Abstract
We have previously shown (Am. J. Respir. Crit. Care Med. 1995;152:1248–1255) that in patients needing mechanical ventilation, the load imposed on the inspiratory muscles is excessive relative to their neuromuscular capacity. We have therefore hypothesized that weaning failure may occur because at the time of the trial of spontaneous breathing there is insufficient reduction of the inspiratory load. We therefore prospectively studied patients who initially had failed to wean from mechanical ventilation (F) but had successful weaning (S) on a later occasion. Compared with S, during F patients had greater intrinsic positive end-expiratory pressure (6.10 ± 2.45 versus 3.83 ± 2.69 cm H2O), dynamic hyperinflation (327 ± 180 versus 213 ± 175 ml), total resistance (Rmax, 14.14 ± 4.95 versus 11.19 ± 4.01 cm H2O/L/s), ratio of mean to maximum inspiratory pressure (0.46 ± 0.1 versus 0.31 ± 0.08), tension time index (TTI, 0.162 ± 0.032 versus 0.102 ± 0.023) and power (315 ± 153 versus 215 ± 75 cm H2O × L/min), less maximum inspiratory pressure (42.3 ± 12.7 versus 53.8 ± 15.1 cm H2O), and a breathing pattern that was more rapid and shallow (ratio of frequency to tidal volume, f/Vt 98 ± 38 versus 62 ± 21 breaths/min/L). To clarify on pathophysiologic grounds what determines inability to wean from mechanical ventilation, we performed multiple logistic regression analysis with the weaning outcome as the dependent variable. The TTI and the f/Vt ratio were the only significant variables in the model. We conclude that the TTI and the f/Vt are the major pathophysiologic determinants underlying the transition from weaning failure to weaning success.
For the majority of mechanically ventilated patients weaning can be accomplished quickly and easily. There is, however, a significant percentage of patients in whom weaning fails. These patients present a great challenge for clinicians since the pathophysiology underlying weaning failure is complex, multifactorial, and to a large extent, not well established (1). Part of the problem probably results from the fact that even excellent physicians often do not accurately judge when a patient is ready to wean (2).
We have previously shown (3) that in patients needing mechanical ventilation, the load imposed on the inspiratory muscles is excessive relative to their neuromuscular capacity. We have therefore hypothesized that weaning failure may occur because at the time of the trial of spontaneous breathing, physicians cannot recognize that there is insufficient reduction of the inspiratory load, since its measurement is not readily available at the bedside.
The present study was done to test this hypothesis in order to better understand the pathophysiologic determinants of weaning failure. To achieve this goal, we prospectively studied patients who initially had failed to wean from mechanical ventilation but had successful weaning on a later occasion. Each patient during weaning success was used as his or her own control during weaning failure. In contrast to studies in which the patients who failed to wean were different from those who succeeded (4, 5), we have chosen this approach in order to eliminate any differences arising from a different patient population. Since our aim was to clarify on pathophysiologic grounds what determines inability to wean from mechanical ventilation, we studied our patients at the end of a failing weaning trial, just before reinstitution of mechanical ventilation. This was done because it has been previously shown that the passive mechanical properties of the respiratory system measured before the weaning trial do not differ between patients who succeed from those who fail (4) and also that there is a progressive deterioration in respiratory mechanics during the spontaneous breathing trial in the patients who fail to wean (5). Consequently, by studying patients at the end of a failing weaning trial, we should be able to find differences that would clarify the pathophysiologic determinants of weaning failure.
Thirty consecutive patients admitted to the intensive care unit for management of acute respiratory failure of different etiology were studied (Table 1). The investigative protocol was approved by the Institutional Ethics Committee. In all cases informed consent was obtained from the patient, the next of kin, or the primary physician. All patients were intubated through a cuffed endotracheal (Portex, 7–10 cm internal diameter [i.d.], cut to a length of 26 cm) or tracheostomy tube (n = 7). The patients had been mechanically ventilated for a period of 8 to 35 d prior to the present investigation and were clinically stable for the preceding 12 h. The patients were divided into three groups: (1) Acute exacerbation of chronic obstructive pulmonary disease (COPD), (group 1, n = 10); diagnosis was based on previous history and routine lung function tests; (2) adult respiratory distress syndrome (ARDS), (group 2, n = 10); all patients met the conventional criteria for the diagnosis of ARDS; and (3) other diseases (five chest trauma, five pneumonia) (group 3, n = 10) (Table 1).
| Group | n | Age (yr) | Duration of MV at Time of Study (d ) | Duration of Discontinuation during Weaning Failure (min) | Days between Weaning Failure and Success | Sex (M/F ) | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| COPD | 10 | 63 ± 7 | 18 ± 12 | 56 ± 26 | 9 ± 4 | 9/1 | ||||||
| ARDS | 10 | 40 ± 16 | 14 ± 8 | 49 ± 19 | 8 ± 4 | 6/4 | ||||||
| Other | 10 | 51 ± 17 | 20 ± 10 | 55 ± 25 | 9 ± 3 | 3/7 |
Flow (V˙) was measured with a heated pneumotachograph (Fleisch No. 2; Lausanne, Switzerland) inserted between the endotracheal tube (ET) and the Y-piece of the ventilator. The equipment dead space (not including the ET) was 75 ml. The pressure drop across the pneumotachograph was measured with a differential pressure transducer (Validyne MP-45, ± 2 cm H2O; Northridge, CA). Changes in lung volume (V) were obtained by numerical integration of the flow signal. Airway pressure (Paw) was recorded from a side port of the ET adapter connected by rigid tubing to a differential pressure transducer (Validyne MP-45, ± 100 cm H2O). All variables were recorded on an 8-channel electrostatic recorder (Gould ES 1000; Gould Instruments, OH). The signals were continuously digitized at 200 Hz, using an analog-to-digital converter connected to a computer (Wyse 486) and stored for later analysis. To reduce the effects of the compliance and resistance of the system connecting the subjects to the ventilator on the measurements of respiratory mechanics (6), a standard low-compliance adult tube was used (2-cm i.d., 60-cm long) and the humidifier was omitted from the inspiratory line. Care was taken to avoid gas leaks around the equipment and the tracheal cuff. The electrocardiogram, heart rate, blood pressure, and arterial oxygen saturation (SaO2 ) were continuously monitored.
Patients were studied at two different stages: (1) within a few hours (mean 4.7, range 2–21) after they had failed an initial T-piece weaning trial, when they where still ventilator-dependent (weaning failure, [F]); and (2) when they were able to breathe spontaneously through the T- piece for at least 6 h and were considered to have been succesfully weaned (weaning success [S]). Fifty-two patients were initially recruited for the study; only 30 patients completed the study and were included in the final analysis. Ten patients died and 12 were extubated before we could measure them (five accidentally and seven during the weekend). All decisions regarding the patient (i.e., when to initiate weaning trials, when a trial fails or succeeds, and when the patient is ready for extubation) were made by the primary physician. The criteria used to define weaning failure were those routinely used in our institution: tachypnea (f > 35 breaths/min), increased accessory muscle activity, diaphoresis, tachycardia, arrhythmias, and cyanosis. Patients meeting these criteria were returned to mechanical ventilation.
Weaning failure. Mechanical ventilation was delivered by a Siemens 300 Servo Ventilator (Siemens-Elema, Solna, Sweden) in the assist-control mode with the ventilator settings prescribed by the primary physicians. Tidal volume (Vt) was set on the ventilator at 0.48– 0.75 L and respiratory frequency (f) at 16–28 breaths/min. The time of mechanical inflation (Ti) was 0.47–0.99 s, while the duration of mechanical expiration (Te) ranged between 1.25 and 2.53 s. Mean inspiratory flow (Vt/Ti) was 0.63–1.39 L s−1. Positive end-expiratory pressure (PEEP) was 0–4 cm H2O. Fractional concentrations of inspired O2 (Fi O2 ) ranged between 0.35 and 0.60. Baseline partial pressures of arterial O2, CO2, and pH were 118 ± 2, 42 ± 8 mm Hg, and 7.42 ± 0.08, respectively (ABL 300). Patients were studied in a semi-recumbent position, and a physician not involved in the study was always present to provide for patient care, if needed.
After inserting the pneumotachograph into the circuit, the patient was allowed to breathe spontaneously through the T-piece. Fi O2 was set at the same level as during mechanical ventilation. All sedative and paralyzing medication had been stopped at least 24 h before the beginning of the study. After 18–110 min from the beginning of the T-piece trial, all patients became unable to sustain spontaneous breathing and reinstitution of mechanical ventilation was required. Twenty-four patients developed tachypnea, 13 exhibited increased accessory muscle activity, 14 diaphoresis, and 17 tachycardia. At this point the arterial Po 2, Pco 2, and pH averaged 91 ± 16, 49 ± 9, and 7.37 ± 0.04, respectively. No patient had a Po 2 less than 60 mm Hg or an SaO2 less than 90%. In order to obtain the basic pattern of spontaneous breathing, we analyzed the recordings of volume and flow in terms of Vt, f, and duty cycle (Ti/Ttot). The reported data are the means of 10–15 consecutive breaths. This analysis of the spontaneous breathing pattern was made 2 min before the reinstitution of mechanical ventilation. Just before the discontinuation of the spontaneous breathing trial, the T-piece was removed and a device consisting of a rigid T-tube with a one-way valve set on the expiratory line was connected to the endotracheal/tracheostomy tube for the measurement of maximum inspiratory pressure (MIP). MIP was measured at the opening of the ET using the 20-s breath-holding technique at FRC (7). Briefly, the patient was allowed to exhale after a normal inspiration through the one-way valve. During this expiration the inspiratory port of the device was manually occluded. During the ensuing inspiratory effort against the occluded airway, the expiratory port was also manually occluded, thus allowing no movement of air in either an inspiratory or an expiratory direction. The airway occlusion was maintained for at least 20 s and the most negative pressure value obtained was considered as MIP. As soon as the mechanical ventilation on assist-control mode had been reinstituted, the patients were sedated (propofol, 2.5 mg/kg intravenously) and hyperventilated in order to abolish respiratory muscle activity (judged by Paw and V˙ tracing contour as representative of passive ventilation and by clinical assessment). With the patient ventilated with control mechanical ventilation and constant inspiratory flow, using the appropriate buttons of the ventilator we tried to simulate the spontaneous breathing pattern obtained prior to the termination of the weaning trial (see above). Inclusion criteria for accepting the simulated breaths as representative of the spontaneous breathing pattern during the T-piece trial were: (1) ± 0.02 L for Vt; (2) ± 0.1 breaths/min for f; and (3) ± 0.02 for Ti/Ttot. The Vt, f, and Ti/Ttot at the end of the T-piece trial amounted to 0.40 ± 0.1 L, 37 ± 8 breaths/min, and 0.37 ± 0.07, respectively (Table 2). The breaths simulated by the ventilator had Vt = 0.40 ± 0.1 L, f = 37 ± 8 breaths/ min, and Ti/Ttot = 0.37 ± 0.07.
| Phase | COPD (n = 10) | ARDS (n = 10) | Others (n = 10) | Combined (n = 30) | p Value | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Vt, ml | F | 342 ± 100 | 465 ± 90† | 407 ± 90 | 401 ± 100 | 0.019 | ||||||
| S | 479 ± 50* | 515 ± 90 | 384 ± 60‡ | 461 ± 90 | ||||||||
| f, b/min | F | 29.4 ± 3.6 | 40.9 ± 9.3† | 39.5 ± 6.3† | 36.6 ± 8.4 | < 0.0001 | ||||||
| S | 24.0 ± 3.3 | 26.9 ± 3.9* | 29.6 ± 5.4* | 26.8 ± 4.8 | ||||||||
| Ti/Ttot | F | 0.31 ± 0.05 | 0.43 ± 0.05† | 0.38 ± 0.06†,‡ | 0.37 ± 0.07 | 0.0042 | ||||||
| S | 0.31 ± 0.02 | 0.36 ± 0.06*,† | 0.33 ± 0.05* | 0.33 ± 0.05 | ||||||||
| f/Vt, b/min/L | F | 97 ± 44 | 92 ± 34 | 103 ± 37 | 98 ± 38 | < 0.0001 | ||||||
| S | 51 ± 11* | 54 ± 11 | 80 ± 24 | 62 ± 21 | ||||||||
| Vt/Ti, L/s | F | 0.54 ± 0.06 | 0.79 ± 0.19† | 0.85 ± 0.24† | 0.73 ± 0.22 | NS | ||||||
| S | 0.65 ± 0.12 | 0.78 ± 0.24 | 0.69 ± 0.20 | 0.71 ± 0.19 |
Respiratory mechanics were assessed during the simulation of spontaneous breathing, with the patient relaxed, by the constant flow occlusion method previously described in detail (6, 8-12). After end-inspiratory occlusion (by the appropriate button of the ventilator), there was an immediate drop in Paw from a maximal (Ppeak) to a lower value (P1), followed by a gradual decrease to a plateau (P2) (9). Five regular breaths afterwards, an end-expiratory occlusion was performed for direct measurement of intrinsic PEEP (PEEPi) (12). The static compliance of the total respiratory system (Cst,rs) was computed as: Vt/(P2 − PEEPi) (12). The minimum and maximum resistances of the respiratory system (Rmin and Rmax ) were calculated by dividing Ppeak − P1 and Ppeak − P2 by the preceding constant V˙. Both Ppeak − P1 and Ppeak − P2 were corrected for the resistance pressure drop caused by the ET. We used the pressure–flow relationship of the ETs reported by Rossi and colleagues (9). After a few breaths had been delivered, the patient was disconnected from the ventilator after a normal inflation and was allowed to expire freely until expiratory flow became nil and the elastic equilibrium volume of the respiratory system had been reached. The difference between the inspired and the expired volume represents the increase in FRC due to PEEPi (ΔFRC). Its value was calculated as the mean of two or three measurements.
The inspiratory work done on the respiratory system (Wi,rs) was calculated as the product of the mean inspiratory airway pressure (Pi) obtained during the simulation of spontaneous breathing by the ventilator and the corresponding Vt. The Pi was calculated as the integral of the Paw curve over time from the onset until the end of inspiratory flow divided by the corresponding Ti. The values of Wi,rs reported are the mean of three breaths. Part of this work was due to the presence of PEEPi (Wpeep i). This work was calculated as Wpeep i = PEEPi × Vt. The rest of the elastic work (Wel) was calculated using a geometric approximation of the static pressure (Pst) − V curve of the respiratory system; that is, although it is known that the Pst is related to the volume of the respiratory system with a relationship of the type: Pst = a × Vb, b is not very different from 1, amounting to 0.947 ± 0.013 in normal subjects (13). Accordingly, we have assumed that this relationship is nearly linear, and so the elastic work of breathing that is represented by the area between the Pst versus volume curve and the corresponding Vt can be approximated by a triangle, i.e., Wel = 1/2 × (Pst − PEEPi ) × Vt. The resistive work of breathing (Wres) was calculated as the difference between Wi,rs and the sum of Wpeep i and Wel. Power was calculated as Wi,rs × f.
The ratio Pi/Pi max, where Pi max is the sum of MIP and PEEPi, was used as an index of the mechanical load imposed on the inspiratory muscles (numerator) to the inspiratory force reserve (denominator). We assessed the ventilatory muscles' energetic demands in terms of the tension–time index (TTI), which was calculated as TTI = Pi/Pi max × Ti/Ttot.
Weaning success. During the successful weaning trials, patients were breathing spontaneously through the T-piece and had been considered by their primary physicians as being ready to extubate. PaO2 was 129 ± 14 mm Hg, PaCO2 was 43.8 ± 8 mm Hg, and pH was 7.41 ± 0.03. The pneumotachograph was inserted into the circuit and the protocol used during F was repeated. The patients were followed for 1 h. All patients were stable during this time period without any indication, objective or subjective, of inability to sustain spontaneous breathing. The pattern of breathing that the patients had at the end of this period was recorded and analyzed as already described. Mechanical ventilation was then resumed, and the patients were sedated (propofol, 2.5 mg/kg intravenously) and subjected to hyperventilation to obtain respiratory muscle relaxation. The pattern of spontaneous breathing was simulated during control mechanical ventilation and the measurements done during F were repeated. All patients were extubated within the next 24 h after completion of the study. No patient required reintubation.
Values are means ± SD. Comparisons between F and S were performed using paired t test. A p value < 0.05 was accepted as statistically significant. Due to the large number of variables compared, a Bonferroni-type adjustment was carried out using Holmes' procedure (14). The p values reported pertain to the variables that remained statistically significant after the Bonferroni-type adjustment. Comparisons between the three groups of patients between F and S were made by repeated measurements two-way ANOVA. A p value < 0.05 was again accepted as statistically significant. A logistic regression model was used with weaning outcome as the dependent dichotomous variable (F was taken as 0 and S as 1). Variables with p < 0.05 were entered into the multivariate analysis based on models that were judged a priori to be clinically sound. This was prospectively determined to be necessary to avoid producing significant results with multiple comparisons (15). To minimize the effect of collinearity, the most clinically relevant variable among related variables (e.g., Rmax instead of Rmin) was entered into the model. The independent variables initially entered into the model were MIP, ΔFRC, TTI, Rmax, f/Vt, power, Wpeep i. A stepwise approach (backward) was used for entering a variable into the model, with 0.05 as the limit for their acceptance or removal. A software package (Statistica for Windows, 6; Stat Soft Inc., Tulsa, OK) was used for analysis.
The ventilatory pattern values during F and S are reported in Table 2. Compared to F, during S patients had lower f, higher Vt and lower Ti/Ttot. The ratio f/Vt was much lower with S, whereas the mean inpiratory flow did not change. During F, patients with COPD had lower f, Vt,Ti/Ttot and Vt/Ti than the other patients; however, the f/Vt ratio was not different. During S these differences disappeared.
The values of respiratory mechanics are reported in Table 3. Both Rmax and Rmin were significantly higher during F, whereas the Cst,rs was the same. During F, patients with COPD had the highest Cst,rs and patients with ARDS had the lowest. During S the Cst,rs of the COPD was essentially the same, while it improved in the patients with ARDS yet remaining significantly lower than in COPD (Table 3). PEEPi was higher during F, as was the ΔFRC. Not suprisingly, patients with COPD had the greatest amount of PEEPi and ΔFRC during both phases, and patients with ARDS had the lowest.
| Variable | Phase | COPD (n = 10) | ARDS (n = 10) | Other (n = 10) | Combined (n = 30) | p Value | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Rmax, cm H2O/L/s | F | 17.73 ± 5.86 | 13.01 ± 3.60 | 11.67 ± 3.02† | 14.14 ± 4.95 | 0.0038 | ||||||
| S | 12.40 ± 1.85 | 10.00 ± 4.04 | 11.18 ± 5.34 | 11.19 ± 4.01 | ||||||||
| Rmin, cm H2O/L/s | F | 11.03 ± 4.54 | 8.85 ± 1.83 | 8.85 ± 3.10 | 9.57 ± 3.39 | 0.026 | ||||||
| S | 8.62 ± 1.86 | 7.50 ± 3.56 | 7.70 ± 4.37 | 7.94 ± 3.33 | ||||||||
| Cst,rs, L/cm H2O | F | 67.02 ± 15.34 | 40.01 ± 7.01† | 51.87 ± 11.38†,‡ | 52.97 ± 15.96 | NS | ||||||
| S | 67.23 ± 11.37 | 48.43 ± 4.17*,† | 44.08 ± 7.66*,† | 53.25 ± 12.97 | ||||||||
| PEEPi, cm H2O | F | 8.17 ± 0.88 | 4.53 ± 2.20† | 5.62 ± 2.37† | 6.10 ± 2.45 | < 0.0001 | ||||||
| S | 6.71 ± 2.41 | 2.20 ± 1.01*,† | 2.58 ± 1.60*,† | 3.83 ± 2.69 | ||||||||
| ΔFRC, ml | F | 523 ± 97 | 164 ± 69† | 295 ± 129† | 327 ± 180 | < 0.0001 | ||||||
| S | 421 ± 138 | 104 ± 46† | 113 ± 74*,† | 213 ± 175 |
Both MIP and Pi max were significantly lower during F, compared with S (Table 4). Patients with ARDS had the highest values of MIP and Pi max during both F and S. The ratio Pi/Pi max was very high during F, averaging 0.49 ± 0.09 in patients with COPD, 0.43 ± 0.08 in patients with ARDS, and 0.44 ± 0.11 in the other patients.
| Variable | Phase | COPD (n = 10) | ARDS (n = 10) | Other (n = 10) | Combined (n = 30) | p Value | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| MIP, cm H2O | F | 33.8 ± 10.5 | 52.9 ± 13.8† | 40.0 ± 9.1‡ | 42.3 ±12.7 | < 0.0001 | ||||||
| S | 46.8 ± 4.7 | 67.3 ± 15.4*,† | 46.5 ± 15.1‡ | 53.8 ± 15.1 | ||||||||
| Pi max, cm H2O | F | 42.0 ± 9.4 | 57.4 ± 15.8† | 45.6 ± 9.2‡ | 48.4 ± 13.3 | 0.0001 | ||||||
| S | 53.5 ± 6.3* | 69.5 ± 15.7*,† | 49.1 ± 11.8‡ | 57.6 ± 14.6 | ||||||||
| Pi/Pi max | F | 0.49 ± 0.09 | 0.43 ± 0.08 | 0.44 ± 0.11 | 0.46 ± 0.1 | < 0.0001 | ||||||
| S | 0.35 ± 0.05* | 0.26 ± 0.09* | 0.31 ± 0.09* | 0.31 ± 0.08 | ||||||||
| TTI | F | 0.149 ± 0.024 | 0.178 ± 0.027 | 0.164 ± 0.035 | 0.162 ± 0.032 | < 0.0001 | ||||||
| S | 0.108 ± 0.009 | 0.094 ± 0.031* | 0.103 ± 0.023* | 0.102 ± 0.023 |
In order to examine the effects of dynamic hyperinflation on the endurance indices of the inspiratory muscles, we plotted the ratio Pi/Pi max against the ΔFRC (Figure 1). ΔFRC was expressed as percentage of predicted (16) inspiratory capacity (IC). A hypothetical critical line was constructed from data derived from normal subjects (17), i.e., at normal FRC the critical pressure above which fatigue may occur is about 50% of Pi max, while at FRC + 1/2 IC this critical pressure is 25–30% of the Pi max. Thus, this critical line represents the critical inspiratory pressures above which fatigue may occur. During F, in all three groups of patients the mean values are near or above the critical line of Pi/Pi max (Figure 1). In contrast, during S the mean Pi/Pi max values were well below the fatiguing threshold (Figure 1, Table 4).
Fig. 1. Pressure–volume diagram similar to that of Roussos and colleagues (17) plotting the ratio Pi/Pi max against dynamic increase in FRC (ΔFRC) expressed as percentage of predicted (16) inspiratory capacity (IC). The ratio Pi/Pi max is the mean inspiratory pressure per breath expressed as a fraction of maximum. Each closed symbol refers to the mean value of each group during weaning failure. Each open symbol refers to the mean value of each group during weaning success. The solid line was constructed from data in normal subjects and represents the critical inspiratory pressures above which fatigue may occur. Namely at normal FRC, the critical inspiratory pressure per breath above which fatigue may occur in normal subjects is about 50% of maximum inspiratory pressure, while at FRC + 1/2 IC this critical pressure is 25–30% of the maximum. COPD = exacerbated chronic obstructive pulmonary disease; ARDS = adult respiratory distress syndrome; F = weaning failure; S = weaning success.
[More] [Minimize]Mean value of TTI was 0.162 ± 0.032 during F and fell to 0.102 ± 0.023 during S (p = 0.00001). In all three groups of patients, during F the mean values of TTI were equal or exceeded the so-called critical value of 0.15 (above which diaphragmatic fatigue may occur in normal subjects [18]) (Figure 2, Table 4). Most importantly, all individual values (range 0.120–0.246) were clustered around this critical value, being far away from the average TTI of the 10 normal subjects (0.01 ± 0.00) (3), breathing with a minute volume comparable to that of our patients (14.4 ± 1.3 versus 14.6 ± 1.2 L/min) (Figures 2 and 3). On the contrary, during S all three groups exhibited significantly lower mean values (Figures 2 and 3).
Fig. 2. Diagram similar to the iso-diaphragmatic tension–time index (iso-TTdi) plot of Bellemare and Grassino (18) constructed from the data of our patients. Ordinate: inspiratory to total cycle duration ratio (Ti/Ttot) at the end of spontaneous breathing trial. Abscissa: mean inspiratory pressure, expressed as a fraction of maximum (Pi/Pi max). The product of each combination of the two variables is the tension–time index (TTI) of global inspiratory muscles. Three iso-TTI isopleths are drawn for reference. Each closed symbol refers to the mean value of each group during weaning failure. Each open symbol refers to the mean value of each group during weaning success. Open diamond represents the average TTI of 10 normal subjects breathing with a minute volume similar to the mean minute volume of our patients. COPD = exacerbated chronic obstructive pulmonary disease; ARDS = adult respiratory distress syndrome; F = weaning failure; S = weaning success.
[More] [Minimize]
Fig. 3. The tension–time index during weaning failure and success. Each closed symbol represents one patient. COPD = exacerbated chronic obstructive pulmonary disease; ARDS = adult respiratory distress syndrome; F = weaning failure; S = weaning success.
[More] [Minimize]No significant changes were observed between F and S in either Wi,rs, Wel, or Wres. On the contrary, both Wpeep i and power decreased during S compared with F (Table 5).
| Variable | Phase | COPD (n = 10) | ARDS (n = 10) | Other (n = 10) | Combined (n = 30) | p Value | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Wi,rs, cm H2O × L | F | 6.70 ± 2.33 | 10.72 ± 3.06 | 7.82 ± 2.09 | 8.42 ± 2.99 | NS | ||||||
| S | 9.23 ± 1.88 | 9.42 ± 3.45 | 5.55 ± 0.97 | 8.07 ± 2.89 | ||||||||
| Wel, cm H2O × L | F | 1.34 ± 0.93 | 2.80 ± 0.96† | 1.72 ± 0.68 | 1.95 ± 1.05 | NS | ||||||
| S | 1.92 ± 0.53 | 2.88 ± 0.96 | 1.69 ± 0.56‡ | 2.17 ± 0.87 | ||||||||
| Wres, cm H2O × L | F | 2.52 ± 0.50 | 5.86 ± 1.57† | 3.91 ± 1.39 | 4.11 ± 1.85 | NS | ||||||
| S | 3.96 ± 0.78 | 5.31 ± 2.24 | 2.93 ± 0.68 | 4.07 ± 1.70 | ||||||||
| Wpeep i, cm H2O × L | F | 2.86 ± 1.12 | 2.02 ± 0.99 | 2.19 ± 0.85 | 2.36 ± 1.03 | 0.016 | ||||||
| S | 3.19 ± 1.18 | 1.23 ± 0.74*,† | 0.93 ± 0.47*,† | 1.79 ± 1.31 | ||||||||
| Power, cm H2O × L/min | F | 184 ± 41 | 447 ± 159† | 313 ± 104 | 315 ± 153 | 0.0026 | ||||||
| S | 220 ± 46 | 262 ± 101* | 164 ± 38 | 215 ± 75 |
Only TTI and f/Vt ratio were found to be significant explanatory variables. The intercept of the model was 20.2 (p = 0.03) and the coefficients of the dependent variables were −150.9 for the TTI (p = 0.008) and −0.035 for the f/Vt ratio (p = 0.04). To access the relative contribution of TTI and f/Vt in determining the weaning outcome, the values of both variables were categorized in 20% percentiles of their respective range. Categories were given the values of 1 to 5 from the lowest to the highest 20% percentile of the respective range, and the logistic regression analysis was repeated with the categorized TTI and f/Vt variables included in the model instead of the original ones. The categorized TTI and f/Vt remained the only significant explanatory variables in the new model. The results of the analysis are presented in Table 6.
| Variable | Coefficient | p Value | Adjusted Odds Ratio | Confidence Intervals | ||||
|---|---|---|---|---|---|---|---|---|
| TTI | 2.33 | 0.005 | 10.28 | 2.1–50.4 | ||||
| f/Vt | 1.16 | 0.035 | 3.19 | 1.86–5.47 |
The main findings of the present study are that: (1) In F the inspiratory load was excessive relative to the neuromuscular capacity; (2) the TTI and the f/Vt ratio are the only explanatory variables that account for weaning failure or success; and (3) a decrease in the load imposed on the respiratory muscles with a concomitant increase in neuromuscular capacity that results in diminished energy demands (↓ TTI) and a more efficient breathing pattern (↓ f/Vt) are the main pathophysiologic changes between F and S.
In our study we simulated the pattern of spontaneous breathing with the respirator under passive conditions (3). This technique has two advantages: (1) it estimates most of the mechanical work necessary to ventilate the respiratory system, contrary to the alternative esophageal balloon technique that, in spontaneously breathing patients, is not able to estimate the elastic and negative work done on the chest wall (19); (2) it allows for the more accurate measurement of PEEPi, since PEEPi measured under static conditions usually has a higher value than dynamic PEEPi (20, 21). Furthermore, overestimation of PEEPi due to expiratory muscle contraction is avoided (22). Expiratory muscle activity could nevertheless lead to a decrease in end-expiratory lung volume in those patients who are not flow-limited. This is a potential limitation of the method we used that could lead to an overestimation of the degree of dynamic hyperinflation in some of our patients. However, part of this problem is minimized by the use of each patient as its own control.
The f/Vt ratio, a measure of rapid, shallow breathing introduced by Yang and Tobin (23), has been validated (24, 25) as a useful index for predicting the weaning outcome. In this context it is calculated at the beginning of a spontaneous breathing trial. Very few studies have examined the time course of this ratio during a failing weaning trial (5, 25), and even fewer have calculated it just prior to the resumption of mechanical ventilation (5). In our study, the f/Vt ratio measured at the end of the T-piece trial was one of the two explanatory variables that accounted for the weaning outcome.
It has been suggested that the adoption of such a rapid, shallow breathing pattern represents the response of the central respiratory controller to loaded breathing (26). Accordingly, Yang and Tobin (23) have reported that patients who fail to wean exhibit rapid shallow breathing upon discontinuation of mechanical ventilation. However, it has been shown (25) that the f/Vt ratio measured during the first minute of unsupported breathing correlates weakly with the f/Vt ratio obtained after 30 or 60 min, tending to decrease in those patients who wean successfully and to increase in those who fail to wean. Furthermore, the adoption of rapid, shallow breathing has been documented to occur after inspiratory muscle fatigue (27), as was likely to occur in our patients during F. When our patients failed to wean, the load on their inspiratory muscles was greater than that during weaning success, as evidenced by the higher values of Pi/Pi max, Rmax, Rmin, and PEEPi. This may have accounted for some of the difference in the f/Vt ratio during F and S. However, differences in load cannot entirely explain the breathing strategies adopted during F and S. If the f/Vt were to represent only a response to loading, it should be strongly correlated with the indices of activity of the respiratory muscles (such as the TTI). Yet, during weaning failure the f/Vt ratio and the TTI were not correlated (r = 0.16, NS). This is further evidenced by the fact that f/Vt and TTI were the two explanatory variables that proved significant in our logistic regression model, implying that both independently determine the weaning outcome. Another possibility is that the rapid, shallow breathing pattern is partly a response of the respiratory center to the anxiety generated during the weaning trial. This is based both on theoretical considerations (1) and on indirect evidence provided by studies where biofeedback and/or hypnosis improved the breathing pattern during weaning from mechanical ventilation (28, 29). This improvement, however, could not be attributed to changes in the load or the mechanical performance of the respiratory muscles, and hence was due to a reduction in the anxiety experienced by the patients.
Whatever the reason(s) may be, the f/Vt ratio is a major determinant of the weaning outcome, being very different in the same patients during weaning failure and success. The fact that f/Vt is a good predictor of weaning outcome (23-25) and plays a major pathophysiologic role in weaning failure strongly supports its clinical use.
Along with f/Vt, TTI was the other explanatory variable that accounted for weaning outcome in our patients. The TTI concept, originating in earlier studies of heart and limb muscles, was applied to the diaphragm by Bellemare and Grassino (18) to determine its endurance characteristics. They found that when TTI was greater than 0.15, the breathing task could not be sustained for long, resulting in diaphragmatic fatigue. Milic-Emili (30) was the first to suggest that the weaning outcome could be explained on the basis provided by the TTI: a TTI greater than 0.15 would indicate impending or ongoing inspiratory muscle fatigue. Indeed, during F the TTI amounted to 0.162 ± 0.032, while during S it was 0.104 ± 0.028. All three groups of patients studied exhibited, during F, TTI values near or above the threshold value of 0.15 (Figure 2). Twenty patients had TTI values greater 0.15, and only 10 had TTI values ranging between 0.12 and 0.15 (Figure 3). In contrast, during S no patient exceeded the critical value of 0.15. This opens the question of whether inspiratory muscle fatigue was responsible for unsuccessful weaning.
Although we do not know for sure, there is substantial evidence indicating that fatigue was impending or present in the majority of our patients during weaning failure. Indeed, during F the Pi/Pi max ratio, which has been directly linked to inspiratory muscle endurance in normal subjects (17), was very high in all three groups of our patients (Table 4, Figure 1). When hyperinflation was taken into account, most of our patients were located above the critical fatiguing threshold. The latter, however, pertains to normal subjects and may not be applicable to our patients. Nevertheless it is reasonable to assume that the fatiguing threshold for normal subjects overestimates the endurance characteristics of patients, since a variety of factors would probably make them more vulnerable to fatigue (e.g., inadequate blood flow to the inspiratory muscles, persistent effects of drugs, or sepsis). More important, the fatiguing threshold was obtained in normal subjects at considerably lower inspiratory flows than those of our patients. As McCool and colleagues (31) and Clanton and coworkers (32) have clearly shown, the increase in inspiratory flow greatly reduces the critical values of the Pi/Pi max ratio, above which fatigue occurs even in normal subjects: an increase in the inspiratory flow from 0.5 to 1 L/s was sufficient to decrease the critical value of Pi/Pi max by 20% (from 0.57 to 0.45). Furthermore, the percentage change in the critical Pi/Pi max ratio was constant for the same percentage change in flow since the relationship between flow and sustainable pressure was linear (33). The fatiguing TTI threshold value of 0.15 was obtained with inspiratory flows ranging from 0.18 to 0.72 L/s (18), which averaged to 50% of the corresponding values in our patients (0.4–1.24 L/s). It can thus be argued that the critical TTI value of our patients should be about 20% lower.
Furthermore, the values of Pi/Pi max obtained during passive mechanical ventilation (which results in minimal chest wall distortion) necessarily underestimated the values that would be obtained during spontaneous breathing, since additional pressure would be exerted as a result of chest wall distortion. Moreover, constant inspiratory flow was used, which results in minimal resistive pressure dissipation (34). Since during spontaneous breathing the inspiratory flow pattern is not square, the Pi should increase. Accordingly, all our patients during F had actual TTI values probably above the fatiguing threshold.
Our results are comparable to those reported in the literature. Pouriat and colleagues (35) found that patients with COPD who failed to wean had a transdiaphragmatic pressure/ maximal transdiaphragmatic pressure (Pdi/Pdimax) ratio of 0.46, well above the 0.40 threshold value over which diaphragmatic fatigue occurs in normal subjects. In patients with COPD who failed to wean, Appendini and coworkers (36) found a Pdi/ Pdimax of 0.47 and a diaphragmatic TTI of 0.17, both greater than the critical fatiguing threshold. In contrast, Jubran and Tobin (5) recently reported that only five out of their 17 patients with COPD who failed to wean had a TTI greater than 0.15. However, they obtained TTI by measuring Pi max at the beginning of the weaning trial and Pi at the end. Since Pi max probably decreased by the end of the trial, they probably underestimated the actual TTI values.
The fact that TTI was excessive during F does not necessarily imply that overt fatigue was present at that time. In fact, the diagnosis of fatigue is hampered by the lack of universally agreed criteria and technical difficulties (37), and our protocol does not allow us to address this issue with certainty. However, an excessive TTI value does not necessarily imply task failure at the time when it is measured. In fact, it predicts that task failure will ensue within a limited time period (Tlim) that is inversely related to the TTI value. In our patients TTI was measured just before reinstituting mechanical ventilation. According to the TTI concept prediction, overt fatigue and task failure, if not present at the time of resumption of mechanical ventilation, would eventually ensue, at least in the majority of our patients, in a limited time period had these patients been allowed to breathe a little more. From a theoretical standpoint, fatigue should be considered rather as a continuum, and what really counts for someones ability to sustain spontaneous breathing is whether one is in the “fatigue process.” We postulate, therefore, that when our patients failed to wean, they were within this “fatigue process.”
In conclusion, the TTI and the f/Vt are the major pathophysiologic determinants of weaning failure or success. A decrease in the load faced by the respiratory muscles with a concomitant increase in their neuromuscular capacity that results in diminished energy demands (↓ TTI) and a more efficient breathing pattern (↓ f/Vt), are the main pathophysiologic changes underlying the transition from weaning failure to success.
The authors thank Dr. Milic-Emili for his valuable suggestions and careful review of the manuscript, Dr. Rani Dafni for statistical advice, and Mr. Athanassios Tsironis, M.Sc., for technical support.
| 1. | Vassilakopoulos T., Zakynthinos S., Roussos C.Respiratory muscles and weaning failure. Eur. Respir. J.9199623832400 |
| 2. | Stroetz R. W., Hubmayr R. D.Tidal volume maintenance during weaning with pressure support. Am. J. Respir. Crit. Care Med.152199510341040 |
| 3. | Zakynthinos S., Vassilakopoulos T., Roussos C.The load of inspiratory muscles in patients needing mechanical ventilation. Am. J. Respir. Crit. Care Med.152199512481255 |
| 4. | Jubran A., Tobin M.Passive mechanics of lung and chest wall in patients who failed or succeeded in trials of weaning. Am. J. Respir. Crit. Care Med.1551997916921 |
| 5. | Jubran A., Tobin M.Pathophysiological basis of acute respiratory distress in patients who fail a trial of weaning from mechanical ventilation. Am. J. Respir. Crit. Care Med.1551997906915 |
| 6. | Kochi T., Okubo S., Zin W. A., Milic-Emili J.Flow and volume dependence of pulmonary mechanics in anesthetized cats. J. Appl. Physiol.641998441450 |
| 7. | 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 |
| 8. | Bates J. H. T., Rossi A., Milic-Emili J.Analysis of the behavior of the respiratory system with constant inspiratory flow. J. Appl. Physiol.58198518401848 |
| 9. | Rossi A., Gottfried S. B., Higgs B. D., Zocchi L., Grassino A., Milic-Emili J.Respiratory mechanics in mechanically ventilated patients with respiratory failure. J. Appl. Physiol.58198518491858 |
| 10. | Eissa N. T., Ranieri V. M., Corbeil C., Chasse M., Robatto F. M., Braidy J., Milic-Emili J.Analysis of behavior of the respiratory system in ARDS patients: effect of flow, volume and time. J. Appl. Physiol.70199627192729 |
| 11. | Gottfried S. B., Rossi A., Higgs B. D., Calverly P. M. A., Zocchi L., Bozic C., Mili-Emili J.Noninvasive determination of respiratory system mechanics during mechanical ventilation for acute respiratory failure. Am. Rev. Respir. Dis.1311985414420 |
| 12. | Rossi A., Gottfried S. B., Zocchi L., Higgs B. D., Lennox S., Calverly P. M. A., Begin P., Grassino A., Milic-Emili J.Measurement of static compliance of total respiratory system in patients with acute respiratory failure during mechanical ventilation. Am. Rev. Respir. Dis.1311985672678 |
| 13. | D'Angelo E., Robatto F. M., Calderini E., Tavola M., Bono D., Torri G., Milic-Emili J.Pulmonary and chest wall mechanics in anesthetized paralysed humans. J. Appl. Physiol.70199126022610 |
| 14. | Holmes S.A simple sequentially rejective multiple test procedure. Scandinavian Journal of Statistics619796570 |
| 15. | Concato J., Feinstein A. R., Holford T. R.The risk of determining risk with multivariable models. Ann. Intern. Med.1181993201210 |
| 16. | Cotes, J. E. 1979. Lung Function, 4th ed. Blackwell Scientific Publications, Boston. 369. |
| 17. | Roussos, C., M. Fixley, D. Gross, and P. T. Macklem. Fatigue of inspiratory muscles and their synergic behavior. J. Appl. Physiol. 46:897–904. |
| 18. | Bellemare F., Grassino A.Effect of pressure and timing of contraction on human diaphragm fatigue. J. Appl. Physiol.53198211901195 |
| 19. | Roussos, C., and E. J. M. Campbell. 1986. Respiratory muscle energetics. In P. T. Macklem and J. Mead, editors. Handbook of Physiology: The Respiratory System, Vol. 3. American Physiology Society, Bethesda, MD. 481–509. |
| 20. | Petrof B. J., Legare M., Goldberg P., Milic-Emili J., Gottfried S. B.Continuous positive airway pressure reduces work of breathing and dyspnea during weaning from mechanical ventilation in severe chronic obstructive pulmonary disease. Am. Rev. Respir. Dis.1411990281289 |
| 21. | Maltais F., Reissmann H., Navalesi P., Hernandez P., Gursahaney A., Ranieri V. M., Sovilj M., Gottfried S. B.Comparison of static and dynamic measurements of intrinsic PEEP in mechanically ventilated patients. Am. J. Respir. Crit. Care Med.150199513181324 |
| 22. | Zakynthinos S. G., Vassilakopoulos T., Zakynthinos E., Roussos C.Accurate measurement of intrinsic positive end-expiratory pressure: how to detect and correct for expiratory muscle activity. Eur. Respir. J.101997522529 |
| 23. | 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 |
| 24. | Epstein S. K.Etiology of extubation failure and the predictive value of the rapid shallow breathing index. Am. J. Respir. Crit. Care Med.1521993545549 |
| 25. | Chatila W., Jacob B., Guaglionone D., Manthous C. A.The unassisted respiratory rate-tidal volume ratio accurately predicts weaning outcome. Am. J. Med.10119966167 |
| 26. | Tobin M. J., Perez W., Guenther S. M., Semmes B. J., Mador M. J., Allen S. J., Lodato R. F., Dantzker D. R.The pattern of breathing during successful and unsuccessful trials of weaning from mechanical ventilation. Am. Rev. Respir. Dis.134198611111118 |
| 27. | Yan, S., P. Sliwinski, A. P. Gauthier, I. Lichros, S. Zakynthinos, and P. T. Macklem. 1993. Effect of global inspiratory muscle fatigue on ventilatory and respiratory muscle responses to CO2. J. Appl. Physiol. 1371– 1377. |
| 28. | Holliday J. E., Hyers T. M.The reduction of weaning time from mechanical ventilation using tidal volume and relaxation biofeedback. Am. Rev. Respir. Dis.141199012141220 |
| 29. | La Riccia P. J., Katz R. H., Peters J. W., Atkinson G. W., Weiss T.Biofeedback and hypnosis in weaning from mechanical ventilators. Chest871985267269 |
| 30. | Milic-Emili J.Is weaning an art or a science? Am. Rev. Respir. Dis.134198611071108 |
| 31. | McCool F. D., McCann D. R., Leith D. E., Hoppin F. G.Pressure-flow effects on endurance of inspiratory muscles. J. Appl. Physiol.601986299303 |
| 32. | Clanton T. L., Ameredes B. T., Thomson D. B., Julian M. W.Sustainable inspiratory pressures over varying flows, volumes and duty cycles. J. Appl. Physiol.69199018751882 |
| 33. | Clanton T. L., Ameredes B. T.Fatigue of the inspiratory muscle pump in humans: an isoflow approach. J. Appl. Physiol.64198816931699 |
| 34. | Coussa M. L., Guerin C., Eissa N. T., Corbeil C., Chasse M., Braidy I., Matar N., Milic-Emili J.Partitioning of work of breathing in mechanically ventilated patients. J. Appl. Physiol.75199317111719 |
| 35. | Pourriat J. L., Lamberto C. H., Hoang P. H., Fournier J. L., Vasseur B.Diaphragmatic fatigue and breathing pattern during weaning from mechanical ventilation in COPD patients. Chest901986703707 |
| 36. | Appendini L., Purro A., Patessio A., Zanaboni S., Carone M., Spada E., Donner C. F., Rossi A.Partitioning of inspiratory muscle workload and pressure assistance in ventilator-dependent COPD patients. Am. J. Respir. Crit. Care Med.154199613011309 |
| 37. | NHLBI WorkshopRespiratory muscle fatigue: report of the respiratory muscle fatigue workshop group. Am. Rev. Respir. Dis.1421990474480 |
