Comparison of Three Methods of Cradual Withdrawal

77 downloads 0 Views 932KB Size Report
Comparison of Three Methods of Cradual Withdrawal from Ventilatory Support during Weaning from. Mechanical Ventilation. LAURENT BROCHARD, ALAIN ...
Comparison of Three Methods of Cradual Withdrawal from Ventilatory Support during Weaning from Mechanical Ventilation LAURENT BROCHARD, ALAIN RAUSS, SALVADOR BENITO, GIORGIO CONTI, JORDI MANCEBO, NOURREDINE REKIK, ALESSANDRO GASPARETTO, and FRAN(:OIS LEMAIRE Medical Intensive Care Unit, Hopital Henri Mondor, Creteil, France; Medical Intensive Care Unit, Hospital Santa Creu i Sant Pau, Barcelona, Spain; and Intensive Care Unit, Universita la Sapienza, Roma, Italy

Several modalities of ventilatory support have been proposed to gradually withdraw patients from mechanical ventilation, but their respective effects on the outcome of weaning from mechanical ventilation are not known. We conducted a randomized trial in three intensive care units in mechanically ventilated patients who met standard weaning criteria. Those who could not sustain 2 h of spontaneous breathing were randomly assigned to be weaned with T-piece trials, with synchronized intermittent mandatory ventilation (SIMV), or with pressure support ventilation (PSV). Specific criteria for performing tracheal extubation were defined for each modality. The number of patients who could not be separated from the ventilator at 21 d (i.e., who failed to wean) was compared between the groups. Patients in whom tracheal intubation was required in a 48-h period fOllowing extubation were also classified as failures. Among 456 mechanically ventilated patients who met weaning criteria, 109 entered into the study (35 with T piece, 43 with SIMV, and 31 with PSV). The three groups were comparable in terms of etiology of disease or characteristics at entry in the study. When all causes for weaning failure were considered, a lower number of failures was found with PSV than with the either two modes, with the difference just reaching the level of significance (23% for PSV, 43% for T piece, 42% for SIMV; p = 0.05). After excluding patients whose weaning was terminated for complications unrelated to the weaning process, the difference became highly significant (8% for PSV versus 33% and 39%, p < 0.025). Also, the probability of remaining on mechanical ventilation, calculated using the Kaplan-Meier estimate, was found to be significantly lower with PSV (p < 0.03). Weaning duration was shorter with PSV (p < 0.05), as was total length of stay in the intensive care unit (p < 0.01). The duration of weaning was explained first by the etiology of the disease and second by the mode of ventilation. We conclude that in this study, the outcome of weaning from mechanical ventilation was influenced by the ventilatory strategy chosen, and the use of PSV resulted in significant improvement compared with other strictly defined weaning protocols using T piece or SIMV.Brochard L, Rauss A, Benito S, Conti G, Mancebo J, Rekik N, Gasparetto A, Lemaire F. Comparison of three methods of gradual withdrawal from venti· latory support during weaning from mechanical ventilation. Am J Respir Crit Care Med 1994; 150:896-903.

After a patient recovers from an acute episode of respiratory failure necessitating the institution of mechanical ventilation, an attempt at discontinuation of mechanical ventilation becomes necessary. In a substantial number of patients, resuming spontaneous breathing without mechanical assistance is poorly tolerated (1, 2). Several modes of partial ventilatory support have been proposed with the aim of gradually decreasing the level of mechanical assistance in order to eventually wean the patient from the ventilator. The most common methods at this stage are intermittent periods of spontaneous breathing on a T piece between periods of controlled or assist-control ventilation (3), synchronized intermittent mandatory ventilation (SIMV) (4, 5), and pressure sup-

(Received in original from April 21, 1993 and in revised form April 1, 1994)

Correspondence and requests for reprints should be addressed to Dr. Laurent 8rochard, Service de Reanimation Medicate, H6pital Henri Mondor, 94010 Creteil-Cedex, France.

Am

J Respir Crit Care Med Vol 1SO. pp 896-903, 1994

port ventilation (PSV)(6-8). Several studies have been performed in the past comparing the efficacy of intermittent mandatory ventilation or SIMV with T-piece trials (3,9-11). Although conflicting results have been published, no clear superiority of one technique over the other has emerged. However; the choice of weaning mode remains a major concern for practitioners, as 10to 50% of mechanically ventilated patients require a prolonged weaning procedure (1-3). PSV, a recent mode of ventilation that has been evaluated mainly in physiological or short-term studies, efficiently reduces the workload imposed on the respiratory muscles (6-8). The level of assistance can be gradually decreased until it only compensates for the additional work imposed by the endotracheal tube and the demand valve of the ventilator, at Whichtime tracheal extubation can be performed (12, 13). PSV might therefore be an optimal support for patients who have difficulties being weaned from the ventilator and offers several potential advantages compared with other modes, although few comparative data are available (6, 10, 14, 15). If one technique has any superiority over the

Brochard, Rauss, Benito, et 01.: Three Methods of Gradual Withdrawal from Ventilatory Support

other, improvement in weaning outcome is more likely to be expected in the subgroup of mechanically ventilated patients who do not tolerate prolonged discontinuation from mechanical ventilation when weaning is attempted. Therefore, we designed a prospective randomized trial to evaluate the efficacy of these three techniques of gradual withdrawal from mechanical ventilation in patients having difficulties tolerating discontinuation of mechanical ventilation.

METHODS Patients Patients were recruited consecutively over an 18-mo period in three medical-surgical intensive care units. All patients had been ventilated for more than 24 h and had recovered from the acute phase of their disease. Discontinuation from mechanical ventilation was attempted when the physician in charge thought that the cause for instituting mechanical ventilation had resolved or improved markedly. In addition, the following were required: body temperature below 38.50 C, hemoglobin higher than 8 mgldl, cardiovascular therapy considered as adjusted by the primary physician in patients with known or suspected cardiac insufficiency (16), relief from sedation, and correction of electrolyte disorders. At this time, several pulmonary function tests were performed, including measurement of maximal inspiratory pressure against an occluded airway and a one-way valve (17), breathing frequency measured during the first 2 min after discoJ1tinuation of mechanical ventilation, and vital capacity (2, 18).At least three of the following criteria were required to continue with the procedure: (1) maximal inspiratory pressure equal to or lower than -25 em H20, (2) breathing frequency below or equal to 35 breaths/min, (3) vital capacity equal to or above 10 mllkg of body weight, and (4) arterial oxygen saturation greater than 90% for an inspired oxygen fraction of 40%. These criteria help to predict the ability of patients to be taken off the ventilator but have a poor positive predictive value (18).Therefore, when at least three criteria were present, the ability of the patient to sustain spontaneous breathing was evaluated during disconnection from the ventilator with supply of supplemental oxygen and humidification of gas on a T piece for a maximum of 2 h. During this period of spontaneous breathing, clinical tolerance of the trial was continuously evaluated. Poor clinical tolerance was diagnosed when breathing frequency was above 35 per minute or increased by 50% or more, when heart rate or systolic blood pressure rose by 20% or more, or when agitation, depressed mental status, or diaphoresis was present. Arterial blood gas tensions were measured (ABL 30; Radiometer, Copenhagen, Denmark) at the end of the 2-h period or at any moment before when clinically required. An arterial Po 2 below 50 mm Hg or an arterial pH of 7.32or below indicated a need for reconnection to mechanical ventilation. When the patient remained stable with no sign of poor tolerance, the endotracheal tube was removed on the same or the next day. When clinical signs of poor tolerance developed at any time before or at the end of the 2-h trial or poor arterial blood gas results were noted as defined above, the patient was put back on the ventilator and considered for randomization into the study. Randomization and initiation of the weaning trial were performed within 24 h after failure of the T-piece trial. Randomization After informed consent was obtained from the patient or next of kin, the patient was randomly assigned on the same day to one of the three following modalities: intermittent periods of T piece, SIMV,or PSV. Randomization was stratified according to center and to etiology of the process responsible for difficult weaning. Because the rate of success and the duration of the weaning period can be influenced by the underlying disease, a stratification was performed, taking into account four different .categories of patients who may present with difficulties to sustain spontaneous ventilation: (1)patients with chronic obstructive pulmonary disease, (2) patients with central nervous system disorders, (3) patients with peripheral neuromuscular disorders, and (4) patients with other disorders, including patients with heart failure and postoperative patients. In each group, stratification was also performed to separate intubated patients and patients who had undergone tracheostomy. The assignment of the

897

ventilatory mode was determined using the closed-envelope technique. These different stratifications were performed as a means to homogenize the case-mix of patients included into each of the three weaning strategies tested. Protocol



The protocol was designed to follow, in each mode, a similar strategy for decreasing ventilatory assistance. For the three modes, the same criteria were used to progress in the weaning procedure with the same basic schedule. T-piece trial. The T-piece method is based on the principle of gradually lengthening the periods of disconnection from the ventilator. During these periods of disconnection, supplemental humidified gas is provided to the patient through a T piece connected to the endotracheal tube. Patients assigned to the T-piece groups had undergone several attempts at disconnection from the ventilator during the day. The fraction of oxygen in the inspired gas was kept similar to that used during mechanical ventilation. Between each period, at least 1 h of mechanical ventilation was reinstituted in the assist-control mode with a high backup minute ventilation. Therefore, the number of periods of disconnection during the day depended on the length of these periods and on nursing staff availability. Durations of these periods could be 5, 15, 30, 60, or 120 min in progressive steps. To select the duration of the first weaning attempt, the duration of the initial tolerance test performed before randomization was used. Because none of the patients had sustained a period longer than 120 min, the duration of the first attempt was selected, among those listed above, as equal or immediately shorter than the duration of the initial tolerance trial. For instance, a patient who sustained 90 min of spontaneous breathing on the initial test was put on a 60-min test at first attempt. During the weaning process, when a preset duration was sustained without evidence of poor tolerance, a longer duration was tried on the next half-day. If the patient was not able to tolerate the duration, the next period was performed with a shorter or with the same duration, depending on whether poor tolerance appeared before or at the end of the disconnection period. Incremental durations of the T-piece periods were recommended two times per day, at approximately 9:00 A.M. and 2:00 P.M. Criteria used to define poor tolerance were the same as those for the initial tolerance test. The number of disconnections during the day could vary from three to eight depending on the step of the weaning process (Le., on the duration of each disconnection). During assist-control ventilation, tidal volume, flow, and a minimal respiratory rate were set on the ventilator, constituting a backup ventilation, but the patient could trigger additional assisted breaths. No period of disconnection was performed after 7:00 P.M. When the duration of disconnection had reached 2 h with adequate gas exchange, tracheal extubation was decided. In most cases, one period of 120-min disconnection was sufficient to decide for extubation, but as many as three 120-mln trials could be performed before extubation was decided (especially for patients with prolonged periods of mechanical ventilation). SIMV. During SIMV, the patient is allowed to breathe spontaneously between mechanical breaths, which are delivered at a low rate. Characteristics of these mechanical breaths are set on the ventilator by the physician and are delivered in synchrony with the spontaneous effort performed by the patient. By contrast, spontaneous breaths are not assisted by the ventilator. The weaning procedure usually consists of gradually decreasing the rate of mechanical breaths on the ventilator. In patients assigned to SIMV, the initial ventilator rate was set at half the total frequency used previously during controlled mechanical ventilation or assist-control ventilation, keeping tidal volume and inspiratory flow constant. Two times per day, in a systematic fashion, a decrease in the ventilator rate by two to four breaths per minute was recommended when the patient breathed at the preceding rate with no sign of poor tolerance. When one or more signs of poor tolerance was present, the ventilator rate was increased back to the preceding level. When the patient was able to tolerate a mechanical rate of four per minute or less over 1 d, tracheal extubation was decided. No other criteria for extubation were required (10, 11, 19). PSv. During PSV, patients are breathing spontaneously, but each inspiration is supported by mechanically driven flow delivered at a preset

898

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE

amount of pressure. No mandatory breath is delivered by the ventilator. Weaning is performed by gradually decreasing the level of pressure delivered for each breath. In patients assigned to PSV, the initial level of pressure support was adjusted until the breathing frequency of the patient ranged between 20 and 30 breaths per minute. Two times per day, in a systematic fashion, a decrease in the level of pressure by 2 or 4 cm H20 was recommended when the patient breathed at the preceding pressure level with no sign of poor tolerance. When one or more signs of poor tolerance appeared, the pressure support level was increased back to the preceding level. When the patient was able to tolerate spontaneous breathing with 8 cm H2 0 of pressure support or less during a 24-h period, tracheal extubation was performed. This criterion for extubation resulted from a previous work showing that this level was generally sufficient to compensate for the additional work imposed by the endotracheal tube and ventilator circuit (13). In all modes of assisted ventilation (Le.,assist-control ventilation, SIM\/, and PSV), a positive end-expiratory pressure of up to 4 cm H2 0 could be applied in every patient. Positive end-expiratory pressure was applied in most of the patients with chronic obstructive pulmonary disease to facilitate the triggering of the ventilator when intrinsic positive alveolar pressure was suspected (20). The ventilator was always set at its maximal triggering sensitivity (minimal pressure level), and the inspired fraction of oxygen was adjusted to maintain arterial oxygen saturation above 90%. During the T-piece trial, neither pressure support nor continuous positive airway pressure was used. The ventilators used in this study were the following: for the three modalities, Servo 900 C (Siemens, Sweden), Cesar (Taema, France), Veolar (Hamilton, Switzerland), 7200 a (Puritan-Bennett, U.S.A.), and Advent (Ohmeda, France) were used; for SIM\/, CPU1 (Ohmeda, France) was also used. . Enteral feeding was always preferable to parenteral nutrition and was administered to each patient via a nasogastric tube at the rate of 30 to 35 kcallkg of body weight.

Statistical Analysis Definitions. The objective of this study was to compare the efficacy of three procedures of gradual withdrawal from mechanical ventilation in the first 21 d after starting the weaning process. During that period, patients were considered as successfully weaned when they had no need for ventilatory assistance for at least 48 h after tracheal extubation or after permanent disconnection from the ventilator in patients with tracheostomies. They were considered as having failed weaning when they could not be separated from the ventilator after 21 d while still being in the selected mode of weaning or when they required tracheal reintubation and/or reconnection to the ventilator in the 48 h after tracheal extubation and/or disconnection from the ventilator. Some patients were withdrawn from the usual weaning procedure, including the 48 h after tracheal extubation, because of the occurrence of intercurrent events. In an intentionto-treat analysis, all of these events were considered as failures. Also, events were divided in two categories. Ischemic cardiac events occurring during the transition from mechanical ventilation to spontaneous breathing (16, 21) and occurrence of nosocomial pneumonia after 72 h in the selected mode were considered as linked with the weaning process and were equivalent to failure of the ventilatory mode. Indeed, in medical patients, nosocomial pneumonia is mainly caused by a prolonged duration of mechanical ventilation (22). Therefore, a high rate of nosocomial pneumonia might potentially result from an excessive length of mechanical ventilation caused by a particular weaning strategy. By contrast, concomitant events interfering with the weaning process but judged unrelated to it, such as stroke, peritonitis, and laryngeal edema, were considered early withdrawals from the protocol and were not included in the failures per se. A second analysis was thus performed to exclude those patients. Last, we also excluded all cases with intercurrent events in a third analysis, without distinguishing those linked or not with the weaning process. Analytical methods. Quantitative values were compared between groups using a one-way analysis of variance, and, when necessary, two-by-two comparisons were made using a Tukey's test. Comparison of qualitative data between groups was made using a x} test. The probability of remaining on mechanical ventilation was calculated

VOL 150

1994

using the Kaplan-Meier estimate, and the comparison between groups was made using the log-rank test (23). For this analysis, the following classifications were made. With regard to the objective of the study, patients could be classified in four categories: success, failure at 21 d, failure before 21 d (l.e., reintubation or early interruptions resulting from the weaning process), and loss of survey (l.e., early interruptions having no direct relation with the weaning process). To construct the curves, all types of failures were considered to be equivalent and classified as remaining on mechanical ventilation. An adjustment for the distribution of etiologies of diseases was also performed to test whether differences between the three groups remained significant after this adjustment. Using the same classification, a Cox proportional-hazards model was performed to determine the dependent variables to explain the length of the weaning process (24). All data are presented as mean ± SO. A P value of 0.05 was considered the limit of significance.

RESULTS During the study, 456 mechanically ventilated patients met positive weaning criteria and underwent a spontaneous-breathing T-piece trial. Among them, 109 eventually failed to tolerate spontaneous breathing over 2 h and were therefore randomly assigned to one of the three modalities of gradual withdrawal from mechanical ventilation; 35 were assigned to T-piece trials, 43 to SIMV, and 31 to PSv. The overall distribution of diagnosis was similar in the three groups (Table 1). Patient characteristics on admission to the intensive care unit and at the beginning of the weaning process are shown in Tables 1 and 2. No statistical difference existed between the groups regarding their characteristics upon admission to the intensive care unit, the duration of mechanical ventilation before weaning, the duration of the initial T-piece tolerance test, and the results of pulmonary function tests performed at the onset of the weaning process, all of which are important predictors of weaning outcome (18, 25). In the group weaned with pressure support, the initial level of pressure at the beginning of the weaning procedure was 18.1 ± 3.8 cm H20; in the group weaned with SIMV, the initial mechanical breath rate was 9.5 ± 3.7 breaths/min; in the T-piece group, the mean first T-piece duration was 38 ± 32 min (or step 2.9 ± 1.2, considering 5 min of disconnection as step 1 and 120 minutes as step 5).

Intention-to-Treat Analysis The percentage of patients who could not be separated from mechanical ventilation at Day 21, for any reason, reached 43% (15/35) with the T piece, 42% (18/43) with SIMV, and 23% (7/31)with PSV. This lower percentage observed with pressure support just reached the level of significance compared with the other modalities (p = 0.05). All causes of failure are detailed in Table 3.

Second Analysis Five episodes of gradual withdrawal from mechanical ventilation with the T piece, two with SIMV, and five with PSV were terminated early for causes considered to have no link with the weaning procedure (Table 3). The outcome of the remaining episodes was compared by several means. First, the total number of failures in this analysis was significantly lower with PSV than with T piece or with SIMV (2/26 [8%] versus 10/30 [33%] and 16/41 [39%], P < 0.02). Also, a day-to-day analysis of the probability of remaining on mechanical ventilation was performed and is shown in Figure 1. After a few days, the curves separated and the probability of remaining on mechanical ventilation was lower with PSV than with the two other modalities (log rank = 7.63, P < 0.03), PSV being significantly different from the two other modalities. Because the

Brochard, Rauss, Benito, et 01.: Three Methods of Gradual Withdrawal from Ventilatory Support

899

TABLE 1 CHARACTERISTICS OF THE PATIENTS ASSIGNED TO THE THREE MODALITIES OF GRADUAL WITHDRAWAL FROM MECHANICAL VENTILATORY SUPPORT ON ADMISSION TO THE INTENSIVE CARE UNIT"

Characteristic Number of patients Number of patients with chronic obstructive lung disease Number of patients with neurological disorders (central or peripheral) Number of patients with other disorders Age, yr Sex ratio, % women Simplified acute physiologic score (25)

T Piece

Synchronized Intermittent Mandatory Ventilation

Pressure Support Ventilation

35

43

31

9(26%)

14(33%)

7(23%)

6 (17%)

7 (16%)

5 (16%)

20 (57%) 54.5 ± 17.5 26 11.8 ± 4.3

22 (51%) 60.1 ± 16.7 32 12.2 ± 3.3

19(61%) 62.9 ± 15.9 40 12.2 ± 4.5

" Data are given as mean ± SO. No significant differences were observed among the three groups.

TABLE 2 MAIN CHARACTERISTICS OF THE PATIENTS ASSIGNED TO THE THREE MODALITIES OF GRADUAL WITHDRAWAL FROM MECHANICAL VENTILATORY SUPPORT AT THE ONSET OF THE WEANING PROCEDURE"

T~ce

Characteristic Duration of mechanical ventilation before weaning, d Tolerance of initial test on T piece, min Inspired fraction of oxygen, % Vital capacity, ml Maximal inspiratory pressure (18), em H2O Breathing frequency, breaths/mint Tidal volume, mit Frequency/tidal volume ratio (20), breaths/min/Lt Minute ventilation, Llmint

(n

=

35)

17 43 37 904

± ± ± ±

31 27 13 214

41 ± 19 32 ± 7 381 ± 126 89 ± 43 11.8 ± 3.5

Synchronized Intermittent Mandatory Ventilation (n = 43)

11 53 36 827

± ± ± ±

10 34 13 232

39 ± 15 32 ± 6 377 ± 111 90 ± 33 11.5 ± 2.6

Pressure Support Ventilation (n = 31)

14 49 37 989

± ± ± ±

17 28 21 543

41 ± 16 30 ± 5 378 ± 106 89 ± 32 10.8 ± 2.8

" Data are given as mean ± SO. No significant differences were observed among the three groups. Values obtained during the first 2 min after discontinuation of mechanical ventilation.

t

proportion of patients with COPD appeared slightly lower in the group ventilated with PSV, an adjustment was made according to this difference. The difference was still significant after adjustment for the etiology of the disease (log rank = 6.18, P < 0.05). No significant difference could be found between the two other modalities. During the 21-d weaning procedure, the probability of being weaned from the ventilator (relative risk) was 2.03 times higher with PSV than with SIMV. The mean duration of the weaning period was calculated in the three groups for the patients who completed the weaning protocol (Le., excluding all patients with early terminations). There was no significant difference between T piece and SIMV (8.5 ± 8.3 d versus 9.9 ± 8.2 d). By contrast, the weaning duration was significantly shorter with pressure support than with the two other modalities pooled together (5.7 ± 3.7 d versus 9:3 ± 8.2 d, P < 0.05). We also compared the total length of stay in the intensive care unit and the length of stay in the intensive care unit after the beginning of the weaning procedure in the same groups of patients. To make these comparisons meaningful, we excluded five outliers (two in the T-piece group with 226 and 195 d of total length of stay in the intensive care unit, one in the intermittent mandatory ventilation group with 151 d, and two in the pressure

support group with 120 and 100 d). We pooled together T piece and SIMV that were not statistically different (28.9 ± 17.7 d versus 26.8 ± 19.0d for the total length of stay in the intensive care unit). We found that the total length of stay in the intensive care unit was significantly shorter with PSV than with the two other modalities (17.5 ± 10.2 d versus 27.8 ± 18.3d, P < 0.01) as well as the length of stay in the intensive care unit after the beginning of the weaning procedure (10.5 ± 6.7 d versus 17.8 ± 13.1 d, P < 0.01). Mortality in the intensive care unit was slightly lower in the pressure support group but without statistically significant difference: 4/31 (13%) for PSV, 8/35 (23%) for T piece, and 10143 (23%) for SIMV. Last, a multivariate analysis was performed to search for factors explaining weaning duration. In the analysis, the Cox proportional-hazards model showed that the length of weaning was first explained by etiology of the disease (p = 0.01), with patients with chronic obstructive pulmonary disease being the most difficult to separate from the ventilator, and then the mode of ventilatory support (p = 0.03), with PSV allowing the shorter weaning duration. The duration of mechanical ventilation before weaning was not a significant parameter, and the frequency/tidal volume ratio added little significant information (p = 0.08).

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE

900

T Piece (n = 35)

Synchronized Intermittent Mandatory Ventilation (n = 43)

Events unrelated to the weaning process Laryngeal edema needing 1 reintubation o Tracheoesophageal fistula 2 Stroke 1 Peritonitis Septic shock without pneumonia 1 Events considered as failures of the weaning process Nosocomial pneumonia 3 > 72 h after start of weaning Ischemic heart failure during 1 weaning 2 Reintubation « 48 h)' Impossibility of weaning 4 after 21 d 16 10 (33%) Number of failures 15 Total number of events

Pressure Support Ventilation (n = 31)

1

2

o o

1 1

The main finding of this study is that the ventilatory management of patients with difficulties being separated from the ventilator has a major influence on the outcome of weaning from mechanical ventilation. In this study, PSV provided a significant benefit compared with SIMV or intermittent periods of T-piece ventilation and assist-control.

o

1

o

o

3

Selection of the Patients 3 5

o

5

2 (8%) 7

(39%)

18

1.0

......

Q.

0.9 0.8 0.7 0.6 0.5

SIMV

0.4 T PIECE

0.3 0.2

...--..

0.1

PSV

0.0 0

5

10

15

Tocompare the number of failures only in patients who terminated the trial without intercurrent events, all patients with early terminations were excluded from the analysis (i.e., including cardiac events and nosocomial pneumonia). The number of failures remained significantly smaller with PSV than with the two other modalities (1/25 [4%] for PSVversus 6/26 [23%] forT piece and 10/35 [29%] for SIMV, p < 0.01).

DiSCUSSION

• Other than for laryngeal edema.

z Oz (!)O -zz:5 c(:::i!: ...... wZ W a: > U-....J Oc( ~t) _z ::!c( mI c(t) mw O:::i!: a:

1994

Third Analysis

TABLE 3 CAUSES OF FAILURE AND EARLY TERMINATION IN THE THREE GROUPS OF PATIENTS ASSIGNED TO A MODALITY OF GRADUAL WITHDRAWAL FROM VENTILATORY SUPPORT

Cause

VOL 150

20

25

DAYS Figure 1. Probability of remaining on mechanical ventilation in patients with prolonged difficulties in tolerating spontaneous breathing. This probability was significantly lower for pressure support ventilation (PSV) than for T piece or synchronized intermittent ventilation (SIMV) (cumulative probability for 21 d, P < 0.03 with the log-rank test).

Many patients do not experience problems being weaned from the ventilator. To select difficult-to-wean patients, we used standard weaning indexes and, as a second step, a 2-h period of T-piece ventilation was performed to assess the ability of the patients to sustain spontaneous breathing during prolonged periods. We reasoned that weaning indexes could be used to exclude nonweanable patients by using low thresholds for these indexes but that many of patients with positive tests may still not be able to tolerate prolonged periods of spontaneous breathing. To assess the latter group, a T-piece trial was then performed. Because the optimal duration of a T-piece trial has never been determined and because our previous experience suggested a high percentage of reintubation with short periods of disconnection, we selected a 2-h period. In one of the three centers, the follow-up of the 2-h T-piece trial was performed prospectively in 169 patients; in this center, 122 patients underwent tracheal extubation at the end of the 2 h while 47 (28%) entered the study. Among the 122extubated patients, 13 needed reintubation because of respiratory distress and eight because of intercurrent events. Therefore, among the entire group, 17% of the extubated patients (or 12% of the whole group) were poorly predicted by the 2-h T-piece trial. For the whole study, the T-piece trial test allowed selection of 24% of all the mechanically ventilated patients in whom discontinuation from mechanical ventilation was attempted. One could argue, however, that some of the patients left on mechanical ventilation and subsequently included in the study because of poor tolerance of the trial have simply been subjected to an excessive load and would have done better after simple extubation. Our previous data concerning the extra work of breathing induced by the presence of an endotracheal tube suggested that the tube itself only constitutes a mild increase in load (13). For instance, there was no significant difference in the value of lung and airway resistances with or without the tube in place (13). Therefore, we hypothesized that the 2-h trial constituted a mild endurance test that may help to assess the ability of the patient to sustain prolonged periods of spontaneous breathing. Further prospective studies are needed to determine whether this approach is optimal. An important aspect of the design of the study was to obtain comparable groups of patients. The different stratifications performed within the randomization helped to make the groups homogenous and. allowed us to study a limited sample of patients. Indeed, regarding the characteristics upon entry to the intensive care unit, the distribution of diagnosis, the duration of the initial

Brochard, Rauss, Benito, et 01.: Three Methods of Gradual Withdrawal from Ventilatory Support

tolerance test, and the weaning indexes appeared similar among the three groups. The number of postoperative patients was slightly higher in the PSV group, and this could influence the results. Indeed, the Cox proportional-hazards model analysis indicated that etiology of the disease was a major factor influencing length and difficulties of weaning. After adjustment for etiology, however, the probability of remaining on mechanical ventilation remained lower for patients ventilated with PSv. In addition, the same trends appeared within the three categories of patients (chronic obstructive pulmonary disease, neuromuscular disorders, postoperative, and others) regarding the superiority of PSV, although the subgroups are too small to make statistical comparisons meaningful. Our results also indicate clearly that protocols comparing modalities of mechanical ventilation should take into account the underlying disease of the patients. Methodological Bias

From the initial level of support supplied to the patients in each group at the beginning of the weaning trial, one can calculate what the shorter duration of weaning could be. Assuming a similar and constant reduction of the level of support in each group, the shorter predicted duration of weaning ranged from 2 to 3 half-days for T piece and SIMV and from 3 to 5 half-days for PSv. Therefore, the way the initial level of support was selected and subsequently decreased in each arm did not introduce a bias in favor of PSV. The absence of blinding after randomization in this study is a potential for bias. The criteria used to appreciate tolerance of ventilatory support and general principles in the decrease of ventilatory support were strictly similar in the three arms. The physicians in charge of the study in each center were responsible for the strict observance of the protocol with regard to the criteria used to change the level of support or to extubate the patient. The rigid criteria used in each arm were designed in part to minimize the subjective part of the medical decisions, and we believe that it is unlikely that this factor had a significant influence on the results (although such an influence cannot be excluded). On the other hand, because very strict criteria were used within each modality, the results of this study are highly influenced by these criteria. Criteria for extubation were different in the three groups, which might have influenced the results. We cannot separate the effects of the particular protocol we used within each modality from the extubation criteria that were used. Data are lacking in the literature concerning precise guidelines on the use of ventilatory modes. We selected setup and extubation criteria for T-piece trials and SIMVthat had been previously reported (3, 9-11,19,26). For SIMV, five patients had to be reintubated or put back on the ventilator within 48 h after extubation, indicating that our decision for extubation, based on clinical tolerance with adequate gas exchange at a low level of mandatory breaths, was not adequate for these patients. It suggested that tracheal extubation was performed too early in the course of weaning in this group or that patients had been previously exhausted by the ventilatory mode used. In the first case, a larger number of patients would not have been separated from the ventilator after 21 d if criteria prolonging the ventilatory length had been used. In such a case, the overall results concerning SIMV would not be changed. The end point to reach and therefore the criteria for extubation may represent different levels of difficulties in the three arms. As discussed above, a T-piece trial may represent a slightly higher workload than a PSV level of 8 em H2 0 for some patients, although individual variations make comparison difficult. It is conceivable that a higher workload may be imposed on the patient during SIMV at 4 breaths/min and that this accounts in part for the differences ob-

901

served in this study. It could explain a longer stay on the ventilator with SIMV and, provided that a high workload may interact with the muscular force reserve of the patient, a higher rate of failure after tracheal extubation. Up to three periods of a 2-h T-piece trial could be attempted before extubation. Among the 23 patients who underwent tracheal extubation, however, 20 were extubated after the first 120-min trial-two after two trials and one after three trials. In no case did repetition of the number of tzo-mln trials change the issue of weaning. This factor probably had no role or a marginal role in the results. Weaning Methods

After the introduction of intermittent mandatory ventilation, this mode of ventilatory support has rapidly been used widely as a weaning technique (3-5). A few comparisons have been made with conventional mechanical ventilation; most included a small number of patients and did not select patients with difficulties in tolerating disconnection from the ventilator (9-11). In a retrospective study involving a large number of patients (and in particular many who required prolonged ventilation), Schachter and coworkers did not find any advantage of intermittent mandatory ventilation over conventional mechanical ventilation (3). No superiority of this mode has ever been demonstrated, and we find in this study no difference between SIMV and a more conventional approach using T piece. Intermittent mandatory ventilation has also been criticized regarding its poor tolerability, particularly in patients with chronic obstructive lung disease. Several criticisms can be made of this mode of ventilation, including the high level of effort required from the patient to open the demand valve and to breathe through the ventilator circuit (27). Therefore, it has been suggested to add a small level of pressure support to compensate for this additional work during the spontaneous cycles (28). Few clinical evaluations of this modality have been made, however, and one can only speculate whether better results might be obtained by combining SIMV and PSV (28, 29). Conventional mechanical ventilation and intermittent periods of disconnection on a T piece had been advocated to offer a gradual withdrawal from ventilatory support in a nonfatiguing pattern (26). Tobin and coworkers have shown that in patients who do not tolerate spontaneous breathing, rapid shallow breathing develops almost instantaneously upon discontinuation of ventilatory support and connection to a T piece (2). Therefore, a training effect of such a technique appears questionable, as the abnormal breathing pattern of these patients probably reflects the inadequacy of lung mechanics and respiratory muscle function to permit successful weaning and corresponds to an attempt to prevent inspiratory muscle fatigue (30). In addition, breathing on a T piece through an endotracheal tube alters the work of breathing in several respects, including changing functional residual capacity and ventilatory requirements (13). Lastly, the time spent off the ventilator may be a crucial point, as too long a spontaneous breathing trial may expose the patient to severe deterioration in arterial blood gases, increase in oxygen consumption, or excessive workload on the respiratory muscles. Therefore, repeated monitoring generally is needed to detect early clinical signs of poor tolerance, imposing an increased workload on nursing staff. PSV efficiently assists each spontaneous breath by decreasing the respiratory workload imposed on the respiratory muscles (6-8). Monitoring of breathing frequency and of accessory muscle activity allows the clinician to find a satisfying workload for the patients, avoiding excessive loading on the one hand and total rest of the respiratory muscles on the other (8). In addition,

902

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE

we previously showed that a pressure support level of 8 cm H20 was most often sufficient to compensate for the additional work caused by the endotracheal tube and the demand valve (13). We thus reasoned that the good tolerance of such a level could be used as a criterion for extubation (13). Potential advantages of this mode of gradual withdrawal from mechanical ventilatory support include the efficacy of workload reduction and the opportunity to judge the ability of the patient to be separated from the ventilator. Although this latter point is probably not different from the T-piece trial, it can be performed without the need for disconnection from the ventilator with its security systems and extensive monitoring.

Reasons for Weaning Failure Failure of weaning from mechanical ventilation can result from various disorders. In this study, separation of a patient from the ventilator was attempted only after correction of the main factors that may cause high ventilatory requirements or reversible respiratory muscle weakness, such as electrolyte disturbances, anemia, left heart failure, and sepsis. Aggressive management of left heart dysfunction was recommended before attempting the protocol, because this factor may impede weaning (16). The main reason that patients do not tolerate discontinuation from mechanical ventilation seems to be the inability of ventilatory muscle function to cope with the imposed workload (30, 31). Weakness of the respiratory muscles is often present because of acute or chronic malnutrition (32) and presumably prolonged mechanical ventilation. These muscles have to face high resistive and elastic loads because of elevated airway and tube resistances, reduced respiratory system compliances, and/or dynamic hyperinflation. Maladaptation of the muscles is probably the cause for rapid shallow breathing, a good indicator of weaning failure (2).When the general causes responsible tor high ventilatory requirement are treated, the strategy for weaning is to provide adequate ventilatory support to avoid excessive workload and to recognize the right moment to perform tracheal extubation. Partial ventilatory support during the period should therefore be instituted with the aims of decreasing respiratory workload until the status of the patient has SUfficientlyimproved and minimizing adverse effects of inadequate ventilatory support. Controlled mechanical ventilation cannot be recommended at that time, because it often requires the patient to be sedated and may promote disuse atrophy of the respiratory muscles. Insufficient ventilatory support or dysynchrony between patient and ventilator may result in worsening of respiratory mechanics or in delay of the extubation/period. Indeed, it can generate discomfort for the patient (14), provide excessive workload and increased oxygen consumption (8, 33, 34), and induce dynamic hyperinflation (35). High levels of oxygen consumption are associated with catecholamine release and modifications of sympathetic tone (8, 16). Therefore, inadequate support not only may result in ventilatory abnormalities but may have numerous systemic deleterious effects. PSV is often found more comfortable than other modalities by patients (6, 14). It has great efficacy in decreasing respiratory muscle workload (6, 8, 34), and its use may be associated with a reduction of dynamic hyperinflation (8, 35, 36). Adjustment of the level of support based on breathing frequency proposed here allows easy management and probably avoids excessive levels of workload (8). It is not known whether adjustment of the level of assistance based on breathing frequency can be recommended with other ventilatory modes. Also, the criteria used here to decide on the time of tracheal extubation in the pressure support group seemed efficient, indicating that the level of assistance required to reach a certain breathing pat-

VOL 150

1994

tern during PSV helps the clinician to assess the ability of the patient to breathe spontaneously. These factors contribute to easy management of the weaning procedure with PSV and probably influenced the outcome of weaning in this study. Summary

We found that a strategy of progressive reduction of PSV until a threshold level was reached offered a benefit over more conventional approaches during gradual withdrawal from mechanical ventilation in difficult-to-wean patients. The results of this study were also influenced by the way each technique was applied, and this factor can not be separated from the intrinsic effects of each ventilatory mode. The presence of chronic obstructive pulmonary disease was found to be a major factor influencing the length of the weaning process. Acknowledgment: The writers thank Christian Brun-Buisson and Alain Harf for their helpful comments. They are indebted to Mrs. Claudie Chastanet for technical assistance in the preparation of the manuscript.

References 1. Thavanainen J, Salmenpera M, Nikki P. Extubation criteria after weaning from intermittent mandatory ventilation and continuous positive airway pressure. Crit Care Med 1983;11:702-7. 2. Tobin MJ, Perez W, Guenther SM, Semmens BJ, Mador MJ, Allen SJ, Lodato RF, Dantzker DR. The pattern of breathing during successful and unsuccessful trials of weaning from mechanical ventilation. Am Rev Respir Dis 1986;134:1111-8. 3. Schachter EN, Tucker D, Beck CJ. Does intermittent mandatory ventilation accelerate weaning? JAMA 1981;246:1210-4. 4. Downs JB, Klein EF, Desautels D, Modell JH, Kirby RR. Intermittent mandatory ventilation: a new approach to weaning patients from mechanical ventilators. Chest 1973;64:331-5. 5. Weisman 1M,Rinaldo JE, Rogers RH, Sanders MH. Intermittent mandatory ventilation. Am Rev Respir Dis 1983;127:641-7. 6. Macintyre NR. Respiratory function during pressure support ventilation. Chest 1986;89:677-83. 7. Brochard L, Pluskwa F, Lemaire F. Improved efficacy of spontaneous breathing with inspiratory pressure support. Am Rev Respir Dis 1987; 136:411-5. 8. Brochard L, Hart A, Lorino H, Lemaire F. Inspiratory pressure support prevents diaphragmatic fatigue during weaning from mechanical ventilation. Am Rev Respir Dis 1989;139:513-21. 9. Hastings PRo Bushnell LS, Skillman JJ. Weintraub RM, Hedley-Whyte J. Cardiorespiratory dynamics during weaning with IMV versus spontaneous ventilation in good-risk cardiac surgery patients. Anesthesiology 1980;53:429-31. 10. Prakash 0, Meij S, Van Der Borden B. Spontaneous ventilation test vs intermittent mandatory ventilation: an approach to weaning after coronary by-pass surgery. Chest 1982;81:403-6. 11. Tomlinson JR, Miller KS, Larch DG, Smith HD, Sahn SA. A prospective comparison of IMV and T piece weaning from mechanical ventilation. Chest 1989;96:348-52. 12. Fiastro JF, Habib MP, Quan SF. Pressure support compensation for inspiratory work due to endotracheal tubes and demand continuous positive airway pressure. Chest 1988;93:499-505. 13. Brochard L, Rua F, Lorino H, Lemaire F, Hart A. Inspiratory pressure support compensates for the additional work of breathing caused by the endotracheal tube. Anesthesiology 1991;75:739-45. 14. FargierJJ, Robert D, Boyer F, Chagny J, Kopp C, Beaulieux J, Moskovtchenko JF, Pouyet M. Positive pressure inspiratory aid v assisted mechanical ventilation after esophageal surgery. J Crit Care 1987;2:101-8. 15. Tokioka H, Saito S, Kosaka F. Comparison of pressure support ventilation and assist control ventilation in patients with acute respiratory failure. Intensive Care Med 1989;15:364-7. 16. Lemaire F, Teboul JL, Cinotti L, Giotto G, Abrouk F, Steg G, MacquinMavier I, Zapol WHo Acute left ventricular dysfunction during unsuccessful weaning from mechanical ventilation. Anesthesiology 1988; 69:171-9. 17. Marini JJ, Smith TC, Lamb V. Estimation of inspiratory muscle strength in mechanically ventilated patients: the measurement of maximal inspiratory pressure. J Crit Care 1986;1:32-8. 18. Yang KL, Tobtn M. A prospective study of indexes predicting the outcome of trials of weaning from mechanical ventilation. N Engl J Med 1991; 324:1445-50. 19. Sassoon CSH, Te IT, Mahutte CK, Light RW. Aiway occlusion pressure:

Brochard, Rauss, Benito, et 01.: Three Methods of Gradual Withdrawal from Ventilatory Support

20. 21.

22.

23.

24. 25.

26. 27.

28.

an important indicator for successful weaning in patients with chronic obstructive pulmonary disease. Am Rev Respir Dis 1987;135:107-13. Smith TC, Marini JJ. Impact of PEEP on lung mechanics and work of breathing in severe airflow obstruction. J Appl PhysioI1988;65:1488-99. Rasanen J, Nikki P, Heikkila J. Acute myocardial infarction complicated by respiratory failure: the effects of mechanical ventilation. Chest 1984;85:21-8. FagonJY, ChastreJ, Novara A, Medioni P, GibertC. Respective influences of initial severity of illness and of mechanical ventilation in the development of ICU acquired pneumonia (abstract). Am Rev Respir Dis 1991;143:A493. Peto R, Pike MC, Armitage P, Breslow NE, Cox DR, Howard SV, Mantel N, McPherson K, Peto J, Smith PG. Design and analysis of randomized clinical trials requiring prolonged observation of each patient. Br J Cancer 1977;35:1-39. Mantel N. Evaluation of survival data and two new rank order statistics arising in its consideration. Cancer Chemother Rep 1966;50:163-70. Le GaIlJR, Loirat P, Alperovitch A, Glaser P, Granthil C, Mathieu 0, Mercier P, Thomas R, Villers D. A simplified acute physiologic score for ICU patients. Crit Care Med 1984;12:975-7. Petty TL. Chronic airflow limitation. In: Petty TL, ed. Intensive and rehabilitative respiratory care. Philadelphia: Lea & Febiger, 1980;222-45. Marini JJ, Smith TC, Lamb VJ. External work output and force generation during synchronized intermittent mechanical ventilation: effect of machine assistance on breathing effort. Am Rev Respir Dis 1988;138: 1169-79. Marini JJ. Weaning from mechanical ventilation. N Engl J Med 1991;

903

324:1496-8. 29. Santak B, Radermacher P, Sandmann W, Falke KJ. Influence on SIMV plus inspiratory pressure support on VAlQ distributions during postoperative weaning. Intensive Care Med 1991;17:136-40. 30. Milic-Emili J. Is weaning an art or a science? Am Rev Respir Dis 1986; 143:1107-8. 31. Pourrlat JL, Lamberto CH, Hoang PH, Fournier JL, Vasseur B. Diaphragmatic fatigue and breathing pattern during weaning from mechanical ventilation in COPD patients. Chest 1986;90:703-7. 32. Arora NS, Rochester OF. Respiratory muscle strength and maximal voluntary ventilation in undernourished patients. Am Rev Respir Dis 1982; 126:5-8. 33. Oh TE, Bhatt S, Lin S, Hutchinson RC, Low JM. Plasma catecholamines and oxygen consumption during weaning from mechanical ventilation. Intensive Care Med 1991;17:199-203. 34. Annat G, Viale JP, Derymez CP, Bouffard YM, Delafosse BX, Motin JP. Oxygen cost of breathing and diaphragmatic pressure time index: measurement in patients with COPD during weaning with pressure support ventilation. Chest 1990;98:411-4. 35. Conti G, Bufi M, Antonelli M, Rocco M, Gasparetto A. Pressure support ventilation (PSV) reverses hyperinflation induced isorhythmic A-V dissociation. Intensive Care Med 1989;15:319-21. 36. Poggi R, Polese G. Brandelese R, Brandi G, Rossi A. Comparison between continuous positive airway and inspiratory pressure support in patients with chronic obstructive pulmonary disease during weaning (abstract). Am Rev Respir Dis 1991;143:A690.