Noninvasive Mechanical Ventilation in the Weaning of Patients

with Respiratory Failure Due to Chronic Obstructive

Pulmonary Disease
A Randomized, Controlled Trial
Stefano Nava, MD; Nicolino Ambrosino, MD; Enrico Clini, MD; Maurizio Prato, MD;
Giacomo Orlando, MD; Michele Vitacca, MD; Paolo Brigada, MD; Claudio Fracchia, MD;
and Fiorenzo Rubini, MD

Background: In patients w i t h acute exacerbations of in the intensive care unit, decreases the incidence of nos-
chronic obstructive pulmonary disease, mechanical venti- ocomial pneumonia, and improves 60-day survival rates.
lation is often needed. The rate of weaning failure is high
in these patients, and prolonged mechanical ventilation Ann Intern Med 1998;128:721-728.
increases intubation-associated complications.

Objective: To determine whether noninvasive ventila-

tion improves the outcome of weaning from invasive me-
E ndotracheal intubation and mechanical ventila-
tion are often needed in patients with acute
exacerbations of chronic obstructive pulmonary dis-
chanical ventilation.
ease. Respiratory muscle weakness, hypercapnia,
Design: Multicenter, randomized trial. hypoxia, and malnutrition are common in these pa-
Setting: Three respiratory intensive care units. tients, and they may receive prolonged mechanical
Patients: Intubated patients w i t h chronic obstructive
ventilation, which has associated risks (1, 2). The
pulmonary disease and acute hypercapnic respiratory
duration of mechanical ventilation is influenced not
failure. only by the cause of disease but also by the venti-
latory strategy chosen (3, 4). Patients receiving in-
Intervention: A T-piece weaning trial was attempted 48 vasive ventilation are prone to developing nosoco-
hours after intubation. If this failed, t w o methods of wean-
mial pneumonia and ventilator-associated pneumonia,
ing were compared: 1) extubation and application of non-
with a high mortality rate (5-8).
invasive pressure support ventilation by face mask and 2)
invasive pressure support ventilation by an endotracheal
Noninvasive ventilation may be a good alterna-
tube. tive approach for patients with chronic obstructive
pulmonary disease. However, in a recent multi-
Measurements: Arterial blood gases, duration of me- center study by Brochard and colleagues (9), only
chanical ventilation, time in the intensive care unit, occur-
29% of patients admitted to intensive care units
rence of nosocomial pneumonia, and survival at 60 days.
were eligible for noninvasive ventilation. Nonethe-
Results: At admission, all patients had severe hypercap- less, noninvasive mechanical ventilation remains the
nic respiratoryfailure (mean pH, 7.18 ± 0.06; mean Paco 2 , "first line intervention in patients with hypercapnic
94.2 ± 24.2 mm Hg), sensory impairment, and similar clin- respiratory failure" (10) in the absence of gross
ical characteristics. At 60 days, 22 of 25 patients (88%) w h o neurologic impairment. Some uncontrolled studies
were ventilated noninvasively were successfully weaned
(11-13) have shown that noninvasive ventilation
compared w i t h 17 of 25 patients (68%) w h o were venti-
may also be useful in facilitating the weaning pro-
lated invasively. The mean duration of mechanical ventila-
t i o n was 16.6 ± 11.8 days for the invasive ventilation
cess in patients who require prolonged mechanical
group and 10.2 ± 6.8 days for the noninvasive ventilation
ventilation.
group (P = 0.021). Among patients w h o received noninva- The aim of our multicenter, randomized study
sive ventilation, the probability of survival and weaning was to evaluate the effectiveness of a new weaning
during ventilation was higher (P = 0.002) and time in the approach in patients with chronic obstructive pul-
intensive care unit was shorter (15.1 ± 5.4 days compared monary disease. This approach consisted of 24 to 48
w i t h 24.0 ± 13.7 days for patients w h o received invasive hours of traditional ventilation delivered through an
ventilation; P = 0.005). Survival rates at 60 days differed endotracheal tube followed by noninvasive pressure
(92% for patients w h o received noninvasive ventilation support ventilation. We compared the results ob-
and 72% for patients w h o received invasive ventilation; tained by using this strategy with those produced by
P = 0.009). None of the patients weaned noninvasively using the same ventilatory mode delivered inva-
developed nosocomial pneumonia, whereas 7 patients
sively. End points of the study were duration of
weaned invasively did.
mechanical ventilation, duration of intensive care
Conclusions: Noninvasive pressure support ventilation unit stay, incidence of nosocomial pneumonia, and
during weaning reduces weaning time, shortens the time 60-day survival rate.
© 1998 American College of Physicians 721

Downloaded From: http://annals.org/ by a University of Chicago Library User on 08/23/2013

 Methods Patients were prospectively recruited from three
hospitals in Montescano, Gussago, and Novi Ligure,
Patients Italy. Of the 68 patients referred to the participating
A two-group, parallel, randomized, controlled centers and intubated for acute respiratory failure,
trial was carried out in patients known to have 50 (74%) were enrolled. Eighteen patients were
chronic obstructive pulmonary disease. All patients excluded at the time of possible randomization:
known to have this disease who were admitted for Eight (12%) had a successful T-piece weaning trial at
acute relapse of their disease at three intensive care the time of randomization, 6 (9%) had altered neu-
units and who needed intubation were eligible for rologic status, and 4 (6%) were hemodynamically
the study. Acute relapse was defined as respiratory unstable.
acidosis (pH < 7.33 while breathing room air); ele- The protocol was approved by each hospital's
vated bicarbonate levels; hypoxemia (Pa02 ^ 45 mm ethics committee. Patients or their relatives gave
Hg while breathing room air); and severe dyspnea informed consent to the study.
in the absence of an objectively documented cause, Study Protocol
such as pneumonia (14, 15) or 1 of 11 nonoperative
respiratory diagnoses (excluding chronic obstructive Intubation was done via the orotracheal route.
pulmonary disease) found in the Acute Physiology, After intubation, all patients were ventilated in con-
Age, and Chronic Health Evaluation (APACHE) trolled mode with intensive care unit ventilators for
III (16). the first 12 hours; during the first 6 to 8 hours, the
patients were sedated and curarized and their air-
Fifty-six percent of patients needed immediate way secretions were suctioned frequently. Standard
endotracheal intubation. Intubation was delayed in settings for controlled ventilation were used: tidal
44% of patients; in these patients, an initial attempt volume, approximately 8 to 10 mL/kg; respiratory
at noninvasive ventilation had failed. To make the rate, 12 to 16 breaths/min; and an inspired oxygen
decision of whether to intubate as objective as pos- fraction (Fi02) as required to obtain an Sa02 of
sible, we established criteria extrapolated from the about 95% (41% ± 0.05%). Patients were then
study by Brochard and coworkers (9). Major criteria given pressure support ventilation (21 ± 2 cm H 2 0)
for intubation were respiratory arrest, loss of con- for an additional 24 to 36 hours. Extrinsic positive
sciousness, psychomotor agitation requiring seda- end-expiratory pressure was added when intrinsic
tion, hemodynamic instability with systolic blood positive end-expiratory pressure was clinically sus-
pressure less than 70 mm Hg or greater than 180 pected. At the end of this brief period of invasive
mm Hg, and a heart rate of 50 beats/min or less ventilation, a T-piece weaning trial was carried out
with loss of alertness. Minor criteria were a respi- in patients who were judged to have reached satis-
ratory rate that was more than 35 breaths/min and factory neurologic status, had a body temperature of
was higher than the value recorded on admission, 37 °C or less, were hemodynamically stable, and had
an arterial pH that was 7.30 or less and was lower an Sa02 of 88% or more for an Fi02 of 40% during
than the value recorded on admission, a Pa0 2 less a brief discontinuation of mechanical ventilation.
than 45 mm Hg despite oxygen supplementation, We considered the T-piece trial to have failed if
deterioration of one or more points in the neuro- patients had any of the following: a respiratory rate
logic score of the scale of Kelly and Matthay (17), more than 35 breaths/min, a Pa0 2 less than 50 mm
and presence of a weak cough reflex with accumu- Hg for an Fi0 2 of 40%, heart rate more than 145
lation of secretions. The presence of one major beats/min or sustained increase or decrease in the
criterion was considered an indication for immedi- heart rate of more than 20%, severe arrhythmia,
ate intubation. In patients undergoing an initial at- systolic blood pressure more than 180 mm Hg or
tempt at noninvasive ventilation, the presence of less than 70 mm Hg, agitation, anxiety, or diaphore-
two minor criteria after 1 hour of treatment or less sis. Patients in whom the weaning trial failed and
was considered to indicate the need for intubation. who did not meet any of the aforementioned exclu-
Patients were excluded from the study if they had sion criteria were randomly assigned to receive non-
concomitant severe diseases, such as neurologic dis- invasive ventilation or to continue the weaning pro-
eases other than hypercapnic encephalopathy; can- cess with invasive pressure support ventilation.
cer and other systemic diseases; cardiac arrest; car- The study was designed to enroll all eligible pa-
diogenic pulmonary edema; cardiogenic shock; aortic tients over a specified 18-month period. Random
aneurysm; acute myocardial infarction; gastrointes- assignment was done by using opaque, sealed, num-
tinal perforation, obstruction, or bleeding; sepsis; bered envelopes. A post hoc analysis was performed
trauma; metabolic coma; diabetic ketoacidosis; drug to confirm the validity of the randomization. To
overdose; coagulopathy; and other hematologic dis- detect any possible differences in the two groups at
eases. Postoperative patients were also excluded. enrollment, the 50 patients were divided according
722 1 May 1998 • Annals of Internal Medicine • Volume 128 • Number 9

Downloaded From: http://annals.org/ by a University of Chicago Library User on 08/23/2013

 to the random assignments, survival or death, and dial artery (ABL 300, Radiometer, Copenhagen,
successful and unsuccessful weaning. Denmark) at admission; 1, 4, 8, and 12 hours after
Immediately after failure of the T-piece weaning the beginning of each ventilatory treatment; and
trial, patients were reconnected to the ventilator in after changes in the ventilatory setting. The neuro-
pressure support ventilation mode until the previous logic score was recorded once daily. Pulmonary
PaC0 2 and pH values were reached (30 to 60 min- function tests were performed with a portable spi-
utes) and the respiratory rate under ventilation was rometer as soon as the patients' clinical condition
30 breaths/min or less. Patients in the noninvasive allowed testing and before discharge from the in-
ventilation group were then extubated and switched tensive care unit in all cases.
to noninvasive pressure support ventilation with a Ventilator-associated pneumonia was defined as
face mask (Gibeck Respiration AB, Upplands- the presence, during mechanical ventilation, of a
Vasby, Sweden) and the same ventilator. The pa-
new and persistent (>48 hours) lung infiltrate on
tients' heart rate, electrocardiograms, Sa0 2 , and
chest radiography combined with at least two of the
blood pressure were monitored continuously, and a
following conditions: fever, peripheral leukocyte
physician was present to intubate patients again, if
count higher than 10 000 cells/mm3, and endotra-
necessary. Patients received ventilation with a level
of pressure support (19 ± 2 cm H 2 0 ) that was ad- cheal secretion obtained by suctioning material from
justed to achieve satisfactory blood gases and a the lower respiratory tract in which a Gram stain
respiratory rate of less than 25 breaths/min. During showed one or more types of bacteria.
the first 48 hours after extubation, noninvasive ven-
tilation was delivered until it was well tolerated (20
to 22 hours per day), spaced by periods of sponta- Statistical Analysis
neous inhalation of oxygen only during meals and to Results are given as the mean ± SD. All statis-
expectorate. tical tests were two-sided. Groups were compared
The level of pressure support was decreased by 2 by using unpaired or paired Mests as appropriate
or 4 cm H 2 0 per day in patients with good toler- when normality assumptions were satisfied and by
ance; patients were allowed to breathe spontane- using the Mann-Whitney test and Wilcoxon test
ously. At least two trials of spontaneous breathing when normality assumptions were not satisfied. Fre-
of gradually increased duration were attempted
quency distributions were compared by using the
each day.
chi-square test. Multiple comparisons were done
To assess weaning from noninvasive ventilation
with an analysis of variance for repeated measures;
as objectively as possible, the following criteria re-
if necessary, post hoc analysis was corrected by us-
corded after at least 3 hours of spontaneous breath-
ing the Bonferroni test. The Kaplan-Meier esti-
ing were established: Sa0 2 of 90% or more with an
mate-of-survival curve was used to determine the
Fi0 2 of 40% or less, pH of 7.35 or more, respiratory
rate less than 35 breaths/min, hemodynamic stabil- probability of success of the two methods over time
ity, absence of severe dyspnea, and depressed neu- and the survival rate at 60 days; survival curves were
rologic status. The absence of even one of these compared by using the log-rank test. Treatment was
criteria was considered failure to wean, and the considered to have failed if 1) the patient could not
patient was reconnected to the ventilator (which be taken off of the ventilator after 60 days (this
was always set at its maximal trigger sensitivity). cut-off was chosen because patients with chronic
Weaning was considered successful if reintubation obstructive pulmonary disease have been shown to
or noninvasive ventilation was not required within have longer weaning times than patients with other
72 hours of suspension of ventilation. conditions (18-22), 2) endotracheal reintubation
In intubated patients who received pressure sup- was required within 72 hours after disconnection
port ventilation, the pressure was titrated to achieve from the ventilator, or 3) death related to mechan-
a breathing frequency of 25 breaths/min or less. ical ventilation occurred. Nosocomial pneumonia,
Pressure support ventilation was initially set at pneumothorax, ischemic cardiac events, or fatal ar-
17.6 ± 2 . 1 cm H 2 0 . For patients who received inva- rhythmia during the weaning process were causes of
sive ventilation, the level of pressure support was death considered to be associated with mechanical
gradually decreased and intermittent trials of spon- ventilation (3). A P value less than 0.05 was con-
taneous breathing were performed twice a day by sidered statistically significant.
using a T-tube circuit or a continuous-flow circuit
with a continuous positive airway pressure of less
than 5 cm H 2 0 (4). The weaning criteria were the
Role of Funding Source
same as those used for noninvasive ventilation.
Arterial blood gases were collected from the ra- Our study did not receive any financial support
1 May 1998 • Annals of Internal Medicine • Volume 128 • Number 9 723

Downloaded From: http://annals.org/ by a University of Chicago Library User on 08/23/2013

 Table 1 . Physiologic Variables* Montescano, 4 in Novi Ligure, and 4 in Gussago),
and 17 were assigned to continue invasive ventila-
Variable Noninvasive Invasive
Pressure Support Pressure Support tion (9 in Montescano, 5 in Novi Ligure, and 3 in
Ventilation Group Ventilation Group Gussago). Patients were also receiving standard
therapy for chronic obstructive pulmonary disease,
Age, y 68.7 ± 8.5 67.0 ± 9.2
APACHE II score 24.3 ± 3.1 23.1 ± 4 . 5 including j32-agonists (42 patients), theophylline (21
Kelly and Matthay scoret 2.8 ± 0.3 2.9 ± 0.5 patients), anticholinergic agents (13 patients), and
Leukocyte count, cells/mm3 8468 ± 1345 8112 ± 1789
Albumin level, g/dL 34.0 ± 10.0 36.0 ± 7.0 diuretics (13 patients).
Respiratory rate, breaths/min 32.7 ± 8.7 30.6 ± 7 . 1 The results of post hoc analysis did not show any
Heart rate, beats/min 101 ± 12 106 ± 11
Systolic blood pressure, mm Hg 133 ± 28 139 ± 2 2 significant differences in the two groups at admis-
PH 7.22 ± 0.07 7.22 ± 0.08 sion (Tables 1 and 2) or at randomization (Table 3)
Paco2, mm Hg 96.3 ± 19.6 91.9 ± 13.8
Pao2: Fio2 ratio 1.48 ± 0.3 1.42 ± 0 . 4 for any of the physiologic variables or for the pres-
FEV1fmL* 501 ± 131 525 ± 129 ence of comorbid conditions.
Percent predicted FEV^ 16.9 ± 10 17.4 ± 9
Vital capacity, mLt 992 ± 208 1089 ± 222
Percent predicted vital capacity* 28.0 ±8 29.2 ± 9 Outcomes
FEV^ vital capacity ratio* 50.7 ±11.7 49.2 ± 13.5
Figure 1 shows the changes in pH and PaC0 2 in
* Data are the mean ± SD. APACHE = Acute Physiology, Age, and Chronic Health the two groups of patients from the time of intuba-
Evaluation; Fio2 = fraction of inspired oxygen.
t See reference 17. tion to discharge from the intensive care unit. Institu-
* Data are given for 14 patients in thenoninvasive ventilation group and 16 patients in tion of invasive ventilation significantly improved
the invasive ventilation group.
blood gas values (P < 0.001), which remained con-
stant throughout the entire period of ventilation,
Results regardless of the mode of ventilation. At discharge,
these values were also similar during spontaneous
Patients breathing with oxygen supplementation.
Twenty-five patients were randomly assigned to Expiratory tidal volume per kilogram was similar
be extubated and to undergo the weaning process in the two groups during mechanical ventilation,
with noninvasive pressure support ventilation, and regardless of the level of pressure support applied
25 remained intubated and received the same mode (mechanical ventilation immediately after random-
of ventilation. Of the 50 patients, 26 were enrolled ization, 8.4 ± 0.5 mL/kg with pressure support of
in Montescano (14 in the group that received non- 19.0 ± 2.0 cm H 2 0 for the noninvasive ventilation
invasive ventilation [1 death] and 12 in the group group and 8.2 ± 0.3 mL/kg with pressure support of
that received invasive ventilation [4 deaths]), 14
were enrolled in Novi Ligure (6 in the group that
received noninvasive ventilation [1 death] and 8 in Table 2. Location before the Intensive Care IJnit, Length
of Hospital Stay, Nuimber of Hospitalizations in
the group that received invasive ventilation [2 the Previous Year, airid Activity Limiits because of
deaths]), and 10 were enrolled in Gussago (5 in the Chronic Obstructive Pulmonary Dise>ase*
group that received noninvasive ventilation [no
Variable Noninvasive Invasive Pressure
deaths] and 5 in the group that received invasive Pressure Support Support
ventilation [1 death]). Ventilation Ventilation
Group Group
As shown in Tables 1 and 2, the two groups of
ri
patients had similar functional and clinical charac-
teristics at admission. Comorbid conditions were Location before the intensive care unit
present in 14 patients who received noninvasive ven- Emergency department 6 6
Other intensive care unit 2 1
tilation and 11 patients who received invasive ven- Other hospital 0 0
tilation. In the noninvasive ventilation group, 5 pa- Hospital floor 17 18
Length of hospital stay before the
tients had hypertension, 4 had rhythm disturbances, intensive care unit
3 had congestive heart failure, 3 had diabetes, 2 had 0 days 7 6
1 day 1 1
sepsis of the urinary tract, and 1 had diverticulosis. 2-4 days 10 12
Four patients had two of the aforementioned con- 5-10 days 6 4
11-30 days 1 2
ditions. In the invasive ventilation group, 4 patients Functional limits because of chronic
had hypertension, 4 had rhythm disturbances, 3 had obstructive pulmonary disease
None 1 2
congestive heart failure, 1 had diabetes, and 1 had Moderate 7 5
hepatic failure. Two patients had two of the afore- Severe 17 18
Hospitalizations in the previous year
mentioned conditions. At the time of hospital ad- 0 8 7
mission, most patients (35 of 50) were receiving 1 11 10
2 6 7
long-term oxygen therapy; 18 of these patients were >2 2 1
randomly assigned to noninvasive ventilation (10 in
724 1 May 1998 • Annals of Internal Medicine • Volume 128 • Number 9

Downloaded From: http://annals.org/ by a University of Chicago Library User on 08/23/2013

 T a b l e 3. Patient Characteiristics at Random ization* ods of weaning. The failure rate among patients
weaned with invasive pressure support ventilation
Variable Noninvasive Invasive
Pressure Support Pressure Support was significantly higher than the rate among pa-
Ventilation Group Ventilation Group tients undergoing noninvasive ventilation (P =
0.002). Furthermore, at day 21 (the time point usu-
pHt 7.31 ±0.05 7.30 ± 0.04
Paco2/ mm Hgt 61.6 ± 11.6 63.5 ± 12.3 ally considered to be a threshold that separates
Pao2:Fio2 ratiot 1.52 ±0.3 1.64 ± 0.4 weanable from unweanable patients [3]), the success
Respiratory rate, breaths/mint 24.3 ±6.1 26.2 ± 4.9
Tidal volume, mL/kgt 5.82 ± 0.077 6.10 ± 0.091 rate was significantly higher in patients who re-
Maximal inspiratory pressure, ceived noninvasive treatment (P = 0.003).
cm H20 -31 ± 12 -34 ± 9
Kelly and Matthay score§ 1.4 ±0.2 1.3 ± 0 . 2 Seven of 25 patients (28%) in the invasive ven-
Systolic blood pressure, mm Hg 125 ±22 119± 16 tilation group and no patients in the noninvasive
Heart rate, beats/min 92 ± 12 89 ± 10
ventilation group developed nosocomial pneumonia.
* Data are the mean ± SD. Fio2 = fraction of inspired oxygen. The attack rate for pneumonia by ventilator day was
t Recorded at the end of the T-piece 1:rial.
* Recorded during the first 5 minutes of the T-piece trial. one patient on day 3, two patients on day 5, one
§ See reference 17.
patient on day 7, one patient on day 8, one patient
on day 16, and one patient on day 22.
Two of 25 patients (8%) in the noninvasive ven-
17.6 ± 2.0 cm H 2 0 for the invasive ventilation tilation group died in the first 60 days. One patient
group; 50% of the ventilatory time, 8.2 ± 0.4 mL/kg in whom weaning attempts failed was discharged
with pressure support of 16.1 ± 1.6 cm H 2 0 for the with a prescription for nasal ventilation to be given
noninvasive ventilation group and 8.1 ± 0.3 mL/kg for 14 to 18 hours per day. Seven patients (28%)
with pressure support of 15.2 ± 1.3 cm H 2 0 for the who were weaned by using invasive ventilation died;
invasive ventilation group; and just before successful 2 were considered unweanable and were discharged
weaning, 8.3 ± 0.3 mL/kg with pressure support of with a prescription for at-home mechanical ventila-
9.7 ± 0.5 cm H 2 0 for the noninvasive ventilation tion through a tracheostomy. The mortality rate at
group and 8.1 ± 0.4 mL/kg with pressure support of 60 days, assessed by using a mortality table, was
9.3 ± 0.3 cm H 2 0 for the invasive ventilation significantly higher in the invasive ventilation group
group). The minute of ventilation per kilogram also
remained similar for both groups (mechanical ven-
tilation immediately after randomization, 198.1 ± 8.3
mL/kg per minute for the noninvasive ventilation
group and 188.8 ± 24.9 mL/kg per minute for the
invasive ventilation group; 50% of the ventilatory
time, 164.3 ± 21.8 mL/kg per minute for the nonin-
vasive ventilation group and 152.8 ± 18.0 mL/kg per
minute for the invasive ventilation group; and just
before successful weaning, 124.5 ± 18.3 mL/kg per
minute in the noninvasive ventilation group and
119.9 ± 21.2 mL/kg per minute in the invasive ven-
tilation group).
Some minor side effects were recorded in the
noninvasive ventilation group. Twenty patients had
cutaneous irritation of the nose; 14 of the 20 had
nose abrasion, which was severe in some cases. Two
patients had gastric distention, and most patients
reported poor sleep, especially in the first few days.
Compared with patients who were weaned by
using the invasive technique, patients who were
weaned by using the noninvasive technique spent
significantly fewer days receiving mechanical venti-
lation (16.6 ± 11.8 days and 10.2 ± 6.8 days; P =
0.021) and in the intensive care unit (24.0 ± 13.7
days and 15.1 ± 5.4 days; P = 0.005). Figure 2 Figure 1. Mean ± SD pH and Paco2 in the two groups of patients
at admission to the hospital; after 24 hours of controlled, intermit-
shows the percentage of patients who could not be tent positive-pressure ventilation (24 h IPPV); at 4 hours after ran-
weaned (because of death associated with mechan- domization to invasive or noninvasive pressure support ventilation;
at 24 hours after randomization; and at discharge. White circles rep-
ical ventilation, reintubation within 72 hours, and resent noninvasive pressure support ventilation; black circles represent inva-
failure to be weaned at 60 days) for the two meth- sive pressure support ventilation.

1 May 1998 • Annals of Internal Medicine • Volume 128 • Number 9 725

Downloaded From: http://annals.org/ by a University of Chicago Library User on 08/23/2013

 shorter, the incidence of nosocomial pneumonia is
lower, and the survival rate at 60 days is higher.
The process of discontinuing mechanical ventila-
tion constitutes a major clinical challenge, especially
in patients with chronic obstructive pulmonary dis-
ease, in whom weaning is particularly difficult. The
rate of weaning failure ranges from 35% to 67%
according to various studies, the modes of ventila-
tion used, and the definition of weaning (18-22).
Noninvasive ventilation has been proposed as an
alternative to intubation in the management of
acute respiratory failure due to chronic obstructive
pulmonary disease. Several studies (10, 23-25) have
Figure 2. Kaplan-Meier curves for patients who could not be shown the efficacy of this method of ventilation in
weaned from mechanical ventilation (defined as weaning failure or reducing the need for intubation; however, only a
death linked to mechanical ventilation) in the two groups. The prob-
ability of weaning failure was significantly lower for the noninvasive venti- few studies were controlled and randomized (9, 26,
lation group (cumulative probability for 60 days, P< 0.01 by the log-rank 27). In their multicenter randomized study, Bro-
test). The vertical line represents day 21, usually considered the threshold
between weanable and unweanable patients. The solid line represents non- chard and coworkers (9) found that noninvasive
invasive pressure support ventilation; the dashed line represents invasive
pressure support ventilation.
ventilation may avoid intubation, but only in se-
lected patients; most patients (69%) in that study
required immediate intubation for safety reasons.
The reasons for needing immediate intubation or
than in the noninvasive ventilation group (92% and failure of noninvasive ventilation are mainly linked
72%; P = 0.009). to hypercapnic-related nervous system disorders,
Causes of death in the noninvasive ventilation such as neurologic depression; psychomotor agita-
group were multiple organ failure (1 patient on day tion requiring sedation; weak cough reflex with dif-
7) and myocardial infarction (1 patient on day 18). ficulty in clearing the airways; and respiratory arrest,
Fatal events in the invasive ventilation group were probably caused by acute fatigue of the respiratory
pneumonia (4 patients, one each on day 8, 10, 18, muscles. In most of these patients, positioning of
and 23), pneumothorax (1 patient on day 11), ar- the endotracheal tube and short-term sedation or
rhythmia (1 patient on day 12), and pulmonary em- curarization allow prompt correction of acidosis
bolism (1 patient on day 21). One death in each and, consequently, improvement of the neurologic
group was not considered to be linked to mechan- status, removal of secretions, and rest for the respi-
ical ventilation (1 case of multiple organ failure and ratory muscles. We undertook our study to assess
1 case of pulmonary embolism). the feasibility of substituting traditional invasive
At discharge from the intensive care unit, pa- ventilation with the noninvasive technique at this
tients in the noninvasive ventilation group and the point.
invasive ventilation group were similar for FEVX Few data are available on the use of noninvasive
(510 ± 111 mL or 17.1% ± 11.0% of the predicted ventilation in the weaning process, and noninvasive
value and 537 ± 92 mL or 17.8% ± 9.0% of the ventilation has been applied only after a long period
predicted value), vital capacity (901 ± 162 mL or of invasive ventilation (11-13). For example, Restrick
27.3% ± 10.0% of the predicted value and 937 ± and associates (13) used noninvasive methods to
151 mL or 29.2% ± 12.0% of the predicted value), wean patients with acute respiratory failure. The
and the ratio of the two measures (56% ± 9% and average time of invasive ventilation was long (40.9
58% ± 11%). days); of interest, two of their patients were success-
fully treated noninvasively only 24 hours after intu-
bation. Our study validates these sporadic reports.
Discussion The main reason for the success of the noninva-
sive technique is probably that the technique avoids
Our study shows that in patients with chronic some of the most important complications of artifi-
obstructive pulmonary disease and acute respiratory cial airways. For example, the early institution of
failure, a new approach to ventilation that combines noninvasive ventilation was associated with a lower
invasive ventilation for not more than 48 hours fol- incidence of nosocomial pneumonia in our study.
lowed by noninvasive pressure support ventilation is Torres and coworkers (5) considered the correlation
more effective than pressure support ventilation de- between several risk factors and the development of
livered invasively for the entire ventilation period. nosocomial pneumonia; presence of chronic ob-
With this approach, the intensive care unit stay is structive pulmonary disease and invasive ventilation
726 1 May 1998 • Annals of Internal Medicine • Volume 128 • Number 9

Downloaded From: http://annals.org/ by a University of Chicago Library User on 08/23/2013

 for more than 3 days were significantly associated the duration of mechanical ventilation, the duration
with an increased risk for nosocomial pneumonia. of intensive care unit stay, and the occurrence of
This finding has been confirmed by other investiga- nosocomial pneumonia and was associated with a
tors (6-8). An endotracheal tube can predispose to better 60-day survival rate than was the same treat-
the development of pneumonia by impairing cough ment delivered invasively. However, further studies
and mucociliary clearance because contaminated se- are needed to assess the clinical efficacy of this
cretions can accumulate above the cuff and leak technique in patients who are sicker or who have
around the cuff or because bacterial binding to the pure hypoxemic respiratory failure.
surface of bronchial epithelium is increased. With a
From Centro Medico di Riabilitazione di Montescano, Montes-
noninvasive ventilatory technique, the risk for aspi- cano, Italy; Centro Medico di Riabilitazione di Gussago, Gus-
ration of colonized or infected oropharyngeal secre- sago, Italy; Fondazione S. Maugeri Istituto di Ricovero e Cura a
tions is probably smaller, there is no tracheal pros- Carattere Scientifico, Pavia, Italy; and Ospedale S. Giacomo,
Novi Ligure, Italy.
thesis, the patient can expectorate freely, and the
vocal cords are not kept open. Acknowledgments: The authors thank Dr. R.C. Stenner for review
Long-term invasive ventilatory support also in- of the English of the manuscript and Marco Pagani for assistance
with statistical analysis.
creases the risk for feeding aspiration. Elpern and
coworkers (28) showed that about 50% of tracheot- Requests for Reprints: Stefano Nava, MD, Division of Pneumol-
omized patients receiving prolonged positive-pres- ogy, Centro Medico di Riabilitazione di Montescano, 27040
Montescano (PV), Italy.
sure ventilation had feeding aspiration and that ad-
vanced age increased the risk for feeding aspiration. Current Author Addresses: Drs. Nava, Brigada, Fracchia, and Ru-
Ventilation delivered by a mask, on the other hand, bini: Pneumology Division, Centro Medico di Riabilitazione di
Montescano, via per Montescano, 27040 Montescano (PV), Italy.
enables patients to eat during spontaneous breath- Drs. Ambrosino, Clini, and Vitacca: Pneumology Division, Cen-
ing periods, probably helping to avoid the risk for tro Medico di Riabilitazione di Gussago, via Pinidolo 23, 25064
food aspiration. Gussago, Italy.
Drs. Prato and Orlando: Servizio di Anestesia e Rianimazione,
The mortality rate at 60 days was significantly Ospedale S. Giacomo, via Raggio 12, 15060 Novi Ligure, Italy.
higher in patients treated only with invasive venti-
lation. The duration of exposure to mechanical ven-
tilation and the different invasiveness of the two References
ventilatory modes may have influenced the outcome
of these patients. Six of the seven deaths (86%) 1. Stauffer JLf Olson DE, Petty TL Complications and consequences of en-
among the intubated patients were considered to dotracheal intubation and tracheotomy. A prospective study of 150 critically ill
adult patients. Am J Med. 1981;70:65-76.
have been linked to mechanical ventilation. 2. Pingleton SK. Complications of acute respiratory failure. Am Rev Respir Dis.
1988;137:1463-93.
Despite the encouraging results of our study, it 3. Brochard L, Rauss A, Benito S, Conti G, Mancebo J, Rekik N, et al.
must be remembered that the study was conducted Comparison of three methods of gradual withdrawal from ventilatory support
during weaning from mechanical ventilation. Am J Respir Crit Care Med.
in a selected group of severely hypercapnic patients 1994;150:896-903.
with chronic obstructive pulmonary disease. Extuba- 4. Esteban A, Frutos F, Tobin MJ, Alia I, Solsona JF, Valverdu I, et al. A
comparison of four methods of weaning from mechanical ventilation. Spanish
tion was done only if patients reached a satisfactory Lung Failure Collaborative Group. N Engl J Med. 1995;332:345-50.
5. Torres A, Aznar R, Gatell JM ( Jimenez P, Gonzalez J, Ferrer A, et al.
neurologic status, a body temperature of 37 °C or Incidence, risk, and prognosis factors of nosocomial pneumonia in mechani-
less, hemodynamic stability, and Sa02 of 88% or cally ventilated patients. Am Rev Respir Dis. 1990;142:523-8.
6. Fagon JY, Chastre J, Do mart Y, Trouillet JL, Pierre J, Darne C, et al.
more for an Fio 2 of 40%. This new weaning tech- Nosocomial pneumonia in patients receiving continuous mechanical ventila-
nique may not be applicable to patients who are tion. Prospective analysis of 52 episodes with use of a protected specimen
brush and quantitative culture techniques. Am Rev Respir Dis. 1989;139:877-84.
sicker, have more severe comorbid conditions, or 7. Craven DE, Kunches LM, Kilinsky V, Lichtenberg DA, Make BJ, Mc-
Crabe WR. Risk factors for pneumonia and fatality in patients receiving
are affected by severe hypoxemic respiratory failure. continuous mechanical ventilation. Am Rev Respir Dis. 1986;133:792-6.
Another potential problem related to the use of 8. Fagon JY, Chastre J, Hance AJ, Montravers P, Novara A, Gibert C.
Nosocomial pneumonia in ventilated patients: a cohort study evaluating at-
noninvasive ventilation is the fact that it may be an tributable mortality and hospital stay. Am J Med. 1993;94:2818.
excessively time-consuming procedure for medical 9. Brochard L, Mancebo J, Wysocki M, Lofaso F, Conti G, Rauss A, et al.
Noninvasive ventilation for acute exacerbations of chronic obstructive pulmo-
and paramedical staff (29). Although this may be nary disease. N Engl J Med. 1995;333:817-22.
10. Meduri GU, Turner RE, Abou-Shala N, Wunderink R, Tolley E. Nonin-
true in the first 2 days of ventilation, it has been vasive positive pressure ventilation via face mask: first-line intervention in
recently shown that after this time, the human re- patients with acute hypercapnic and hypoxemic respiratory failure. Chest.
1996;109:179-93.
sources needed are significantly less in patients re- 11. Goodenberger D M , Couser Jl Jr, May JJ. Successful discontinuation of
ceiving noninvasive ventilation than in patients un- ventilation via tracheostomy by substitution of nasal positive pressure venti-
lation. Chest. 1992;102:1277-9.
dergoing invasive ventilation (30). 12. Udwadia ZF, Santis GK, Steven MH, Simonds AK. Nasal ventilation to
facilitate weaning in patients with chronic respiratory insufficiency. Thorax.
In conclusion, in a selected population of pa- 1992;47:715-8.
tients with severe hypercapnic acute respiratory fail- 13. Restrick LJ, Scott AD, Ward EM, Feneck RO, Cornwell WE, Wedzicha
JA. Nasal intermittent positive-pressure ventilation in weaning intubated pa-
ure who needed prompt intubation, early extubation tients with chronic respiratory disease from assisted intermittent, positive-
and application of pressure support ventilation by a pressure ventilation. Respir Med. 1993;87:199-204.
14. Seneff MG, Zimmerman JE, Knaus WA, Wagner DP, Draper EA. Pre-
noninvasive route as a bridge to weaning reduced dicting the duration of mechanical ventilation. The importance of disease and

1 May 1998 • Annals of Internal Medicine • Volume 128 • Number 9 727

Downloaded From: http://annals.org/ by a University of Chicago Library User on 08/23/2013

 patient characteristics. Chest. 1996;110:469-79. 23. Brochard L, Isabey D, Piquet J, Amaro P, Mancebo J, Messadi AA, et
15. Seneff MG, Wagner DP, Wagner RP, Zimmerman JE, Knaus WA. Hos- al. Reversal of acute exacerbations of chronic obstructive lung disease by
pital and 1-year survival of patients admitted to intensive care units with acute inspiratory assistance with a face mask. N Engl J Med. 1990;323:1523-30.
exacerbation of chronic obstructive pulmonary disease. JAMA. 1995;274: 24. Meduri GU, Abou-Shala N, Fox RC, Jones CB, Leeper KV, Wunderink
1852-7. RG. Noninvasive face mask mechanical ventilation in patients with acute
16. Knaus WA, Wagner DP, Draper EA, Zimmerman JE, Bergner M, Bastos hypercapnic respiratoryfailure. Chest. 1991;100:445-54.
PG, et al. The APACHE III prognostic system. Risk prediction of hospital 25. Vitacca M, Rubini F, Foglio K, Scalvini S, Nava S, Ambrosino N. Non-
mortality for critically ill hospitalized adults. Chest. 1991;100:1619-36. invasive modalities of positive pressure ventilation improve the outcome of
17. Kelly BJ, Matthay MA. Prevalence and severity of neurological dysfunction acute exacerbations in COLD patients. Intensive Care Med. 1993;19:450-5.
in critically ill patients. Influence on need for continued mechanical ventila- 26. Bott J, Carroll MP, Conway JH, Keilty SE, Ward EM, Brown A M .
tion. Chest. 1993;104:1818-24. Randomised controlled trial of nasal ventilation in acute ventilatory failure due
18. Lemaire F. Difficult weaning. Intensive Care Med. 1993; 19 Suppl 2:S6973. to chronic obstructive airways disease. Lancet. 1993;341:1555-7.
19. Sporn H, Morganroth M. Discontinuation of mechanical ventilation. Clin 27. Kramer N, Meyer TJ, Meharg J, Cece RD, Hill NS. Randomized, prospec-
Chest Med. 1988;9:113-26. tive trial of noninvasive positive pressure ventilation in acute respiratory fail-
20. Goldstone J, Moxham J. Assisted ventilation. 4. Weaning from mechanical ure. Am J Respir Crit Care Med. 1995;151:1799-806.
ventilation. Thorax. 1991;46:56-62. 28. Elpern EH, Scott MG, Petro L, Ries MH. Pulmonary aspiration in mechan-
21. Conti G, De Blasi R, Pelaia P, Benito S, Rocco M, Antonelli M, et al. ically ventilated patients with tracheostomies. Chest. 1994;105:563-6.
Early prediction of successful weaning during pressure support ventilation in 29. Chevrolet JC, Jolliet P, Abajo B, Toussi A, Louis M. Nasal positive
chronic obstructive pulmonary disease patients. Crit Care Med. 1992;20:366-71. pressure ventilation in patients with acute respiratory failure. Difficult and
22. Nava S, Rubini F, Zanotti E, Ambrosino N, Bruschi C, Vitacca M, et al. time-consuming procedure for nurses. Chest. 1991;100:775-82.
Survival and prediction of successful ventilator weaning in COPD patients 30. Nava S, Evangelisti I, Rampulla C, Compagnoni ML, Fracchia C, Rubini
requiring mechanical ventilation for more than 21 days. Eur Respir J. 1994;7: F. Human and financial costs of noninvasive mechanical ventilation in patients
1645-52. affected by COPD and acute respiratory failure. Chest. 1997;111:1631-8.

More thorough and scientific medicines employed, even equally free of charge, by
doctors of the first reputation, would not have brought nearly so many poor people
together and, above all, would not have done them so much good. There would have
been lacking the main instruments of healing—prevention, respect, faith, and grati-
tude. Man is composed of a soul and a body and it is the former that governs the
latter. The wounded who have received consolation, the sick who have been per-
suaded to hope are already in a state to be cured; their blood circulates better, their
nerves are strengthened, sleep returns, and the body revives. Nothing is more effi-
cacious than confidence.

Duff Cooper
Talleyrand
Stanford, CA: Standford Univ Pr; 1967.

Submitted by:
Henryk Kafka, MD, FRCPC
Belleville, Ontario, Canada

Submissions from readers are welcomed. If the quotation is published, the sender's name will be acknowl-
edged. Please include a complete citation, as done for any reference.—The Editor

12% 1 May 1998 • Annals of Internal Medicine • Volume 128 • Number 9

Downloaded From: http://annals.org/ by a University of Chicago Library User on 08/23/2013