The evaluation of patients' readiness for liberation from mechanical ventilation starts with the resolution of respiratory failure and/or the disease entity that prompted the initiation of mechanical ventilation as well as the presence of a basic level of physiological readiness (Table 1). Prediction based on clinical 'gestalt' alone is frequently inaccurate. In a study by Stroetz and Hubmayr 9, intensivists could not accurately predict patient tolerance of a 1-hour period on minimal pressure support. Also, Afessa and colleagues 10 reported that critical care practitioners could not accurately forecast 3-day and 7-day weaning outcome for mechanically ventilated patients in a medical intensive care unit (ICU). Thus, there is a sound rationale that predicting readiness of patients to be successfully liberated from mechanical ventilation needs to be based on objective weaning predictors that can be applied in clinical decision making.

A large spectrum of weaning predictors has been studied, which can be divided into simple weaning indices, simple measures of load and capacity, integrative weaning indices, and complex predictors requiring special equipment. A recent expert panel 11 sponsored by the American College of Chest Physicians, Society of Critical Care Medicine, and the American Association for Respiratory Care developed evidence-based weaning guidelines and noted that only eight variables had some predictive capacity: minute ventilation (V_E), negative inspiratory force, maximum inspiratory pressure, tidal volume (V_T), breathing frequency (f), the ratio of breathing frequency to tidal volume (f/V_T), P_0.1/P_Imax (ratio of airway occlusion pressure 0.1 s after the onset of inspiratory effort to maximal inspiratory pressure), and CROP (integrative index of compliance, rate, oxygenation, and pressure) (Table 2). However, another recent study by Conti and colleagues 12 in 2004 showed that vital capacity, V_T, P_0.1, V_E, f, maximum inspiratory pressure, f/V_T, P_0.1/P_Imax and P_0.1 × f/V_T are poor predictors of weaning outcome in a general ICU population. These conflicting results have been attributed to many factors, such as the measurement techniques that differ from one study to another, different timing of measurements made by different investigators, and lack of objective criteria to determine tolerance of the trial 13141516.

The ultimate goal of mechanical ventilatory support is liberation from the ventilator. In the majority of patients this is a simple process. A quick and direct method of testing readiness for liberation from mechanical ventilation is simply to initiate a trial of unassisted spontaneous breathing in the form of either a T-piece trial, a CPAP trial, or a pressure supported ventilation (PSV) trial. The decision on which trial to use remains largely a matter of physician preference; however, an important underlying rule is 'DO NOT EXHAUST' 44.

The classical approach for readiness testing is the spontaneous breathing trial with either a T-piece or a CPAP trial 45. T-piece or CPAP trials can be used for two objectives: to test readiness for liberation from mechanical ventilation; or to implement a weaning protocol that involves increasing periods of unsupported spontaneous breathing interspersed with periods of full ventilatory support 46. Wysocki and colleagues 46 showed that the reduced breathing variability in the breathing pattern was a marker of poor outcome from mechanical ventilation. This is in contrast to previous findings by Engoren 26 and El-Khatib and colleagues 27. However, these discrepancies can be explained by the differences in the experimental set-ups as Engoren and El-Khatib and colleagues were assessing variability in breathing pattern during minimal ventilatory support while Wysocki and colleagues were assessing breathing pattern during a T-piece trial and when patients were off any ventilatory support. A subject of considerable interest is the duration of the T-piece or CPAP trials that best reflect patient readiness for liberation from mechanical ventilation. Large randomized controlled trials indicate that the mean time for a T-piece trial failure is approximately 40 to 50 minutes 236. The proper trial duration for those conducted on CPAP or PSV is not well known. However, there is convincing evidence 23647 that a spontaneous breathing trial with either T-piece, CPAP, or PSV should be conducted for 30 to 120 minutes in order to test readiness for liberation from mechanical ventilation. It is likely that patients requiring more prolonged mechanical ventilation will require longer spontaneous breathing trials 48.

One advantage of using PSV instead of a T-piece or CPAP trials is its capacity to overcome the work of breathing imposed by the ventilator and the endotracheal tube 4950, particularly after a significant stay on mechanical ventilation during which it is anticipated that the tube radius is narrow or there are secretions lining the endotracheal tube lumen 5152. One major issue remains the level of pressure support required to overcome the imposed work of breathing while avoiding either over-assisting, which can result in excessive numbers of extubation failures, or under-assisting, which can result in unnecessarily prolonged mechanical ventilation. Studies have shown that the appropriate PSV level differs substantially from patient to patient 4950. Some studies have recommended the use of 5 to 7 cmH₂O of pressure support to offset the imposed work of breathing imposed by the breathing circuit, ventilator, and endotracheal tube resistances 24.

Esteban and colleagues 3 showed in patients who failed a first spontaneous breathing trial that the median duration of weaning was 5 days for synchronized intermittent mandatory ventilation (SIMV), 4 days for PSV, and 3 days for T-piece trials. Other investigations indicated that the degree of respiratory muscle rest on SIMV is not proportional to the level of ventilatory support 5354. Combining PSV and SIMV reduced the pressure-time product and the work of breathing of the intervening and mandatory breaths, compared to SIMV alone 55. Nevertheless, there is strong evidence that in patients who fail their initial attempt at spontaneous breathing (Table 3), SIMV weaning is inferior to repetitive and scheduled T-piece trials or PSV trials and serves to prolong the duration of mechanical ventilation 23.

Alongside T-piece, CPAP, and PSV trials, other modern forms of assisted spontaneous breathing (automatic tube compensation (ATC), proportional assist ventilation (PAV) and neurally adjusted ventilation assist (NAVA)) have been developed and brought to technical fruition in recent years. While PSV provides a constant pressure support level that can lead to either under-compensation and/or over-compensation of the pressure drop across the endotracheal tube, ATC provides a programmable pressure that is intended to compensate for the endotracheal tube resistance during inspiration and expiration according to the actual gas flow 56. As such, ATC compensates continuously for the endotracheal tube and, hence, is sometimes referred to as electronic extubation 57. Kuhlen and colleagues 56 showed that in weaning from mechanical ventilation, patients' work of breathing during spontaneous breathing trials is reduced by the application of PSV of 7 cmH₂O whereas the workload during automatic tube compensation corresponded to the values during trials of spontaneous breathing through a T-piece. Cohen and colleagues 57 showed that significantly more patients undergoing a spontaneous breathing trial with ATC were able to maintain spontaneous breathing for 48 hours after discontinuation of mechanical ventilation and extubation than with CPAP alone. Although not much research has been done on PAV and NAVA as newer weaning modes, some researchers have postulated a potential role of PAV and NAVA in weaning patients from mechanical ventilation 5859. Much research is needed in this area.

When a patient fails a spontaneous breathing trial (Table 3), this should prompt the clinicians to conduct a systematic search for the reversible factors that may be responsible for weaning failure. Some of these factors may be simple to correct while others may be impossible to correct.

There is very strong evidence that weaning/liberation protocols can be successfully implemented in ICUs while expediting weaning and improving the success of liberation from mechanical ventilation 9091929394. These protocols, whether they are physician-directed or implemented by respiratory therapists and ICU nurses, can enhance clinical outcomes and reduce costs of caring for critically ill patients. Marelich and colleagues 95 evaluated the effect of a single ventilator weaning protocol used in medical and surgical ICUs on the duration of mechanical ventilation and the incidence of ventilator associated pneumonia. They reported that a multidisciplinary weaning protocol was effective in reducing the duration of mechanical ventilatory support (from a median of 124 hours to a median of 68 hours) without any adverse effects on patient outcome. Also, they indicated that the weaning protocol was associated with a 50% decrease in incidence of ventilator-associated pneumonia. Similarly, Dries and colleagues 96 showed that protocol-driven weaning reduces the use of mechanical ventilation from 0.47 to 0.33 ventilator days/ICU days and reduces by 71% the incidence of ventilator-associated pneumonia in surgical ICU patients. Using a handheld computer, Iregui and colleagues 97 showed that respiratory care practitioners employing a weaning protocol programmed on a handheld computer can wean patients from mechanical ventilation more efficiently compared with the use of a paper-based weaning protocol. Lellouche and colleagues 98 compared usual care for weaning with a computer-driven weaning protocol based on a feedback system that uses the ventilatory pattern and end-tidal CO₂ to drive the appropriate pressure support levels. They reported that weaning duration was reduced in the computer-driven group from a median of 5 to 3 days and the total duration of mechanical ventilation from 12 to 7.5 days. The computer-driven weaning protocol also decreased median ICU stay from 15.5 to 12 days 98. In cardiac surgery patients, Petter and colleagues 99 showed that during adaptive support ventilation, a ventilatory mode providing automatic adjustments of the ventilator settings, the outcome of tracheal extubation was comparable to PSV, with fewer ventilator setting manipulations and a smaller number of alarms. They suggested that adaptive support ventilation may simplify postoperative ventilatory support without delaying extubation. Much research is needed to identify any role for the modern modes of assisted spontaneous breathing in expediting and improving liberation from mechanical ventilation.

Whereas many controlled trials conducted in North America have demonstrated the effectiveness of nursing-based and respiratory therapist-based protocols in the early liberation from mechanical ventilation and reducing the length of stay in the ICU 100101102103, weaning from mechanical ventilation in Europe has generally been less common and mainly physician-directed. Blackwood and colleagues 104 interviewed a sample of ten consultant physicians in two European ICUs and reported that although local physicians were supportive in theory, introduction of protocolized weaning is likely to be difficult because of the breadth of information required for successful decision making.

Consultants' views in this study were not consistent with North American findings that physicians' caution may unnecessarily prolong weaning. Two recent European studies reported conflicting results on the utility of protocolized weaning in affecting the outcome and duration of mechanical ventilation. Blackwood and colleagues 105, using a physician-driven protocol, reported that protocolized weaning did not reduce the duration of mechanical ventilation and was not associated with an increased rate of re-intubation or ICU mortality, while Tonnelier and colleagues 106, using a nurses' protocolized-directed weaning strategy, showed a shorter duration of mechanical ventilation and a shorter length of stay in the ICU, with no difference in extubation failure or incidence of ventilator-associated pneumonia in patients requiring ventilatory support for greater than 48 hours. However, it should be pointed out that most (if not all) of the above clinical studies supporting these weaning approaches did not routinely use daily spontaneous breathing trials in the control group.

In 2006, Goodman 107 described how a weaning protocol is designed and introduced into the critical care unit of a district general hospital. He indicated that a multi-professional group interested in weaning needs to work to formulate a protocol, to implement the protocol into the ICU, and to maintain an ongoing auditing system.

While each institution must customize the weaning protocols to its local practice, there are important general concepts that may ease the process of implementation and enhance success 105. First, protocols should not be viewed as static constructs, but rather dynamic tools in evolution, which can be modified to accommodate new data and/or clinical practice guidelines. Second, institutions must be prepared to commit the necessary resources (for example, technology and personnel) to develop and implement weaning protocols.

Patients with limited ventilatory reserve or chronic pulmonary disease can benefit from non-invasive positive pressure ventilation (NPPV) in the period post-extubation and after liberation from mechanical ventilation. Ferrer and colleagues 108 showed that earlier extubation with NPPV resulted in shorter duration of mechanical ventilation and length of stay, less need for tracheostomy, and lower incidence of complications, and improved survival in 21 patients who had failed a weaning trial for 3 consecutive days. In patients with an exacerbation of COPD, Nava and colleagues 109 showed that NPPV facilitated extubation within 48 hours after intubation, decreasing the period of ventilatory support, the ICU stay, and the incidence of nosocomial pneumonia as well as increasing survival. A recent meta-analysis 110 of five studies and 171 patients reported that the use of NPPV facilitated weaning with a consistent positive effect of non-invasive weaning on mortality. However, caution should be exercised not to prolong non-beneficial use of NPPV, particularly in non-COPD patients, as this may worsen outcomes 111. In a group of 221 patients with similar baseline characteristics and who were randomly assigned to either NPPV (114 patients) or standard medical care (107 patients), Esteban and colleagues 111 showed that NPPV did not reduce mortality or the need for re-intubation among patients receiving mechanical ventilation who had respiratory failure after extubation. The mortality rate tended to be higher among the patients assigned to NPPV than among those assigned to standard medical therapy, suggesting that unnecessarily prolonged use of NPPV may worsen outcomes by delaying necessary re-intubation.

Extubation (that is, removal of the endotracheal tube) is the second stage in the liberation from mechanical ventilation process. It should be considered and attempted after patients have passed a spontaneous breathing trial as part of the liberation process. However, even with the success of a spontaneous breathing trial, the possibility of an extubation failure (that is, the need for re-intubation within 48 to 72 hours) is 10% to 20% 112113. Some of the factors that have been implicated in such extubation failures include cough strength, propensity for aspiration, volume of secretions and suctioning frequency, level of consciousness, and potential for upper airway stridor 114115116. Decreasing the risk for extubation failure necessitates the thorough assessment of patients' ability to cough and clear airway secretions 114115 as well as the use of steroids for patients who are at risk of post-extubation laryngeal edema 116.

The role of tracheostomy in liberation from mechanical ventilation has been the subject of several studies. The benefits commonly ascribed to tracheostomy 117118119120121 include increased patient comfort, decreased airway resistance, better secretion removal, improved oral hygiene, less laryngeal damage, and ability to eat and speak. Boynton and colleagues 119 showed in an observational prospective cohort study that the median duration of weaning was 3 days in surgical patients who underwent tracheostomy prior to any active weaning attempts versus 6 days in patients in whom initial weaning attempts were made with the endotracheal tube in place. The frequency of fatigue and pneumonia were also lower in the patients who underwent early tracheostomy. Diehl and colleagues 120 and Davis and colleagues 121 reported that the work of breathing decreased by 26% to 55%, the pressure-time index decreased by 55%, the airway resistance decreased from 9.4 to 6.3 cmH₂O/l/s, and the auto-PEEP decreased from 2.9 to 1.6 cmH₂O following tracheostomy. They postulated that the rigid nature of the tracheostomy tube represents reduced imposed work of breathing compared with the longer, thermoliable endotracheal tube and concluded that tracheostomy can substantially reduce the mechanical workload of ventilator-dependent patients.

Each institution should develop its own protocol for liberation from mechanical ventilation that is based on solid scientific data and take into consideration the level of personal and technical support in the ICU. The algorithm presented in Figure 1 reflects our institutional protocol for liberation from mechanical ventilation, which is not meant as a guideline but just an example of a simple bedside approach to the liberation from mechanical ventilation. An essential component of our evaluation for readiness for a spontaneous breathing trial involves the assessment of respiratory system resistance, particularly in patients who have received mechanical ventilation for more than 7 days. Measurement of the respiratory system resistance (which includes endotracheal tube resistance) is performed with an end-inspiratory occlusion maneuver 122 during mechanical ventilation and prior to initiation of the spontaneous breathing trial. This will particularly give us an insight into the endotracheal tube condition (that is, endotracheal tube resistance), which can be a major workload during the subsequent spontaneous breathing trial. For resistance values ≤ 9 cmH₂O/l/minute, patients can receive a spontaneous breathing trial with a T-piece, otherwise patients will undergo spontaneous breathing trials with CPAP or PSV.

Successful liberation from mechanical ventilation in the ICU depends on the application of skilled judgment, decision making, and medical and nursing interventions. Most patients do not require a prolonged period of gradual withdrawal of mechanical ventilation, which carries the risks of ventilator-induced lung injury, nosocomial pneumonia, airway trauma, and increased cost of care. On the other hand, overly aggressive and premature discontinuation of ventilatory support can precipitate ventilatory muscle fatigue, gas-exchange failure, and loss of airway protection. The last two decades of research have provided the information necessary to make adequate clinical decisions in liberating patients from mechanical ventilation.

Key points with regard to liberation from mechanical ventilation are: most mechanically ventilated patients can be liberated from mechanical ventilation after a short spontaneous breathing trial; most weaning predictors may not be sufficiently accurate for liberation decision-making, although they are helpful in identifying causes of respiratory failure; the overall medical management of patients who continue to require ventilatory support should be continuously re-evaluated to ensure that all factors contributing to ventilator-dependence are assessed; ventilatory support strategies should be aimed at maximizing patient comfort and unloading of the respiratory muscles; the duration of spontaneous breathing trials can be anywhere from 30 to 120 minutes; and the duration of mechanical ventilation can be reduced by using clinical protocols that can be executed by respiratory therapists and ICU nurses and not necessarily by ICU physicians.

ATC = automatic tube compensation; COPD = chronic obstructive pulmonary disease; CPAP = continuous positive airway pressure; CROP = integrative index of compliance, rate, oxygenation, and pressure; f = breathing frequency; f/V_T = ratio of breathing frequency to tidal volume; ICU = intensive care unit; NAVA = neurally adjusted ventilation assist; NPPV = non-invasive positive pressure ventilation; P_0.1/P_Imax = ratio of airway occlusion pressure 0.1 s after the onset of inspiratory effort to maximal inspiratory pressure; PAV = proportional assist ventilation; PEEP = positive end expiratory pressure; PSV = pressure support ventilation; RSBI = rapid shallow breathing index; SIMV = synchronized intermittent mandatory ventilation; V_E = minute ventilation; V_T = tidal volume.

The authors declare that they have no competing interests.