Abstract
The discrepancy in results from different studies regarding outcome of weaning from mechanical ventilation may be due to several factors such as the differences in patient populations and weaning indexes used. In order to analyze the clinical characteristics and weaning indexes in patients undergoing a 2-h T-piece weaning trial and the relationship between the etiology of acute respiratory failure (ARF) and the outcome of this weaning trial, we prospectively studied 217 patients receiving mechanical ventilation who met standard weaning criteria. Successful weaning occurred in 57.6% (125 of 217) of patients: 13 of 33 (39.4%) patients with chronic obstructive pulmonary disease (COPD), 27 of 46 (58.7%) neurologic patients, and 85 of 138 (61.6%) patients with ARF. Ventilatory support was reinstituted in 31.8% (69 of 217) patients: 20 of 33 (60.6%) of patients with COPD, four of 46 (8.7%) neurologic patients, and 45 of 138 (32.6%) patients with ARF (p < 0.001). Reintubation was required in 23 of 148 (15.5%) patients: 15 of 42 (35.7%) neurologic patients, and eight of 93 (8.6%) patients with ARF, whereas no patient with COPD was reintubated (p < 0.001). Using a discriminant analysis, the following variables were selected as the best predictors of outcome: (1) in the whole population, days of mechanical ventilation before weaning trial (DMV), frequency-to-tidal volume ratio (f/Vt), maximal inspiratory pressure (MIP), airway occlusion pressure (P0.1), maximal expiratory pressure (MEP), and vital capacity (VC); (2) in patients with ARF, DMV, P0.1/MIP, MIP, f/Vt, and age; (3) in patients with COPD, f/Vt, P0.1, P0.1/MIP, MIP, age, and DMV; (4) in neurologic patients, MIP, MEP, and f/Vt · P0.1. Using these predictors, 74.6% of the whole population, 76.1% of patients with ARF, 93.9% of patients with COPD, and 73.9% of neurologic patients were accurately classified as weaning successes or failures. The highest rate of reintubation occurred in neurologic patients. In this group, the ability to cough and clear respiratory secretions, objectively reflected by MEP, may help in clinical decision-making.
The purpose of weaning indexes is to differentiate between those patients who can maintain spontaneous breathing indefinitely and those who are unable to do so in order to avoid both the premature discontinuation of ventilatory support and unnecessarily long periods of mechanical ventilation. Many factors influence the weaning outcome: the functional parameters used as indexes of weaning, the criteria used to define weaning failure or success, the moment at which the patients are studied, different clinical practice from unit to unit (i.e., sedation, analgesia, muscle paralysis) and, probably, the differences in patient populations. Weaning indexes have been established from patients with different clinical conditions (chronic obstructive pulmonary disease [COPD], heart failure, neurologic disease, etc.) and are frequently inaccurate (1, 2). The weaning process is still considered to be on the frontier between art and science.
Classic respiratory parameters such as vital capacity (VC), maximal inspiratory pressure (MIP), and expired volume per minute (V˙e) are useful in patients who have been receiving ventilatory support for a short period of time (3), but their value as weaning predictors in prolonged mechanical ventilation, in COPD, and also in elderly patients has not been demonstrated (4-7). These weaning parameters have been reported to delay extubation in some patients (8), whereas in others, reintubation will be necessary within the first hours (9, 10). New indexes integrating different physiologic variables have recently been proposed in an attempt to achieve greater accuracy in predicting the weaning outcome (5, 7, 11, 12). The frequency-to-tidal volume ratio (f/Vt) described by Yang and Tobin (7) has shown greater predictive power than MIP or V˙e. The study by Sassoon and Mahutte (13) has shown a high sensitivity and specificity for both airway occlusion pressure (P0.1) and f/Vt and for their combination (P0.1·f/Vt). However, some recent reports have suggested that the accuracy of these indexes may depend on the underlying disease (14).
Moreover, two recent investigations (10, 15) report that a two-step process of screening patients with a set of objective criteria and followed by a 2-h spontaneous breathing period (i.e., a weaning trial) before deciding either extubation or reconnection to mechanical ventilation is very useful to decide the discontinuation of mechanical ventilation in most patients. This strategy can also reduce the number of days of mechanical ventilation and the complications rate. It may reduce ICU costs as much as 25%, according to the study by Ely and coworkers (15). Nevertheless, a considerable number of patients, at least 25% (9, 10), still fail the weaning trial and need either reintubation or resumption of mechanical ventilation.
The aim of the present study was to determine the clinical and functional characteristics of patients undergoing a 2-h weaning trial and also the relationship between the etiology of the acute respiratory failure and the outcome of this trial.
Over a 2-yr period (1994–1995), we prospectively studied all patients who were intubated and mechanically ventilated for more than 48 h in our 16-bed medical-surgical intensive care unit. Patients were classified depending on the etiology of disease: (1) patients with COPD, (2) patients with central nervous system disease (neurologic), and (3) patients with acute respiratory failure of various etiologies (ARF). COPD was diagnosed according to the clinical history (tobacco smoking, chronic productive cough, dyspnea, and wheezing), physical examination, and laboratory tests (including chest radiography, arterial blood gas determinations showing hypoxemia, hypercapnia, and high plasma bicarbonate levels, and, in some patients, pulmonary function tests). The underlying disease in neurologic patients was: ischemic stroke in 11, hypertensive intracerebral hemorrhage in four, subarachnoid hemorrhage in three, head trauma (including contusion, brain hemorrhages, shearing lesions, and subdural and epidural hematomas) in 18, meningoencephalitis in three, metabolic encephalopathy in three, and postneurosurgical states as a result of brain tumor in four. ARF occurred as a result of cardiac surgery in 43 patients, abdominal surgery in 33, pneumonia in 13, acute respiratory distress syndrome in 13, sepsis in 18, multiple trauma in eight, heart failure in eight, and acute gastrointestinal bleeding in two. All patients were monitored with pulse oximetry and electrocardiography and were controlled hemodynamically from an arterial (radial or femoral) line. The ventilatory mode used in all patients before beginning the study was volume assist-control ventilation, with a constant inspiratory flow pattern.
Patients were enrolled in the study when the underlying cause of respiratory failure had improved and the arterial oxygen saturation was equal to or greater than 90% for an inspired oxygen fraction equal to or lower than 0.4, with a PEEP level equal to or lower than 5 cm H2O. The inclusion criteria were: body temperature below 38° C, hemoglobin equal to or higher than 8 g/dl, cardiovascular pharmacologic therapy (including inotropic agents, vasodilators, and/or diuretics) considered appropriate by the primary physician when cardiac insufficiency and/or ischemia was known or suspected, correction of electrolyte disorders, no intravenous sedatives (including benzodiazepines, opiates, propofol, and barbiturates) given for at least 48 h before the weaning trial, and adequate neurologic status. The neurologic status was evaluated using the Glasgow coma scale, and 11 points (patient conscious with normal motor response, at least in one limb, and able to obey commands, but unable to answer verbally because of intubation) were necessary for inclusion in the study. These criteria did not exclude patients with a focal neurologic deficit such as hemiplegia. Patients with a tracheostomy were excluded. Informed consent was obtained from all patients.
After enrollment in the study, patients were disconnected from the ventilator, O2 was added to achieve arterial oxygen saturation equal to or above 90% as measured by pulse oximetry, and patients breathed spontaneously for 3 to 5 min through a T-tube circuit. Tidal volume (Vt) and respiratory frequency (f) were measured with a spirometer during this period. We measured these parameters in the first 3 to 5 min instead of the first minute after disconnection, as originally described by Yang and Tobin (7). MIP was also measured, and the most negative value of three efforts was selected. When at least two of the following criteria were present (f ⩽ 35 breaths/min, Vt ⩾ 5 ml/kg of body weight, and MIP lower than −20 cm H2O), patients were considered ready to be weaned. Ability to sustain prolonged spontaneous breathing was then continuously evaluated by allowing the patient to breathe spontaneously on the T-piece for 2 h (T-piece trial) with a supplementary supply of humidified oxygen if necessary. Before starting this T-piece trial, P0.1, maximal expiratory pressure (MEP), MIP, and VC were measured to predict weaning outcome (see Procedures below). If the 2-h T-piece trial was clinically well tolerated, patients were extubated. If clinical tolerance to T-piece trial was poor, patients were reconnected on partial or total ventilatory support. Clinical tolerance to a spontaneous breathing trial was considered poor when f was greater than 35 breaths/min or increased by 50% or more, when heart rate was above 140 beats/min or increased by 20% or more or if arrhythmias appeared, when systolic blood pressure was lower than 80 mm Hg or greater than 160 mm Hg or when patients showed agitation, depressed mental status or diaphoresis. These clinical data were registered in a questionnaire by the attending physician. The decision of reconnection to mechanical ventilation was left to the attending physician. Nurses were aware of this protocol, and they significantly contributed to its implementation. This rapid weaning trial was considered a failure when patients needed reintubation within 48 h or did not tolerate spontaneous breathing and required reconnection to mechanical ventilation. Weaning was considered successful if spontaneous breathing was sustained for more than 48 h after extubation.
Measurements of functional respiratory parameters before starting the T-piece trial were performed in a semirecumbent position. Airflow was measured with a heated Fleisch no. 2 pneumotachograph (Metabo, Epalinges, Switzerland), which was attached to the oral end of the endotracheal tube. The pneumotachograph was calibrated with a 1-L syringe connected to a differential pressure transducer (MP45 ± 2.25 cm H2O; Validyne Corp., Northridge, CA). Airway pressure (Paw) was measured from a side port proximal to the endotracheal tube by means of a differential pressure transducer (Validyne MP45 ± 225 cm H2O). From the airflow and Paw signals we calculated: f, Vt, P0.1, MEP, MIP, f/Vt, V˙e, VC, P0.1/MIP, and P0.1 · f/Vt. Using a 12-bit analog-to-digital converter (2901; Data Translation, Marlboro, MA), the airflow and Paw signals were continuously digitized at 128 Hz, and were acquired by a microcomputer (IBM 55Sx; Greenock, UK) to perform subsequent calculations. The airflow signal was used to calculate the components of the breathing pattern. A unidirectional valve (Hans Rudolph, Kansas City, MO) was connected to the proximal end of the pneumotachograph to measure P0.1.P0.1 was obtained by performing selective occlusions in the inspiratory limb by manually inflating a latex balloon during expiration. Values of P0.1 are the average of five measurements obtained at random during a 60- to 90-s period. MEP was measured by occluding the expiratory port of a unidirectional valve (Hans Rudolph). In this way, patients could inspire but not expire. During this maneuver, the patients were coached to actively perform an expiratory effort against the occluded airway. At the end of 25 to 30 s of occluded expiration, the most positive pressure developed was recorded as the MEP for that run. MIP was measured according to the method described for uncooperative patients (16) and was obtained by occluding the inspiratory port of the unidirectional valve (Hans Rudolph, Kansas City, MO). This allows the patients to expire but not to inspire. At the end of 25 to 30 s of occluded inspiration, the most negative pressure developed was recorded as the MIP for that run. VC was measured according to the “stacking” method (17): VC values were calculated by adding the expiratory reserve volume and the inspiratory capacity obtained from the airway selective occlusions used to calculate maximal pressures. After selective airway occlusions, patients were reconnected to ventilatory support for 1 or 2 min, if necessary, until heart rate and arterial saturation returned to basal levels.
The respiratory parameters evaluated as weaning indices were: f, Vt, f/Vt, V˙e, P0.1, MIP, MEP, VC, P0.1/MIP, P0.1·f/Vt, age, and days of mechanical ventilation before weaning trial (DMV). Data obtained were compared between successful weaning and weaning failure groups using a one-way analysis of variance. Multiple comparisons were carried out among etiologic groups (COPD, ARF, and neurologic) by means of Tukey's test. A p value < 0.05 was considered as statistically significant. All data are presented as mean ± SD.
A discriminant analysis was performed in order to classify the cases according to the independent variable recorded. Discriminant analysis allows the classification of the population under study into groups (in this case, success of weaning/ failure of weaning). The independent variables were included stepwise into an equation until the best combination of variables for classification was obtained. Discriminant analysis included the overall cohort. The independent variables analyzed were f/Vt, V˙e, P0.1, MIP, MEP, VC, P0.1/MIP, P0.1·f/Vt, age, and DMV. The discriminant score allowed patients to be classified in the highest probability group (success or failure of weaning). The efficiency of the discriminant function was then evaluated as the percentage of correctly classified cases (number of patients correctly predicted group/number of patients in the actual group).
We also assessed the accuracy of f/Vt and P0.1 as weaning predictors by using recommended threshold values, i.e.: f/Vt ⩽ 100 and P0.1 ⩽ 4.5 (7, 18). The accuracy of f/Vt and P0.1 is represented as sensitivity (SE = TP/TP+FN), specificity (SP = TN/TN+FP), positive (PPV = TP/TP+FP) and negative (NPV = TN/TN+FN) predictive value, and diagnostic accuracy (DA = TP+TN/TP+TN+FP+FN), where TP(true positive) is f/Vt ⩽ 100 and P0.1 ⩽ 4.5 and weaning success, TN (true negative) is f/Vt > 100 and P0.1 > 4.5 and weaning failure, FP (false positive) is f/Vt ⩽ 100 and P0.1 ⩽ 4.5 and weaning failure and FN (false negative) is f/Vt > 100 and P0.1 > 4.5 and weaning success.
We included a total of 217 patients: 33 patients with COPD, 46 neurologic patients, and 138 patients with ARF. The mean age was 57.6 yr (range, 16 to 83 yr), 151 were male and 66 were female. After the 2-h T-piece trial, weaning was successful in 125 patients (57.6%): 13 COPD, 27 neurologic, and 85 ARF. Intolerance to the 2-h T-piece trial was detected in 69 patients (31.8%): 20 patients with COPD, four neurologic patients, and 45 patients with ARF. Finally, 23 (15 neurologic patients and eight patients with ARF) out of 148 patients, 15.5%, who had tolerated the 2-h T-piece trial and were extubated, needed reintubation within 48 h. The evolution of patients according to the etiology of respiratory failure is shown in Figure 1. Reintubation was required in 35.7% (15 of 42) of the neurologic patients and in 8.6% (eight of 93) of the patients with ARF, but it was unnecessary in any of the patients with COPD (p < 0.001). The reasons for reintubation in the neurologic group were: inability to clear secretions in five patients, fever with a decreased consciousness level in three, atelectasis with hypoxemia in three, and angor pectoris, acute bronchospasm, pulmonary aspiration of gastric contents, and upper airway obstruction, each in one patient. The reasons for reintubation in the ARF group were: hypoxemia in two patients and atelectasis with hypoxemia, sepsis, bronchopulmonary infection, cardiorespiratory arrest, inability to clear secretions, and upper airway obstruction, each in one patient.
Fig. 1. Distribution of the 217 patients according to their evolution and etiology. Of 33 patients with COPD, 13 (39.4%) were successfully extubated and 20 (60.6%) required progressive withdrawal of mechanical ventilation; 46 had a neurologic disease, 27 of whom (58.7%) were successfully extubated, four (8.7%) required progressive withdrawal from mechanical ventilation, and 15 (32.6%) were reintubated; 138 had ARF of various etiologies, with 85 (61.6%) being successfully extubated, 45 (32.6%) requiring progressive withdrawal of mechanical ventilation, and eight (5.8%) being reintubated. *p < 0.01, COPD versus other groups; **p < 0.001, neurologic versus other groups.
[More] [Minimize]The mean time of the spontaneous breathing period before reconnection to mechanical ventilation, excluding reintubated patients who, by definition, tolerated a 120-min spontaneous breathing trial, was 39 min (range, 2 to 115 min). In these patients, the weaning trial failed in the first 30 min in 44 of 69 (64%) patients (12 COPD, 28 ARF, and 4 neurologic), between 30 and 60 min in eight of 69 (12%) patients (one COPD and seven ARF), between 60 and 90 min in seven of 69 (10%) patients (two COPD and five ARF), and between 90 and 120 min in 10 of 69 (14%) patients (five COPD and five ARF) (see Figure 2). We did not find any relationship between the following parameters: f, Vt, f/Vt, MIP, MEP, P0.1, and DMV and time to failure in the weaning trial. The reasons for weaning trial failure were: tachypnea > 35 breaths/min in 28 patients (22 ARF and 6 COPD), diaphoresis plus agitation in 23 (11 ARF, 10 COPD, and two neurologic), tachypnea plus tachycardia in nine (six ARF, two COPD, and one neurologic), hypertension plus tachycardia in seven (four ARF, two COPD, and one neurologic), and cardiac arrhythmias in two patients with ARF. Among the 37 patients reconnected because of tachypnea, 22 (59%) already had a high f/Vt ratio.
Fig. 2. Distribution of patients according to the time of spontaneous breathing in the weaning trial before reconnection to mechanical ventilation.
[More] [Minimize]The clinical characteristics and functional respiratory parameters obtained in all patients are shown in Table 1. There were no significant differences between weaning failure or success regarding age and minute volume, whereas significant differences were observed in the other parameters. The same parameters are shown in Table 2 when patients were analyzed according to the etiology of the disease. The variables selected by discriminant analysis and the efficiency of discriminant function, evaluated as the percentage of correctly classified cases, are shown in Table 3. The variables with more weight selected by the discriminant analysis were, in order of importance: DMV, f/Vt, MIP, P0.1, MEP, and VC in the whole population; DMV, P0.1/MIP, MIP, f/Vt, and age when only patients with ARF were considered; f/Vt, P0.1, P0.1/MIP, MIP, age, and DMV in patients with COPD and MIP, MEP, and f/Vt·P0.1 in neurologic patients. The discriminant function correctly predicted the weaning outcome in 74.6% (162 of 217) of the total group, in 76.09% (105 of 138) of patients with ARF, in 93.9% (31 of 33) of patients with COPD, and in 73.9% (34 of 46) of neurologic patients. The sensitivity, specificity, predictive power, and diagnostic accuracy for f/Vt and P0.1 are shown in Table 4.
| All Patients (n = 217 ) | ||||
|---|---|---|---|---|
| SW (n = 125) | FW (n = 92) | |||
| Age, yr | 55 ± 17 | 59 ± 15 | ||
| DMV | 7 ± 7 | 12 ± 12† | ||
| f, breaths/min | 24 ± 6 | 29 ± 8† | ||
| Vt, ml | 432 ± 143 | 378 ± 134‡ | ||
| V˙ e, L | 11 ± 10 | 10 ± 3 | ||
| MIP, cm H2O | 65 ± 21 | 53 ± 17† | ||
| MEP, cm H2O | 53 ± 25 | 37 ± 17† | ||
| P0.1, cm H2O | 3.6 ± 1.5 | 5 ± 2.4† | ||
| f/Vt, breaths/min/L | 65 ± 30 | 88 ± 44† | ||
| VC, ml | 1,634 ± 642 | 1,297 ± 512† | ||
| f/Vt · P0.1, cm H2O breaths/min/L | 241 ± 177 | 452 ± 363† | ||
| P0.1/MIP, % | 6.3 ± 3.2 | 10.3 ± 5.6† | ||
| ARF (n = 138) | COPD (n = 33) | Neurologic (n = 46 ) | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| SW (n = 85) | FW (n = 53) | SW (n = 13) | FW (n = 20) | SW (n = 27 ) | FW (n = 19) | |||||||
| Age, yr | 58 ± 16 | 58 ± 15 | 61 ± 11 | 68 ± 8 | 47 ± 20 | 55 ± 18 | ||||||
| DMV | 6 ± 7 | 13 ± 13† | 5 ± 3 | 9 ± 10 | 10 ± 8 | 13 ± 10 | ||||||
| f, breaths/min | 25 ± 7 | 29 ± 7‡ | 19 ± 5 | 30 ± 8† | 25 ± 6 | 29 ± 8 | ||||||
| Vt, ml | 428 ± 125 | 404 ± 145 | 422 ± 209 | 291 ± 90‡ | 452 ± 164 | 401 ± 105 | ||||||
| V˙ e, L | 10 ± 3 | 11 ± 3 | 7.6 ± 2 | 8.4 ± 3 | 15 ± 20 | 11 ± 3 | ||||||
| MIP, cm H2O | 64 ± 21 | 52 ± 18‡ | 60 ± 19 | 59 ± 16 | 70 ± 19 | 58 ± 13‡ | ||||||
| MEP, cm H2O | 55 ± 27 | 36 ± 16† | 48 ± 21 | 48 ± 17 | 50 ± 23 | 31 ± 8† | ||||||
| P0.1, cm H2O | 3.7 ± 1.6 | 4.9 ± 2.3† | 3.2 ± 1.2 | 5.6 ± 3.1‡ | 3.7 ± 1.5 | 4.3 ± 2.1 | ||||||
| f/Vt, breaths/min/L | 64 ± 32 | 84 ± 37‡ | 59 ± 32 | 110 ± 50‡ | 59 ± 22 | 82 ± 52 | ||||||
| VC, ml | 1,598 ± 632 | 1,206 ± 507† | 1,491 ± 851 | 1,289 ± 431 | 1,808 ± 525 | 1,542 ± 526 | ||||||
| f/Vt · P0.1, cm H2O breaths/min/L | 252 ± 189 | 425 ± 314† | 217 ± 193 | 595 ± 431‡ | 223 ± 119 | 413 ± 457 | ||||||
| P0.1/MIP, % | 6.1 ± 3.2 | 10.2 ± 6.1† | 6.3 ± 3.8 | 12.6 ± 8.7‡ | 5.7 ± 2.8 | 7.8 ± 4.3 | ||||||
| Group | Variables | Standardized Canonical Discriminant Function Coefficients | Percent of “Grouped” Cases Correctly Classified | |||||||
|---|---|---|---|---|---|---|---|---|---|---|
| SW | FW | Total | ||||||||
| All Patients, | DMV | 0.531 | 76.8% | 71.7% | 74.65% | |||||
| n = 217 | f/Vt | 0.480 | (96/125) | (66/92) | (162/217) | |||||
| MIP | 0.449 | |||||||||
| P0.1 | 0.405 | Cut point: FW < 0.1 < SW | ||||||||
| MEP | 0.322 | |||||||||
| CV | 0.312 | |||||||||
| Patients with ARF, | DMV | 0.693 | 78.8% | 71.7% | 76.09% | |||||
| n = 138 | P0.1/MIP | 0.459 | (67/85) | (38/53) | (105/138) | |||||
| MIP | 0.396 | |||||||||
| f/Vt | 0.263 | Cut point: FW < 0.16 < SW | ||||||||
| Age | 0.251 | |||||||||
| Patients with COPD, | f/Vt | 2.314 | 100% | 90% | 93.94% | |||||
| n = 33 | P0.1 | 1.804 | (13/13) | (18/20) | (31/33) | |||||
| P0.1/MIP | 0.955 | |||||||||
| MIP | 0.951 | Cut point: FW < −0.18 < SW | ||||||||
| Age | 0.752 | |||||||||
| DMV | 0.288 | |||||||||
| Neurologic patients, | MIP | 0.584 | 74.1% | 73.7% | 73.91% | |||||
| n = 46 | MEP | 0.525 | (20/27) | (14/19) | (34/46) | |||||
| f/Vt · P0.1 | 0.470 | Cut point: FW < −0.1 < SW | ||||||||
| Sensitivity | Specificity | Positive Predictive Value | Negative Predictive Value | Diagnostic Accuracy | ||||||
|---|---|---|---|---|---|---|---|---|---|---|
| f/Vt ⩽ 100 | ||||||||||
| Total group | 0.90 | 0.36 | 0.66 | 0.73 | 0.67 | |||||
| COPD | 0.92 | 0.65 | 0.63 | 0.93 | 0.76 | |||||
| Neurologic | 0.93 | 0.26 | 0.64 | 0.71 | 0.65 | |||||
| ARF | 0.89 | 0.28 | 0.64 | 0.71 | 0.66 | |||||
| P0.1 ⩽ 4.5 cm H2O | ||||||||||
| Total group | 0.75 | 0.55 | 0.70 | 0.62 | 0.67 | |||||
| COPD | 0.77 | 0.80 | 0.71 | 0.84 | 0.79 | |||||
| Neurologic | 0.81 | 0.37 | 0.65 | 0.58 | 0.63 | |||||
| ARF | 0.73 | 0.53 | 0.71 | 0.55 | 0.65 |
There were no significant differences in ICU mortality rates regarding the cause of respiratory failure: 21.2% (seven of 33) in COPD, 15.2% (seven of 46) in neurologic, and 14.7% (18 of 138) in ARF group rates. Mortality rates among patients failing the trial of spontaneous breathing was 24.6% (17 of 69). ICU mortality among patients requiring reintubation was 34.8% (eight of 23), and it was significantly higher than mortality in successfully extubated patients, 5.6% (seven of 125), p < 0.001.
This study has confirmed that a 2-h T-piece weaning trial is often failed in patients who satisfy classic weaning criteria (at least two of these three: f ⩽ 35 breaths/min, Vt ⩾ 5 ml/kg of body weight, and MIP lower than −20 cm H2O), especially for COPD. However, we do not know how many patients who do not satisfy the above-mentioned criteria will go on to fail this weaning trial, simply because they were excluded up front. The predictive value of respiratory functional parameters is strongly influenced by the etiology of respiratory failure. In patients with COPD, f/Vt and P0.1 are the most accurate weaning predictors. The highest percentage of reintubations occurred in neurologic patients, and respiratory functional parameters did not predict extubation outcome in this group.
The protocol used in this study differentiated well between good tolerance to spontaneous breathing leading to successful extubation and the need to continue under mechanical ventilation. Classic weaning criteria (i.e., at least two of the following: f ⩽ 35 breaths/min, Vt ⩾ 5 ml/kg, MIP < −20 cm H2O) together with the addition of a 2-h T-piece trial, identified a group of 148 extubated subjects (of whom 23, or 15.5%, needed subsequent reintubation) and another group of 69 patients (31.8%) who were reconnected to mechanical ventilation because of signs of poor clinical tolerance. The addition of a 2-h T-piece trial thus allowed a correct selection of 194 out of 217 patients (89%): 125 who were successfully extubated plus 69 who were reconnected to mechanical ventilation. Indeed, only 125 of 217 patients (58%) satisfying two of three classic weaning criteria were successfully extubated, thus indicating the poor accuracy of these criteria in predicting weaning outcome. These results coincide with other reports where about 60% of patients were successfully extubated after adding a 2-h period of T-piece trial to classic weaning indices such as Vt or MIP (10, 19). However, using these classic weaning criteria as entrance criteria does not reliably provide the true negative rate (failed criteria and failed weaning) or the false negative rate (failed criteria and not failed weaning) because when the screening criteria predicted failure, the patients simply were not enrolled in the protocol. This is clearly a limitation of this and similar studies (9, 10, 19), and some patients could perhaps have been safely extubated despite not fulfilling these criteria.
A spontaneous breathing period is routinely used to evaluate the withdrawal of mechanical ventilation (9, 10, 19, 20). Gandı́a and Blanco (20) considered that a 2-h weaning trial duration was not excessively prolonged because their patients had been receiving mechanical ventilation for many days as a consequence of severe lung disease. Indeed, in our population, this 2-h T-piece trial represented only 0.9% of the total time that patients were receiving mechanical ventilation. During this period, patients need close clinical surveillance and it is, therefore, advisable to keep it as short as possible, without paying the price of an unduly high reintubation rate. In a preliminary communication, Esteban and coworkers (21) found no differences in reintubation rate when a 30-min weaning trial was compared with the traditional 120-min trial (13.5 versus 13.4%, respectively), suggesting that 30 min is better because it is shorter. In their study, however, the rate of failed spontaneous breathing trial was much lower (12.2% in a 30-min trial and 15.6% in a 120-min trial) than in our study, where it was 31.8%. The reason may be a different case mix of patients and/or a different duration of mechanical ventilation and/or a different clinical approach before the decision of weaning was taken. In our study, as Figure 2 shows, the weaning trial failed during the first 30 min of spontaneous breathing in most patients (44 of 69, 64%). Nevertheless, a considerable number of patients (25 of 69, 36%) showed clinical intolerance to the T-piece during the last 90 min. If we consider shortening the T-piece trial duration, all these aspects should be born in mind.
We do not know whether patients who did not tolerate a 2-h T-piece trial would have been safely extubated without needing subsequent reintubation with a shorter protocol duration or different strategy. Indeed, Esteban and coworkers (19) showed that patients randomized to pressure support ventilation set at 7 cm H2O were more likely to pass a weaning trial than were those randomized to a T-piece trial (without any increase in reintubation rate). This suggests that some patients failing the T-piece could have been extubated successfully with pressure support. De Haven and coworkers (22) found that 105 out of 589 extubated patients had a preextubation respiratory rate > 30 breaths/min, which was considered a reason for failure. Of these 105 patients, 97 were successfully extubated, thus indicating that tachypnea per se is sensitive but not sufficiently specific. Lastly, reports on unplanned extubation make it clear that many patients who appear to be ventilator-dependent are actually extubatable. In a recent study in our institution (23), we found that after an episode of unplanned extubation, the only independent variables associated with the need for reintubation were the number of days of mechanical ventilation and type of support (whether or not patients were being weaned from mechanical ventilation) at the time of unplanned extubation. When patients were in the weaning period only five out of 32 needed reintubation, whereas reintubation was necessary in 22 out of 27 who had unplanned extubation during full mechanical ventilatory support (p < 0.001). These data indicate that some patients are receiving mechanical ventilation for a longer period of time than necessary.
From the little data available in the literature, the reintubation rate ranges from 4 to 19% (9, 10, 15, 19, 24). We do not yet know what can be considered an “optimal” reintubation rate. A very low rate could reflect an unduly prolonged duration of mechanical ventilation, whereas a very high rate could reflect precocious extubation, which may be a significant source of associated complications such as nosocomial pneumonia (25) and ICU and hospital mortality (19, 24). It is conceivable, however, that the need for reintubation is a marker of the severity of the illness rather than of mortality per se (26). Our data also showed an increase in ICU mortality rate for reintubated patients in comparison with successfully extubated patients (34.8 versus 5.6%, p < 0.001), thus emphasizing the need to improve our understanding of the extubation failure.
The 15.5% reintubation rate we observed using the 2-h T-piece trial was similar to that previously reported by others (9, 10, 24). Ely and colleagues (15) recently found a very low reintubation rate (4%) in medical-coronary patients who met screening of the respiratory function and tolerated trials of unassisted breathing. If we had excluded the neurologic group, only 6% (eight of 125) of patients would have been reintubated, confirming that a two-step approach using objective screening criteria along with 2-h spontaneous breathing can yield a very low reintubation rate. As well as the etiology of the disease, other factors that could influence the reintubation rate include the previous duration of mechanical ventilation before weaning was attempted, and the different clinical approach in managing these patients, i.e., extubation did not occur as an automatic step after passing a spontaneous breathing trial in some studies (15).
Weaning may be influenced by the underlying disease. In agreement with the literature (27), our results show that 60.6% of patients with COPD needed progressive withdrawal of mechanical ventilation. On the other hand, ventilatory support was resumed in only 8.7% of neurologic patients because of intolerance to the T-piece trial.
The number of days receiving mechanical ventilation (DMV) before the weaning trial was started was significantly higher in the weaning failure group (12 ± 12) than in the successful weaning group (7 ± 7), p < 0.001. The discriminant analysis performed in all patients and in the subgroup of patients with ARF emphasizes the importance of DMV in weaning outcome, as this is the most important variable selected in both groups. The length of time receiving mechanical ventilation before the weaning trial could indirectly indicate the severity of the initial injury, leading to intubation and mechanical ventilation and also the different clinical approach regarding practices such as sedation, analgesia, and muscle paralysis.
Of course, using f and Vt as screening criteria tends to favor the exclusion of patients with rapid and/or shallow breathing, thus making it difficult to know the true test characteristic of f/Vt. Another limitation of our study is that the apparent accuracy of f/Vt is perhaps improved using tachypnea as a criterion for weaning failure because 59% of patients who were reconnected because of tachypnea already had a high f/Vt. Moreover, rapid shallow breathing during a weaning trial with the patients receiving partial ventilatory support does not necessarily preclude successful extubation (14, 28). Gandı́a and Blanco (20) found that the breathing pattern in successfully extubated patients was not stable after discontinuation of ventilatory support. An increase in V˙e and respiratory frequency was observed 15 min after beginning the weaning trial. Therefore, prediction of the outcome of a weaning trial could theoretically be less accurate if the criteria used were determined too early. This may explain why the addition of a 2-h T-piece trial is very useful to clinically evaluate these patients. Additionally, an interesting study by Epstein (14) showed that extubation failed in patients with a f/Vt < 100 because of a process different from or in addition to the underlying illness, thus suggesting that the pathophysiologic basis for weaning failure is probably different from that responsible for extubation failure. The findings in our neurologic group seem to confirm this contention.
Our results also suggest that the usefulness of f/Vt as a weaning index depends on the etiology of respiratory failure. In patients with COPD, f/Vt > 100 is highly accurate to predict a weaning trial failure because f/Vt was the first index selected by discriminant analysis, and extubation did not fail in any patient with COPD. Furthermore, the diagnostic accuracy to predict weaning success obtained with a threshold f/Vt value lower than 100 is higher in patients with COPD (0.76) than in neurologic patients or patients with ARF (0.65 and 0.66, respectively). However, one patient with COPD and a f/Vt > 100 was safely extubated because the tolerance to the 2-h T-piece trial was clinically satisfactory. The explanation may be that factors other than the physiologic work of breathing and respiratory muscle capacity such as psychological or mental stress (29), sex, endotracheal tube size (30), or age (31) also influence the pattern of breathing. This again emphasizes the importance of a spontaneous breathing trial because when a f/Vt index predicts failure, some patients could be successfully extubated if the spontaneous breathing trial was clinically well tolerated.
The value of P0.1 as an index to predict the weaning outcome is not widely accepted, probably because its measurement requires additional equipment. There is also high interindividual and intraindividual variability and the P0.1 value also depends on end-expiratory lung volume (32), which makes comparisons difficult. In the study by Sassoon and coworkers (5) performed in 12 patients with COPD, a P0.1 higher than 6 cm H2O showed a sensitivity and specificity of 100%. Herrera and colleagues (33) found that a P0.1 higher than 4.2 cm H2O had a sensitivity of 78% and a specificity of 100%. Sassoon and Mahutte (13) recently showed that the predictive power of P0.1 in patients with respiratory failure of various etiologies who failed weaning is nearly as high as both the f/Vt index and the product f/Vt·P0.1. In our study, when analyzing the overall population and the COPD and ARF subgroup, we found significantly lower values of P0.1 in successfully extubated patients in comparison with patients who failed weaning, but no differences in P0.1 values were observed in neurologic patients. These findings and the data shown in Table 4 suggest that patient population could explain the different results obtained with P0.1 as a predictor of weaning outcome.
Another interesting finding in our study was that the weaning trial failed in a high percentage of neurologic patients because of extubation failure (35.7%). The need for reintubation in these patients was neither clinically suspected nor suggested by abnormal physiologic indexes. Hence, in patients with mental status changes or neurologic impairment, other tools need to be developed and prospectively evaluated in order to decide the appropriate time to withdraw mechanical ventilation. Our data suggest that from a clinical point of view, it is important to evaluate the ability to cough and clear secretions, especially in neurologic patients. In our study, only MIP and MEP values were significantly lower in those patients in whom weaning failed. Additionally, the discriminant analysis selected MEP, MIP, and f/Vt·P0.1 as predictors of weaning outcome, with a diagnostic accuracy of 74%. It has been suggested that MEP is related to the ability to cough (34). This can be assessed objectively by measuring MEP. Alterations of expiratory muscles and coordination between inspiratory and expiratory muscles can make coughing and clearing of respiratory secretions difficult, possibly leading to progressive hypoxemia and hypercapnia, and ultimately warranting reintubation. Furthermore, MEP values in neurologic patients and patients with ARF who failed the weaning trial were significantly lower than in the successful weaning trial group, but this was not so in patients with COPD. One possible explanation could be that most patients with COPD present a constant expiratory muscular activity (35), which helps adequate clearance of secretions.
Because in some circumstances clinical judgment alone is not sufficiently accurate to predict weaning outcome (36), it seems reasonable, from a practical standpoint, that weaning predictors and clinical judgment should be combined. Indeed, the accuracy of some weaning predictors can be improved by clinical evaluation (for example in patients with ARF or COPD) and, conversely, the use of some weaning predictors can improve the accuracy of clinical judgment (for example, in neurologic patients). In conclusion, the addition of a period of spontaneous breathing, such as a 2-h T-piece trial to the classic weaning criteria is very useful to determine the weaning trial outcome in a general ICU patient population. The respiratory parameters studied as weaning predictors vary considerably depending on the underlying disease. In patients with COPD, f/Vt and P0.1 are the most accurate weaning predictors. Finally, neurologic patients presented the highest percentage of reintubations. In this group, MEP can evaluate the ability to cough and clear respiratory secretions, which may help in clinical decision-making.
The writers thank Mrs. Carolyn Newey for helping in the editing of the English manuscript and Mr. Ignasi Gich and Mrs. Ela Bak for statistical advice.
Supported in part by Grant FISss 95/0605.
| 1. | Tahvanainen J., Salmenpera M., Nikki P.Extubation criteria after weaning from intermittent mandatory ventilation and continuous positive airway pressure. Crit. Care Med.111983702707 |
| 2. | Morganroth M. L., Morganroth J. L., Nett L. M., Petty T. L.Criteria for weaning from prolonged mechanical ventilation. Arch. Intern. Med.144198410121016 |
| 3. | Tomlinson J. R., Miller K. S., Lorch D. G., Smith L., Reines H. D., Sahn S. A.A prospective comparison of IMV and T-piece weaning from mechanical ventilation. Chest961989348352 |
| 4. | Menzies R., Gibbons W., Goldberg P.Determinants of weaning and survival among patients with COPD who require mechanical ventilation for acute respiratory failure. Chest951989398405 |
| 5. | Sassoon C. S. H., Te T. T., Mahutte C. K., Light R. W.Airway occlusion pressure: an important indicator for successful weaning in patients with chronic obstructive pulmonary disease. Am. Rev. Respir. Dis.1351987107113 |
| 6. | Krieger B. P., Ershowsky P. F., Becker D. A., Gazeroglu H. B.Evolution of conventional criteria for predicting successful weaning from mechanical ventilatory support in elderly patients. Crit. Care Med.171989858861 |
| 7. | Yang K. L., Tobin M. J.A prospective study of indexes predicting the outcome of trials of weaning from mechanical ventilation. N. Engl. J. Med.324199114451450 |
| 8. | DeHaven C. B., Hurst J. M., Branson R. D.Evaluation of two different extubation criteria: attributes contributing to success. Crit. Care Med.1419869294 |
| 9. | Brochard L., Rauss A., Benito S., Conti G., Mancebo J., Rekik N., Gasparetto A., Lemaire F.Comparison of three methods of gradual withdrawal from ventilatory support during weaning from mechanical ventilation. Am. J. Respir. Crit. Care Med.1501994896903 |
| 10. | Esteban A., Frutos F., Tobin M. J., Alı́a I., Solsona J. F., Vallverdú I., Fernández R., de la Cal M. A., Benito S., Tomás R., Carriedo D., Macı́as S., Blanco J.for the Spanish Lung Failure Collaborative GroupA comparison of four methods of weaning patients from mechanical ventilation. N. Engl. J. Med.3321995345350 |
| 11. | Jabour E. R., Rabil D. M., Truwit J. D., Rochester D. F.Evaluation of a new weaning index based on ventilatory endurance and the efficiency of gas exchange. Am. Rev. Respir. Dis.1441991531537 |
| 12. | Fernández R., Cabrera J., Calaf N., Benito S.P0.1/Pimax: an index for assessing respiratory capacity in acute respiratory failure. Intensive Care Med.161990175179 |
| 13. | Sassoon C. S. H., Mahutte C. K.Airway occlusion pressure and breathing pattern as predictors of weaning outcome. Am. Rev. Respir. Dis.1481993860866 |
| 14. | Epstein S. K.Etiology of extubation failure and the predictive value of the rapid shallow breathing index. Am. J. Respir. Crit. Care Med.1521995545549 |
| 15. | Ely E. W., Baker A. M., Dunagan D. P., Burke H. L., Smith A. C., Kelly P. T., Johnson M. M., Browder R. W., Bowton D. L., Haponik E. F.Effect on the duration of mechanical ventilation of identifying patients capable of breathing spontaneously. N. Engl. J. Med.335199618641869 |
| 16. | Marini J. J., Smith T. C., Lamb V.Estimation of inspiratory muscle strength in mechanically ventilated patients: the measurement of maximal inspiratory pressure. J. Crit. Care119863238 |
| 17. | Marini J. J., Rodriguez R. M., Lamb V. J.Involuntary breath-stacking: an alternative method for vital capacity estimation in poorly cooperative subjects. Am. Rev. Respir. Dis.1341986694698 |
| 18. | Conti G., de Blasi R., Pelaia P., Benito S., Rocco M., Antonelli M., Bufi M., Mattia C., Gasparetto A.Early prediction of successful weaning during pressure support ventilation in chronic obstructive pulmonary disease patients. Crit. Care Med.201992366371 |
| 19. | Esteban A., Alı́a I., Gordo F., Fernández R., Solsona J. F., Vallverdú I., Macı́as S., Allegue J. M., Blanco J., Carriedo D., León M., de la Cal M. A., Taboada F., Gonzalez de Velasco J., Palazón E., Carrizosa F., Tomás R., Suárez J., Goldwasser R. S.for the Spanish Lung Failure Collaborative GroupExtubation outcome after spontaneous breathing trials with T-tube or pressure support ventilation. Am. J. Respir. Crit. Care Med.1561997459465 |
| 20. | Gandı́a F., Blanco J.Evaluation of indexes predicting the outcome of ventilator weaning and value of adding supplemental inspiratory load. Intensive Care Med.181992327333 |
| 21. | The Spanish Lung Failure Collaborative GroupMulticenter, prospective comparison of 30 and 120 minute trials of weaning from mechanical ventilation (abstract). Am. J. Respir. Crit. Care Med.155199720A |
| 22. | De Haven C. B., Kirton O. C., Morgan J. P., Hart A. M. L., Shatz D. V., Civetta J. M.Breathing measurements reduces false negative classification of tachypneic preextubation trial failures. Crit. Care Med.241996976980 |
| 23. | Betbesé A. J., Pérez M., Bak E., Rialp G., Mancebo J.A prospective study of endotracheal unplanned extubation in intensive care unit patients. Crit. Care Med.26199811801186 |
| 24. | Epstein S. K., Ciubotaru R. L., Wong J. B.Effect of failed extubation on the outcome of mechanical ventilation. Chest1121997186192 |
| 25. | Torres A., Gatell J. M., Aznar E.Re-intubation increases the risk of nosocomial pneumonia in patients needing mechanical ventilation. Am. J. Respir. Crit. Care Med.1521995137141 |
| 26. | Epstein S. K., Ciubotaru R.Independent effects of etiology of failure and time to reintubation on outcome for patients failing extubation (abstract). Am. J. Respir. Crit. Care Med.1581998489493 |
| 27. | Esteban A., Alı́a I., Ibáñez J., Benito S., Tobin M. J.the Spanish Lung Failure Collaborative GroupModes of mechanical ventilation and weaning: a national survey of Spanish hospitals. Chest106199411881193 |
| 28. | Lee K. H., Hui K. P., Chan T. B., Tan W. C., Lim T. K.Rapid shallow breathing (frequency-tidal volume ratio) did not predict extubation outcome. Chest1051994540543 |
| 29. | Mador J. M., Tobin M. J.Effect on alterations in mental activity on the breathing pattern in healthy subjects. Am. Rev. Respir. Dis.1441991481487 |
| 30. | Epstein S. K., Ciubotaru R. L.Influence of gender and endotracheal tube size on preextubation breathing pattern. Am. J. Respir. Crit. Care Med.154199616471652 |
| 31. | Krieger B. P., Isber J., Breitenbucher A., Throop G., Ershowsky P.Serial measurements of the rapid-shallow-breathing index as a predictor of weaning outcome in elderly medical patients. Chest112199710291034 |
| 32. | Fernández R., Blanch L. L., Artigas A.Respiratory center activity during mechanical ventilation. J. Crit. Care61991102111 |
| 33. | Herrera M., Blasco J., Venegas J., Barba R., Doblas A., Márquez F.Mouth occlusion pressure (P0.1) in acute respiratory failure. Intensive Care Med.111985134139 |
| 34. | Truwit J. D., Marini J. J.Evaluation of thoracic mechanics in the ventilated patient: primary measurements. J. Crit. Care31988133150 |
| 35. | Ninane V., Yernault J. C., De Troyer A.Intrinsic PEEP in patients with chronic obstructive pulmonary disease: role of expiratory muscles. Am. Rev. Respir. Dis.48199310371042 |
| 36. | Stroetz R. W., Hubmayr R. D.Tidal volume maintenance during weaning with pressure support. Am. J. Respir. Crit. Care Med.152199510341040 |
Acknowlegment :
