Clinical review: The ABC of weaning failure - a structured approach

About 20% to 30% of patients are difficult to wean from invasive mechanical ventilation. The pathophysiology of difficult weaning is complex. Accordingly, determining the reason for difficult weaning and subsequently developing a treatment strategy require a dedicated clinician with in-depth knowledge of the pathophysiology of weaning failure. This review presents a structural framework ('ABCDE') for the assessment and treatment of difficult-to-wean patients. Earlier recognition of the underlying causes may expedite weaning from mechanical ventilation.


Introduction and focus
In most patients, mechanical ventilation can be dis continued as soon as the underlying reason for acute respiratory failure has been resolved. However, 20% to 30% of patients are considered diffi cult to wean from mechanical ventilation. Weaning failure is defi ned as the failure to pass a spontaneous-breathing trial or the need for reintubation within 48 hours following extubation [1]. Th e risks associated with post-extubation distress and reintubation are considerable. Th erefore, clinicians often choose to use a two-step diagnostic approach before extu bation. First, weaning parameters are assessed (screening), and, second, a weaning trial is initiated. Th e validity of the diff erent screening tools (for weaning), such as the rapid shallow breathing index and maximal inspiratory pressure, has been discussed in detail elsewhere [2]. Diff erent techniques have been used for weaning and these include gradual reduction in mandatory rate during intermittent mandatory ventilation, gradual reduction in pressure support, and spontaneous breathing through a T-piece. Esteban and colleagues [3] concluded that trials of spontaneous breathing resulted in faster liberation from mechanical ventilation compared with weaning using pressure support or intermittent mandatory ventilation. However, other studies failed to prove the superiority of spontaneous-breathing trials over pressure support ventilation as a weaning technique [4,5]. Th ese diff erences in outcome most likely are related to the diff erent designs of these studies, patient characteristics, and durations of mechanical ventilation before initiation of weaning. In-depth discussion of weaning techniques is beyond the scope of this paper and has been published earlier [2].
Th e pathophysiology of weaning failure is complex and often multifactorial. Accordingly, determining the reason and subsequently developing a treatment strategy require a dedicated clinician with in-depth knowledge of the pathophysiology of weaning failure. Several reviews have been published on weaning failure, but most focus on only one aspect of weaning and therefore are of limited help to clinicians. Complex clinical problems probably benefi t from a structured approach. An ' ABC approach' is valuable in training juniors in trauma management [6]. In this paper, we present an opinion-based ' ABC approach' for diffi cult weaning. Unlike the ABC in trauma medicine ('treat fi rst that kills fi rst'), our 'weaning ABC' does not emphasize a specifi c order of assessment but is purely alphabetical and has the advantage of 'forcing' the clinician to systematically review the most likely causes for failed weaning. With the understanding of the barriers that impede successful weaning in a specifi c patient, a tailored treatment strategy that may reduce duration of mechanical ventilation can be designed. Th e following topics should be evaluated in a diffi cult-towean patient (Figure 1): airway and lung dysfunction, brain dysfunction, cardiac dysfunction, diaphragm dysfunction, and endocrine dysfunction. Th e algorithm presented in Figure 1 shows possible strategies to confi rm the role of any of these items and options to reduce the contribution to weaning failure. For each patient, diagnostics as described in the white box should be performed to assess the reasons(s) for diffi cult weaning. Endocrine dysfunction is probably relatively rare and therefore is not included in the fi rst line of evaluation. Possible treatment/interventions are mentioned but, of course, need to be individualized. If the fi rst-line evaluation does not improve weaning, proceed to the next level (within the aff ected column failure. Th e most prevalent causes for impaired mechanics are summarized in Table 1.

Resistance in weaning
Resistance of the upper airway should be considered in diffi cult weaning. Rumbak and colleagues [7] found that tracheal obstruction caused by tracheal injury may contri bute to weaning failure in patients who were on invasive mechanical ventilation for more than 4 weeks. Tracheal injury includes tracheal stenosis, tracheo malacia, and granulation tissue formation. In intubated patients, the endotracheal tube could increase airway resistance. Indeed, resistance of an endotracheal tube removed from a patient is signifi cantly higher than the resistance of a new tube when tested in a laboratory setting and this is due to contami nation of the tube with airway secretions [8]. No correla tion between the duration intubation and the resistance of the tube was found. However, in general, it is a mis conception that, after extubation, upper airway resistance decreases. In fact, the work of breathing in patients mechanically ventilated for ± 5.5 days increased after extubation, and this increase was probably due to edema of the upper airways [9]. Increased resistance of the small airways is a characteristic of chronic obstructive pulmonary disease (COPD) and asthma but has been demonstrated in acute respiratory distress syndrome (ARDS), and this latter increase was due to edema of bronchial walls. Vassilakopoulos and colleagues [10] showed, in a heterogeneous group of patients, that airway resistance is signifi cantly higher when failing a weaning trial compared with a successful trial in these same patients. Jubran and Tobin [11] found that, in COPD patients failing a spontaneous-breathing trial, airway resistance signifi cantly increases during this trial (9 ± 2 cm H 2 O up to 15 ± 2 cm H 2 O; P <0.05), whereas in weaning-success patients, airway resistance did not increase signifi cantly. Th ese studies show that increased airway resistance may contribute to weaning failure.
Increased airway resistance is associated with the development of intrinsic positive end-expiratory pressure (PEEPi). PEEPi develops if expiratory time is shorter than the time needed to empty the lungs to the elastic equilibrium volume of the total respiratory system. Accordingly, PEEPi may develop because of increased fl ow resistance, expiratory fl ow limitation, high breathing frequency, and loss of elastic recoil of the lung. Indeed, in COPD patients with weaning failure, PEEPi is higher compared with those successfully weaned [11]. PEEPi acts as an inspiratory threshold before inspiratory fl ow can be generated and result in elevated work of breathing. In addition, pulmonary hyperinfl ation resulting from PEEPi places the diaphragm at a suboptimal position on the length-tension curve, impair ing the ability to generate negative pressure. Moreover, PEEPi is associated with patient-ventilator asynchrony and, in particular, with ineff ective triggering (inspiratory eff ort without ventilator support). Recently, de Wit and colleagues [12] showed that, in patients with a high incidence of ineff ective triggering (>10% of all breaths) (as derived from ventilator waveform tracings), duration of mechanical ventilation is signifi cantly longer compared with patients with a lower incidence of ineff ective triggering. Although the causes for ineff ective triggering are unknown in this study, it demonstrates the clinical relevance of patient-ventilator asynchrony. Measurement of airway resistance and PEEPi may help clinicians to understand the reason for failed weaning.

Resistance: diagnostic approach
Flexible bronchoscopy is the gold standard for diagnosing upper airway disease. Patients should be disconnected from the ventilator during bronchoscopy so that the presence of tracheomalacia can be assessed. Tracheo malacia may be treated with (nocturnal) non-invasive ventilation or placement of endotracheal stents [7]. Manage ment of the other causes for central airways obstruction is usually evident but is complex in nature. Th e fl ow-time and pressure-time loops provide a qualitative assessment of the presence of increased airway resistance and PEEPi ( Figure 2). When ventilating with a constant fl ow, a more quantitative assessment of total airway resistance can be obtained by measuring peak pressure, plateau pressure, and inspiratory fl ow: R,aw = (P peak -P plat ) / fl ow (normal is less than 5 cm H 2 O/L per second).
PEEPi is obtained by subtracting PEEP appl from PEEP tot . Unfortunately, this technique is seldom satisfactory in spontaneously breathing patients. In those patients, the only way that PEEPi can be measured adequately is by using an esophageal balloon.

Resistance: therapeutic strategies
In COPD patients being weaned with pressure support ventilation, appropriate setting of the cycle-off criterion is of importance to limit PEEPi and the work of breathing, as demonstrated by Chiumello and colleagues [13]. Changing the cycling-off criterion from 5% to 40% signifi cantly reduced PEEPi and the work of breathing. Th ese eff ects were most prominent at low levels of pressure support [13]. Applied PEEP may reduce inspira tory work of breathing imposed by PEEPi [14]. Applied PEEP should match the level of PEEPi, as estimated by the expiratory occlusion technique.
Patients with asthma and COPD should be treated optimally to reduce bronchoconstriction. Th e eff ect of bronchodilators in patients without a history of COPD has not been tested, but if airway resistance appears elevated, latent obstructive airway disease may be present. Attention should be paid to the optimal delivery of inhaled medications as this will greatly infl uence its eff ect [15]. Clinical studies have shown that, in patients with COPD, the bronchodilating eff ect does not further increase after four doses of 100 μg of albuterol via metered dose inhaler. Th e eff ect of 400 μg of albuterol lasted for at least 60 minutes. Th e frequency of administration depends on the severity of bronchoconstriction and its clinical eff ect. Th e eff ect of the long-acting beta-2 agonist salmeterol (four doses of 25 μg) lasts for at least 6 hours [16]. However, it should be noted that the maximal bronchodilating eff ect of salmeterol is achieved after more than 30 minutes and therefore is less suitable in the acute setting. As much more experience is available for albuterol in mechanically ventilated patients, it remains the beta-2 agonist of choice. . Indeed, when expiration is interrupted (red solid arrow) after the next inspiration, airway pressure rises (middle panel), refl ecting total PEEP (applied PEEP and intrinsic PEEP). To measure respiratory resistance, inspiratory hold is applied (red dotted arrow), resulting in rapid decay in airway pressure from peak to P1 and a subsequent slow decay to plateau pressure (P,plat).
Heunks and van der Hoeven Critical Care 2010, 14:245 http://ccforum.com/content/14/6/245 Compliance Respiratory compliance is determined by the compliance of the chest wall and lungs (Table 1). In patients with ARDS, compliance was signifi cantly lower at the time of weaning failure compared with weaning success [10]. Th e static compliance of the respiratory system can be calculated after measuring inspiratory and expiratory plateau pressure and tidal volume ( Figure 2): compliance respiratory system = TV / (P plat -PEEP tot ) (normal is 60 to 100 mL/cm H 2 O). Optimizing compliance by reducing edema of the lung and chest wall, resolving atelectasis and thoracocentesis, and removing ascites may facilitate weaning.

Gas exchange
Most weaning patients still have considerable distur bances in gas exchange at the time of weaning and may develop hypoxemia or hypercapnia or both during a spontaneous-breathing trial [17]. Weaning-failure patients are frequently unable to increase minute ventilation in response to high arterial partial pressure of carbon dioxide (PaCO 2 ) because of respiratory muscle weak ness, altered respiratory mechanics, and so on. In these patients, it is important to limit instrumental dead space (for instance, resulting from heat and moisture exchangers [18]) as much as possible.

Delirium and depression in weaning
Brain dysfunction in diffi cult-to-wean patients is related mostly to delirium. Only one paper specifi cally studied the role of delirium in weaning, and the paper showed that impaired cognitive function was associated with a more than fourfold higher risk of failed extubation [19]. Other psychological disturbances, such as anxiety and depression, may interfere with successful weaning [20]. Sleep disruption frequently occurs in intensive care unit (ICU) patients, but no data showing the eff ects of disturbed sleep architecture on weaning are available.

Diagnostic approach
Th e confusion assessment method for ICU (CAM-ICU) is a well-validated screening tool for delirium in mecha ni cally ventilated patients [21] and is available in numerous languages, but other screening tools have been validated as well [22]. Diagnosing other cognitive disorders in the ICU is best performed by psychologists or psychiatrists. In general wards, reducing risk factors for delirium reduces the incidence of delirium [23]. Midazolam is considered a risk factor for the development of delirium [24]. A recent multicenter trial indicated that sedation with dexmedetomidine reduced the incidence of delirium and reduced the duration of mechanical ventilation compared with sedation with midazolam [25].

Therapeutic strategies
High levels of sedatives are associated with increased time spent on the ventilator. A daily wakeup call and, if appropriate, subsequent spontaneous-breathing trials are associated with reduced time spent on the ventilator [26]. Th e role of sedatives in failed-spontaneous breathing trials or failed extubation is less clear. Depression is associated with weaning failure in patients admitted to a long-term weaning facility [27], and preliminary data suggest that pharmacological treatment of depression favors weaning from mechanical ventilation [20]. Sleep may be improved by limiting noise and light during sleep hours and adequately treating pain/discomfort. Th e eff ect of ventilator mode on sleep quality has been a subject of discussion [28,29], and diff erences in outcome are probably related to the level of pressure support.

Cardiac dysfunction
Th e transition from mechanical ventilation to spon taneous breathing imposes an additional load on the cardiovascular system because of intrathoracic pressure changes, which aff ect ventricular preload and afterload and increased oxygen consumption by the respiratory muscles.

Cardiac function during weaning
In patients with COPD but without cardiac disease, weaning was associated with a signifi cant reduction in left ventricle ejection fraction and this reduction was probably due to increased left ventricular afterload [30]. Several studies have specifi cally investigated the eff ects of weaning with spontaneous-breathing trials on cardiac performance. In patients with COPD (FEV 1 [forced expiratory volume in 1 second] 1.0 ± 0.2 L) but without a history of cardiac disease, weaning was associated with a reduction in left ventricle ejection fraction (54% ± 12% versus 47% ± 13% during mechanical ventilation and spontaneous breathing, respectively; P <0.01) [30]. As the authors did not fi nd evidence for myocardial ischemia when using technetium radionuclide angiography in these patients, it was proposed that increased left ventricle afterload imposed by disconnection from the ventilator caused the reduction in left ventricle ejection fraction. Moreover, in diffi cult-weaning patients with COPD and a history of heart disease, spontaneousbreathing trials resulted in elevated pulmonary artery occlusion pressure and left ventricular end diastolic pressure, suggesting reduced ventricular compliance [31]. Elevated left ventricular end diastolic pressure during weaning may induce pulmonary and bronchial wall edema and thus increase the work of breathing.
In a heterogeneous group of patients with a failedweaning trial, a progressive decline in mixed venous oxygen saturation (SvO 2 ) was found [32]. Th e decrease in SvO 2 resulted from the inability to improve cardiac output and therefore oxygen transport. Increased afterloads of both the right ventricle and left ventricle were found in these patients.
Brain natriuretcic peptide (BNP) is a hormone released from the myocardium upon stretch. In patients with failed-spontaneous breathing trials and in failedextubation patients, changes in BNP are signifi cantly higher than in patients with successful extubation [33]. Th is response of BNP does support a role for cardiac failure in selected diffi cult-weaning patients.

Diagnostic approach
Th e fi rst step to assess cardiac dysfunction as a cause for weaning failure is electrocardiography at the fi nal stages of the weaning trial to detect ischemia. Subsequently, SvO 2 could be used as a screening tool for cardiac dysfunction in diffi cult weaning. It should be noted that reduction in SvO 2 is the normal response to increased loading. In healthy subjects performing moderate exercise, SvO 2 decreases below 50% [34]. Accordingly, a decrease in SvO 2 should not be regarded as a sign of cardiac failure. On the other hand, in weaning-failure patients, without a reduction in SvO 2 , cardiac failure is highly unlikely but rather is the result of hypoventilation, lack of motivation, or cognitive dysfunction. Accordingly, in patients who do exhibit a decrease in SvO 2 during failed weaning, cardiac failure may play a role and measurement of pulmonary artery occlusion pressure and cardiac output by means of a Swan-Ganz catheter [35] or echocardiography [36] must be considered.

Treatment strategies
In diffi cult-to-wean patients with evidence of cardiac failure, afterload reduction and ultimately the use of inotropes must be considered. Notably, amelioration of pulmonary mechanics will reduce left ventricle afterload. No studies have compared the diff erent classes of inotropes in these patients. Levosimendan is a new inotropic used in the treatment of cardiac failure [37]. One of the mechanisms of action is enhanced calcium sensitivity of the cardiac contractile proteins, resulting in elevated pressure generation. In a preliminary study [38], levosimendan was tested in 12 patients who had heart failure (left ventricular ejection fraction of 28% ± 5%) and who were diffi cult to wean from mechanical ventilation (ventilator dependency for more than 10 days). Within 51 hours of levosimendan treatment, 7 of the 12 patients could be weaned from the ventilator. Th e other 5 patients could not be weaned at all and died in the ICU. Th ough promising, levosimendan has not been tested in placebocontrolled trials or directly compared with the older inotropes in weaning-failure patients. Interestingly, it was recently shown that levosimendan improves contractility of human diaphragm muscle fi bers as well [39].

Diaphragm/respiratory muscle function
Initiating a spontaneous-weaning trial puts an acute load on the inspiratory muscles. Th e elevated load is imposed on a weakened respiratory muscle pump. When considering the inspiratory muscles as a cause for weaning failure, it should be kept in mind that dysfunction of the ventilatory pump may result from a lesion anywhere between the aff erent chemo-receptors of the respiratory centers and the contractile proteins.

Drive, weakness, and fatigue
Impaired respiratory drive is an uncommon cause of weaning failure. In fact, in most weaning-failure patients, respiratory drive is increased [17]. Occasionally, clinically relevant diaphragm dysfunction results from damage to the phrenic nerve(s). Th e most frequent disorder aff ecting the phrenic nerves is critical illness polyneuropathy. As biopsy studies have shown that critical illness polyneuropathy and critical illness myopathy often coexist [40], the term 'ICU-acquired weakness' has gained popularity. Th ere is convincing evidence that contractility of the respiratory muscles is impaired in mechanically ventilated patients. Laghi and colleagues [41] found that, in a heterogeneous group of mechanically ventilated patients (± 29 days), transdiaphragmatic pressure obtained after magnetic stimulation of the phrenic nerves was only approximately 35% compared with healthy subjects [42]. Levine and colleagues [43] found profound muscle fi ber atrophy accompanied by activation of the proteolytic ubiquitin-proteasome pathway in the diaphragm of patients mechanically ventilated for 18 to 96 hours. Th us, there is little discussion of whether respiratory muscles are weak in ICU patients, but the development of muscle fatigue after a failed-weaning trial is the subject of debate. Laghi and colleagues [41] assessed diaphragm force before and 30 minutes after a failed-weaning trial. Sur pris ingly, these patients did not develop inspiratory muscle fatigue. Possible explanations for these un expected fi ndings include early reconnection of the patient to the ventilator, recruitment of other inspiratory muscles, or muscle recovery from fatigue within 30 minutes after the weaning trial. Although ICU-acquired weakness is the most frequent neurological disorder in the ICU, other neurological disorders may be associated with diffi cult weaning (reviewed in [44]).

Diagnostic approach
Th e diagnostic approach of diaphragm dysfunction is sophisticated, and an in-depth neurological examination should be performed by a neurologist. Th e most frequently used test to evaluate respiratory drive in mechanically ventilated or spontaneously breathing patients is the airway occlusion pressure at 100 ms (P 0.1 ). To this end, the inspiratory limb of the ventilator is occluded and the drop in airway pressure is continuously monitored. In normal subjects, P 0.1 is about 0.5 to 1.5 cm H 2 O. It should be noted that P 0.1 depends not only on respiratory drive but also on inspiratory muscle capacity. Accordingly, a reduced P 0.1 is not a very specifi c sign of impaired drive, but in patients with normal/high P 0.1 , impaired drive is highly unlikely.
Generalized skeletal muscle weakness will guide the clinician to respiratory muscle weakness but does not prove inspiratory muscle weakness, and vice versa. A global assessment of inspiratory muscle function can be obtained by measuring maximal inspiratory pressure. Lower limits for maximal inspiratory pressure in healthy subjects younger than 65 years old are −75 cm H 2 O in men and −50 cm H 2 O in women [45]. Values that are more negative virtually exclude signifi cant inspiratory muscle weakness (good negative predictive value). Values that are less negative do not prove muscle weakness (poor positive predictive value), as this method depends on cooperation and lung volume. Accordingly, maximal inspiratory pressure may be useful in evaluating the diff erential diagnosis of diffi cult weaning. As this test does not evaluate the endurance of the inspiratory muscles, the value of maximal inspiratory pressure as a weaning predictor is much more controversial [46].
Th e defi nite diagnosis of respiratory muscle fatigue after a trial of spontaneous breathing is not feasible in routine clinical care. Th e development of hypercapnia during a weaning trial is not a very specifi c fi nding, as hypercapnia may result from causes other than muscle fatigue, such as central hypoventilation and rapid shallow breathing. Th e absence of hypercapnia in a weaningfailure patient virtually excludes inspiratory muscle fatigue.

Treatment strategies I. Mechanical ventilation is a double-edged sword for the respiratory muscles
Unloading prevents the development of fatigue, but inactivity is associated with the development of weakness. Very few studies have assessed the eff ect of inspiratory muscle training on muscle strength and in particular on clinical outcome such as duration of ventilatory assist. In an observational study, 10 diffi cultto-wean patients (mechanical ventilation for 34 ± 31 days before the start of the study) were subjected to inspiratory muscle training with a pressure threshold training device [47]. Training bouts consisted of three to fi ve sets of six inhalation maneuvers through the device, 5 to 7 times a week. In 44 ± 43 days, inspira tory pressure could be increased from 7 ± 3 cm H 2 O at inclusion up to 18 ± 7 cm H 2 O at the time of successful weaning. It should be noted that this is a small uncon trolled study and the training intensity was set at rather subjective criteria. However, from a physiological point of view, it seems reasonable to train the inspiratory muscles, as part of a generalized training program [48], in patients with inspiratory muscle weakness and prolonged weaning from mechanical ventilation. When considering a training schedule, clinicians should keep in mind that the diaphragm operates mostly at its submaximal force level. Th erefore, when the clinician decides that inspira tory muscle training may facilitate weaning from mecha nical ventilation, from a physiological point of view, an endurance training strategy is most appropriate.

II. Antioxidants modulate respiratory muscle function in healthy subjects
In a large, randomized, non-blinded trial, surgical ICU patients received the antioxidants alpha-tocopherol and ascorbic acid or only standard care from ICU admission until discharge [49]. Antioxidant supplementation was associated with a reduction in ventilator-dependent days from 4.6 to 3.7 (P <0.05). Further trials are needed before antioxidant treatment can be recommended.

III. Eff ects of tight glycemic control on clinically relevant outcome parameters
Th e eff ects of tight glycemic control on clinically relevant outcome parameters (mortality and days of mechanical ventilation) in critically ill patients are complex and well beyond the scope of this review. Notably, in a recent large prospective multicenter study [50], intensive glycemic control did not aff ect the duration of mechanical ventilation. Although that study did not investigate the prevalence of neuropathy and myopathy, the clinical relevance of such fi ndings would be controversial as tight glycemic control did not favor weaning and reduced survival in this trial.

IV. Eff ect of growth hormone treatment
Th e eff ect of growth hormone treatment (0.07 to 0.13 mg/kg of body weight per day) has been reported in two parallel studies published in 1999 [51]. Growth hormone treatment is associated with increased mortality (39% versus 20% in the Finnish study and 44% versus 18% in the European study). In addition, growth hormone increased time spent on the ventilator. However, the successful use of growth hormone supplementation in ICU patients who are more chronic has been described [52]. Future studies are needed to address this issue, but today the safety of growth hormone supplementation in weaning patients is unknown.

Endocrine and metabolic dysfunction
Th e role of endocrine disorders in diffi cult weaning has gained little interest in the literature. Huang and Lin [53] found that, in 93 selected diffi cult-to-wean patients, 70 fulfi lled criteria for adrenal insuffi ciency. Cortisol supplementation in these patients reduced duration of weaning compared with placebo (3.4 ± 2.3 days versus 6.5 ± 4.7 days, respectively; P <0.05). Th e pathophysiological mechanisms for improved clinical outcome in these cortisol-supplemented patients are, however, unknown.
For a variety of reasons, including decreased central drive [54] and respiratory muscle weakness [55], hypothy roidism may delay weaning from mechanical ventilation. In a retrospective study [56], 3% of patients (4/140) admitted to a long-term weaning facility were diagnosed with new-onset hypothyroidism. After thyroid hormone supplementation, three patients could be weaned from mechanical ventilation. Although the eff ect of thyroid hormone supplementation in diffi cult-weaning patients is anecdotal, it is reasonable in weaning-failure patients with a laboratory diagnosis of hypothyroidism.
Malnutrition frequently occurs in mechanically ventilated patients and is associated with poor prognosis [57] but also with reduced muscle mass and as such contributes to diffi cult weaning. Nutritional status should be evaluated by determining body mass index, plasma albumin concentration, and nitrogen balance [58]. Ideally, energetic needs should be determined by indirect calorimetry to prevent under-and overfeeding.
Electrolyte abnormalities, including very low plasma levels of phosphate and magnesium, have been shown to aff ect skeletal muscle function. Although these should be corrected in diffi cult-weaning patients, no studies have investigated the role electrolyte abnormalities in weaning failure. Other metabolic disturbances that increase the work of breathing and that therefore may be associated with diffi cult weaning include metabolic acidosis and fever.

Conclusions
As outlined, the reasons for failing a weaning trial are diverse. When a patient does not pass a weaning trial, structural evaluation could help to identify factors that played a role in that specifi c patient. Our opinion-based ' ABC' algorithm could be helpful. It should be noted that the algorithm has several loose ends as our understanding of weaning failure is far from complete.