Weaning from mechanical ventilation allows patients to resume their spontaneous breathing.1 Almost 40–50% of the total duration of mechanical ventilation is spent on the weaning process.2 Delayed weaning not only exposes the patient to increased cost of intensive care but also increased risk of complications.3–5 Hospital mortality increases with prolonged mechanical ventilation, in part because of complications like ventilator-associated pneumonia (VAP) and airway trauma.5 On the other hand, premature weaning is associated with difficulty in re-establishing artificial airway, compromised gas exchange, high incidence of VAP and increased mortality.6 The major factor in successful weaning is resolution of the precipitating illness. Other factors include the comorbid illnesses, cause of acute respiratory failure (ARF), protocol and the method of weaning. Among these, the method of weaning is an important variable because of the potential to intervene. The major weaning studies have been conducted using spontaneous T-piece trials and pressure support (PS) ventilation.7,8 In these studies, readiness to wean has been assessed by an initial 2h T-piece trial; patients who tolerate this trial are extubated whereas those failing this trial are randomized to different weaning methods. The reintubation rates of the initial spontaneous breathing trials (SBTs) have ranged from 10 to 20%.7–11

Since the inception of our respiratory intensive care unit (RICU), it has been the practice to wean patients by gradual PS reduction without employing an initial SBT. No study has compared the efficacy and safety of a weaning method without an initial SBT as the initial strategy. We hypothesized that weaning would be more physiological once PS is gradually decreased, and could potentially result in better outcomes than initial SBTs. The aim of this randomized controlled trial (RCT) was to examine the efficacy and safety of two different weaning methods viz. gradual reduction of PS without an initial SBT versus SBTs using low-level PS.

The study was conducted between January 2008 and June 2009 in the RICU of this institute, and was approved by the Ethics Committee (PGIMER Ethics Committee; VS/1353). An informed consent was taken from all the patients or their relatives. All data in the RICU were entered prospectively into a computer program specifically designed for this purpose as previously described.12

Post-extubation

All patients were monitored for 48h following extubation. Supplemental oxygen or noninvasive ventilation (NIV) was used as clinically indicated. Patients were followed up till the hospital discharge. The entire protocol is schematically shown in Fig. 1.

Figure 1.

Algorithm depicting the entire protocol, from randomization of the patient to implementation of the two methods of weaning.

(0.2MB).

During the study period, 184 patients were mechanically ventilated in the RICU. Thirty-two died prior to the onset of weaning, eight patients were ventilated for <24h, and 20 patients did not provide written consent and were not included in the study (Fig. 2). Sixty-four males and 60 females with a median (IQR) age of 35 (24–50) years were included in the trial. The mean (SD) SOFA score at admission was 5.4 (3.8). The causes of respiratory failure requiring ventilatory support were acute respiratory distress syndrome (ARDS) in 58, poisoning (including organophosphate compounds, snake bite, inhalational toxins and others) in 24, chronic obstructive pulmonary disease in 10, neuromuscular disorders (including Guillain–Barre syndrome, myasthenic crisis) in 10 and other causes (interstitial lung disease, acute exacerbation of asthma, bronchogenic carcinoma, diffuse alveolar haemorrhage [3 patients], lung collapse, pneumothorax, congestive cardiac failure [2 patients], acute coronary syndrome and pericardial tamponade) in 14 patients.

Figure 2.

CONSORT diagram demonstrating the flow of participants through each stage of the trial.

(0.21MB).

Of the 124 patients, 61 were assigned to the PS group and 63 to the SBT group. Four patients (3 PS group, 1 SBT group) developed concomitant events unrelated to the weaning process (cerebrovascular accident [n=2], acute myocardial infarction [n=1] and massive gastrointestinal bleed [n=1]). These were considered as early withdrawals and were not included in the analysis. The baseline characteristics in both the groups were comparable except the PaO2/FiO2 ratio, which was significantly higher in SBT group (Table 1). The mean (SD) pressure support in the PS group at the beginning of the weaning procedure was 17.2 (3.6) cm of H2O. The median (IQR) duration of ventilation prior to weaning was 80.2 (50.5–175.6)h.

The results of the study suggest that weaning by gradual reduction of PS without an initial SBT was associated with better outcomes (in terms of higher weaning trial successes, shorter ICU stay and trend towards quicker time to extubation) than weaning by PS-supported SBTs. This study was conducted in the RICU of a tertiary referral institute manned by intensivists and nursing staff well trained in ventilatory strategies. Our patient profile included medical patients with respiratory failure of various etiologies. Randomization ensured comparability between the two groups. The only significant difference was a higher PaO2/FiO2 ratio in the SBT group, which is unlikely to affect the results as the other parameters were well matched. We also ensured strict adherence to the protocol including the criteria used for altering the level of PS or extubation.

In the SBT arm, trials were conducted every 24h as once daily SBTs with a rest period of 24h are associated with best outcomes.7,8 Two trials have shown that once daily SBT is associated with higher weaning success compared to multiple trials.7,8 Esteban et al. showed that 30min SBTs are as effective as 120min trials in achieving successful extubation with similar reintubation and mortality rates.11 Also, longer duration of SBTs may delay the recovery of muscle function as reflected by the longer ICU stays in the 120min group in their study.11 Most patients who fail an initial SBT do so in the first 20min with the success rate being similar in 30 and 120min trials.9,11 Hence, the duration of the SBT was limited to 1h in our study to further reduce the monitoring time and the workload on the RICU staff. We used PS of 7cm H2O for SBT as it was shown that the number of successful extubation was 10% higher in SBT performed with PS of 8cm H2O than with T-piece with similar reintubation rates.9 Brochard et al. found that a PS of 8cm of H2O was sufficient to compensate for the additional work of breathing caused by the endotracheal tube and demand valve.16 Another advantage of SBT with a specified PS over a T-piece trial is that it does not require disconnection from the ventilator, and ensures safety in a resource constrained ICU.

Two large studies have investigated gradual withdrawal of ventilatory support.7,8 However, these studies included patients by an initial 2h SBT with those failing SBT randomized to PS reduction, synchronized intermittent mandatory ventilation (SIMV) or SBTs. These trials established that SIMV was associated with longer duration of weaning than the other two methods. Studies using initial T-piece trials have produced failure rates ranging from 12 to 32%.7–11 Our study is different from these studies in that we directly randomized patients to gradual PS reduction without initial SBT, and the results suggest that this strategy is as effective as PS-supported SBT. However, a larger RCT is required to confirm the results of our study. The PS group in our study had lesser duration of weaning and shorter ICU stay consistent with earlier results.7 However, one study reported that the duration of weaning was significantly shorter with T-piece trials than PS.8 One possible reason could be the restrained manner in which PS was reduced in this study wherein a respiratory rate cutoff of <25breaths/min was required for reduction of PS whereas intolerance to T-piece trial was considered at 35breaths/min.8 Also, prior to extubation, the patients had to tolerate a PS of 5cm of H2O for 2h. This may have led to slower reduction of PS and hence a longer duration of weaning. Further, this study did not consider the failure of T-piece trials requiring reinstitution of ventilation in their outcomes.

The failure rate of SBT was 30% in our study similar to previous reports of 26–42%.4,7,8,11 The reintubation rate of 8% is consistent with earlier reintubation rates of 4–19%.4,7–9,17 The median time to failure of the first weaning trial was 43 and 25min in the PS and SBT arms, respectively. The time to failure of the weaning trial is shorter in the SBT group because PS effectively assists each spontaneous breath and hence reduces the respiratory workload imposed on the respiratory muscles.18–20 The short duration of trial failure re-emphasizes that 60min SBTs are as effective as the 120min trials. The PS group also had shorter ICU stay compared to the SBT group consonant to quicker weaning. The occurrence of VAP in reintubated patients was 70% in our study compared to 17% in those without reintubation, which is similar to previous observation.17 In the multivariate model, the baseline SOFA score, the duration of ventilation prior to weaning and the weaning method significantly influenced successful weaning trials. Vallverdu et al. in a study on factors affecting the weaning outcome in diverse cause of mechanically ventilated patients also found duration of ventilation as a significant factor associated with weaning success.10 Kollef et al. found that baseline APACHE II score was one of the factors that predicted successful weaning.5

Our study has number of limitations. We used an unconventional method to assess the readiness to wean, i.e. CPAP of 5cm H2O. As the study compares the standard SBT method to gradual PS reduction, there was a need for an initial short trial of CPAP to objectively select patients for randomization into the trial. The study has a small sample size and includes diverse causes of respiratory failure. However, our study population was more homogenous than previous studies as we included only medical patients. The absence of blinding is a potential for bias in the study but blinding was not possible as this was an intervention study in critically ill patients that required close monitoring following intervention. Another limitation of the study is the inclusion of patients with tracheostomy (number similar in both arms), and the definitions for extubation and reintubation in this group of patients. Similarly, the use of NIV following extubation is another source of bias although its use was at the discretion of the attending physician and the number in which it was applied was small. The strength of the study is the evaluation of a hitherto uninvestigated method of weaning.

In conclusion, weaning by gradual reduction of PS without an initial SBT was associated with higher success rates, quicker weaning, and a shorter ICU stay versus once daily PS-supported SBTs. However, RCTs with a larger sample size are required to confirm the results of our findings.

The authors have no conflicts of interest to declare.