surgery have focused on coronary artery bypass surgery predominantly⁵  or exclusively.^1–3,6–23 In one study, half the patients underwent valvular surgery.²⁴  Thus, it is important to study the effect of different weaning modalities in this group, rather than simply extrapolating from studies of patients undergoing coronary artery bypass grafting.

We hypothesized that in patients undergoing elective fast-track cardiac valvular surgery, ASV would be associated with a shorter duration of ventilation compared with physician-directed weaning and carried out an unblinded, randomized, controlled, parallel-arm trial to test this hypothesis.

The study was approved by the Joint Chinese University of Hong Kong-New Territories East Cluster Clinical Research Ethics Committee and registered at the Centre for Clinical Trials of The Chinese University of Hong Kong, Hong Kong, China (CUHK CCT00308). Written informed consent was obtained from all participants.

The study was carried out in the 22-bed adult multidisciplinary intensive care unit (ICU) of a 1,400-bed tertiary referral university hospital. The ICU is the sole adult ICU in the hospital. Cardiac surgical patients receive 1:1 nursing at all times. The unit was staffed with six doctors (of whom approximately half were intensive care specialists) during the day and two resident doctors (one of whom was an anesthesiology or intensive care specialist), with a further intensive care specialist available for consultation at night.

All patients scheduled for elective cardiac valvular surgery between May 2012 and March 2013 were screened for eligibility. Patients undergoing isolated valve surgery and those undergoing valve surgery combined with another cardiac surgical procedure were included. Preoperative exclusion criteria were acute or chronic obstructive pulmonary disease, serum creatinine concentration greater than 200 μmol/l, serum aspartate transaminase concentration greater than 80 U/l, left ventricular ejection fraction less than 30%, history of seizures or stroke, and age older than 80 yr or younger than 18 yr. Postoperative exclusion criteria were chest tube drainage greater than 500 ml/h, reoperation, myocardial infarction, need for high-dose inotropes or vasopressors or intraaortic balloon pump, and refractory hypoxemia with an arterial oxygen tension to fractional inspired oxygen concentration ratio less than 150 mmHg.

Patients were randomized to receive and be weaned using ASV or be ventilated and weaned as directed by duty clinicians (control group). Simple randomization with 1:1 allocation was carried out according to a computer-generated sequence using sealed opaque envelopes which were opened after the patient’s arrival to the ICU. The clinical trial design was a parallel-arm design. The CONSORT diagram (fig. 1) details the results of screening and randomization.

Patients were premedicated with oral midazolam 0.1 to 0.2 mg/kg unless contraindicated. Anesthesia was induced with various combinations of IV fentanyl 2 to 10 µg/kg, IV midazolam 20 to 50 μg/kg, supplemented with propofol up to 50 mg, and/or sevofluvane up to end-tidal concentration of 6.0%. Paralysis was achieved using IV rocuronium 1 mg/kg. Patients were intubated after induction and paralysis and were ventilated with oxygen-enriched air using volume-controlled ventilation with automatic adjustment of flow to minimize airway pressure, with tidal volume of 8 to 10 ml/kg actual body weight and positive end-expiratory pressure of 5 cm H₂O at a frequency of 12 breaths/min. These were adjusted to maintain an end-tidal carbon dioxide tension of 30 to 35 mmHg. During cardiopulmonary bypass, ventilation was stopped, and 5 to 10 cm H₂O of continuous positive pressure was applied to the lungs. Anesthesia was maintained with a combination of IV propofol (0 to 8 mg kg⁻¹ h⁻¹) and IV remifentanil (0 to 0.2 µg kg⁻¹ min⁻¹) to keep the bispectral index scale between 40 and 60. Boluses of IV fentanyl 50 to 100 µg were given as needed for intense stimulation and on rewarming. Atracurium was infused at 0.3 mg/kg per minute IV to provide muscle relaxation. IV morphine 0.1 to 0.3 mg/kg was given to provide intraoperative and postoperative analgesia. For hemodynamic fluctuations unrelated to depth of anesthesia, IV administration of phenylephrine 50 to 100 μg, ephedrine 6 mg, metoprolol 1 mg, glyceryl trinitrate infusion, atropine 0.3 mg, inotropes, or fluid boluses were used according to the clinical situation.

Surgery was carried out using standard procedures through a median sternotomy by a single surgical team. Mechanical or bioprosthetic valves were used for valve replacement at the discretion of the surgical team.

After surgery, the patients were transferred to the ICU sedated with propofol (1 to 2 mg kg⁻¹ h⁻¹). On arrival in the ICU, paralysis was reversed if necessary (based on train of four) with neostigmine, and sedation was stopped. Patients were managed according to standard unit protocols except for mechanical ventilation. In brief, arterial blood pressure, central venous pressure, electrocardiogram, and pulse oximetry were monitored continuously. Normal saline, lactated Ringer’s solution, and gelatin-based colloid solutions were used for fluid resuscitation. Hemoglobin concentrations were maintained 8 g/dl or greater. Dopamine or epinephrine was used to maintain mean arterial pressure 60 mmHg or greater and glyceryl trinitrate and sodium nitroprusside to treat hypertension (mean arterial pressure ≥100 mmHg). The bedside nurses assessed analgesic requirements and gave boluses of 1 to 2 mg morphine to a total of 10 mg on patient request if fully awake or if still sedated when hypertensive and restless.

All patients were ventilated with a Hamilton-G5 ventilator (Hamilton Medical GA, Rhäzuns, Switzerland; software version 2.1X). Patients randomized to the ASV group were managed according to the algorithm in figure 2. If the patient was not successfully weaned within 8 h, the weaning protocol was discontinued, and ventilatory management was directed by duty physicians.

Ventilation of patients in the control group was managed at the duty physicians’ discretion. Patients in both groups were assessed for extubation when the bedside nurse observed that they met the following criteria: responsive and cooperative; Fio₂ less than 40%; Pao₂/Fio₂ more than 150; positive end-expiratory pressure less than or equal to 5 cm H₂O; hemodynamically stable; urine output exceeds 0.5 ml kg⁻¹ h⁻¹; the last hour chest tube drainage less than 100 ml; no uncontrolled arrhythmia; and rectal temperature is above 36.0°C. If these criteria were fulfilled, the physician on duty was notified to assess and extubate the patient.

The primary endpoint of the study was duration of mechanical ventilation in the ICU. Secondary endpoints were duration of intubation in the ICU, duration of ICU stay, duration of hospital stay, postoperative complications, and mortality. Other data collected are listed in tables 1–7.

The sample size was calculated by PASS 11.0 (NCSS, Kaysville, UT). From a previous study of ASV in patients undergoing coronary artery bypass surgery and managed postoperatively in our unit,¹  we estimated the duration of ventilation would be 300 min with an interquartile range of 205 to 365 min. Based on this estimate, an effect size of 90-min reduction in duration of ventilation, power of 80%, a trial designed to demonstrate superiority and statistical significance threshold of P value less than 0.05, we calculated that a total of 56 patients would be required. Allowing for a 20% dropout rate, we planned to recruit 68 patients in total.

Parametric continuous data are expressed as mean (SD), nonparametric continuous data are presented as median (interquartile range), and categorical data as number (%).

The Shapiro–Wilk test was used to check data for normality. Student t test (independent sample) and Mann–Whitney U test were used for continuous parametric and nonparametric variables, respectively. In particular, the difference in primary outcome (duration of mechanical ventilation) was analyzed using the Mann–Whitney U test, and the median difference (95% CIs) in duration of mechanical ventilation was estimated using quartile regression. Chi-square or Fisher exact tests were used for categorical data. To compare the ventilator settings and arterial blood gas between both groups during several time points, generalized estimating equation analysis with gamma distribution and exchangeable correlation was applied. All tests were two tailed, and P value less than 0.05 was considered statistically significant. Statistical analysis was carried out using SPSS 18.0 for Windows (IBM, Chicago, IL). Data were analyzed on an intention-to-treat basis and per-protocol basis.

Details of number of patients screened, randomized, excluded postoperatively, and analyzed are shown in the CONSORT diagram (fig. 1). Thirty-four patients were randomized to each group. Four patients in the ASV group were excluded postoperatively because of hemodynamic instability (one patient), reoperation (one patient), and seizures (two patients). Three patients in the control group were similarly excluded because of cardiac arrest, reoperation, and seizure (one patient each). As a result, 30 patients in the ASV group and 31 patients in the control group were included in the intention-to-treat analysis. Protocol violations occurred in five patients in the ASV group (propofol not stopped promptly) and three patients in the control group (propofol not stopped promptly in two patients and ventilated using a different ventilator in one), leaving 25 patients (ASV group) and 28 patients (control group) in the per-protocol analysis. All analyses presented were based on complete data except for hemodynamic data and ventilatory data.

Demographic, preoperative, and intraoperative parameters were comparable in the two groups except for an increased total dose of propofol and a trend to increased aortic cross clamp time and complex surgery in the ASV group and an increased prevalence of preoperative atrial fibrillation in the control group (tables 1 and 2).

Duration of mechanical ventilation and intubation was statistically significantly shorter in the ASV group than that in the control group. On an intention-to-treat analysis, duration of mechanical ventilation was reduced from a median of 342 min to a median of 205 (table 3). Median difference in duration of mechanical ventilation was 137 min (95% CI, 26 to 248).

On per-protocol analysis, use of ASV was associated with statistically significantly fewer manual ventilator adjustments (median of 2 vs. 4) and ventilator alarms (11 vs. 27) but one additional blood gas sample (6 vs. 5) (table 4).

There were no differences between the two groups (per-protocol analysis) in terms of respiratory, ventilatory, or hemodynamic variables except that peak pressures and tidal volumes were slightly higher in the control group (tables 5 and 6). Hemodynamic data for one patient in the per-protocol ASV group and inspired oxygen data for one patient in the per-protocol control group were lost due to a computer error. There were no statistically or clinically significant differences in the incidence of postoperative complications (table 7). Two patients in the ASV group required reintubation: one for an esophageal perforation and one because of a seizure. The patient with the esophageal perforation died on the 28th postoperative day.

In the control group, all but three patients were ventilated in volume-preset SIMV with pressure support with constant flow (SIMV breaths). Two were ventilated in constant-flow volume-preset assist control mode and one in pressure-preset SIMV with pressure support. All patients were subsequently changed to pressure support mode after a median of 207.5 (104 to 300) min. The majority of patients underwent an unassisted breathing trial before extubation, but four were extubated from pressure support ventilation.

Our results demonstrate that closed-loop weaning of fast-track cardiac surgical patients after valvular surgery using ASV is statistically significantly faster than weaning directed by physicians. This is in keeping with previous results in fast-track patients who had undergone coronary artery bypass grafting, in whom weaning in ASV mode is as fast or faster than weaning using a variety of other modes, such as SIMV, pressure-regulated volume control and pressure control. However, the median time to extubation in the ASV arms of these studies is 2.7 to 5 h.^1–3  This is long compared with extubation times in cardiac surgical patients in which early extubation has been targeted,^5,13,19  raising the possibility that the studies should be interpreted not as ASV decreasing the time to extubation compared with other modes, but increasing it less. Further support for this possibility comes from the study by Gruber et al..¹  In this study, protocolized weaning was abandoned in those patients who failed to be weaned using ASV or pressure-regulated volume control with automode within 8 h. Weaning was then left to the clinician’s discretion. All these patients were successfully extubated within 30 min of abandoning protocolized weaning. In our study, control group weaning was at the discretion of the physician, suggesting that ASV led to a real improvement in speed of weaning.

The use of a control group in which physicians direct weaning has the advantage that it more closely reflects the real-life clinical situation. However, there are a number of disadvantages. First, the impossibility of blinding ventilation studies leaves open the possibility that the clinicians deliberately slowed weaning in the control group. We believe that this is unlikely. The ICU in which the study was conducted is the sole adult ICU for a 1,400-bed tertiary referral university hospital. Despite this, the ICU only has 22 staffed beds. There is, therefore, constant pressure to wean patients as soon as possible so that they can be discharged if necessary. Second, the clinicians may have been too busy to wean the patients in the control group. Again, we believe this is unlikely. Postoperative cardiac surgical patients are usually admitted to the ICU during office hours, when the medical staffing of the ICU was high both in terms of number of doctors and their seniority.

The most common mode of weaning in the control group was SIMV. Studies in patients ventilated for more than 1 week have demonstrated that this is an inferior mode of weaning,^25,26  and it is possible that the results of our study simply reflect the inferiority of SIMV rather than the superiority of ASV. However, the population of patients ventilated for 1 week is very different to our study population, and these data are not directly applicable to our patients.

Our study was a single-center study and, although adequately powered, involved a relatively small number of patients. Therefore, caution should be applied in extrapolating our results to other settings. In particular, we believe that the level of staffing may be an important factor in determining the degree of benefit from closed-loop weaning. A multicenter study of closed-loop weaning using SmartCare^® (Draeger Medical, Lubeck, Germany) found that closed-loop weaning resulted in more rapid weaning,²⁷  whereas a similar study carried out in an extremely well-staffed unit demonstrated no benefit.²⁸ 

There was an imbalance in baseline characteristics between our groups with a higher prevalence of preoperative atrial fibrillation in the control group. Although we cannot exclude the possibility that this difference explains the longer ventilation time in the control group, we think this is unlikely, and any effect of atrial fibrillation will have been balanced, to some extent, by a higher dose of propofol and the trend to longer aortic cross clamp time and more complex surgery in the ASV group.

The peak airway pressure was slightly, but statistically significantly, higher in the control group. This is consistent with the use of volume-preset SIMV with pressure support with constant flow during SIMV breaths in the majority of control group patients. In ASV, the inspiratory pressure is constant within a breath.

A recent meta-analysis of closed-loop weaning concluded that there was no benefit from closed-loop weaning in the surgical subgroup of patients.²⁹  However, the analysis included studies of a more heterogeneous group of surgical ICU patients. In particular, a substantial weighting was given to a study of non–fast-track cardiac surgical patients in which 50% of patients were not extubated by 16 h. The authors of that study speculated that patients were left intubated because the unit policy was to discharge patients from the ICU on the morning after surgery at the earliest.³⁰  When the focus is limited to ASV and fast-track cardiac surgical patients, the results of this current study and two previous studies show a reduction in ventilation time compared with other modes of weaning,^1,2  with one further study showing no difference.³ 

Use of ASV was also associated with a reduction in the number of manual changes of ventilator settings and the number of ventilator alarms, at the cost of one additional arterial blood gas analysis. Finding adequate numbers of staff is a problem in many ICUs and a ventilatory mode that reduces workload is an advantage.

Length of ICU and hospital stay was not altered by use of ASV. This is not surprising as there are many other factors which determine length of stay. However, in the context of a postoperative care pathway designed to maximize efficient use of intensive care resources, use of ASV may provide an important operational advantage. Van Mastrigt et al. randomized cardiac surgical patients to short intensive care stay (8 h) or conventional (overnight) ICU stay. The ICU readmission rate and postoperative morbidity were similar in the two groups, but the improvement in quality of life was higher in the short stay group and the cost was lower. Of the 300 patients randomized to short stay, 161 were able to be discharged from ICU within 8 h. The median time to extubation was 6.5 h.²⁰  Among our ASV group, the median time to extubation was 4.9 h, and 80% of patients were extubated in less than 8 h. This suggests that the proportion of patients who are ready for discharge at 8 h may be increased by use of ASV and thus the savings may be greater.

Postoperative complications did not differ between groups, and the incidence of complications was within the range expected for cardiac surgery. This suggests that the shorter ventilation time associated with use of ASV does not come at the cost of increased complications and that fast-track postoperative care of patients undergoing cardiac valvular surgery can be safely carried out.

In conclusion, use of closed-loop weaning with ASV results in more rapid weaning of patients after cardiac valvular surgery than did physician-directed weaning, without an increase in morbidity. This may facilitate early discharge from the ICU, which has been shown to decrease cost and improve quality of life without increased postoperative morbidity.²⁰ 

The authors thank Hamilton Medical GA, Rhäzuns, Switzerland, for the loan of two ventilators.

Support was provided solely from institutional and/or departmental sources.

Dr. Gomersall has received funding for travel to speak at two conferences from Hamilton Medical GA, Rhäzuns, Switzerland. The other authors declare no competing interests.